FA15 Gems Gemology PDF

FALL 2015
VOLUME LI
THE QUARTERLY JOURNAL OF THE GEMOLOGICAL INSTITUTE OF AMERICA
Colombian Trapiche Emeralds

Large Colorless HPHT-Grown Synthetic Diamonds
Diamonds from the Letšeng Mine
Fall 2015
VOLUME 51, No. 3
EDITORIAL
221 Trapiche and More...
Duncan Pay
FEATURE ARTICLES
222 Colombian Trapiche Emeralds: Recent Advances in Understanding
Their Formation
Isabella Pignatelli, Gaston Giuliani, Daniel Ohnenstetter, Giovanna Agrosì,
pg. 254 Sandrine Mathieu, Christophe Morlot, and Yannick Branquet
Proposes a model for trapiche emerald formation based on petrographic, spectroscopic, and
chemical examination.
260 Large Colorless HPHT-Grown Synthetic Gem Diamonds from
New Diamond Technology, Russia
Ulrika F.S. D’Haenens-Johansson, Andrey Katrusha, Kyaw Soe Moe,
Paul Johnson, and Wuyi Wang
pg. 270 Examines a new source of colorless and near-colorless gem-quality HPHT synthetic diamonds
using spectroscopic and gemological analysis.
280 Letšeng’s Unique Diamond Proposition

Russell Shor, Robert Weldon, A.J.A. (Bram) Janse, Christopher M. Breeding, and Steven B. Shirey
Explores the history, geology, and current production of this unique source of large diamonds.
NOTES AND NEW TECHNIQUES

300 Origin Determination of Dolomite-Related White Nephrite through
Iterative-Binary Linear Discriminant Analysis
Zemin Luo, Mingxing Yang, and Andy H Shen
A technique for origin identification based on statistical analysis and LA-ICP-MS spectrometry.
pg. 293
REGULAR FEATURES
312 Lab Notes
Unusual graining structure in pink diamond • Yellow HPHT-treated rough diamond • Color-
zoned emerald • Orange faceted eosphorite • Large faceted hibonite • Dyed and natural
green jadeite • Coated kornerupine beads • Assembled cultured blister pearl with unusual
pg. 325 feature • Three large natural abalone pearls • Irradiated green-blue CVD synthetic diamonds
• Polished freeform topaz imitating diamond rough
323 2015 G&G Challenge Winners
324 G&G Micro-World
Cleavage system in pink diamond • Rough diamond fragment with green cleavage surface
• Blue gahnite inclusions in cat’s-eye heliodor • Trapiche pezzottaite • Modified Rheinberg
illumination • Glassy melt inclusions in Montana sapphires • Molybdenite in topaz
331 Letters
334 Gem News International
Large star emerald • Natural faceted red rutile • Rare “star and cross” quartz from Brazil
• Serpentine cabochon with olive-green color • Characterization of Thai-Myanmar petri-
fied woods • Coated rock crystal imitation of ruby • Color-change cubic zirconia as
peridot imitation • Coral inclusions in plastic • Hydrogrossular garnet imitating jadeite
• Impregnated and dyed turquoise • Errata • Tino Hammid (1952–2015)
Editorial Staff
Editor-in-Chief Editors, Lab Notes Contributing Editors
Duncan Pay Thomas M. Moses James E. Shigley
gia.edu/gems-gemology
dpay@gia.edu Shane F. McClure Andy Lucas
Donna Beaton Subscriptions
Managing Editor Editors, Micro-World Copies of the current issue may be purchased for
Stuart D. Overlin Nathan Renfro Editor-in-Chief Emeritus $29.95 plus shipping. Subscriptions are $79.99 for one
soverlin@gia.edu Elise A. Skalwold Alice S. Keller year (4 issues) in the U.S. and $99.99 elsewhere. Cana-
John I. Koivula dian subscribers should add GST. Discounts are avail-
Editor Customer Service able for group subscriptions, GIA alumni, and current
Jennifer-Lynn Archuleta Editors, Gem News Martha Erickson GIA students. To purchase print subscriptions, visit
jennifer.archuleta@gia.edu Emmanuel Fritsch (760) 603-4502 store.gia.edu or contact Customer Service. For insti-
gandg@gia.edu tutional rates, contact Customer Service.
Gagan Choudhary
Technical Editors
Christopher M. Breeding Database Coverage
Tao Z. Hsu
tao.hsu@gia.edu Gems & Gemology’s impact factor is 0.778, accord-
Editorial Assistants
ing to the 2013 Thomson Reuters Journal Citation
Jennifer Stone-Sundberg Brooke Goedert Reports (issued July 2014). G&G is abstracted in
Erin Hogarth Thomson Reuters products (Current Contents: Phys-
ical, Chemical & Earth Sciences and Science Cita-
tion Index—Expanded, including the Web of
Production Staff Knowledge) and other databases. For a complete list
Creative Director Photographer Production Supervisor of sources abstracting G&G, go to gia.edu/gems-
gemology, and click on “Publication Information.”
Faizah Bhatti Robert Weldon Richard Canedo
Image Specialists Video Production Production Specialist Manuscript Submissions
Gems & Gemology, a peer-reviewed journal, welcomes
Kevin Schumacher Larry Lavitt Juan Zanahuria the submission of articles on all aspects of the field.
Eric Welch Pedro Padua Please see the Author Guidelines at gia.edu/gems-
Nancy Powers Multimedia Specialist
gemology or contact the Managing Editor. Letters on
Illustrator Betsy Winans Lynn Nguyen articles published in G&G are also welcome. Please
Peter Johnston note that Field Reports, Lab Notes, Gem News Inter-
national, and Micro-World entries are not peer-reviewed
sections but do undergo technical and editorial review.
Editorial Review Board Copyright and Reprint Permission

Abstracting is permitted with credit to the source. Li-
Ahmadjan Abduriyim A.J.A. (Bram) Janse Benjamin Rondeau
braries are permitted to photocopy beyond the limits
Tokyo, Japan Perth, Australia Nantes, France
of U.S. copyright law for private use of patrons. In-
Timothy Adams E. Alan Jobbins George R. Rossman structors are permitted to reproduce isolated articles
San Diego, California Caterham, UK Pasadena, California and photographs/images owned by G&G for noncom-
mercial classroom use without fee. Use of photo-
Edward W. Boehm Mary L. Johnson Kenneth Scarratt graphs/images under copyright by external parties is
Chattanooga, Tennessee San Diego, California Bangkok, Thailand prohibited without the express permission of the pho-
James E. Butler Anthony R. Kampf Andy Shen tographer or owner of the image, as listed in the cred-
Washington, DC Los Angeles, California Wuhan, China its. For other copying, reprint, or republication
permission, please contact the Managing Editor.
Alan T. Collins Robert E. Kane Guanghai Shi
Gems & Gemology is published quarterly by the
London, UK Helena, Montana Beijing, China
Gemological Institute of America, a nonprofit educa-
John L. Emmett Stefanos Karampelas James E. Shigley tional organization for the gem and jewelry industry.
Brush Prairie, Washington Lucerne, Switzerland Carlsbad, California Postmaster: Return undeliverable copies of Gems &
Emmanuel Fritsch Lore Kiefert Elisabeth Strack Gemology to GIA, The Robert Mouawad Campus,
Nantes, France Lucerne, Switzerland Hamburg, Germany 5345 Armada Drive, Carlsbad, CA 92008.
Our Canadian goods and service registration number
Eloïse Gaillou Ren Lu Wuyi Wang
is 126142892RT.
Paris, France Wuhan, China New York, New York
Any opinions expressed in signed articles are under-
Gaston Giuliani Thomas M. Moses Christopher M. Welbourn stood to be opinions of the authors and not of the
Nancy, France New York, New York Reading, UK publisher.
Jaroslav Hyršl Nathan Renfro
Prague, Czech Republic Carlsbad, California
About the Cover

The lead article in this issue is an in-depth study on the geological formation and growth of Colombian trapiche
emeralds. The cover photo shows a 58.83 ct crystal specimen (left) and a 22.74 ct cabochon (right), both displaying
the material’s signature six-rayed pattern, from the renowned Peñas Blancas emerald region. The host rock, seen in
the background, is carbonaceous shale containing veins of calcite, pyrite, and minute crystals of low-grade emer-
ald. Photo by Robert Weldon/GIA, courtesy of Jose Guillermo Ortiz and Colombian Emeralds Co.
Printing is by L+L Printers, Carlsbad, CA.
GIA World Headquarters The Robert Mouawad Campus 5345 Armada Drive Carlsbad, CA 92008 USA
© 2015 Gemological Institute of America All rights reserved. ISSN 0016-626X
Trapiche and More ...
Colombia’s unique trapiche emeralds exert a fascination for many gemologists. In fine
specimens, the combination of rich, gemmy green color with the six spokes’ geometry is
visually arresting.
In our lead article, we’re delighted to present a comprehensive paper on these intriguing
gems by a team of researchers headed by Isabella Pignatelli and Gaston Giuliani. They
review the geology of trapiche emerald, provide 3-D petrographic examination of their
crystals along with spectroscopic and chemical analyses, and propose a new formation
model informed by their recent work. We hope this paper will serve as the primary refer-
ence on these one-of-a-kind gems for many years to come. Take a look at our cover
photo and decide whether trapiche is a gemological curiosity or an object of beauty in its
own right.
In our next paper, GIA’s Ulrika D’Haenens-Johansson and her coauthors investigate the gemological
properties of colorless to near-colorless HPHT-grown synthetic diamonds produced specifically for gem
use by a Russian firm, New Diamond Technology. Their analysis of 44 faceted samples ranging up to 5.11
ct provides convincing evidence of the quality improvements—and increased size—of synthetics grown
by this method.
Meanwhile, Lesotho’s Letšeng-la-Terae mine is currently enjoying a renaissance. Rising prices for large, high-
quality diamonds have overcome the twin challenges of this unique mine’s remote location and low ore
grade. In our third article, GIA’s Russell Shor and
Robert Weldon and their co-authors profile Letšeng A gemological curiosity or an object of
to explain how the current owners have adapted their beauty in its own right?
operations to promote recovery of the mine’s biggest
gems and increase profitability. Coauthors Bram Janse, Christopher “Mike” Breeding, and Steven Shirey
offer insight into this mine’s geology, its location on the Kaapvaal craton, and the distinctive characteristics
of its diamonds.
Next, Zemin Luo, Mingxing Yang, and Andy Shen from the China University of Geosciences in Wuhan
apply a statistical analysis method based on linear discriminant analysis (LDA) and LA-ICP-MS trace-
element data to the provenance of dolomite-related white nephrite. The authors demonstrate the
technique’s high accuracy and potential for origin determination with many other gems.
We hope you enjoy our second Micro-World column. We had a fantastic response to the first installment
and are extremely pleased to welcome several new contributors. We also want to thank our Lab Notes and
Gem News International (GNI) contributors—these sections always refresh me with their diversity.
Finally, I’d like to welcome two new editors to our GNI section: Gagan Choudhary of the Gem Testing
Laboratory in Jaipur, India, and GIA’s own Dr. Mike Breeding—we thank them for their commitment and
participation.
Welcome to our Fall issue!
Duncan Pay | Editor-in-Chief | dpay@gia.edu
EDITORIAL GEMS & GEMOLOGY FALL 2015 221

FEATURE ARTICLES
COLOMBIAN TRAPICHE EMERALDS:

RECENT ADVANCES IN UNDERSTANDING
THEIR FORMATION
Isabella Pignatelli, Gaston Giuliani, Daniel Ohnenstetter, Giovanna Agrosì, Sandrine Mathieu,
Christophe Morlot, and Yannick Branquet
Colombia is the traditional source of the world’s finest emeralds, including the famed trapiche crystals,
with their distinctive texture resembling a wheel with six spokes. This gemological curiosity, found ex-
clusively in the black shales of the country’s western emerald zone, is linked to the peculiar structural
geology of the deposits. The study presents a review and update on Colombian trapiche emeralds, fol-
lowed by a three-dimensional examination of the crystals combined with spectroscopic and chemical
analyses. The proposed formation model incorporates the structural geology of the deposits with the
formation of trapiche and non-trapiche emeralds. The fluid accumulation at the faults’ tip in the black
shales leads to maximum fluid overpressure and sudden decompression and formation of the emerald-
bearing vein system. The authors show that trapiche emerald growth starts at the beginning of the de-
compression that is responsible for local supersaturation of the fluid. The hydrothermal fluid comes in
contact with the black shale matrix, favoring the formation of emerald seed crystals. During the growth
of these seeds, textural sector zoning occurs, sometimes associated with chemical sector zoning, along
with displacement of the matrix. Displacement growth occurs because the emeralds continue their
growth, pushing the matrix material away from the growing faces. An overgrowth, generally of gem
quality, can form after decompression, surrounding the core, the arms, and the dendrites, restoring the
emeralds’ euhedral habit.
C
olombian emeralds continue to set the stan- side of the Eastern Cordillera Basin. Named after the
dard as the finest and most spectacular crys- Spanish word for the cogwheels used in sugar mills,
tals unearthed (Ringsrud, 2013). Unique these mineral curiosities are prized by collectors. The
mineralogical curiosities include the emerald gas- texture on a section perpendicular to the c-axis is
tropods extracted from the Matecaña mine in characterized by a central core, six arms, and den-
Gachalà (Vuillet et al., 2002) and the famous gem- drites between the arms and around the core (figure
trapiche emeralds (figure 1) such as the 80.61 ct Star 1). An overgrowth may also be present. The trapiche
of the Andes (Ward, 1993). texture can be observed in other minerals, such as ru-
The past 25 years have seen major advances in our bies, tourmalines, chiastolites, and garnets (box A).
understanding of the formation of Colombian emer- Without a link between field observation and miner-
alds (Ottaway et al., 1994; Cheilletz et al., 1994), but alogical studies, there is still no consensus on the
the genesis of trapiche emerald remains unresolved causes of the growth mechanism and texture acqui-
despite analytical and geological advances. This type sition, or on the geological conditions necessary for
of emerald is very rare and has been recovered only the formation of trapiche emerald.
occasionally, from just a few mines on the western This article provides an update on Colombian
trapiche emerald, with complete historical, geologi-
See end of article for About the Authors and Acknowledgments.
cal, mineralogical, gemological, and crystallographic
GEMS & GEMOLOGY, Vol. 51, No. 3, pp. 222–259, background. The review opens debate on the key
http://dx.doi.org/10.5741/GEMS.51.3.222. question of the geological conditions necessary for
© 2015 Gemological Institute of America the formation of gem-quality trapiche material. The
222 COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015

Figure 1. Looking down
the crystal axis of a
backlit 58.83 ct trapiche
emerald from Peñas
Blancas (also featured
on the cover of this
issue). Photo by Robert
Weldon/GIA, courtesy
of Jose Guillermo Ortiz,
Colombian Emerald Co.
ongoing research into the mineralogy, chemical com- Colombian emeralds. Colombia offers a unique case
position, and crystallography of trapiche emerald is because other gem trapiches have not been studied
connected with field expeditions conducted by intensively in connection with their geological set-
French and Canadian research teams during the ting. For the gemologist, this knowledge offers a bet-
1990s that proposed the hydrothermal sedimentary ter understanding of the mineralogy of trapiche
model now recognized by the scientific and explo- emerald and a deeper appreciation of the complex ge-
ration communities (Maya et al., 2004; Mantilla ologic processes required to form such a peculiar
Figueroa et al., 2007). The scope is to propose a co- mineral texture. Geologic terms that might be unfa-
herent mechanism for the formation of trapiche, miliar to the reader are defined in the glossary and
linking it for the first time to the genetic model of italicized upon first mention within the text.
COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015 223

BOX A: TRAPICHE VS. TRAPICHE-LIKE MINERALS
Win (2005) distinguished three types of trapiche miner- The symmetry affects the number of arms and dendrites
als as a function of their appearance due to coloring el- in the trapiche texture, as well as the directions where
ements, other mineral inclusions, or intergrowth of the the dendrites develop. Because high symmetry implies
same mineral. Schmetzer et al. (2011) improved this dis- that a certain number of directions are equivalent, the
tinction, defining only two groups: “trapiche” minerals positions of the dendrites can be predicted taking into
and “trapiche-like” minerals (figure A-1). account the crystal system. For example, in emerald
Trapiche minerals are characterized by crystallo- there are six dendrites respectively along the positive
graphically equivalent growth sectors that are separated and negative sides of the a-axis, whereas in chiastolite
by more or less sharp boundaries of inclusions (figure A- the four dendrites are along the a- and b-axes.
1, left). The boundaries intersect in a central point or ex- To the best of our knowledge, no trapiche texture
tend from the edge of a central core. Gems that belong has been reported in minerals with low symmetry (or-
to this group include emeralds from Colombia, corun- thorhombic, monoclinic, or triclinic), even though the
dum (e.g., Müllenmeister and Zang, 1995; Schmetzer et development of dendrites on the corners of minerals
al., 1996, 1998; Sunagawa, 1999; Garnier et al., 2002a, with euhedral morphology such as olivine is already
2002b), tourmalines (Hainschwang et al., 2007; Schmet- known (Faure et al., 2003, 2007). This suggests that at
zer et al., 2011), chiastolites (Rice and Mitchell, 1991; least two of the three basic vectors a, b, and c have to
Rice, 1993), and garnets (Harker, 1950; Atherton and be equal (or, in the case of chiastolite, nearly equal) to
Brenchley, 1972; Wilbur and Ague, 2006). obtain the trapiche texture, and thus the role of symme-
We also mention two examples of non-Colombian try cannot be neglected.
trapiche emerald crystals, one from the Brazilian state The presence of transformed organic matter or
of Goiás (DelRe, 1994) and the other from the Manan- graphite is observed in trapiche minerals formed in differ-
jary area in Madagascar (Johnson and Koivula, 1998). ent geological environments—for instance, metamorphic
The Brazilian crystal showed fibrous arms and a tapered deposits such as ruby (Garnier et al., 2002a,b), chiastolite,
core. The specimen from Madagascar presented the garnet (Rice and Mitchell, 1991), and tourmaline (Schmet-
spoke-like texture associated with Colombian emerald. zer et al., 2011). The role of carbonaceous materials in
The core was limited to a center point, while the den- trapiche formation was raised by Rice and Mitchell (1991)
drites contained black material. These two gems refer and Rice (1993), referring to Burton (1986), who stated that
more to trapiche than trapiche-like emeralds, but they graphite can control the fluid composition and dictate the
cannot be compared with Colombian samples because mechanism and timing of mineral growth.
their geological formation is unknown. In trapiche-like minerals, the texture is caused by
Two other features observed in trapiche crystals the distribution of color-inducing elements or of inclu-
should be mentioned: symmetry and inclusions of or- sions in alternating portions of the crystal (figure A-1,
ganic matter. Trapiche minerals are always character- right). This is the case with sapphire (Koivula et al.,
ized by high symmetry: cubic for garnet, hexagonal for 1994; Khotchanin et al., 2010; Kiefert, 2012), quartz
emerald, and trigonal for corundum and tourmaline. (Win, 2005), and aquamarine (Koivula, 2008; Befi, 2012).
The only exception is chiastolite, which belongs to the Cordierite-indialite intergrowths also show a trapiche-
orthorhombic system but is pseudotetragonal (a ≈ b). like texture, as described by Rakovan et al. (2006).
Figure A-1. Trapiche vs. trapiche-like

corundum. Left: Trapiche ruby origi-
nating from marble in Mong Hsu
(Myanmar). The trapiche texture is
characterized by dendrites and arms,
though the specimen lacks a central
core. Photo by V. Garnier. Right: A
pinacoidal section of a trapiche-like
blue sapphire from the Changle alkali
basalt (China), showing a hexagonal
core and outer zones with a fine oscil-
latory zoning. Photo by G. Giuliani.

74º W 72º W
San ssif
N
Ma
0 100 km
tand
er
in
as
Sierra
a
a B
er
Nevada
del Cocuy Figure 2. Simplified geo-
ill
en
logical map of the East-
d
al ern Cordillera Basin in

or
gd
Colombia. The emerald
l C
Ma
6º N Peñas Blancas deposits are hosted by

ra
Lower Cretaceous sedi-

le
Coscuez
nt
La Pita
dd
mentary rocks forming

Ce
Muzo two mineralized zones

Mi
Tunja
Yacopi
located on the eastern
and western border of
m
Macanal
in
riu
the basin, respectively.

ino
as
icl
Chivor
B
The western border con-
nt
aa
d
tains the mining districts
n
let
Gachalá
la
Vil
of La Glorieta–Yacopi,
re
Bogotá
m
riu
Muzo, Coscuez, La Pita,

Fo
ino
and Peñas Blancas,

l
tic
s
o
an
while Gachalá, Chivor,

an
y
cu
Co
Ll
and Macanal lie on the

asi t
e-
su rardo
n
am
4º N eastern border.
et
bb
Gi
Qu
Tertiary
Upper Cretaceous
thrust
Emerald deposits Lower Cretaceous
normal fault Triassic-Jurassic
fold axis strike-slip fault Basement and plutons
GEOLOGY OF COLOMBIAN EMERALDS: the Lower Cretaceous (135–116 Ma). The sediments
AN UPDATE are characterized by a succession of beds of sand-
Geological Setting. Located in the Eastern Cordillera stones, limestones, black shales, and evaporites.
Basin, the Colombian emerald deposits consist of The deposits from the eastern belt are contained
two belts. On the western side lie the mining dis- in the Berriasian limestone–black shale horizon (135–
tricts of Muzo, Coscuez, La Pita, Peñas Blancas, and 130 Ma) of the Guavio Formation, which is overlain
La Glorieta–Yacopi. On the eastern side are Chivor, by siliceous black shales of the Valanginian Macanal
Gachalá, and Macanal (figure 2). The Eastern Formation (130–122 Ma). In the Chivor mining dis-
Cordillera is a slightly folded belt overthrusting the trict, the host rocks comprise the Guavio Formation
Llanos Foreland Basin to the east and the Middle shale sequences that contain intercalations of lime-
Magdalena Basin to the west. The belt resulted from stone lenses and gypsum beds, as suggested by phan-
the tectonic inversion at the Middle Miocene (15 tom nodules, mesh and chevron textures, and
Ma), during the Andean tectonic phase, of the central coquina limestone grading to black shales interca-
part of the subsiding marine basin. The major part lated with olistostromes (Branquet, 1999).
of the Eastern Cordillera is formed by thick folded The deposits from the western belt are contained
and faulted Mesozoic sedimentary series. Emerald in the black shales and intercalated dolomitic lime-
mineralization is hosted in the Neocomian series of stones of Valanginian-Hauterivian age (130–116 Ma)

corresponding to the Rosablanca and Paja Forma-
tions. The lithostratigraphic column, from bottom to
top, is formed by dolomitic limestones from the
Rosablanca Formation, calcareous carbon-rich black
shales from the Hauterivian that are the main emerald-
bearing horizon, siliceous black shales (Hauterivian),
and mudstones of the Barremian-Aptian Paja Forma-
tion (116–108 Ma).
The ages of emerald formation were obtained in-
directly by argon-argon dating of syngenetic green
muscovite. From this analysis, the first date was 65 Figure 3. Primary multiphase fluid inclusion trapped
Ma in the eastern zone, at the Cretaceous-Tertiary in an emerald from El Oriente mine in the Chivor dis-
boundary when the dinosaurs disappeared (Cheilletz trict. The cavity, 180 µm long, contains (from right to
et al., 1997). The second date was between 38 and 32 left) two cubes of sodium chloride (halite), a rounded
gas bubble, and two minute calcite crystals, with salt
Ma in the western zone, at the time of the Eocene-
water occupying 75 vol.% of the cavity. The cavities
Oligocene boundary (Cheilletz et al., 1994). usually contain only one cube of salt, but sometimes
two or three cubes crystallize during fluid cooling.
Emerald Mineralization. In the two mineralized belts, Photo by Hervé Conge.
emerald-bearing veins are spatially associated with
stratiform breccias and white or black albitites, the
latter resulting from the albitization of the black shales. liani et al., 1993; Mantilla Figueroa et al., 2007). Stage
The veins are parallel, en echelon, or conjugate arrays 2 is characterized by faulting and folding associated
with extensional vein sets and hydraulic breccias
filled with calcite and dolomite, pyrite, muscovite,
albite, bitumen, and the precipitation in drusy cavi-
In Brief ties of fluorite, apatite, parisite, dolomite, emerald,
• Trapiche emeralds are found in the black shales of a and quartz (Hall, 1993).
few Colombian mines, in the western belt of the East-
The Colombian emerald deposits were previously
ern Cordillera Basin. They are characterized by a tex-
ture, visible perpendicular to the crystal’s c-axis, that thought to be associated with mafic or granitic intru-
usually consists of a central core, six arms, and den- sions (Beus and Mineev, 1972; Ulloa, 1980). Detailed
drites between the arms and around the core. geochemical studies undertaken in the 1990s led to a
• While trapiche emeralds are formed from the same hydrothermal model involving the circulation of hot
fluid as gem-quality Colombian emeralds, their texture basinal brines (Kozlowski et al., 1988; Ottaway, 1991;
results from a complex growth history characterized by Giuliani et al., 1991; 1992; 1995; 2000; Cheilletz et
fluid pressure variations. al., 1994; Ottaway et al., 1994; Banks et al., 2000;
• The formation of trapiche emeralds is controlled by the Mantilla Figueroa et al., 2007). The fluids trapped by
structural geology of the deposits and by the crystal’s emerald are three-phase or multiphase fluid inclu-
symmetry, which determines the number and the de- sions (figure 3) characterized by the presence of a cube
velopment of growth sectors. (or occasionally two or three cubes) of halite (NaCl).
At room temperature, the fluid inclusion cavities
contain 75 vol.% salty water—in other words, a brine
forming two successive stages of extensional vein (liquid H2O, 10 vol.% gas corresponding to the vapor
systems (Cheilletz and Giuliani, 1996). Stage 1 is ac- bubble, and 15 vol.% halite daughter mineral). The
companied by a bedding-parallel vein system filled mineralizing solutions are basinal brines that inter-
by fibrous calcite, pyrite, and green muscovite. It is acted with evaporites. The H2O-NaCl-CO2-(Ca-K-Mg-
associated with the formation of décollement fault Fe-Li-SO4) fluids (Banks et al., 2000) are NaCl-saturated
planes that focused the hydrothermal fluids and in- [~40 wt.% equivalent NaCl] and were trapped at ap-
duced formation of albite and calcite in the black proximately 300–330°C (Roedder, 1984; Ottaway et
shales. The sodium metasomatism led to leaching of al., 1994). The CO2 densities of fluid inclusions range
major (Si, Al, K, Ti, Mg), trace (Ba, Be, Cr, V), and rare- from 0.02 to 0.25 g/cm3 and indicate that emerald
earth elements from the enclosing black shales (Beus crystallized under variable fluid pressure. The in-
and Mineev, 1972; Beus, 1979; Ottaway, 1991; Giu- crease of CO2 pressure caused fluid overpressure in

Emerald deposits
and occurrences Figure 4. Tectonic settings
Anticlines C Coscuez of Colombia’s western
Thrusts A SSW NNE emerald zone (modified
Tear faults 0 200 m from Branquet et al.,
Peñas Blancas
Albian and Late 1999a). A: Structural lo-
t
Cretaceous
aul
cation map of the emer-
e
sediments
as f
iclin
ald deposits and the
anc
fault
Early Cretaceous
ant
line
sediments
s Bl
D Quipama mine (Muzo deposit) regional structures, from
uez
antic
apa
a
Peñ
NNW SSE Rodriguez and Ulloa
Cosc
a Ch
ez (1994). B: Lithostrati-
de l
cu
Otanche
Cos
Alto
graphic column of the
e
on
Coscuez emerald-bearing forma-
lt z
fau
0 200 m tions. From bottom to
ero
C top: lowermost Creta-
Min
E ceous rocks, Valanginian-
Rio
Tequendama mine (Muzo deposit)

Rio NNW SSE Hauterivian dolomitic
Ito
co limestone, Hauterivian
fau
lt calcareous carbon-rich
E black shale, Hauterivian
Val. Hauterivian Barremian
Tequendama siliceous black shale, and

D 0 200 m
Barremian-Aptian mud-
Quipama Muzo
stones. The red dotted line
N
represents the décolle-
B ment zone. C through E:
0 5 km
Emerald deposits of
Coscuez, Quipama, and
Tequendama mines
(Muzo district).
the black shales and consequent fracturing and brec- posits such as Chivor present extensional structures
ciation (Ottaway et al., 1994). The high-salinity brines branched on a brecciated evaporitic level that acted
interacted with calcareous shales rich with black or- as a local, gravity-driven detachment (Branquet et al.,
ganic matter. Sulfate ions (SO42–) in minerals of evap- 2015). These tectonic structures are synchronous
oritic origin were reduced by organic matter in the with the circulation of the hydrothermal fluids and
black shale to form hydrogen sulfide (H2S) and bicar- emerald deposition.
bonate (HCO3–), which are responsible for the crystal- On the western side, the deposits measure about
lization of pyrite and carbonates, respectively, with 100 meters across and display numerous folds,
bitumen in the veins being derived from the organic thrusts (Pogue, 1916; Scheibe, 1926), and tear faults
matter. (Laumonier et al., 1996). At Muzo, thrusts are evi-
The thermal reduction of sulfate by organic mat- denced by the presence of calcareous black shales
ter, at 300°C, released the chromium, vanadium, and over siliceous black shales. All the tectonic contacts
beryllium in the black shale, which in turn enabled are marked by centimeter- to meter-thick hydrother-
emerald formation (Ottaway et al., 1994; Cheilletz et mal breccias called “cenicero” (ashtray) by the local
al., 1994). miners (Scheibe, 1926). These white or red breccias
outline the thrust planes, which are associated with
Structural Setting and Type of Mineralization. De- intense hydraulic fracturing due to overpressured flu-
tailed structural mapping and geometric analysis ids (Giuliani et al., 1990; Ottaway et al., 1994; Bran-
suggest that structural controls on mineralization are quet et al., 1999b). The breccias are cataclasites, with
drastically different between the western and eastern clasts of calcareous black shales and white albitite
sides of the Colombian emerald belt (Branquet, 1999; within a carbonate-albite-pyrite cement. Multistage
Branquet et al., 1999a,b). Muzo and Coscuez are char- brecciation corresponds to successive fault-fluid flow
acterized by compressive structures formed along pulses, and dilatant sites result from shear-fracturing
tear faults (figure 4), whereas the eastern emerald de- synchronous to the thrust-fault propagation. Each

Unmineralized zone (“cambiado”)
Banco Emerald-bearing zone (“capa esmeraldifera”)
Miguel Ruiz
Albitite zones. All these tectonic structures are associated
Veins, breccia, “cenicero” with fluid circulation in the calcareous carbon-rich
“Cenicero” black shale, inducing intense albitization, carbonati-
Alluvium zation, and pyritization. The siliceous black shales,
House called “cambiado” by the local miners, have no min-
River eralization (Oppenheim, 1948).
From a geological study in 1914–1915, Pogue
(1916) described the trapiche texture but did not pro-
vide the precise location of the source. Scheibe (1926,
Banco Jerena 1933) identified the disseminations of trapiche emer-
130m
ald in the black shales from the Banco Amarillo.
Banco Tequendama (“Banco” refers to a mineralized level, and “amarillo”
is Spanish for yellow; here the yellowish color is due
to pyrite alteration.) The geological map of Banco de
la Republica (Barriga Villalba, 1948) indicates that
Banco Amarillo was formed by the breccia zone, al-
Banco Itoco bitites, and the emerald-bearing veins (figure 5). The
100 m
30 m
geological map by Laumonier et al. (1996) shows that
nte
Rio It
oco
a preserved remnant of the Banco Amarillo level (fig-
e
rdi
130 m
gua
20 m ure 6) is located under the administration building of

El A
160 m
ada
the Muzo mine and corresponds to a klippe of the

ebr
90 m
Banco
Qu
Los Chulos calcareous black shale that overthrust the siliceous

co
Banntral 105 m
black shale. Since 1948, this emerald-bearing forma-
Ce
120 m
tion has been exploited further to the southwest up
Banco 10 m to the Gallinazo, Zincho, and Malvinas mining
130 m Limón
Banco El Banco
140 m workings (compare figures 5 and 6). Furthermore, Ot-
Rio It
Aguardiente La Polvorera Banco Amarillo taway (1991) reported that “more recently trapiche
oco
130 m
130 m have been found in the shale adjacent to larger bodies
Banco El Gallinazo Banco El Cascarón of Cenicero.”
280 m 230 m
140 m In the Coscuez deposit, the lithostratigraphic col-
umn is formed, from bottom to top (Branquet, 1999),
Banco by dolomitic limestone forming the peak of El Reten,
Las Animas im
as N
s An calcareous carbon-rich black shale, and siliceous
a La
eb
rad black shale. The folds and thrusts were guided by the
Qu
0 200 m
Coscuez tear fault, which acted as a vertical conduit
for the mineralizing fluids developed in the calcare-
ous carbon-rich black shale (figure 4B). Breccia, formed
by opening of dilatant sites related to fluid pressures
Figure 5. Geological map of the Muzo mines, with the and hydrothermal replacement, are similar to those
height of each mining site reported in bold. Modified described for the Muzo deposit (Branquet et al.,
from Barriga Villalba (1948).
1999b). Trapiche emeralds have been found at
Coscuez but have never been described by geologists.
In conclusion, the deposits on the western side
pulse is associated with (1) emerald-bearing banded formed as the consequence of a compressive phase
carbonate vein-like structures present throughout the characterized by folding and thrusting along tear
breccia; (2) emerald-bearing thrust-associated carbon- faults at the time of the Eocene-Oligocene boundary.
ate veins occurring in the wall rocks formed of cal- These complex structures imply the existence of a
careous black shales, called “cama” by the local basal regional décollement fault at a level of evapor-
miners; (3) emerald-bearing carbonate dykes escaping ites (Branquet et al., 1999b). The fluid circulation is
from the breccia zone and crosscutting the wall rocks; linked to the thrust-fault propagation.
(4) en echelon sigmoidal tension gashes; and (5) drag On the eastern side, the mines are scattered along
folds indicating shearing in the roof of the breccia a regional white-brecciated level that contains emer-

ald (Branquet, 1999; Branquet et al., 1999a, 2015). The
brecciated level in the Chivor area, more than 10 km
A
long and 1–10 meters thick, is stratiform (i.e., parallel N
Minabuco
to the sedimentary strata) and largely made of hy-
drothermal breccia formed by fragments of the hang-
ing wall (carbonated black shale, limestone, and 0 150 m
whitish albitite), cemented by carbonates and pyrite.

Its formation is related to the dissolution of an evap-
Tequendama mining zone
co
bu
oritic horizon. All the mineralized structures are
i na
Rio mining zone in 1995
aM
branched from the brecciated level. In the Chivor Ito
co
rad
mines, emerald is located in centimeter- to decime-
eb
Qu
ter-thick carbonate- and pyrite-bearing listric faults,
meter-wide extensional fractures injected with hy- B
drothermal breccia, and extensional sets of fractures
Ito
co
in the calcareous carbon-rich black shales of the
f au
lt
Tequendama pit
Macanal Formation. According to a member of the
family that has owned the Chivor mines for many
years, trapiche emeralds have never been found there Puerto Arturo pit
or anywhere else in the eastern zone of the Eastern
Cordillera (D. Oswaldo, pers. comm., 2014). Never-
theless, Nassau and Jackson (1970) claimed to have
studied a thousand trapiche emeralds from these
mines, and their article raised the question of Chivor Palo Blanco pit
Rio
origin. In conclusion, the brecciated level, the hy- Quipama mining zone ien
te
Ito
drothermal fluid circulation and emerald formation uard
co
a Ag
occurred at the same time, at the Cretaceous-Tertiary rad
eb
boundary (~65 Ma), during an extensional tectonic Qu
event linked to evaporite dissolution and driven by Administration

gravity (Branquet et al., 1999a, 2015).
Therefore, trapiche emeralds are only found in the
deposits on the western side (Muzo, Coscuez, and
Peñas Blancas) of the Eastern Cordillera Basin. The fa ult Gallinazo
trapiche emerald–bearing deposits are formed along nte
die
uar
thrust and fault planes associated with fluid circula- Ag Zincho as
Anim
ad a Las
tions, intense hydraulic fracturing of the enclosing Quebr
black shales forming breccias, and carbonate veins. Fault
Trapiche and non-trapiche emeralds are associated Thrust

s t
vina aul
with these compressive structures. Mal as f
lvin
Ma Tear fault
EMERALD: STRUCTURE AND HABIT
Fold axis
Emerald is a gem variety of beryl, a cyclosilicate with
the ideal formula Be3Al2Si6O18. Its structure is char- U4 Sedimentary
bedding
acterized by six-membered rings of silica tetrahedra Siliceous black shales Calcareous carbon-rich
U2 U3 black shale (emerald zone) A
lying in planes parallel to (0001). The rings are linked (unmineralized zone) Section
B
laterally and vertically by two kinds of coordination U1 Alluvium
polyhedra, both distorted: BeO4 tetrahedra and AlO6
octahedra (figure 7). The stacking of these six-mem-
Figure 6. Geological map of the Muzo mines, where
bered rings forms large open channels parallel to the U1 through U4 represent the different tectonic units.
c-axis with nonuniform diameter, consisting of cav- U1, U2, and U4 are composed of barren siliceous
ities with a diameter around 5.1 Å separated by bot- black shales, while U3 consists of the emerald-bearing
tlenecks with a diameter of about 2.8 Å. The calcareous carbon-rich black shale. Modified from
channels may be filled by alkali ions (such as Na+, Laumonier et al. (1996).

A Ti4+ (Groat et al., 2014). Variable amounts of neutral
H2O and CO2 molecules (Wood and Nassau, 1968)
and noble gases such as argon, helium (Damon and
Kulp, 1958), xenon, and neon (Giuliani et al., 2005)
are also normally present in the channels.
Emerald is defined by Schwarz and Schmetzer
(2002) as “the yellowish green, green or bluish green
beryl which reveals distinct chromium and/or vana-
Be2+
dium absorption bands in the red and blue violet
Si4+
ranges of their absorption spectra.” The quantitative
Al3+
ranges of the Cr and V substitutions are between 25
O2– ppm (Wood and Nassau, 1968) and 3.4 wt.% (Andri-
anjakavah et al., 2009) for Cr2O3, and between 34
ppm (Zwaan et al., 2012) and 2.44 wt.% (Rondeau et
al., 2008) for V2O3. Emerald crystallizes in the hexag-
onal system with cell parameters a = 9.218(2) Å, c =
B
5.1 Å
9.197(2) Å, α = β = 90°, and γ =120°, and its space
group is P6/mcc (Artioli et al., 1993). Its typical habit
is prismatic (figure 8), characterized by eight faces
–
and their corresponding growth sectors: six {1010}
first-order prismatic faces and two pinacoidal {0001}
– –
faces. Small additional {1012} and {1122} faces may
2.80 Å 2a
also be present in some crystals.
Be2+
Si4+
2b TRAPICHE COLOMBIAN EMERALD: A REVIEW
Al3+
Colombian trapiche emeralds were first described by
O2– the French mineralogist Emile Bertrand (1879), in a
2a meeting at the Société Géologique de France, in
which he presented “curious crystals of emerald…
from Muso, New Granada.” All 40 samples displayed
a nearly colorless hexagonal center surrounded by a
green portion. The latter is described as having stri-
2.80 Å ations parallel to the sides of the central hexagon and
with “modifications” in the directions tangential to
the sides of the hexagon. No further information was
added about the composition or the nature of the
Figure 7. The structure of beryl, as seen in an apical “modifications.” Bergt (1899) described a cut trapiche
view (A) and a lateral view (B). In the apical view, emerald from Muzo received by the geologist Stübel
hexagonal silicate rings stacked parallel to the c-axis in 1868. Similar emeralds were reported by Codazzi
(normal to the drawing) are held together by Al3+ (oc-
(1915) in an accurate listing of Muzo’s minerals.
tahedral site) and Be2+ (tetrahedral site). The lateral
These emeralds were reported to have a cyclic twin-
view, which is perpendicular to the apical view,
shows the hexagonal silicate rings and the bottleneck ning, as observed in aragonite. This contradicted the
(2b site) and open-cage (2a site) structures. From optical observations of Pogue (1916), who stated that
Charoy (1998). these emeralds were not twinned. According to
Pogue, the presence of re-entrant angles was due to
“the effect of solution, the disposition of carbona-
ceous inclusions and the crystallizing forces, as
Li+, K+, Rb+, and Cs+) that are needed to balance the shown also, for example, in chiastolite.” It is worth
positive charges when cation substitutions occur in noting that the Pogue study contained the first men-
the structure. For example, Be2+ can be substituted tion of inclusions of carbonaceous matter in the
with Li+ (Aurisicchio et al., 1988), whereas Al3+ is emeralds, “arranged in a six-rayed centering about a
generally replaced by Fe2+, (Mg2+, Mn2+), Cr3+, V3+, or tapering hexagonal core.”

_
Prismatic growth sectors {1010}
Hexagonal prismatic Pinacoidal growth

habit sectors {0001}
Figure 8. The habit of
(0001) emerald crystals, char-
acterized by eight main
_
_ (0110) faces and their corre-
(1100) sponding growth sec-
–
tors: six {101 0}
first-order prismatic
faces and two pina-
_ coidal {0001} faces.
(1010) Adapted from Rakovan
et al. (2006).
During a geological study of Muzo in 1914–15, also observed. The core of the emeralds had the shape
Scheibe (1926, 1933) collected several emeralds from of two opposite hexagonal pyramids with their ver-
the Banco Amarillo in the calcareous carbon-rich tices located in the middle of the crystal. Sometimes
black shales of the Lower Cretaceous (again, see fig- these pyramids were so unevenly developed that the
ure 5). These emeralds appeared very different from core resembled a column. The core was richer in in-
those usually found in the veins, and for this reason clusions, some of them more darkish to black, than
they were analyzed in detail by Bernauer (1933). Dif- the rest of the crystal. In fact, it can be surrounded
ferent morphologies were observed, as shown in fig- by the dark inclusions (figure 9B), entirely composed
ure 9: euhedral hexagonal prismatic emeralds (figure of the dark inclusions (figure 9D), or formed by sev-
9A), but also crystals with signs of corrosion along eral parallel crystals separated by dark inclusions (fig-
the edges of the prism (figures 9B to 9E). The emer- ure 9C). In some emeralds the inclusions can be
alds showed dark, fibrous inclusions between the whitish instead of dark, due to the presence of albite
prismatic edges, starting from the middle of the crys- and kaolin.
tal and enlarging toward the prism’s corners. The in- Although Bernauer (1933) suggested that these
clusions, which seemed to be emphasized by the trapiche emeralds were due to cyclic twinning as in
corrosion, were composed of quartz, muscovite, car- cordierite, chrysoberyl, aragonite, or other minerals,
bonates, pyrite, and a dark carbonaceous matter he emphasized that the growth rates in the different
(probably with an organic or bituminous origin), crystallographic directions played an important role
sometimes with biotite and kaolin. Multiphase in- in the incorporation of the dark inclusions and, thus,
clusions with liquid, vapor, and solid phases were in the formation of these emeralds. Nevertheless,
Figure 9. Different mor-

phologies of trapiche
emeralds from the
Muzo mine, described
by Bernauer (1933). A:
Euhedral hexagonal
prismatic crystals with
no sign of corrosion. B
through E: Crystals
showing different de-
A B C D E grees of corrosion along
the prism’s edges.

Barriga Villalba (1948) described the unusual trapiche identify them as natural or synthetic. The presence
emeralds from Muzo as aragonite-type twins of sci- of channels, three-phase fluid inclusions, and solids
entific interest but no gemological value. Studying (organic matter, possibly graphite, as well as quartz
the mineralogical features of Muzo emeralds, Barriga and albite) confirmed their natural origin. Chaudhari
Villalba (1948) pointed out the special aspects of the (1969) described the emeralds with the unusual gear-
core in crystals of different sizes. The gem core was shaped pattern as composite crystals made of two
formed by pyramidal and hexagonal cones presenting distinct entities: a hexagonal pale green core and a
a three-dimensional geometric shape that tapered dark green overgrowth on the prism faces. Around
smoothly from the pinacoid to the vertex. These gem the core, the sectors were separated by a very fine-
cones, sometimes of high quality, were extracted grained clay-like matrix (containing abundant
from the emerald crystal. quartz). Some of the sectors were irregular and
The 1960s saw further studies on Colombian formed by stringers that grew from the core to the
trapiche emerald, with the discovery in 1963 of new faces. Small stringers of emeralds at 60° angles to
material in the Peñas Blancas mines (Tripp and Her- each other in the matrix were also described, but
nandez, 1970). In 1964, McKague introduced the term only in the gaps between the sectors. According to
“trapiche” to name the unusual Colombian emerald Chaudhari, the sawtooth contact between the matrix
whose crystal habit was similar to the cane-crushing and the sectors with the stringers indicated that sev-
gears used by farmers. According to McKague, eral changes occurred in the chemical and physical
trapiche emerald was characterized by four morpho- conditions during trapiche growth.
logical elements: Nassau and Jackson (1970) marked a real advance
in the understanding of trapiche formation after
1. A central deep green hexagonal prism tapered
Bernauer (1933). Nevertheless, this work raised a ques-
toward one end and without inclusions
tionable point on the origin of trapiche material. The
2. Six trapezoidal-shaped prisms extending from
– authors claimed to have studied a thousand trapiche
the {1010} faces of the central prism and con-
emeralds from Chivor and only two from Muzo. Be-
taining opaque inclusions
tween 1963 and 1970, trapiche emeralds came only
3. A colorless fine-grained beryl occurring be- from the Peñas Blancas and Muzo mines. Leiper (1967)
tween and within the six trapezoidal prisms had not specified the provenance of his samples, lead-
but also in the central prism, with opaque min-
ing to a series of misunderstandings in subsequent bib-
erals (probably altered in limonite) observed on
liographic citations. But Leiper’s photos of the trapiche
and between the beryl grains
emeralds resembled those presented by Tripp and Her-
4. An overgrowth separated from the six prisms nandez (1970) for the Peñas Blancas mine. Samples
by scattered patches of opaque inclusions from Peñas Blancas were given to Leiper by M. Ander-
McKague was the first to propose a genetic model ton, manager of the Chivor emerald mine, and this is
for the formation of trapiche emeralds. In this model, probably the source of the confusion over the origin
the central prism formed first. A change in pressure, of the stones studied by Nassau and Jackson. Finally,
temperature, and chemical composition or a combi- Sinkankas and Read (1985) clarified the provenance of
nation of these parameters caused the formation of the trapiche emeralds described by Nassau and Jack-
the trapezoidal prisms. A further change in the sys- son, asserting that they were from Peñas Blancas.
tem led to the overgrowth’s formation. At the end, a Examinations of cross-sections of single crystals
more drastic change occurred and the trapezoidal have shown considerable variation in the structure
prisms were partially converted into fine-grained of trapiche emerald; see figure 10. Nassau and Jack-
–
beryl at the intersections of the {1010} faces. son (1970) used the terms “core,” “arms,” and “two-
In 1967, Leiper examined unusual trapiche emer- phase region.” The hexagonal core may or may not
alds reportedly from the Chivor mine. The core and the be present (figure 10C), and it is usually of good-qual-
six arms branching from it were emerald of good quality emerald, but it is possible to find samples with
ity, while the material between the arms was a whitish- one end as in figure 10B and the other as in figure
greenish beryl. This low-quality material was compared 10C. The core usually has a taper, as previously
to the fine-grained beryl described by McKague, noted by Pogue (1916), Barriga Villalba (1948), and
usually called “moralla” by the local miners. McKague (1964). The emerald formed on the prism
In 1968, Schiffman analyzed trapiche crystals to faces of the core are named “arms,” because their

Two decades later, Van der Giessen (1994) reported
several Colombian trapiches, among them a special
sample with two hexagonal cores surrounded by eight
trapezoidal sectors (figure 11). This sample seems to
be a rare case of contact twinning in trapiche emer-
alds. According to the author, the formation of the
A B C trapiche is related to sector zoning. This peculiar ex-
ample is perhaps closer to the unusual double-
trapiche emeralds from Muzo reported by Hsu (2013),
which consist of twelve fine arms intersecting in a
central point. These samples are described as “com-
posed of two six-ray stars shifted slightly from each
other,” though one might think they are twinned.
D E F
Ohnenstetter et al. (1998) studied Colombian
trapiche emeralds from Muzo, Coscuez, and Peñas
Figure 10. Schematic cross-sections of different trapiche
emeralds described by Nassau and Jackson (1970).
Blancas. The cores were colorless, and the trapezoidal
growth sectors were colored with vivid green over-
growths. Cathodoluminescence underlined the chem-
ical zoning due to variations of vanadium (V) and
shape is not always trapezoidal (as described in pre-
chromium (Cr). The first published microprobe data
vious works). Moreover, the arms may be asymmet-
on trapiche emerald confirmed that the green zones
rical (figure 10F). The material observed between the
were V-rich (0.5 < V2O3 < 0.79 wt.%) and relatively Cr-
arms as well as between the core and the arms can
poor (0.35< Cr2O3 < 0.39 wt.%). The colorless hexago-
be white/gray or dark. The white to gray material is
nal prism was a Cr- and V-free beryl. Scanning electron
formed by a mixture of emerald and albite, while the
microscope observation of sections perpendicular to
dark material is carbonaceous, even if it contains al-
the c-axis showed streaks along the trapezoidal growth
bite. Trapiche emeralds with a core completely com-
sectors and a feathery aspect to the inclusions in the
posed of this dark material have also been reported.
arms of the crystals. The arms were formed mainly of
Nassau and Jackson (1970) rejected the hypothesis
albite and beryl, with some apatite and pyrite. The au-
of twinning to explain trapiche formation. They pro-
thors thought that trapiche emeralds resulted from a
posed two mechanisms, one for the samples purport-
edly from Chivor and one for the Muzo samples. In
the “Chivor” samples, the central core formed first,
Figure 11. Schematic cross-section of an exceptional
followed by beryl growth on its prism faces. Simulta- trapiche emerald examined by Van der Giessen
neously, a dendritic growth of both beryl and albite (1994). The emerald is characterized by two hexago-
under eutectic conditions occurred at the corners of nal cores surrounded by eight trapezoidal sectors.
the central core. The presence of beryl-albite can be
also related to an abrupt increase in the growth rate of
the arms. In fact, this favors the trapping of the foreign
phase (albite in this case) at the interface between the
growing “arms.” According to the authors, all Chivor
samples are grown “in the same time, in the same or
similar environments, in a relatively small region.”
To explain the formation of Muzo samples, they sug-
gested instead an abrupt decrease in the growth rate
of the arms and/or an increased concentration of car-
bonaceous material in the growth medium.
In 1971, O’Donoghue synthesized the work of
Nassau and Jackson (1970) and remarked that the
whitish gray or dark inclusions are not randomly dis-
tributed but follow the “hexagonal axis,” i.e., the
<100> directions in the hexagonal system (a-axis).

BOX B: CRYSTAL NUCLEATION, GROWTH, AND TEXTURE
Sunagawa (1987, 1999) proposed that crystal morphology such a system of albite and beryl has not been investigated
depends on the degree of supersaturation (also called un- up to now. In the same way, Sunagawa et al. (1999) also
dercooling) that conditioned the driving force and ex- considered the dendritic morphology and mineral compo-
plained the evolution of crystal habit by changes in sition (corundum and carbonate) of trapiche ruby that was
growth mechanism. Figure B-1 presents the driving force formed under eutectic growth conditions.
versus the growth rate anisotropy. The driving force is Despite the fact that Colombian trapiche emerald is
characterized by two critical points (X and Y) where the related to basinal fluid circulation, the association of al-
predominant growth mechanism changes: spiral growth bite-beryl in a H2O-NaCl ±(CO2) fluid seems to be simi-
for a low driving force, two-dimensional nucleation lar to paragenesis in low-temperature H2O-saturated
growth (2DNG) for an intermediate driving force, and an pegmatites. In such a magmatic system, supersaturation
adhesive growth mechanism for a high driving force. The is the driving force of the crystallization, and so the rate
expected morphologies change from polyhedral to hopper of nucleation and growth is driven by supersaturation in
or skeletal and then to fractal, spherulitic, and dendritic. the solution. In a binary diagram of temperature and
Crystals demonstrating rapid growth textures are composition, the displacement of the liquidus (the tem-
found in many rock types and geological environments: perature above which a magma is molten) may occur in
pillow basalts (Bryan, 1972), granites and pegmatites response to changes in pressure, temperature, or compo-
(London, 2008), magmatic olivine (Faure et al., 2003), sition (London, 2008). A decrease in pressure such as a
chiastolites and garnets in metamorphic environments drop in the H2O fluid pressure in an H2O-saturated melt
(Burton, 1986; Rice, 1993), metasomatic tourmalines will provoke a decompression, providing the driving
(Byerly and Palmer, 1991), and finally in trapiche emer- force of crystallization with dendritic textures (London,
alds and rubies (Sunagawa et al., 1999). 2008). In the case of Colombian emerald, the decompres-
As stated by Nassau and Jackson (1970) for Colombian sion phenomena by a drop of H2O (and CO2) fluid pressure
emerald, trapiche starts to grow as a beryl in the core, and at the tip of the faults induced a huge supersaturation that
then as coeval albite and beryl in the dendrites and provided the driving force to initiate crystallization of the
stringers in the arms. Based on this mineralogical evidence dendrites of trapiche emerald composed of albite and
with the simultaneous growth of both albite and beryl, the emerald. The predominance of albite over emerald in the
growth of trapiche emerald could be interpreted as a eu- dendrites indicates a great undercooling that promoted
tectic binary system (Nassau and Jackson, 1970), though the crystallization first of albite and then emerald.
Figure B-1. This diagram shows

INTERFACE
the three types of crystal growth
SMOOTH ROUGH mechanisms (Sunagawa, 1999):
spiral growth (blue curve), two-
dimensional nucleation growth
(2DNG; red curve), and adhe-
Dislocation Nucleation sive growth (green curve). The
GROWTH RATE
nature of the interfaces is classi-

fied as either rough or smooth.
The growth rates are deter-
mined by the state of the inter-
faces, which are classified as
Skeletal
Polyhedral either rough or smooth. The in-
creasing intensity of the driving
Hopper Dendritic Spherulitic Fractal force is presented between the
critical points X and Y, where
Spiral growth 2DNG Adhesive growth
nucleation occurred. A rough
interface is expected above Y
X Y DRIVING FORCE and smooth interface below X,
where dislocation occurred.
rapid and skeletal growth, complicated by interactions Sunagawa et al. (1999) compared the formation of
between the hydrothermal fluids and the black shales. trapiche emeralds to trapiche rubies in light of crystal

growth mechanisms, presenting evidence of many ple to another, but also within different portions of
parallels (box B). Two main differences were pointed the same sample (from pale to deep green). The sam-
out: the typical absence of the core in rubies, and the ples were prepared either as polished sections for elec-
dissimilar geological settings. However, the forma- tron microprobe analysis (EPMA), scanning electron
tion mechanism seems to be the same for both min- microscopy (SEM), and cathodoluminescence analy-
erals: if a core is present, it is formed at first by sis, or cut and polished on both sides to a thickness
layer-by-layer growth under small driving force con- of 150–200 µm for microscopic examination. The
ditions. When the driving force increases, a dendritic main source and features of the crystals, and the dif-
multiphase growth occurs around the core. Then, a ferent analytic methods applied for each trapiche
decrease of the driving force leads to layer-by-layer emerald, are reported in table 1.
growth and the formation of the ruby or emerald sec- EPMA was performed at the University of Lor-
tors. The contemporaneousness of the two last stages raine in Nancy’s SCMEM laboratory on a fully auto-
is not ruled out by Sunagawa et al. (1999). mated Cameca SX100. The detection limits for trace
In the past decade and a half, the main features of elements in mass percentage were 715 ppm for Mn,
Colombian trapiche emeralds have been revisited in 500 for Na and Si, 300 for Mg and Al, 110 for K, 40
an exhaustive work by Hochleitner (2002); Giuliani et for Ti, and 100 for V, Cr, and Fe. Data reduction was
al. (2002) added that they are only found on the west- performed with the PAP program (Pouchou and Pi-
ern side of the Eastern Cordillera Basin as material choir, 1991). Chemical formulas of emerald were cal-
“disseminated in black shales or albitized black shales culated on the basis of 3 Be and 18 O apfu. BeO was
near emerald-bearing veins.” Garnier et al. (2002) determined by stoichiometry and H2O by the equa-
compared the geological setting of trapiche rubies tion derived from experimental data (Giuliani et al.,
from Mong Hsu in Myanmar with that of Colombian 1997): H2O (in wt.%) = (0.84958 × Na2O (in wt.%) +
trapiche emeralds; the results underscored the differ- 0.8373).
ences in deposit type, fluid circulation, and P-T con- Cathodoluminescence (CL) and back-scattered
ditions of formation. Hainschwang et al. (2007) and electron (BSE) images were recorded at the SCMEM
Schmetzer et al. (2011) compared Zambian trapiche laboratory using an FEG JEOL J7600F SEM with an ac-
tourmalines to Colombian trapiche emeralds. Their celerating voltage of 15 kV and a beam current of 10
studies suggested similar formation, depending on the to 20 nA for the X-ray analysis and 2 nA for the CL
growth rates of different sectors of the crystals. Both images. This SEM is equipped with a Gatan MonoCL2
minerals contain dark carbonaceous inclusions along system with three diodes measuring the intensity of
the a-axis, but the main visual difference is the tour- light in different wavelength regions, so the observed
maline’s three-sectored core due to the trigonal system colors correspond to those of the visible spectra. CL
of crystallization (space group R3m). images were also obtained under a Technosyn cold
CL device, where the electrons are generated by an
NEW SCIENTIFIC ADVANCES ON THE electric discharge between two electrodes under low
COLOMBIAN TRAPICHE EMERALDS gas pressure.
Material and Methods. The twelve trapiche emeralds Trapiche emeralds were investigated by X-ray dif-
examined in this study were provided in 1998 by fraction topography (XRDT) at the University of Bari,
Omar Bustos Santana, manager of Sociedad Esmer- Italy, to obtain the spatial distribution and full char-
alda Ltda. (figure 12; table 1). The samples came from acterization of the crystal defects in the whole sample
three Colombian mines: Coscuez (figures 12A–12C), volume (Lang, 1959). Topographic images were taken
Peñas Blancas (figure 12D), and Muzo (figures 12E– in transmission geometry using a conventional Lang
12L). These samples were stubby and small; their di- camera (Rigaku ME110ED) with monochromatic
ameters varied from approximately 4.0 mm to 1.2 cm, MoKα1 radiation (λ = 0.709 Å) and a micro-focus X-ray
generally with lengths of a few millimeters, with the tube. To minimize the X-ray absorption according to
exception of one sample approximately 1.1 cm long the Beer-Lambert law, the optimum kinematic diffrac-
(figure 12I). The habit of the trapiche emeralds could tion condition μt ≈ 1 (μ = linear absorption coefficient;
not be well defined owing to the presence of a whitish t = crystal thickness) must be followed. Diffraction
or dark material; nevertheless, the pinacoidal {0001} contrast was recorded on Kodak SR photographic film.
–
faces and first-order prism {1010} faces were recogniz- The structural defects were characterized by applying
able in many of them. The color varied from one sam- the extinction criteria to their diffraction contrasts, ac-

A B C D
E F G H
I J K L
Figure 12. The Colombian trapiche emeralds studied in this work originate from the Coscuez (A–C), Peñas Blancas
(D), and Muzo (E–L) mines. The sample sizes are reported in table 1. Shown are samples T2 (A), T3 (B), T4 (C), T5
(D), T6 (E), T7 (F), T8 (G), T9 (H), T10 (I), T11 (J), T12 (K), and T13 (L).
cording to kinematic and dynamic X-ray diffraction lyze emeralds with satisfactory results and above all
theories (Authier and Zarka, 1994). without color modification.
X-ray computed tomography (CT) scanning is a
non-destructive technique used in this study to reveal RESULTS
3D interior details of trapiche emeralds. Images were In figure 13A, two sections of a trapiche emerald, per-
made with a Phoenix Nanotom S scanner, using a res- pendicular and parallel to the c-axis, are illustrated
olution of 3.4 μm/voxel and a nanofocus X-ray tube schematically: one can distinguish the central core,
tension value of 100 kV. Virtual cross-sections from the arms, the dendrites, and the overgrowth. It is
all axes were extracted from the volume to observe worth noting that the trapiche emeralds are more de-
the physical structure (e.g., inclusions and porosity) veloped perpendicular to the c-axis rather than along
and to detect the presence of phases with different the c-axis. For this reason, it is rare to observe a sec-
densities in the samples. The Nanotom scanner pro- tion parallel to the c-axis, such as the example shown
duces files with voxel (3D pixel) resolutions between in figure 12I. Such sections are usually not available
30.0 and 0.6 µm as a function of sample size. X-ray for scientific study, as trapiches are cut as cabochons
computed tomography has already been used for geo- that display their particular texture for jewelry.
science applications (Breeding et al., 2010; Tsuchiyama The spatial relationships between the core, arms,
et al., 2005; Cnudde and Boone, 2013; Jia et al., 2014) and dendrites are visible in three perpendicular sec-
but never previously for colored gems, because the tions obtained by X-ray computed tomography (fig-
X-ray exposure can strongly modify their colors. Here ure 14). Each aspect of the trapiche texture is
the technique was applied for the first time to ana- described in detail below.

TABLE 1. Features and properties of the 12 trapiche emeralds investigated in this study.
Locality Sample Geometric pattern Diameter Section Trapiche Composition (wt.%) Solid Microscopy SEM EPMA XRDT CT CL
(mm) c-axis matrix inclusions
(V2O3) (Cr2O3)
Coscuez T2 Hexagonal crystal 8 ‖ None c: 0.13–0.15 0.00–0.04 Ab, Qtz, Brl, Ank, 3 3 3 3 3
with core, arms, Dol
dendrites and d: 0.16–0.20 0.02–0.03
small overgrowth a: 0.22–0.62 0.02–0.03
og: 0.62–0.90 0.05–0.90
Coscuez T3 Center, arms 6 ‖ AbBS Ab, Dol, Brl, Cal, 3 3

Qtz, Fap, Fe(O,OH),
Ank, Py
Coscuez T4 Small core, arms, Arm up to ‖ AbBS c: 0.12–0.14 0.12–0.14 Ab, Cal, Brl, Tur, Rt, 3 3 3 3
dendrites 10 mm Mca, Fap, Po, Zrn,
d: 0.20–0.30 0.18–0.19 Dol, Kln
a: 0.20–0.30 0.15–0.20
Peñas T5 Hexagonal green 4 ‖ AbBS c: 0.68–0.72 0.30–0.40 Kln, Fe(O,OH), Ab, 3 3 3 3

Blancas core with green + clays Rt, Brl, Zrn, Po, Cal,
arms, dendrites d: 1.17 0.68 Dol, Fap, Mca
a: 0.90–1.16 0.0–0.90
Muzo T6 Center, Arm up to ‖ None 3 3

asymmetric arms 3.5 mm
Muzo T7 Small core, arms, 10 ‖ None Mca, Kln, Tur, Dol, 3 3

small overgrowth Cal, Ank, Rt,
Fe(O,OH), Qtz, Fl,
Fap, Ab, Py
Muzo T8 Hexagonal crystal 10 ‖ None c: 0.02–0.03 0.00–0.02 Ab, Qtz, Mca, Rt, 3 3 3 3 3
with hexagonal Clays in Dol, Brl
core and deep fractures a: 0.02–0.06 0.00–0.06
green overgrowth og: 0.62–0.64 0.31–0.72
Muzo T9 Hexagonal green 5 ‖ AbBS Brl, Mnz, Qtz, Pt, 3 3 3

core and green + clays Fap, Ab, Mca, Rt,
arms Fe(O,OH), Po
Muzo T10 Bipyramidal core, 10 ⊥ None c: 0.15–0.26 0.02–0.12 Ab, Kln, Fap 3 3 3
arms
a: 0.15–0.23 0.02–0.09
Muzo T11 Large green core, 8, with ‖ AbBS c: 0.59–0.75 0.49–0.62 Brl, Qtz, Rt, Ab, 3 3 3 3
small arms core ~3 mm + clays Kln, Dol, Py
d: 0.02–1.12 0.02–0.99
a: 0.91–1.09 0.73–0.89
Muzo T12 Small hexagonal 12 ‖ AbBS

core, arms
Muzo T13 Core, arms 5 ‖ Ab
c = core; d = dendrite; a = arm; og = overgrowth; AbBS = albitized black shale, Ab = albite; Cal = calcite; Qtz = quartz; Brl = beryl (emerald); Ank = ankerite;
Dol = dolomite; Fap = F-apatite; Py = pyrite; Fe(O,OH) = iron oxides and hydroxides; Tur = tourmaline; Rt = rutile; Mca = muscovite; Po = pyrrhotite;
Zrn = zircon; Kln = kaolinite; Fl = fluorite; Mnz = monazite; ‖ = section parallel to crystal’s c-axis; ⊥ = section perpendicular to crystal’s c-axis.

c
A overgrowth A B
a3
_
(1010) arm
dendrites
core
_ 00)
arm
( 11
)
10
core
arm
(0 1 _
arm
a rm
arm arm
core
a rm a2
arm
es
(01 _
_ 0)
C D rit
0
d
10
(11
arm core arm en

)
arm d
_ core
(1010) core
Dendrites of
a1
whitish or dark material
1 mm
B
core Figure 14. X-ray computed tomography images show-
ing the spatial relationships between the core, arms,
and dendrites in a trapiche emerald from Muzo.
arm arm
pyramid’s vertex the pyramid’s vertex (figure 13B). This is confirmed

by X-ray computed tomography: Figure 15A shows
that the size of the core decreases from the top to the
Figure 15. X-ray computed tomography images of a

trapiche emerald from Muzo. A: The size of the core
decreases from the top toward the bottom of the crys-
Figure 13. A: Schematic diagrams of trapiche emerald tal. B: The dendrites around the core develop laterally
sections, perpendicular to the c-axis (top left) and par- and penetrate into the arms. C: The spatial distribu-
allel to it (top right). The sections show trapiche emer- tion and morphology of solid inclusions in the sam-
ald’s central core, arms, dendrites, and overgrowth ple. D: The elevated porosity (in blue) of the whole
zones. B: Aspect and variations of the core, which has sample in a section perpendicular to the c-axis.
a hexagonal bipyramidal shape. Its size varies as a
function of the sections perpendicular to the c-axis:
largest if the section is cut at the base of the pyramid,
smallest near the pyramid’s vertex.
Core. The central core may or may not be present in

a trapiche emerald. The core is bounded by pina-
coidal faces [0001] (see figures 12A, 12D, 12G, 12J,
and 12L) and has the shape of two opposite hexago-
nal pyramids (figures 12I and 13) corresponding to
the pinacoidal growth sectors. The presence and the
A B
size of the core depend on the position of the cross- 1 mm 1 mm
section along the bipyramidal shape, as illustrated

in figure 13B. If the trapiche’s growth is homoge-
neous in both positive and negative directions of the
c-axis, the two hexagonal pyramids will develop
similarly. If the growth is inhomogeneous (due to
limited space in the growth medium, for example),
one pyramid will be more developed than the other,
as in figure 12I. Moreover, in the sections perpendi-
cular to the c-axis, the core is largest if the section C D 1 mm
is cut at the base of the pyramid, and smallest near 1 mm

Figure 16. Core images
of several trapiche
emeralds. Samples
showing the upper part
A B of the cabochon cut
1.75 mm 1.75 mm
with the hexagonal
arm
core (A), and with only
de a central point (B); opti-
nd cal (C) and SEM (D)
rit
core
arm es images of a sample
core
de with no core, the den-
nd
rit drites and arms cross-
es
ing in the central point;
an example of a hexag-
onal dark core (E); signs
of fracturing and disso-
C 500 μm
D 100 μm
lution in a core, attest-
ing to a complex
growth history (F).
E 2 mm F 1 mm
bottom of the trapiche. When a section is cut be- as demonstrated by the density contrast in the tomo-
tween the vertices of two opposite pyramids, the core graphic images (figures 14 and 15B).
is absent and the arms and dendrites intersect in a Optical observations of the sections perpendicular
central point, as in figures 12B, 12E, 16C, and 16D. to the c-axis under cross-polarized light confirm that
This explains why some trapiche samples show a the core cannot be completely extinct, but a kind of
hexagonal core on one end (figure 16A) but no core wavy extinction, as in quartz, is made visible by
on the other end (figure 16B). The color of the core turning the stage of the microscope. It follows that
can vary from very light to deep green, reflecting dif- the core may be plastically deformed during or after
ferent quantities of the chromophores V3+ and Cr3+ its formation. This particular optical feature was
(see the “Trace-Element Analysis” section below). mentioned by Bernauer (1933), who wrote of “listas
The core is surrounded by dendrites that can also onduladas,” and Chaudhari (1969), who described “a
replace it completely or partially, as described by sort of grating structure” in the core.
Bernauer (1933). The replacement gives the core a dark Signs of dissolution are observed in the core of
aspect, as shown in figure 16E. The dendrites around two samples. For example, figures 17A and 17B
the core develop laterally and penetrate into the arms, show that the core’s edges are not straight, due to

A B
•A
B
core
Brl
core
de
dendrite nd
rit
e
Cal
arm
arm
400 μm 600 μm
Figure 17. A: An SEM image of the spatial relationship between the core, dendrites, and arms in the sample from
figure 12C. The inset shows the whole sample, along with the locations of images A and B. The wavy contact be-
tween the core and the dendrites indicates a previous dissolution of the core by the fluids. The dendrites are
formed mainly by calcite and albite sometimes associated with beryl. B: A microscopic image of the dendrites and
the arms of emerald (Brl), which formed simultaneously, though the arms were affected by dissolution and frac-
turing (see the aspect of the core and arms). The microscopic image shows that the fracture crosscutting the arm is
filled by calcite (Cal) and ± albite associated with tiny beryl crystals (around ten microns).
etching, and are partially covered by the arms. In quartz, carbonates, fluorite, and phyllosilicates.
figure 16F, dissolution also occurred during the Some of the fractures are filled by fluid inclusions,
core’s complex growth history. During its forma- indicating fluid circulation and consequent dissolu-
tion, the core underwent a deformational event tion of the edges and corners of the core. The effects
marked by two main fractures coeval with the den- of dissolution are marked by the modification of the
drites’ development. The fractures are filled by the core’s shape from hexagonal to oval, and by the ir-
same minerals found in the dendrites: albite, pyrite, regular etched borders. During a second stage, core
A B
Figure 18. Optical im-

ages of growth arms in
trapiche emeralds. A
and B: The arms are
characterized by bun-
dles of straight disloca-
tions perpendicular to
–
the {101 0} faces. C: Di-
500 μm 500 μm rection of propagation of
the dislocations in two
C D adjacent growth sectors.
Along these dislocation
zones and at the contact
of the dendrites, the
crystals trap solids and
cavities of fluid inclu-
sions. D: Primary fluid
inclusion trails are par-
allel to the dislocations.
100 μm 100 μm

A B Figure 19. These multiphase
fluid inclusions trapped in
the arms of trapiche emer-
Cb V
alds contain liquid (L) and
vapor (V) phases as well as
solid-phase daughter miner-
Syl L
H als (H = halite, Cb = carbon-
ates, Syl = sylvite). A: This
primary multiphase inclu-
sion exhibits stretching and
necking-down phenomena.
10 μm 100 μm
B: Alignment of multiphase
C L D fluid inclusions along the
dislocation plane of the
V arm. C: A multiphase fluid
H
inclusion trapped at the end
of a dendrite (on the right)
Cb and parallel to the disloca-
tions. D: A large cube of
H
halite in a fluid inclusion
cavity formed at the edge of
the dendrites. E: Trails of
20 μm
multiphase fluid inclusions
20 μm
affected by necking-down.
E F Re-equilibration of the cavi-
ties occurred during cooling,
V and some of the primary
Cb fluid inclusions became
monophase (L or V) or
L
biphase (V + L or L + H) cav-
ities. F: Typical primary
multiphase fluid inclusion
H in trapiche emerald, similar
to those trapped by non-
100 μm 20 μm trapiche emerald.
growth resumed and the pseudo-hexagonal shape In each arm, bundles of straight dislocations de-
–
was restored. velop perpendicular to the {1010} faces (figures 18A
Rare multiphase fluid inclusions are observed in and 18B). This indicates that they formed during the
the core. They are similar to those found in the arms, growth, in the direction of minimum elastic energy
indicating that the fluid composition was the same. (Authier and Zarka, 1994). The direction of propaga-
However, multiphase fluid inclusions in the core are tion of the dislocations in two adjacent growth sectors
pseudo-secondary because they are often associated can be observed in figure 18C. Only in the trapiche of
with fracturing. figure 12E do the dislocations appear slightly curved
on the border of the largest arm. This is probably due
Arms. Six arms with nearly identical size surround to a change in the direction of minimum energy dur-
the core. They appear elongated in the 〈110〉 directions
–
ing the “anomalous” growth of this arm.
–
and are bounded by the {1010} faces. Sometimes a few Solid and multiphase fluid inclusions are found in
arms are larger than others (figures 12E and 12F), mak- the arms. They are parallel to the dislocations (figures
ing the trapiche asymmetric. 18D, 19B, and 19E) and all located in proximity to the
The color of the arms, like that of the core, can dendrites. The solid inclusions are similar to those
vary from pale (figures 12E and 12F) to intense found in the dendrites (see “Dendrites” below). The
green (figures 12D, 12H, 12K, and 12L) as a function fluid inclusion cavities follow the extension of the
of chemical composition (see “Trace-Element dendrites in the arms. They are multiphase fluid in-
Analysis”). clusions containing daughter minerals, mainly halite

A B
Figure 20. Elongated
voids observed in the
arms, perpendicular to
the growth front and par-
allel to the dislocations.
A: An elongated fluid in-
clusion cavity that
formed parallel to the dis-
location plane and after
subsequent trapping of a
100 μm 100 μm solid inclusion (the black
C D solid on the upper part of
the cavity). B: Trails of
voids parallel to the dislo-
cation planes. C: Mono-
phase fluid inclusions
with highly irregular den-
dritic morphologies in the
arms. D: Flat monophase
fluid inclusions that de-
veloped along the disloca-
100 μm
100 μm
tion planes.
but also carbonates and sylvite, liquid and vapor presence of solid inclusions. The voids resemble those
phases (figures 19A, 19C, and 19F). They belong to the described in Colombian emerald-bearing veins
H2O-NaCl ± (CO2) system already characterized for (Touray and Poirot, 1968) and in trapiche tourmaline
Colombian emerald-bearing veins (Roedder, 1963; (Schmetzer et al., 2011). Nevertheless, these voids
Touray and Poirot, 1968; Kozlowski et al., 1988; Ott- may also have formed by dissolution along disloca-
away, 1991; Giuliani et al., 1991; Cheilletz et al., tion bundles if the supersaturation decreased after
1994). Sometimes, the multiphase fluid inclusions can crystal growth (Scandale and Zarka, 1982; Authier
be concentrated in certain portions of the arms, form- and Zarka, 1994).
ing trails (figure 19E). The arms are sometimes affected by fractures,
Evidence of heterogeneous trapping is observed in filled by albite and small euhedral beryl crystals (fig-
some multiphase fluid inclusions at the contact be- ure 17B), that are synchronous with the arms’ forma-
tween dendrites and arms. This trapping is character- tion. The growth of the arms is not perturbed,
ized by the presence of a large cubic crystal of halite because there is no discontinuity between the albite
(figure 19D) occupying more than 15 vol.% of the cav- in the fractures and the albite in the dendrites on
ity and up to 70 vol.% (see “Emerald Mineralization” both sides of the arms.
above). In some emeralds, the arms have a fibrous texture
Monophase fluid inclusions with highly irregular like that of beryl in the dendrites (figure 21A), but in
dendritic morphologies (figures 20C and 20D) are pres- others they are not fibrous and appear similar to the
ent in the arms. They do not contain solid or vapor core. Their different textures can reveal important
phases. Similar inclusions were described in quartz information about the growth rates of the arms.
crystals by Invernizzi et al. (1998) and interpreted as
earlier-formed inclusions re-equilibrated at conditions Dendrites. In trapiche emeralds, dendrites surround
of high internal underpressure. The dendritic mor- the core (figures 21B and 21F) and develop along the
phology “results from the closure of the original in- a-axis from the corners toward the outer edge with a
clusion void and preservation of fluid in long dendritic herringbone texture (figure 21C). In particular, the
dissolution channels” (Invernizzi et al., 1998). dendrites may be larger near the edge, where they as-
The arms contain elongated voids perpendicular sume a characteristic fan shape (figures 21B and 21E).
to the growth front and thus parallel to the disloca- In some samples, the herringbone texture is not de-
tions (figures 20A and 20B). Figure 20A shows that veloped and the fan shape starts directly on the cor-
the formation of the elongated voids is related to the ners of the core. In this case, the dendrites are larger

Figure 21. A: An SEM
image of dendrites con-
taining stringers of
core beryl intersecting at 60°
core angles. B: The dendrites
A B
100 μm 500 μm are enlarged near the
edge of the crystal,
C where they assume a
characteristic chevron
shape (see also images
core
C–E). C: An SEM image
of the typical herring-
bone texture of den-
drites. The crystal
formed on an albitized
200 μm
black shale matrix (on
the left side of the
image). D: This SEM
image shows the pres-
ence of dendrites in the
arms. E: Aspect of the
dendrites from the core
to the edge of the emer-
ald crystal in a special
section where the core
is represented by a cen-
D 200 μm
tral point. The arms in-
clude albitized black
shale (right side of the
image), part of the ma-
trix that supports the
emerald. F: An SEM
image of dendrites sur-
rounding the core.
core
E 500 μm F 200 μm
and contain stringers of emerald intersecting at 60° albitites as well as in the dendrites (figure 22D), explain-
angles (figure 21A), as described by Chaudhari (1969). ing why they often appear dark (figures 12B, 12C, and
The dendrites may be present even in the arms (fig- 12K). The X-ray computed 3D tomography image (fig-
ures 12D and 12F) and parallel to the extension of the ure 15D) of the trapiche emerald from figure 12H per-
arms (figure 21D). mitted the calculation of the dendrites’ volume. The
The dendrites are developed on albitized and calci- sample’s total volume is 64.75 mm3, of which 18% is
tized black shales (figures 22A, 22B, 22D, and 22E), and occupied by the dendrites and 81% by the core and the
sometimes rounded remnants of albitite are included arms, with the remaining 1% representing the porosity.
in the calcite-bearing dendrites (figure 22C). In addition, The dendrites contain mainly the albite and
variable quantities of organic matter are found in the emerald assemblage, which explains why the den-

A B Figure 22. A: Microscopic
Ca Brl image of dendrites formed
Po
Brl on albitized (± calcitized)
Ab
black shales (BS). The
BS
chevron-like habits charac-
teristic of trapiche contain
albite (Ab), emerald (Brl),
Ca and calcite (Cal). B: SEM
Zrn image of the albitized black
Ab Py shale with albite, zircon
Cal
(Zrn), calcite, pyrite (Py),
Brl 500 μm 50 μm and pyrrhotite (Po). C: SEM
image of sample T4 showing
C Abt dendrite arm dendrite
the arm formed by emerald
Cal Ab Brl with dendrites on both
sides. The dendrites are
formed by calcite ± albite
Abt Brl 500 μm
and emerald. The arm and
the dendrites formed on al-
D E bitite (Abt), a rounded rem-
Rt nant of which is included in
Ab the dendrite on the left side.
Po Om
Ab D: SEM image of the albitite
containing zircon, emerald,
Brl pyrrhotite with framboidal
Om
polycrystalline aggregates,
Zrn rutile (Rt), and organic mat-
ter (Om). E: Morphology
and size of albite crystals
25 μm
from the albitite, showing
40 μm
the rock’s high porosity.
drite zones were also called “two-phase regions” by Overgrowth. Overgrowth is not always present in
Nassau and Jackson (1970). The high quantity of al- trapiche emeralds. Its color is different from that of
bite (with or without calcite) makes the dendrites the adjacent arms, and it can be paler (figure 12F) or
whitish, as shown in figures 12D and 12J. Quartz and darker (figures 12A and 12G). The color difference in-
dolomite complete the main paragenesis (figures 23A dicates variations of V2O3 and Cr2O3 contents in the
and 23B). Other solid inclusions characteristic of growth medium during the formation of the over-
Colombian emeralds are found in minor amounts growth (see “Trace-Element Analysis”). If many
(figure 23C): rutile, carbonates (dolomite, ankerite, chemical changes occurred, the overgrowth becomes
siderite), muscovite, fluorapatite, pyrite, pyrrhotite, zoned, as in the trapiche shown in figure 12G. The
zircon, fluorite, tourmaline, and minerals containing contact between the overgrowth and the arms is
rare-earth elements such as monazite and parisite. often marked by the presence of inclusions.
Muscovite and rutile are V-bearing, and tourmaline X-ray Diffraction Topography. Even though X-ray dif-
usually shows complex zoning (figure 23D). The fraction topography is particularly suitable to inves-
pyrrhotite forms framboidal polycrystalline aggre- tigate the strain associated with extended defects and
gates typical of sedimentary sulfides. to reconstruct the growth history of minerals (Agrosì
Secondary alteration can affect the dendrites. The et al., 2006; Agrosì et al., 2013), it has not been used
formation of clay minerals (mostly kaolin) and iron until now to study trapiche emerald. This is probably
oxides/hydroxides modifies their original dark/whitish because the technique requires the cutting of thin
color to a brownish one (figures 12H and 12L). The slices to minimize X-ray absorption, a drawback
typical evidence of alteration is the pseudomorphic given the rarity of trapiche material. X-ray topo-
replacement of framboidal sulfides by iron oxides/ graphic images of trapiche emeralds are shown in fig-
hydroxides. The alteration is favored by the elevated ure 24. The samples’ limited thickness kept us from
porosity of the dendrites, which is visible in the tomo- preparing slices with different orientations, and only
graphic images of figure 15D. slices perpendicular to the c-axis were cut.

A B Figure 23. SEM images of in-
Fl cavity
Brl Brl clusions found in dendrites
Mca Qtz at the contact with trapiche
emerald arms. A: Aspect of
Mca
the cavities filled by solids
Qtz in the dendrites. The assem-
blages are formed by rutile
Qtz (Rt) + quartz (Qtz), and
Rt Brl quartz + emerald (Brl) +
Qtz Ab
Dol
muscovite (Mca). B: Other
mineral assemblages
formed by dolomite (Dol) +
Ab
Ab albite (Al) + quartz + mus-
100 μm Mca 50 μm covite (Mca). C: Aspect of a
dendrite cavity filled by al-
C D bite, calcite (Cal), ankerite
Rt (Ank), muscovite, quartz,
Brl Cal
pyrrhotite (Po), fluorapatite
Fap (Fap), and rutile. The differ-
Ank ent solids are coeval, and
Ab the limits of some minerals
Ab show the presence of small
Cal
fluid inclusion cavities, in-
Ab
dicating that a fluid was
Po bathing the whole cavity
Qtz
Tur during dendrite formation.
D: An SEM image of a tour-
Mca
maline (Tur) crystal in-
10 μm
50 μm cluded in calcite.
Topographic images taken with diffraction vec- No diffraction contrast corresponding to the
tors g of type hki0 show a strong diffraction contrast growth sector boundaries can be seen in the topo-
for the arms, while the core is almost invisible (fig- graphic images. Thus, the boundaries between the
ures 24A and 24C). In all arms, the presence of bun- adjacent arms can be recognized only by means of
dles of dislocations running perpendicular to the the lack of contrast corresponding to the dendrites or
prismatic growth fronts is confirmed by the topo- by the directions of propagation of the dislocations
graphic images (figure 24). It was impossible to char- in each arm (for example figures 24E, 24F, and 18C).
acterize these dislocations, because the orientation This indicates that in trapiche emeralds there is a
of the slices did not permit us to obtain topographic good interconnection between the different arms.
images with reflections suitable to determine the Topographic images taken with diffraction vectors
–
Burgers vectors applying the extinction criterion. inclined with respect to the c-axis (e.g., g = 1231) show
The extinction criterion can be explained only refer- that the hexagonal core is also visible (figures 24B and
ring to the diffraction contrast origin. The contrasts 24D). The fact that both core and arms are visible in
observed in the topographic images are due to the these topographic images indicates crystalline continu-
different intensities of diffraction between regions ity between them, despite the presence of the dendrites.
characterized by strain fields associated to the de- The presence of dislocations in the core cannot be
fects (kinematical effects) and almost perfect regions excluded, because the diffraction contrast observed
(dynamical effects). The diffraction contrasts can be (figure 24B) is not solely due to the fractures (slightly
extinct if the scalar product between the diffraction visible in figures 24A and 24C). It is worth noting that
vector g and the vector representative of the strain the contrast of these dislocations is always extinct on
field associated to the defect is equal to zero. In the the topographic images with g = hki0. On the basis of
case of dislocations, the extinction criterion is based the extinction criterion, the b vector should thus be
on the scalar product g × b = 0, where b represents parallel to the c-axis, and one may suppose that the
the Burgers vector. It follows that the dislocations dislocations in the core are of screw type. This could
are out of contrast when g is perpendicular to b. be verified by a topographic study of the sections par-

– –
g = 1010 g = 1231
A 3 mm
B
Figure 24. A to D: MoKα1
X-ray topographic im-
ages of the trapiche
emerald from figure 12G,
taken with the diffrac-
–
tion vectors g = 1010 (A),
– –
g = 1231 (B), g = 0110 (C)
– –
and g = 1321(D). E to F:
X-ray topographic im-
– – – ages of the trapiche
g = 0110 g = 13 21
emerald from figure 13E,
taken with the diffrac-
–
tion vectors g =1010 (E)
C D ––
and g = 3121 (F).
–
g = 10 10 ––
g = 31 21
E 500 μm F
allel to the c-axis, but unfortunately it was not possi- show variations in the reddish hue. Cold CL images
ble to cut sections parallel and perpendicular to the obtained on the Coscuez crystal indicate that the
c-axis from the same sample. For this reason, the pres- core and arms have the same CL colors, but the den-
ence of screw dislocations has not been confirmed. drites formed by a fine intergrowth of albite, calcite,
and emerald (figure 25D) have different CL colors for
Cathodoluminescence. CL images of the trapiche each: orange, yellow, and dark crimson for albite, cal-
emerald samples exhibited several distinct features. cite, and emerald, respectively. These CL colors
The cold CL images obtained on the Peñas Blancas could be explained by the presence of Mn in calcite
crystal (see figure 12D) show a crimson color due to (Ohnenstetter et al., 1998) and of Fe substituting Al
the presence of Cr3+ and V3+ ions acting as the main in albite (Geake and Walker, 1975; White et al., 1986;
CL activators (figure 25C). The cold CL image of the Götze et al., 2000).
Muzo crystal (figure 25D) illustrates the color homo-
geneity of the core and the different textural and Trace-Element Analysis. Six sections of trapiche
color aspect of the arms. These fibrous dendrites emerald from Coscuez, Peñas Blancas, and Muzo

TABLE 2. Representative chemical composition of trapiche emeralds from Coscuez, Muzo, and Peñas Blancas,
obtained by electron microprobe analysis.
Coscuez Muzo Peñas Blancas
Oxides (wt.%) T4-6 T4-1 T4-14 T11-1 T11-32 T11-23 T8-2 T8-11 T8-4 T8-13 T5-1 T5-10 T5-11
a d c a d c og og a c a d c
SiO2 66.28 65.16 66.81 65.22 66.52 65.87 65.97 66.90 65.57 66.17 65.55 66.26 65.25
Al2O3 17.26 17.27 17.61 15.19 17.18 16.77 16.62 17.69 17.56 17.80 16.28 15.22 15.77
Cr2O3 0.17 0.18 0.12 0.80 0.95 0.62 0.29 0.72 0.05 0.02 0.90 0.68 0.30
Fe2O3 0.20 0.12 0.08 0.37 0.08 0.12 0.07 0.04 0.03 0.03 0.19 0.56 0.16
V2O3 0.27 0.27 0.14 1.08 0.82 0.67 0.64 0.30 0.05 0.03 0.87 1.17 0.68
BeO calc.a 13.79 13.56 13.90 13.57 13.84 13.71 13.73 13.92 13.64 13.77 13.64 13.79 13.58
MgO 0.58 0.53 0.55 1.28 0.33 0.65 0.99 0.29 0.73 0.69 0.71 1.40 1.47
Na2O 0.56 0.50 0.45 0.98 0.27 0.49 0.97 0.25 0.45 0.43 0.61 0.59 0.76
H2O calc.b 1.31 1.26 1.22 1.67 1.07 1.31 1.66 1.04 1.22 1.20 1.35 1.34 1.48
Total 100.42 98.85 100.88 100.16 101.06 100.21 100.94 101.15 99.30 100.14 100.10 101.01 99.45
Si 6.000 6.000 6.000 6.000 6.000 6.000 6.000 6.000 6.000 6.000 6.000 6.000 6.000
Al 1.841 1.874 1.864 1.647 1.826 1.800 1.782 1.870 1.894 1.902 1.756 1.624 1.709
Cr 0.012 0.013 0.009 0.058 0.068 0.045 0.021 0.051 0.004 0.001 0.065 0.049 0.022
Fe3+ 0.014 0.008 0.005 0.026 0.005 0.008 0.005 0.003 0.002 0.002 0.013 0.038 0.007
V3+ 0.020 0.020 0.010 0.080 0.059 0.049 0.047 0.022 0.004 0.002 0.064 0.085 0.050
Be2+ b 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000
Mg 0.078 0.073 0.074 0.176 0.044 0.088 0.134 0.039 0.100 0.093 0.097 0.092 0.041
Na 0.098 0.089 0.078 0.175 0.047 0.087 0.171 0.043 0.080 0.076 0.108 0.104 0.135
Total 11.063 11.078 11.040 11.161 11.050 11.077 11.159 11.027 11.083 11.077 11.103 11.089 11.125
V/Cr 1.58 1.49 1.16 1.34 0.86 1.07 2.19 0.41 0.99 1.49 0.96 1.71 2.25
a
Amount of Be inferred from stoichiometry zones of the trapiche emerald: a = arm; d = dendrite; c = core; og = overgrowth.
b
Calculated following the equation H2O = (0.84958)(Na2O) + 0.8373 (Giuliani et al., 1997).
were analyzed by EPMA and with elemental compo- dendrites, arms, and overgrowths (tables 1–3). Over-
sitional EDX maps in order to evaluate chemical all, the emeralds contained relatively low concentra-
composition, zoning, and color variations in the core, tions of Fe2O3 , from 200 to 5700 ppm. The Na2O
A C
0.5 cm Figure 25. A and B: SEM and
CL images of a Muzo sam-
dendrite ple show that under CL, the
core is homogenous and ap-
pears crimson-colored due
core
to the presence of Cr3+ and
arm V3+ ions acting as the main
125 μm CL activators. The arms, on
the other hand, are fibrous
and show variations in the
reddish hue. C: The cold CL
B D image of a Peñas Blancas
core
sample shows the homo-
geneity of its red color due
to the presence of the V and
Cr chromophores. D: The
dendrite
cold CL image of a Coscuez
arm
sample reveals different CL
colors: orange for albite, yel-
low for calcite, and dark
arm crimson for emerald.
0.5 cm

TABLE 3. Representative chemical composition of trapiche emeralds from Coscuez and Muzoa, obtained by
electron microprobe analysis.
Coscuez Muzo
Oxides (wt.%) T11-25 T11-20 T11-33 T11-28 T11-1 T11-4 T10-1 T10-8 T10-12 T10-7 T10-5 T10-3 T10-16
c c d d a a center cS cN aS aS aS aN
SiO2 66.36 65.37 66.07 65.59 65.22 65.34 66.50 65.50 66.21 66.38 66.57 65.71 66.29
Al2O3 16.49 16.15 15.87 16.32 15.19 16.15 18.20 16.20 17.26 17.68 17.37 17.41 17.93
Cr2O3 0.60 0.49 0.87 0.63 0.80 0.73 0.09 0.12 0.05 0.05 0.11 0.03 0.03
Fe2O3 0.10 0.11 0.19 0.09 0.37 0.21 0.08 0.51 0.10 0.13 0.23 0.26 0.20
V2O3 0.72 0.59 0.89 0.58 1.08 0.87 0.15 0.26 0.13 0.20 0.21 0.21 0.20
BeO calc.b 13.81 13.60 13.75 13.65 13.57 13.60 13.84 13.92 13.78 13.81 13.85 13.67 13.79
MgO 0.75 0.98 1.21 0.67 1.28 0.86 0.67 1.48 0.90 0.57 0.87 0.82 0.67
Na2O 0.66 0.78 0.86 0.55 0.98 0.64 0.56 1.13 0.63 0.39 0.69 0.67 0.56
H2O calc.c 1.39 1.49 1.56 1.30 1.67 1.38 1.31 1.79 1.87 1.10 1.42 1.40 1.31
Total 100.88 99.56 101.27 99.38 100.16 99.78 101.40 100.91 100.93 100.31 101.32 100.18 100.98
Si 6.000 6.000 6.000 6.000 6.000 6.000 6.000 6.000 6.000 6.000 6.000 6.000 6.000
Al 1.757 1.747 1.699 1.760 1.647 1.748 1.935 1.749 1.843 1.883 1.845 1.874 1.913
Cr 0.043 0.036 0.062 0.046 0.058 0.053 0.006 0.009 0.004 0.004 0.008 0.002 0.014
Fe3+ 0.007 0.008 0.013 0.006 0.026 0.015 0.005 0.035 0.007 0.009 0.016 0.018 0.014
V3+ 0.052 0.043 0.065 0.043 0.080 0.064 0.011 0.019 0.009 0.014 0.015 0.015 0.015
Be2+ b 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000 3.000
Mg 0.101 0.134 0.164 0.091 0.176 0.118 0.090 0.202 0.122 0.077 0.117 0.112 0.090
Na 0.116 0.139 0.151 0.098 0.175 0.114 0.098 0.201 0.111 0.068 0.121 0.119 0.098
Total 11.076 11.107 11.154 11.043 11.161 11.111 11.146 11.215 11.096 11.056 11.121 11.139 11.144
V/Cr 1.19 1.20 1.02 0.91 1.34 1.18 1.66 2.15 2.58 3.97 1.90 6.95 6.62
a
Analyses on Coscuez samples were performed on a section perpendicular to the c-axis. Analyses on Muzo samples were conducted on a section
parallel to the c-axis.
b
Amount of Be inferred from stoichiometry zones: a = arm; c = core; d = dendrite; S = southern part of the crystal; N = northern part of the crystal.
c
Calculated following the equation H2O = (0.84958)(Na2O) + 0.8373 (Giuliani et al., 1997).
concentrations were between 1400 and 15,500 ppm, tact zone between the arm and the overgrowth was
which corresponds to a calculated H2O in the chan- sharp and distinguished by a Cr-rich zone (figure
nels between 0.95 and 2.15 wt.%, following the equa- 26F). The Al anomalies detected at the border and
tion proposed by Giuliani et al. (1997). In terms of within the fractures of the crystal were due to
chromophores, V was the main element; the highest argillites that formed during a late meteoric water
concentrations in V2O3 and Cr2O3 were 11,700 and stage.
9,700 ppm, respectively. The V/Cr ratio was between The distribution of the different chemical ele-
0.42 and 6.9 (tables 2 and 3). Some of the emeralds ments in the various maps were coupled to the chem-
were Cr-free, with V2O3 content up to 1.17 wt.% ical distribution of Cr and V (in wt.%). EPMA
(sample in figure 12D; table 2). The Mg contents were analyses presented in table 2 show that the arms were
between 2,300 and 14,800 ppm. richer in Cr and V than the core. The contact zone be-
Chemical compositional EDX maps of trapiche tween the arms and the overgrowth was sharp and in-
from Muzo (sample T8 in figure 12G) were obtained dicated by a Cr-rich zone in the EDX image (figure
on a section perpendicular to the c-axis (figure 26A). 26F). The Cr2O3 and V2O3 contents calculated by EDX
This sample contained a colorless to very light green reached 0.72 and 0.64 wt.%, respectively.
core rimmed by dendrites and six colorless arms (fig- Sample T10 is a section parallel to the c-axis of a
ure 12G). The arms were rimmed by deep green over- V-rich greenish trapiche emerald from Muzo (figure
growth zones. All the chemical elements of emerald 27A). The chemical composition of the center, the
were distributed homogenously in the EDX-map of core, and the rims are reported in table 3. The spatial
the core and arms. The dendrites were characterized distribution of V and Cr in the longitudinal section
by Na anomalies that correspond to minute crystals showed a positive correlation between these two el-
of albite formed during their growth. The overgrowth ements for the center and the core (figure 27B). The
zones were richer in Mg and Na, with important Cr arms from both sides of the crystal had a restricted
and V anomalies, than the core and arms. The con- V2O3 range (table 3).

A B Figure 26. A: An SEM view
Na
of sample T8 from Muzo
(shown in figure 12G). B–F:
EDX elemental maps show
the distribution of sodium
(B), magnesium (C), alu-
minum (D), vanadium (E),
and chromium (F). Brighter
color corresponds to a
higher concentration of the
element being imaged.
3 mm Image B shows a very high
concentration of Na in the
C D dendrites where albite is
Mg Al
present. Image C shows a
slight increase of Mg in the
overgrowth zone. Image D
shows important concen-
trations of Al in the frac-
tures filled by clay
minerals during late mete-
oric fluid circulation. Im-
ages E and F show that
there is no chemical differ-
ence in the distribution of
V and Cr between the core
E F and the arms: Both areas
V are very poor in chro-
Cr
mophores (Cr and V <600
ppm), although the over-
growth zone is very rich in
V and Cr: This area has the
“green grass” color of Muzo
emeralds. The emerald
zone in contact with the
arms is Cr-rich (up to 7200
ppm), while the outer zone
is V-rich (up to 6400 ppm).
DISCUSSION tween them are filled by dendrite material is ex-

The formation of trapiche emerald was approached cluded by the presence of the dendrites around the
by a combination of previous and new experimental core (see, for example, figures 14 and 15). Moreover,
data on the trapiche texture with the constraints of X-ray topographic images show crystalline continu-
tectonics and basinal fluid circulation in a sequence ity between the arms and the core, indicating they
of black shales of the western emerald zone in are not several crystals but portions of the same
Colombia’s Eastern Cordillera. emerald. The trapiche emerald is thus a unique crys-
tal formation, as already stated by Nassau and Jack-
Growth History. While many hypotheses have been son (1970). Furthermore, the presence of multiphase
proposed, how and why trapiche emeralds formed fluid inclusions in the trapiche emeralds similar to
will be elucidated on the basis of new experimental those observed in non-trapiche Colombian emeralds
data presented in this work. These data further con- suggests that they formed from the same fluid in the
firm that trapiche emeralds are neither composite system, H2O-NaCl ± (CO2).
(Chaudhari, 1969) nor twinned crystals (Codazzi, The chemical composition of the interstitial fluid,
1915; Bernauer, 1933). The possibility that trapiche the temperature (approximately 300–330°C), and the
emeralds are made of three intersecting crystals elon- bulk composition of the rocks (i.e., albitized black
–
gated in the <110> directions and that the spaces be- shales) are the same for trapiche emerald and emerald-

Figure 27. Variation of
vanadium (V2O3) and
C
0.14 B chromium (Cr2O3) con-
A
14 9
tents (in wt.%) in the
core 13
16 15 trapiche emerald from
12 5 8
Muzo presenting a sec-
11
tion parallel to the c-
Cr2O3 (wt.%)
0.10 44
2
1 0.0
North
arm 81
x–
1
axis (see table 3). A:
0 . 8
0.6 2 0
The different points of
10 y= R=
arm
10
center 11 T10a south analyses are reported
1 0.06
profile 5 6 7 13
2 T10a north from 1 to 16 in the core
4 3 2 12
7 T10c south
(pts. 1, 8–10, 11–13, and
South
4
9 15 6 T10c north
8 15) and the two arms
3 T10 center
0.02
14 16 (pts. 2–7, 14, and 16). B:
0.5 cm
0.12 0.16 0.20 0.24 0.28 Variation of the V2O3
V2O3 (wt.%) and Cr2O3 contents in
the core and the arms
of the sample.
bearing veins. The difference in the texture between and Cr in the core and the arms (tables 2 and 3). Dif-
these two is due to extreme variation of pressure in ferent amounts of chromophore elements can result
the history of their formation. That trapiche formed in different colors of various sectors, as in sample T4
before regular emerald is evidenced by field observa- (figure 12C), where the core is very pale green and the
tion. Trapiche emerald developed on albitized black arms are green (richer in both Cr and V). The higher
shale that is crosscut by regular-emerald-bearing veins quantity of V and Cr also suggests that the arms grow
and tension gashes. Some of the trapiche have an over- faster than the core. The faster growth causes the de-
growth zone that undoubtedly grew during the for-
mation of non-trapiche emerald. The tectonic
Figure 28. Schematic representation of trapiche tex-
decompression increased the degree of supersatura- ture in Colombian emeralds, formed by textural sec-
tion, allowing the formation of trapiche emeralds, tor zoning and displacement growth.
whereas emerald overgrowths occurred at a lower de-
gree of supersaturation, until finally, at equilibrium Matrix inclusions perpendicular
conditions, euhedral non-trapiche emerald grew in the to the growth front
veins. Fan-shaped
Stringers
dendrite
The trapiche emeralds started growing at the be- of emerald
Matrix
ginning of the local decompression. The hydrother-
mal fluid came in contact with the albitized black Arm
shale matrix, favoring the formation of seed crystals

of emerald. During the growth of these seeds, tex-
tural sector zoning—i.e., the accumulation of inclu- Arm Arm
sions derived from the matrix along the interfaces of
growth sectors (Andersen, 1984)—took place. This is
proved by the presence of dendrites all around the
Core
core (pinacoidal growth sectors) and between the
arms (prismatic growth sectors).
Textural zoning can also occur with chemical sec-
tor zoning, in which the growth sectors have differ- Arm Arm
ent chemical compositions. In both cases, the growth

sectors can be easily recognized. They are separated
by inclusions in textural sector zoning, and by com- Arm
position and color differences in chemical zoning.
This is the case with some trapiche samples analyzed
in this study showing different concentrations of V

pletion of Al in the fluid near the prismatic growth carbon (bitumen), which can float in the fluid as a
fronts, allowing the incorporation of metal elements syngenetic material and concentrate in the dendrites.
in the octahedral site to allow the growth’s front to Overpressure can explain the presence of fractures
advance. The faster growth of arms is supported by in the core and arms (see “Results” for details) filled
the higher quantity of defects perpendicular to the with the same materials observed in the matrix. It
growth fronts such as bundles of dislocations and can also account for the plastic deformation of the
solid and multiphase fluid inclusions (figures 20 and core that results in an undulatory extinction under
24). cross-polarized light in some samples.
The overpressure and associated fluid supersatu- Overpressure can also cause a change in the
ration allow emerald to continue its growth and favor chemical composition of the growth medium, lead-
the displacement growth pushing the matrix mate- ing to the dissolution of the emerald not yet in equi-
rial from the growing faces (Yardley, 1974; Carstens, librium with these new conditions. Consequently,
1986; Riche and Mitchell, 1991). Two factors are nec- partial dissolution affects some portions of growing
essary for displacement growth: high fluid pressures crystals. This kind of “selective” dissolution (etch-
and slow growth (Yardley, 1974). Because the core ing) has been described by Bernauer (1933) on some
grows slower than the arms, it pushes away the ma- trapiche emeralds from Muzo. It has been observed
trix material that accumulates around it as dendrites by Lyckberg (2005a,b) on the prismatic faces of he-
(figure 28). liodor from Karelian, Russian, and Ukrainian peg-
The matrix material is also displaced by the pris- matites, and by Tempesta et al. (2011) on beryl
matic growth sectors. Nevertheless, the matrix can- crystals from Minas Gerais, Brazil.
not be moved away as fast as these sectors grow. Thus After decompression and hydraulic fracturing, the
the matrix is trapped, forming dendrites within degree of supersaturation of the fluid decreased con-
and/or along the arms and perpendicular to the siderably. This can favor the formation of an over-
growth fronts (i.e., parallel to the elongation direction growth surrounding the core, the arms, and the
of the arms; see figures 12D and 21D). It should be dendrites. The lower degree of supersaturation can
noted that the matrix material is trapped primarily also explain the formation of elongated voids in the
along the a-axis, forming dendrites whose size is a arms by dissolution along the dislocation bundles
function of the quantity of displaced material (figure (Scandale and Zarka, 1982; Authier and Zarka, 1994).
12). The crystallographic orientation of the dendrites
can be explained by the Berg effect (1938). Because the Geological History. The emerald-bearing veins in
driving force is higher on the corners than on the Colombia’s western emerald zone formed in a con-
faces, the emerald can grow so fast as to trap most of text of fluid pressure variations in tear faults and
the matrix material along the a-axis. The fan-shaped thrusts affecting the Lower Cretaceous black shales.
dendrites observed in some trapiche emeralds are Formed along tear faults were numerous veins and
common in trapiche chiastolites and garnets (Riche hydraulic breccias. They resulted from the fracturing
and Mitchell, 1991; Rice, 1993) and are also known of the wall rocks during a tear fault increment (a dis-
as “re-entrants” (Pogue, 1916). Fan-shaped dendrites placement along the fault plane, likely inducing a
are also considered nucleation zones for minor seismic event). This increment is responsible for de-
growth (Rice, 1993), explaining why in some den- velopment of dilational jogs along several fault seg-
drites the stringers of emerald develop parallel to the ments. As a result, the jogs caused a large fluid
edges of prismatic sectors and intersect at 60° angles pressure gradient and “pumped” the surrounding hy-
(figure 21A). This 60° angle is not arbitrary and indi- drothermal fluids. Blocks of black shales, white al-
cates that the emerald grows parallel to the two pris- bitite, and albitized and/or carbonatized black shales
matic sectors delimiting the fan-shaped dendrite. The simultaneously collapsed into the jog and were ce-
rapid growth along the a-axis is also proved by the fact mented by carbonates and albite.
that the stringers of emerald have a tree-like, fibrous Along the thrust, the fault propagation is achieved
texture (figures 25A and 25C). by shear-tensional failure caused by hydraulic fractur-
During displacement growth, the black shales ing under supra-lithostatic fluid pressure and small
played a key role in the thermal reduction of sulfate differential effective stresses (box C). The polygenetic
by oxidizing the organic matter. Oxidation generated nature of the fragments of the breccia such as in the
CO2, creating fluid overpressure in the black shale. tear faults, and the rotations and partial dissolution of
Such exothermic reactions produced hydrothermal the elements, clearly indicate bulk transport of mate-

BOX C: GEOLOGICAL MODEL FOR THE PROPOSED MECHANISM OF
TRAPICHE EMERALD FORMATION
A Failure and thrust increment B

τ
Post-failure discharge,
abrupt decompression
A' and high fluid flow
B
A σ3 σ3
σn
σ'3 σ'1
B Nucleation and B
τ Failure envelope for thrust A' growth of cores, A'

initiation in black shales dendrites and arms
Fluid pressure σ1 σ1
build-up
σn A A
σ'3 σ '1
C D
τ
Fluid pressure progressive drop NNW TEQUENDAMA MINE SSE
900 - B
σn
σ'3 σ'1 A
800 -
σ3 Crystallization of trapiche overgrowth
and non-trapiche emeralds
B 700 -
σ1
600 -
100 m Hydrothermal breccias Thrust

(hydraulic and fluidized types) (arrow for kinematics)
Siliceous black shales (permeability barrier) Fluidized breccia

Carbonated black shale (fluid-saturated rock mass) Wall-rock dislocation (mainly by implosion)
and hydrothermal alteration
Albitites
Fluidized breccia injections
Hydrothermal alteration (mainly albitization with
carbonatization and pyritization) Hydrothermal fluid flow
Figure C-1. Geological model for the proposed mechanism of the trapiche emerald formation from the western
emerald zone. A: Thrust propagation and failure. B: Nucleation and growth of core, dendrites, and arms of
Tequendama mine. In the Mohr’s circle diagrams in A, B, and C, τ = shear stress; σn = normal stress; σ1 and
trapiche emerald. C: Formation of trapiche overgrowths and regular emerald in veins. D: Cross-section of the
σ3 = the main stresses.
rial during fluid flow (fluidization phenomena). The thrust veins, which are horizontal tension gashes in
cement is formed by carbonates, pyrite, and albite, cre- this compressive tectonic context. The sealing of
ating the hydrothermal breccia called “cenicero.” The faults by carbonate deposition and cementation of the
fluidized breccias are associated with emerald-bearing fluidized breccia depends on the elapsed time between

Trapiche emerald (figure 29) formed when the fluid
pressure exceeded the lithostatic pressure (box C). The
The formation of Colombian trapiche emerald is closely
local decompression induced the supersaturation of
related to the structural geology history in the western
the hydrothermal fluid and an increase of the driving
zone. The correlation between the formation of tear
faults under compressional stress and the growth of force of the emerald crystal growth (stage 1 under
trapiche emerald can be elucidated in the following “Emerald Mineralization”). The emerald seed crystals
three steps (figure C-1: A, B and C): grew on albitized carbon-rich black shale (see box C,
figure C-1) or albitite (the type seen at Peñas Blancas;
• A: Thrust propagation and failure. The incremental
thrust propagation is related to the highest fluid pres- Vuillet et al., 2002), as argued by the presence of solid
sures, which caused the failure and propagation of inclusions and matrix around the trapiche cores
the fault (from point A to point A'). The failure oc- (Bernauer, 1933; Nassau and Jackson, 1970). Stage 1 is
curred in the hybrid-shear-extensional mode, the also associated with the formation of tension gashes
thrust presenting a dilatant component. Physically, and thrust-fluidized breccias (“cenicero”) and thrust
the Mohr’s circle moved to the left and touched the veins.
failure envelope of the black shale material. The hydrothermal fluid bathed each whole cavity,
• B: Abrupt decompression of fluid pressure is related and the precipitation of carbonates, pyrite, and albite
to the opening of dilatant sites (i.e., high permeability sealed the fault open spaces (hydrothermal breccia),
and porosity zones) along the thrust. Hypersaline Na- and emerald deposition occurred with its carbonate-
bearing fluids responsible for strong albitization of pyrite assemblage in the thrust veins (stage 2). Dur-
the hosting carbon-rich black shales are “pumped”
ing this last episode of emerald crystallization in the
into those dilatant zones with high flow rates. As a
veins, the degree of supersaturation of the fluid de-
result, high fluid pressure gradients also triggered flu-
idization processes responsible for the formation of creased considerably up until chemical equilibrium.
polygenic breccia. Physically, the Mohr’s circle de- During this time, the trapiche emeralds probably ac-
scribes an incipient rightwards motion. This decom- quired their final deep-green overgrowths, also found
pression caused an increase of the driving force of the in non-trapiche emerald-bearing thrust veins.
emerald crystal growth. The emerald seed crystals
grew on albitized black shales. During the growth of CONCLUSIONS
these seeds, textural sector zoning of trapiche emer- Trapiche emeralds occur in association with black
ald took place with the dendrites all around the core shales in the western emerald zone of Colombia’s
and between the arms.
Eastern Cordillera Basin. They formed in a geological
• C: Formation of trapiche overgrowths and non- context of fluid pressure variations along faults and
trapiche emeralds during a progressive decrease of
thrusts affecting the Lower Cretaceous sedimentary
pressure and flow (i.e., lower fluid pressure gradient
than in B). Non-trapiche emeralds are mainly hosted
series. The fluid accumulation at the faults’ tip led to
within sub-horizontal extensional veins and flu- maximum fluid overpressure and subsequent decom-
idized hydrothermal breccias. pression. The rocks failed, and trapiche emerald
formed in the albitized black shale host rock, followed
In diagram D in figure C-1, a cross-section of the
Tequendama mine following the geological map in fig- by the crystallization of emerald in carbonate veins.
ure 6 illustrates the relationship between structures and Trapiche emeralds from Muzo, Coscuez, and
trapiche growth. The cross-section shows the different Peñas Blancas were investigated by several analytical
lithologies and tectonic structures described in the text. techniques in order to propose a model of formation
The thrusts and tear faults (represented in figure 6) are for trapiche texture linked to the structural geology
associated with the breccias, sub-horizontal extensional of the deposits. Petrographic and SEM observations
veins, and potential zones related to trapiche emerald. combined with X-ray topography and tomography
images confirmed that trapiche emerald is a single
crystal formed by different zones: a core (pinacoidal
growth sectors), dendrites, and arms (prismatic
growth sectors), sometimes surrounded by an over-
the drop and the following build-up of fluid pressure. growth. The color of these zones is due to the pres-
The thrusts propagate in a succession of faulting, fluid ence of V and lower concentrations of Cr, although
flow (multistage events), hydrothermal alteration, and both elements were detected by EPMA even in col-
emerald deposition (Branquet et al., 1999b). orless zones such as some of the cores.

mal fluid came in contact with the black shale
matrix, favoring the formation of emerald
crystal seeds.
2. Textural sector zoning, sometimes associated
with chemical sector zoning, took place during
the growth of these seeds. The inclusions of
black shale matrix were incorporated between
the growth sectors, forming the dendrites
around the core and between the arms.
3. The emeralds continued their growth, displac-
ing the matrix material from the growing faces.
As a result, the matrix may also be trapped in
the arms perpendicular to their growth fronts,
even if it is preferentially incorporated along
the a-axis. The dendrites between the arms
have a typical fan-shape near the borders of the
samples, because the emerald grows faster
along the a-axis than along 〈110〉 (see again fig-
–
ure 13), and thus a higher quantity of matrix
Figure 29. These trapiche emeralds, all from Peñas material can be trapped.
Blancas, range from 5.09 to 22.74 ct. Photo by
Robert Weldon/GIA. Courtesy of Jose Guillermo 4. An overgrowth, generally of gem quality, may
Ortiz, Colombian Emerald Co. be formed after decompression. This over-
growth surrounding the core, arms, and den-
drites and restoring the emeralds’ euhedral
The formation of the trapiche texture in Colom- habit. During overgrowth formation, non-
bian emeralds is controlled both by the peculiar trapiche emerald formed in the carbonate veins.
“structurally controlled” genesis of the deposits and For the first time, the development of the
by the crystal symmetry, determining which trapiche texture has been related to the geological
growth sectors develop and how many of them. conditions of Colombian emerald deposits. Al-
From this, a few crucial points on trapiche forma- though some formation models were previously pro-
tion can be summarized:
posed for trapiche minerals, they were extrapolated
1. The growth of trapiche emerald started during from the deposit context and were not supported by
the local decompression responsible for the geological events responsible for the supersaturation
supersaturation of the fluid. The hydrother- of the mineralizing fluids.
ABOUT THE AUTHORS Bari, Italy. Ms. Mathieu specializes in quantitative analyses using
Dr. Pignatelli, a geologist and expert in mineralogy and crystal- EDX and WDX spectrometry in complex systems at GéoRes-
lography, is currently a postdoctoral researcher at the University sources 7359 research unit, Vandœuvre, France. Mr. Morlot is
of California, Los Angeles. Dr. Giuliani is director of research at responsible for X-ray tomography applied to Geosciences at
the Institute of Research for Development (IRD) in Toulouse GéoRessources 7359 unit, Vandœuvre, France. Dr. Branquet is
(GET) and the French National Center of Research an assistant professor of earth sciences at the University of Or-
(CRPG/CNRS) in Vandœuvre. He is recognized for his work on léans in France.
the geology, geochemistry, and genesis of several colored stone
deposits. Dr. Ohnenstetter is a retired geologist based in Mar- ACKNOWLEDGMENTS
inville-sur-Madon, France, formerly with CRPG/CNRS, who spe- The authors would like to thank Omar Bustos Santana, who col-
cializes in petrology and cathodoluminescence spectroscopy lected and provided the trapiche emerald samples used in this
and mineralogy. Dr. Agrosì is a researcher with the Department study. We thank also the manuscript reviewers for their helpful
of Earth and Geo-environmental Sciences at the University of and constructive suggestions.

GLOSSARY
Albitite: a magmatic or metasomatic (see metasomatism) most commonly calcite or quartz.
rock composed almost wholly of coarse- to fine-grained Eutectic system: a mixture of two species that melt and
crystals of albite. solidify at constant temperature. The eutectic system be-
Basinal brine: fluids in basins that are enriched Na, Ca, and haves in fact as a pure body from the standpoint of fusion.
Mg sulfates. In basins, processes of sediment or basin de- Evaporites: any sedimentary rock such as gypsum, anhy-
watering give rise to large volumes of fluids moving drite, or rock salt (such as halite or sylvite) that is formed
through different types of rocks (generally sedimentary). by precipitation from evaporating seawater.
The temperature increases to 80°–200°C or higher based
Extensional vein: equivalent of tension gash.
on the depth of fluid circulation in the basin. After reacting
with different rocks, the fluids are enriched in Na, Ca, and Fluidization phenomena: transport of rock fragments and
Mg sulfates, and their salinity can increase tremendously. breccia by a fluid (liquid or gas) in a fault structure. During
fluid flow, the heavier fragments of the breccia are trans-
Black shale: a fine-grained clastic sedimentary mudrock
ported, suspended, rotated, sorted, and cemented by the
composed of clay mineral flakes and other fragments,
matrix. The fragments and matrix are suspended and
mainly quartz and calcite.
transported as a fluid-supported pulp.
Breccia: a rock composed of broken fragments of minerals
Framboidal: Referring to a texture of roughly spherical ag-
or rock cemented together by a fine-grained matrix.
gregates of discrete rounded equi-angular euhedral micro-
Burgers vectors: a quantity, usually represented by an crystallites approximately 0.5 μm in diameter, with an
arrow, describing the magnitude and the direction of the average aggregate size ranging from 5 to 20 μm.
lattice distortion resulting from a dislocation.
Hydraulic breccia: a rock formation resulting from hy-
Cataclasite: a rock consisting of angular clasts in a fine- draulic fracturing, in which the hydrostatic pressure is
grained matrix that formed by fracturing and pulverization greater than lithostatic pressure. Hydraulic fracturing may
of the rock during faulting. create a breccia that is filled with vein material. Such vein
Clast: a fragment of pre-existing mineral or rock. systems may be quite extensive, and can form in the shape
Coquina limestone: slightly cemented sedimentary rock of tabular dipping sheets, diatremes, or laterally extensive
formed almost completely of sorted fossils, generally shell mantos controlled by boundaries such as thrust faults,
and shell fragments. competent sedimentary layers, or cap rocks.
Daughter mineral: the new phases formed in a fluid inclu- Incremental thrust propagation: the propagation of a dis-
sion cavity that have precipitated from the solution within. located part of the terrane over an area in relatively small
overlapping increments.
Décollement plane: a low-angle dipping fault or shear zone
corresponding to a gliding plane between two rock masses. Internal over- and underpressure: When trapped by a min-
These planes are developed in compressional tectonic set- eral, pressure and temperature conditions preserved by
tings (involving folding and overthrusting) and extensional fluid inclusion cavities can sometimes suffer re-equilibra-
settings. Also known as a basal detachment fault. tion during growth. As a result, the cavities can suffer in-
ternal overpressure when the internal pressure in the
Dendrite: A portion of the trapiche texture having saw-
inclusion exceeds the external confining pressure, or inter-
tooth contact with the core and the arms and formed by
nal underpressure when the internal pressure is lower than
dendritic growth (i.e., a high degree of supersaturation, see
the external confining pressure.
box B). In agreement with the literature, the word can also
be used to refer to feather-edge morphologies due to the in- Jogs (or bends): a gap or segment between two fault plane
corporation of matrix material (as in this study). with the same strike and slip direction. Jogs offset perpen-
dicular to the slip direction.
Detachment: see décollement plane
Diffraction vector g: In a crystal of any structure, vector g Klippe: in thrust fault terrains, the remnant of a nappe after
is normal to the planes (hkl) and has a length inversely pro- erosion has removed the connecting portions.
portional to the distance between the planes. Layer-by-layer growth: a growth mechanism based on two-
Dolomitic limestone (or dolomite): an anhydrous carbon- dimensional nucleation.
ate mineral with the formula (Ca, Mg)(CO3)2. Listric faults: faults in which the fault plane curves; the dip is
Drag folds: curvature that represents the bending of rocks be- steeper near the surface and shallower with increased depth.
fore they break. Drag folds occur in conjunction with faults. Lithostatic pressure: the pressure or stress imposed on a
Drusy cavity: a hollow space within a rock with a crust of layer of soil or rock by the weight of material. It is also called
tiny crystals. overburden pressure, confining pressure, or vertical stress.
En echelon vein: also known as tension gash arrays, these Metasomatism: the chemical alteration of a rock by hy-
veins are characteristic of ductile-brittle shear zones. They drothermal and/or other fluids. The chemical elements of
are sets of short, parallel planar lenses on the surface of a the mother rock undergo fluid transport, accompanied by
rock. En echelon veins originate as tension fractures and are a modification (metasomatic process) of chemical compo-
subsequently filled through the precipitation of a mineral, sition to form a daughter rock.

Meteoric water: the water derived from precipitation that controlled by tectonic (e.g. normal faults as in passive mar-
forms ice melts and bodies of water such as rivers and lakes. gin or rift basin) and thermal re-equilibration of thinned
Mohr’s circle: A diagram that graphically illustrates in two crust.
dimensions the complex relationships between compo- Supersaturation (or undercooling): the difference between
nents of normal and shear stress. the concentration C∞ at the equilibrium temperature TE of
Nappe: a large sheetlike body that moved more than two a saturated solution and the concentration C at the growth
kilometers above the thrust fault from its original location. temperature TG (Sunagawa, 2005). Supersaturation is criti-
cal because it is the driving force for crystal nucleation and
Necking-down: a re-equilibration process of the morphol-
growth.
ogy of a fluid cavity during cooling. This post-entrapment
phenomenon changes the shape of the cavity by splitting Tear faults: small-scale local strike-slip faults. Strike-slip
it into smaller fluid inclusions. Necking-down leads to a faults with left-lateral motion are known as sinistral
distribution of the phases (vapor, liquid, solid) that is dif- faults, those with right-lateral motion are known as dex-
ferent from the original fluid inclusion assemblage. tral faults. Each is defined by the direction of movement
of the ground on the opposite side of the fault from an ob-
Olistostrome: a sedimentary deposit composed of a chaotic
server. Tear faults are associated with other structures
mass of heterogeneous material, such as blocks and mud.
such as folds, thrust faults, and normal faults. In a thrust
The mass accumulates as a semi-fluid body via submarine
belt, the tear faults strike perpendicular to the thrust front
gravity sliding or slumping of the loose materials.
and fold axis. As the thrust sheet propagates into the fore-
Overthrusting: large total displacement of thrust faults. land, the tear faults start to segment the thrust sheets,
Phyllosilicates: silicate minerals with sheet-like structures. eventually acting as lateral boundaries that separate the
Pinacoidal faces: two equivalent faces related by an inver- thrust units.
sion center. Tension gash: veins characteristic of ductile-brittle shear
Scalar product: the product of the magnitudes of two vec- zones, classically disposed en echelon when non-coaxial
tors and the cosines of the angles between them. strain is applied to rocks (see en echelon vein). Also called
Shear zone: a zone of strong deformation with a high strain extensional veins.
rate surrounded by rocks with a lower state of finite strain. Thermal reduction of sulfate: redox reactions in which sul-
It is characterized by a length-to-width ratio of more than fate is reduced by hydrocarbons, either bacterially (bacterial
5:1. Shear zones form a continuum of geological structures, sulfate reduction) or inorganically (thermochemical sulfate
ranging from brittle to brittle-ductile (or semi-brittle), duc- reduction).
tile-brittle, and ductile shear zones. Thrust fault: a fault with a very low dip angle and a very
Space group: A set of symmetry operations of a three- large total displacement of the terranes, called overthrust
dimensional crystal pattern. or detachment. Thrust faults are characteristic of compres-
Stringers: oblique growth of emeralds at 60° to each other sive tectonics. This process results in an outlier of exotic,
occurring in the dendrites. They are also observed in other often nearly horizontally translated strata called “nappe”
trapiche minerals such as chiastolite and garnet, where overlying autochthonous strata.
they form at 90° and 60°, respectively (Rice, 1993). Undulatory extinction: the consequence of tectonic con-
Subsiding marine basin: areas generally characterized by straints, seen as a darker zone during the rotation of the
thick accumulation of marine sediments. Subsidence is microscope stage. Also called wavy extinction.
REFERENCES
Agrosì G., Bosi F., Lucchesi S., Melchiorre G., Scandale E. (2006) Artioli G., Rinaldi R., Ståhl K. (1993) Structure refinements of
Mn-tourmaline from island of Elba (Italy): Growth history and beryl by single-crystal neutron and X-ray diffraction. American
growth marks. American Mineralogist, Vol. 91, Nos. 5–6, pp. Mineralogist, Vol. 78, pp. 762–768.
944–952, http://dx.doi.org/10.2138/am.2006.1978 Atherton M.P., Brenchley P.J. (1972) A preliminary study of the
Agrosì G., Tempesta G., Scandale E., Harris J. W. (2013) Growth structure, stratigraphy and metamorphism of some contact
and post-growth defects in a diamond from Finsch mine rocks of the Western Andes, near the Quebrada Venado Muerto,
(South Africa). European Journal of Mineralogy, Vol. 25, No. Peru. Geological Journal, Vol. 8, pp. 161–178, http://dx.doi.org/
4, pp. 551–559, http://dx.doi.org/10.1127/0935-1221/2013/ 10.1002/gj.3350080114
0025-2301 Aurisicchio C., Fioravanti G., Grubessi O., Zanazzi P.F. (1988)
Andersen T.B. (1984) Inclusion patterns in zoned garnet from Magerøy, Reappraisal of the crystal chemistry of beryl. American Min-
north Norway. Mineralogical Magazine, Vol. 48, No. 346, pp. 21– eralogist, Vol. 73, Nos. 7–8, pp. 826–837.
26, http://dx.doi.org/10.1180/minmag.1984.048.346.03 Authier A., Zarka A. (1994) X-ray topographic study of the real
Andrianjakavah P.R., Salvi S., Béziat D., Rakotondrazafy A.F.M., structure of minerals. In A.S. Marfunin, Ed., Composition,
Giuliani G. (2009) Proximal and distal styles of pegmatite-re- Structure and Properties of Mineral Matter, Springer-Verlag,
lated metasomatic emerald mineralization at Ianapera, south- Berlin, pp. 221–233.
ern Madagascar. Mineralium Deposita, Vol. 44, No. 7, pp. Banks D.A., Giuliani G., Yardley B.W.D., Cheilletz A. (2000) Emer-
817–835, http://dx.doi.org/10.1007/s00126-009-0243-5 ald mineralisation in Colombia: fluid chemistry and the role

of brine mixing. Mineralium Deposita, Vol. 35, No. 8, pp. 699– Chaudhari M.W. (1969) An unusual emerald. Schweizerische Min-
713, http://dx.doi.org/10.1007/s001260050273 eralogische und Petrographische Mitteilungen, Vol. 49, No. 3,
Barriga Villalba A.M. (1948) Estudio cientifico de la Esmeraldas de pp. 569–575.
Colombia. In Esmeraldas de Colombia, Banco de la Republica Cheilletz A., Giuliani G. (1996) The genesis of Colombian emer-
Publisher, Bogotá, Colombia Parte II, pp. 97–133 [in Spanish]. alds: a restatement. Mineralium Deposita, Vol. 31, No. 5, pp.
Befi R. (2012) Gem News International: Trapiche aquamarine from 359–364, http://dx.doi.org/10.1007/BF00189183
Namibia. G&G, Vol. 48, No. 2, pp. 143–144. Cheilletz A., Féraud G., Giuliani G., Rodriguez C.T. (1994) Time-
Berg W.F. (1938) Crystal growth from solutions. Proceedings of the pressure and temperature constraints on the formation of
Royal Society, Vol. 164, No. 916, pp. 79–95, http://dx.doi.org/ Colombian emeralds; an 40Ar/39Ar laser microprobe and fluid
10.1098/rspa.1938.0006 inclusion study. Economic Geology, Vol. 89, No. 2, pp. 362–
Bergt W. (1899) Die älteren Massengesteine, kristalli nen Schiefer 380, http://dx.doi.org/10.1180/10.2113/gsecongeo.89.2.361
und Sedimente, p. 46. In F. Bernauer Ed., Compilación de los Cheilletz A., Giuliani G., Branquet Y., Laumonier B., Sanchez A.J.,
Estudios Geológicos Oficiales en Colombia -1917 a 1933, No. Féraud G., Arthan T. (1997) Datation K-Ar et 40Ar/39Ar à 65 ± 3
4, pp. 199–221 (in German). Ma des gisements d’émeraude du district de Chivor-Macanal:
Bernauer F., Ed. (1933) Las llamadas maclas multiples de esmeralda argument en faveur d’une déformation précoce dans la Cordil-
de Muzo y sus anomalias opticas. Compilación de los Estudios lère Orientale de Colombie. Comptes Rendus de l’Académie
Geológicos Oficiales en Colombia -1917 a 1933, No. 4, pp. des Sciences, Paris, Sér. IIa, Vol. 324, pp. 369–377 (in French).
199–221 (in Spanish). Cnudde V., Boone M.N. (2013) High-resolution X-ray computed
Bertrand E. (1879) Compte-rendu de la séance du 13 Février 1879 tomography in geosciences: a review of the current technology
[Report of the meeting of February 13, 1879]. Bulletin de la So- and applications. Earth-Science Reviews, Vol. 123, pp. 1–17,
ciété Minéralogique de France, Vol. 2, p. 31. http://dx.doi.org/10.1016/j.earscirev.2013.04.003
Beus A.A. (1979) Sodium - a geochemical indicator of emerald Codazzi R.L. (1915) Los minerales de Muzo. Contribución al Estu-
mineralization in the Cordillera oriental, Colombia. Journal of dio de los Minerales de Colombia, Bogotá, pp. 3–7 (In Spanish).
Geochemical Exploration, Vol. 11, No. 2, pp. 195–208, Damon P.E., Kulp J.L. (1958) Excess helium and argon in beryl and
http://dx.doi.org/10.1016/0375-6742(79)90023-2 other minerals. American Mineralogist, Vol. 43, Nos. 5–6, pp.
Beus A.A., Mineev D.A. (1972) Some geological and geochemical 433–459.
features of the Muzo and Coscuez emerald zone, Cordillera DelRe N. (1994) Lab Notes: Emerald, trapiche from a new locality.
oriental, Colombia. INGEOMINAS, Informe 1689, 50 pp. G&G, Vol. 30, No. 2, pp. 116–117.
Branquet Y. (1999) Etude structurale et métallogénique des gise- Faure F., Trolliard G., Nicollet C., Montel J.M. (2003) A development
ments d’émeraude de Colombie: contribution à l’histoire model of olivine morphology as a function of the cooling rate and
tectono-sédimentaire de la cordillère orientale de Colombie. the degree of undercooling. Contributions to Mineralogy and
PhD Thesis, Université de Lorraine, Nancy, France [in French]. Petrology, Vol. 145, No. 2, pp. 251–263, http://dx.doi.org/
Branquet Y. Laumonier B., Cheilletz A., Giuliani G. (1999a) Emer- 10.1007/s00410-003-0449-y
alds in the eastern Cordillera of Colombia: two tectonic settings Faure F., Schiano P., Trolliard G., Nicollet C., Soulestin B. (2007)
for one mineralisation. Geology, Vol. 27, No. 7, pp. 597–600, Textural evolution of polyhedral olivine experiencing rapid
http://dx.doi.org/10.1130/0091-7613(1999)027%3C0597: cooling rates. Contributions to Mineralogy and Petrology, Vol.
EITECO%3E2.3.CO;2 153, No. 4, pp. 369–492, http://dx.doi.org/10.1007/s00410-006-
Branquet Y., Cheilletz A., Giuliani G., Laumonier B., Blanco O. 0154-8
(1999b) Fluidized hydrothermal breccia in dilatant faults during Garnier V., Ohnenstetter D., Giuliani G., Blanc P., Schwarz D.
thrusting: The Colombian emerald deposits. In Fractures, Fluid (2002a) Trace-element contents and cathodoluminescence of
and Mineralization. Geological Society, London, 1999, Special “trapiche” rubies from Mong Hsu, Myanmar (Burma): geolog-
Publication 155, pp. 183–195. ical significance. Mineralogy and Petrology, Vol. 76, Nos. 3–4,
Branquet Y., Giuliani G., Cheilletz A., Laumonier B. (2015) pp. 179–193, http://dx.doi.org/10.1007/s007100200040
Colombian emeralds and evaporites: tectono-stratigraphic sig- Garnier V., Ohnenstetter D., Giuliani G., Schwarz D. (2002b)
nificance of a regional emerald-bearing evaporitic breccia level. Rubis trapiches de Mong Hsu, Myanmar. Revue de Gemmolo-
13th SGA Biennial Meeting, Nancy, France, Proceedings Vol. gie AFG, Vol. 144, pp. 5–12 (in French).
4, pp. 1291–1294. Geake J.E., Walker G. (1975) Luminescence of Minerals in the
Breeding C.M., Shen A.H., Eaton-Magaña S., Rossman G.R., Near Infrared and Raman Spectroscopy of Lunar and Terres-
Shigley J.E., Gilbertson A. (2010) Developments in gemstone trial Minerals. Academic Press, New York, pp. 73–89.
analysis techniques and instrumentation during the 2000s. Giuliani G., Rodriguez C.T., Rueda F. (1990) Les gisements d’émer-
G&G, Vol. 46, No. 3, pp. 241–257, http://dx.doi.org/10.5741/ aude de la Cordillère Orientale de la Colombie: nouvelles don-
GEMS.46.3.241 nées métallogénies. Mineralium Deposita, Vol. 25, pp. 105–111
Bryan W.B. (1972) Morphology of quench crystals in submarine (in French).
basalts. Journal of Geophysical Research, Vol. 77, No. 29, pp. Giuliani G., Cheilletz A., Dubessy J., Rodriguez C.T. (1991) H2O-
5812–5819, http://dx.doi.org/10.1029/JB077i029p05812 NaCl-CaCl2-bearing fluids in emeralds from the Vega San Juan
Burton K.W. (1986) Garnet-quartz intergrowths in graphitic mine, Gachalá district, Colombia. Plinius, No. 5, p. 90.
pelites: the roles of the fluid phase. Mineralogical Magazine, Giuliani G., Sheppard S.M.F., Cheilletz A., Rodriguez C.T. (1992)
Vol. 50, pp. 611– 620, http://dx.doi.org/10.1180/minmag.1986. Fluid inclusions and 18O/16O, 13C/12C isotope geochemistry:
050.358.06 contribution to the genesis of emerald deposits from the Ori-
Byerly G.R., Palmer M.R. (1991) Tourmaline mineralization in the ental Cordillera of Colombia. Comptes Rendus de l’Académie
Barberton greenstone belt, South Africa: early Archean meta- des Sciences, Paris, Vol. 314, pp. 269–274.
somatism by evaporite-derived boron. Contributions to Min- Giuliani G., Cheilletz A., Dubessy J., Rodriguez C.T. (1993) Emerald
eralogy and Petrology, Vol. 107, No. 3, pp. 387–402, deposits from Colombia: chemical composition of fluid inclu-
http://dx.doi.org/10.1007/BF00325106 sions and origin. Proceedings of the 8th IAGOD Symposium,
Carstens H. (1986) Displacive growth of authigenic pyrite. Journal Ottawa, pp. 159–168.
of Sedimentary Petrology, Vol. 56, No. 2, pp. 252–257. Giuliani G., Cheilletz A., Arboleda C., Carrillo V., Rueda F., Baker
Charoy B. (1998) Cristallochimie du béryl: l'état des connaissances. J. (1995) An evaporitic origin of the parent brines of Colombian
In L'émeraude, Association Française de Gemmologie-CNRS- emeralds: fluid inclusion and sulphur isotope evidence. Euro-
ORSTOM, Groupe Rosay, Paris, France, pp. 47–54 (in French). pean Journal of Mineralogy, Vol. 7, pp. 151–165.

(1997) Fluid composition, δD of channel H2O, and δ18O of lat-
Giuliani G., France-Lanord C., Zimmermann J-L., Cheilletz A. (1996) Mise en évidence d’une tectonique compressive Eocène-
Oligocène dans l’Ouest de la Cordillère orientale de Colombie,
tice oxygen in beryls: Genetic implications for Brazilian, d’après la structure en duplex des gisements d’émeraude de
Colombian, and Afghanistani emerald deposits. International Muzo et Coscuez. Comptes Rendus de l’Académie des Sci-
Geology Review, Vol. 39, No. 5, pp. 400–424, http://dx.doi.org/ ences, Paris, Série IIa, Vol. 323, pp. 705–712 (in French).
10.1080/00206819709465280 Leiper H. (1967) Rare trapiche emerald crystals show unique twin-
Giuliani G., France-Lanord C., Cheilletz A., Coget P., Branquet Y., ning evidence. Lapidary Journal, Vol. 21, No. 4, pp. 565.
Laumonier B. (2000) Sulfate reduction by organic matter in London D. (2008) Pegmatites. The Canadian Mineralogist, special
Colombian emerald deposits: chemical and stable isotope (C, publication No. 10, Mineralogical Association of Canada,
O, H) evidence. Economic Geology, Vol. 95, No. 5, pp. 1129– Québec.
1153, tp://dx.doi.org/10.2113/95.5.1129 Lyckberg P. (2005a) Le béryl héliodore de Karélie (Finlande). Le
Giuliani G., Jarnot M., Neumeier G., Ottaway T., Sinkankas J., Règne Minéral, Vol. 62, pp. 33–39 [in French].
Staebler G. (2002) Emeralds of the World. ExtraLapis English ——— (2005b) Gem beryl from Russia and Ukraine. In Beryl and
No. 2: The Legendary green beryl. Lapis International Pub- Its Color Varieties, Lapis International Publishing, East Hamp-
lisher, LLC, East Hampton, CT. ton, CT, pp. 49–57.
Giuliani G., Marty B., Banks D. (2005) Noble gases in fluid inclu- Mantilla Figueroa L.C., Silva A., Serrano J., Gomez J., Ramirez J.,
sions from emeralds: Implications for the origins of fluids and Meza J., Pelayo Y., Ortega L., Plata L., Peña E. (2007) Investa-
constraints on fluid-rock interactions. Abstract, 18th Biennial gación petrográfica y geoquímica de las sedimentitas del
Meeting of European Current Research on Fluid Inclusions, Cretácico inferior (K1) y sus manifestaciones hidrotermales
Siena, Italy, CD-ROM. asociadas; Planchas 169, 170, 189, 190 (Cordillera oriental):
Götze J., Krbetschek M.R., Habermann D., Wolf D. (2000) High- imlicacions en la búsqueda de Esmeraldas. INGEOMINAS in-
resolution cathodoluminescence studies of feldspar minerals. forme, Bogotá, Colombia, 330 pp. (In Spanish).
In Cathodoluminescence in Geosciences. Springer Verlag, Maya M., Buenaventura J., Salinas R. (2004) Estado del
Berlin, pp. 245–270. conocimiento de la exploracion de Esmeraldas en Colombia.
Groat L., Giuliani G., Marshall D., Turner D. (2014) Emerald. In Ge- INGEOMINAS informe, Bogotá, Colombia, 89 pp. (In Spanish).
ology of Gem Deposits, Mineralogical Association of Canada, McKague H.L. (1964) Trapiche emeralds from Colombia. G&G,
Short Course Series 44, Tucson, pp. 135–174. Vol. 11, No. 7, pp. 210–223.
Hainschwang T., Notari F., Anckar B. (2007) Trapiche tourmaline Müllenmeister H.J., Zang J. (1995) Ein Trapiche-Rubin aus Myan-
from Zambia. G&G, Vol. 43, No. 1, pp. 36–46, http://dx.doi.org/ mar (Burma). Lapis, Vol. 20, No. 12, p. 50 (in German).
10.5741/GEMS.43.1.36 Nassau K., Jackson K.A. (1970) Trapiche emeralds from Chivor and
Hall M. (1993) Mineralogía y geoquímica de las vetas esmer- Muzo, Colombia. American Mineralogist, Vol. 55, Nos. 3–4, pp.
aldíferas de Muzo, Departamento de Boyacá con implicaciones 416–427.
en la prospección futura de esmeraldas en otras partes de O’Donoghue M.J. (1971) Trapiche emerald. The Journal of Gem-
Colombia. Compilación de los estudios geológicos oficiales en mology, Vol. 12, No. 8, pp. 329–332.
Colombia. Tomo XIII, pp. 1–326 (In Spanish). Ohnenstetter D., Giuliani G., Bustos O. (1998) Emeraudes trapiches
Harker A. (1950) Metamorphic Textures. Methuen and Co., London. colombiennes. In L’émeraude. Connaissances actuelles et
Hochleitner R. (2002) The emerald: mineralogically a beryl! In Emer- prospective. Association Française de Gemmologie, pp. 119–124
alds of the World, ExtraLapis English No. 2: The Legendary green (in French).
beryl. Lapis International, LLC, East Hampton, CT, pp. 11–17. Oppenheim V. (1948) The Muzo emerald zone, Colombia. Eco-
Hsu T. (2013) Gem News International: Rare double-trapiche nomic Geology, Vol. 43, No. 1, pp. 31–38, http://dx.doi.org/
emerald. G&G, Vol. 49, No. 1, p. 53. 10.2113/gsecongeo.43.1.31
Invernizzi C., Vityk M., Cello G., Bodnar R. (1998) Fluid inclusions Ottaway T.L. (1991) The geochemistry of the Muzo emerald de-
in high pressure/low temperature rocks from the Calabrian Arc posit, Colombia. Master’s thesis, University of Toronto, Canada.
(Southern Italy): the burial and exhumation history of the sub- Ottaway T.L., Wicks F.J., Bryndzia L.T., Kyser T.K., Spooner E.T.C.
duction-related Diamante-Terranova unit. Journal of Metamor- (1994) Formation of the Muzo hydrothermal emerald deposit
phic Geology, Vol. 16, No. 2, pp. 247–258, http://dx.doi.org/ in Colombia. Nature, Vol. 369, No. 6481, pp. 552–554,
10.1111/j.1525-1314.1998.00138.x http://dx.doi.org/10.1038/369552a0
Jia L.C., Chen M., Jin Y. (2014) 3D imaging of fractures in carbon- Pogue J.E. (1916) The emerald deposits of Muzo, Colombia. Trans-
ate rocks using X-ray computed tomography technology. Car- actions of the American Institute of Mining Engineers, Vol. 55,
bonates and Evaporites, Vol. 29, No. 2, pp. 147–153. pp. 383–406.
Johnson M.L., Koivula J.I., Eds. (1998) Gem News: Beryl from Mada- Pouchou J.L., Pichoir F. (1991) Quantitative analysis of homogeneous
gascar…and trapiche beryl. G&G, Vol. 34, No. 2, pp. 137–138. or stratified microvolumes applying the model “PAP.” In K.F.J.
Khotchanin K., Thanasuthipitak P., Thanasuthipitak T. (2010) Heinrich and D.E. Newbury, Eds., Electron Probe Quantitation,
Characteristics of trapiche blue sapphire from Southern Viet- Springer Science-Business Media, New York, pp. 31–76.
nam. Chiang Mai Journal of Science, Vol. 37, No. 1, pp. 64–73. Rakovan J., Kitamura M., Tumada O. (2006). Sakura ishi (cherry
Kiefert L. (2012) Gem News International: Unusual trapiche sap- blossom stones): Mica pseudomorphs of complex cordierite-
phire. G&G, Vol. 48, No. 3, p. 229. indialite intergrowths from Kameoka, Kyoto Prefecture,
Koivula J.I. (2008) Gem News International: Two unusual aqua- Japan. Rocks and Minerals, Vol. 81, No. 4, pp. 284–292,
marines. G&G, Vol. 44, No. 3, pp. 275–276. http://dx.doi.org/10.3200/RMIN.81.4.284-292
Koivula J.I., Kammerling R.C., Fritsch E., Eds. (1994) Gem News: Rice A.H.N. (1993) Textural and twin sector-zoning and displace-
“Trapiche” purple-pink sapphire. G&G, Vol. 30, No. 3, p. 197. ment of graphite in chiastolite and -pyralspite and grandite gar-
Kozlowski A., Metz P., Jaramillo H.A.E. (1988) Emeralds from nets in the variscides of South-West England. Read at the
Somondoco, Colombia: chemical composition, fluid inclusion Annual Conference of the Ussher Society, pp. 129–131.
and origin. Neues Jahrbuch für Mineralogie Abhandlungen, Vol. ——— (2007) Chemical disequilibrium during garnet growth:
59, pp. 23–49. Monte Carlo simulations of natural morphologies: Comment.
Lang A.R. (1959) The projection topograph: a new method in X- Geology, Vol. 35, No. 1, pp. 124–125, http://dx.doi.org/10.1130/
ray diffraction microradiography. Acta Crystallographica, Vol. G23441C.1
12, No. 3, pp. 249–250, doi:10.1107/S0365110X59000706. Rice A.H.N., Mitchell J.I. (1991) Porphyroblast textural sector-zon-
Laumonier B., Branquet Y., Cheilletz A., Giuliani G., Rueda F. ing and matrix displacement. Mineralogical Magazine, Vol. 55,

No. 380, pp. 379–396, http://dx.doi.org/10.1180/minmag. Sunagawa I., Bernhardt H.J., Schmetzer K. (1999) Texture formation
1991.055.380.08 and element partitioning in trapiche ruby. Journal of Crystal
Ringsrud R. (2013) Emeralds: A Passionate Guide. GVP Publish- Growth, Vol. 206, No. 4, pp. 322–330, http://dx.doi.org/10.1016/
ing, Oxnard, CA. S0022-0248(99)00331-0
Rodríguez E.M., Ulloa C.M. (1994) Plancha 189-La Palma, Bogotá, Tempesta G., Scandale E., Agrosì G. (2011) Striations and hollow
Ingeominas, scale 1:1000000. channels in rounded beryl crystals. Periodico di Mineralogia,
Roedder E. (1963) Studies of fluid inclusions II: freezing data and Vol. 79, No.1, pp. 75–87.
their interpretation. Economic Geology, Vol. 58, No. 2, pp. 163– Touray J.C., Poirot J.P. (1968) Observations sur les inclusions flu-
211, http://dx.doi.org/10.2113/gsecongeo.58.2.167 ides primaires de l’émeraude et leurs relations avec les inclu-
Roedder E. (1984) Fluid Inclusions. Mineralogical Society of Amer- sions solides. Comptes Rendus de l’Académie des Sciences,
ica. Reviews in Mineralogy, Vol. 12, BookCrafters Publisher, Paris, Sér. D, Vol. 266, pp. 305–308 (in French).
Chelsea, MI. Tripp E.J., Hernandez L.H. (1970) The complete trapiche emerald
Rondeau B., Fritsch E., Peucat J.-J., Nordrum F.S., Groat L. (2008) picture. The cutting of gems from trapiche crystals. Lapidary
Characterization of emeralds from a historical deposit: Byrud Journal, Vol. 24, No. 1, pp. 97–104.
(Eidsvoll), Norway. G&G, Vol. 44, No. 2, pp. 108–122, Tsuchiyama A., Uesugi K., Nakano T., Ikeda S. (2005) Quantitative
http://dx.doi.org/10.5741/GEMS.44.2.108 evaluation of attenuation contrast of X-ray computed tomog-
Scandale E., Zarka A. (1982) Sur l’origine des canaux dans les raphy images using monochromatized beams. American Min-
cristaux. Journal of Applied Crystallography, Vol. 15, No. 4, eralogist, Vol. 90, No. 1, pp. 132–142, http://dx.doi.org/10.2138/
pp. 417–422, http://dx.doi.org/10.1107/S0021889882012291. am.2005.1552
Scheibe R. (1926) Die Smaragdlagerstätte von Muzo (Kolombien) Ulloa C. (1980) Ambiente geológico de los yacimientos esmer-
und ihre nähere Umberbung. Neues Jahrbuch für Mineralogie aldíferos en Colombia. Boletin Sociedad Geológica Perú, Vol.
Abhandlungen, Vol. 54, pp. 419–447 (in German). 65, pp. 157–170.
Scheibe R. (1933) Informe geológico sobre la mina de esmeraldas Vuillet P., Giuliani G., Fischer J.C., Chiappero P.J. (2002) Emeraudes
de Muzo. Compilación de los Estudios Geológicos Oficiales en de Gachala, Colombie: historique, genèse et découvertes paléon-
Colombia, No. 1, pp. 169–198 (in Spanish). tologiques. Le Règne Minéral, No. 46, pp. 5–18 (in French).
Schiffman C.A. (1968) Unusual emeralds. Journal of Gemmology, Ward F. (1993) Emeralds. Gem Book Publishers, Bethesda, MD.
Vol. 11, No. 4, pp. 105–114. White W.B., Masako M., Linnehan D.G., Furukawa T., Chan-
Schmetzer K., Hänni H.A., Bernhardt H.J., Schwarz D. (1996) drasekhar B.K. (1986) Absorption and luminescence of Fe3+ in
Trapiche rubies. G&G, Vol. 32, No. 4, pp. 242–250, single-crystal orthoclase. American Mineralogist, Vol. 71, Nos.
http://dx.doi.org/10.5741/GEMS.32.4.242 11–12, pp. 1415–1419.
Schmetzer K., Beili Z., Yan G., Bernhardt H.J., Hänni H.A. (1998) Wilbur D.E., Ague J.J. (2006) Chemical disequilibrium during garnet
Element mapping of trapiche rubies. Journal of Gemmology, growth: Monte Carlo simulations of natural crystal morpholo-
Vol. 26, No. 5, pp. 289–301. gies. Geology, Vol. 34, No. 8, pp. 689–692, http://dx.doi.org/
Schmetzer K., Bernhardt H.J., Hainschwang T. (2011) Chemical and 10.1130/G22483.1
growth zoning in trapiche tourmaline from Zambia – a re-evo- Win K.K. (2005) Trapiche of Myanmar. Australian Gemmologist,
lution. Journal of Gemmology, Vol. 32, No. 5-8, pp. 151–173. Vol. 22, No. 6, pp. 269–270.
Schwarz D., Schmetzer, K. (2002) The definition of emerald: the Wood D.L., Nassau K. (1968) Characterization of beryl and emer-
green variety of beryl colored by chromium and/or vanadium. ald by visible and infrared absorption spectroscopy. American
In Emeralds of the World, ExtraLapis English No. 2: The Leg- Mineralogist, Vol. 53, Nos. 5–6, pp. 777–800.
endary green beryl. Lapis International, LLC, East Hampton, Yardley B.W.D. (1974) Porphyroblasts and “crystallization force”:
CT, pp. 74–78. Discussion of some theoretical considerations. Geological So-
Sunagawa I. (1987) Morphology of minerals. In I. Sunagawa, Ed., ciety of America Bulletin, Vol. 85, No. 1, pp. 61–62,
Morphology of Crystals, Part B. Terra Scientific Publishing http://dx.doi.org/10.1130/0016-7606(1974)85<61:PACFDO>
Company, Tokyo, pp. 509–587. 2.0.CO;2
Sunagawa I. (1999) Growth and morphology of crystals. Forma, Zwaan J.C., Jacob D.E., Häger T., Cavalcanti Neto M.T.O., Kanis
Vol. 14, pp. 147–166. J. (2012) Emeralds from the Fazenda Bonfim region, Rio Grande
Sunagawa I. (2005) Crystals: Growth, Morphology and Perfection. do Norte, Brazil. G&G, Vol. 48, No. 1, pp. 2–17,
Cambridge University Press, Cambridge, UK. http://dx.doi.org/10.5741/GEMS.48.1.2

FEATURE ARTICLES
LARGE COLORLESS HPHT-GROWN

SYNTHETIC GEM DIAMONDS FROM
NEW DIAMOND TECHNOLOGY, RUSSIA
Ulrika F.S. D’Haenens-Johansson, Andrey Katrusha, Kyaw Soe moe, Paul Johnson, and Wuyi Wang
The Russian company New Diamond Technology is producing colorless and near-colorless HPHT-grown
synthetic diamonds for the gem trade. Forty-four faceted samples synthesized using modified cubic
presses were analyzed using a combination of spectroscopic and gemological techniques to characterize
the quality of the material and determine the means of distinguishing them from natural, treated, and
alternative laboratory-grown diamonds. These samples, with weights ranging from 0.20 to 5.11 ct, had
color grades from D to K and clarity grades from IF to I2. Importantly, 89% were classified as colorless
(D–F), demonstrating that HPHT growth methods can be used to routinely achieve these color grades.
Infrared absorption analysis showed that all were either type IIa or weak type IIb, and photoluminescence
spectroscopy revealed that they contained Ni-, Si-, or N-related defects. Their fluorescence and phos-
phorescence behavior was investigated using ultraviolet excitation from a long-wave/short-wave UV
lamp, a DiamondView instrument, and a phosphorescence spectrometer. Key features that reveal the
samples’ HPHT synthetic origin are described.
I
n the last few years, available colorless and near- the absence of color), clarities, and sizes. Addition-
colorless gem-quality synthetic diamonds have ally, producers must be able to manufacture a high
improved dramatically with advances in growth enough volume to satisfy demand at a price point
technologies and a heightened understanding of the that makes them competitive with natural diamond
underlying processes. Although the majority of these and alternative diamond products.
products have been grown using the well-established New Diamond Technology (NDT), which to-
chemical vapor deposition (CVD) technique (see gether with Inreal and Nevsky Brilliant is part of a
Martineau et al., 2004; Wang et al., 2007; 2012), the Russian diamond group, has grown synthetic dia-
high-pressure, high-temperature (HPHT) method, as- monds for industrial and technological uses. In the
sociated more with attractive fancy color samples, past year, the company has produced large high-qual-
has recently emerged as a method for producing siz- ity, colorless samples for the gem trade. NDT says it
able colorless and near-colorless synthetic diamonds has developed HPHT technology that enables the
(D’Haenens-Johansson et al., 2014). The commercial growth of multiple synthetic diamonds in different
viability of synthetic diamonds depends on a variety reaction layers within the same cell, resulting in
of factors, such as desirable colors (or, in this case, much larger production runs compared to other
known methods, reaching up to 200 samples (with
cross-sectional sizes of 2.5–3.0 mm) across two layers
GemS & GemoloGy, Vol. 51, No. 3, pp. 260–279,
in a single run. At present, NDT can simultaneously
http://dx.doi.org/10.5741/GemS.51.3.260. grow up to 16 colorless “rough” crystals approaching
© 2015 Gemological Institute of America 10 ct apiece.
260 LARGE COLORLESS HPHT SYNTHETIC DIAMONDS GEMS & GEMOLOGY FALL 2015
A B
Figure 1. A: These three colorless HPHT-grown synthetic diamonds demonstrate some of the sizes achieved by
New Diamond Technology. From left to right, samples NDT04, NDT02, and NDT01 weigh 2.02, 2.20, and 2.30 ct,
respectively. Photo by Ulrika D’Haenens-Johansson. B: Sample NDT-A, a 5.11 ct specimen with K color, is the
largest faintly colored synthetic diamond studied by GIA to date. The visual characteristics of these samples are
summarized in table 1. Photo courtesy of New Diamond Technology.
It was reported in May 2015 that NDT had pro- BACKGROUND ON

duced a 32.26 ct “rough” synthetic diamond, which HPHT SYNTHETIC DIAMOND GROWTH
was subsequently faceted into a 10.02 ct square-cut Most HPHT growth of single-crystal diamond, irre-
emerald shape and graded by the IGI Hong Kong lab- spective of the press design, is based on the tempera-
oratory as E color and VS1 clarity (International ture-gradient method first developed by Bovenkerk et
Gemological Institute, 2015). This is believed to be al. (1959) for General Electric. The growth capsule is
the largest colorless synthetic diamond to date from filled with the ingredients necessary for laboratory di-
either HPHT or CVD growth. New Diamond Tech- amond growth: a source of carbon (such as graphite),
nology’s crystals have the potential to be faceted and a sacrificial diamond “seed” that acts as a template
polished into gems that are significantly larger than for diamond formation, and a metallic solvent/cata-
those currently available from other HPHT synthetic lyst (usually Fe, Ni, Co, or their alloys). The catalyst
producers such as AOTC, which limit their polished enables diamond growth at lower temperatures and
sizes to less than 1 ct, usually about 0.5 ct (D’Hae- pressures than would be otherwise possible, alleviat-
nens-Johansson et al., 2014; AOTC, 2015). ing some of the technological requirements. The cap-
In October 2014, NDT loaned GIA a suite of 44 sule is then exposed to pressures of 5–6 GPa and
colorless to faintly colored polished gem-quality temperatures of 1300–1600°C. The design of the sys-
HPHT-grown synthetic diamonds ranging from 0.20 tem creates a temperature gradient, with the source
to 5.11 ct, representative of their 2014 production carbon area hotter than the area where the seed is lo-
(see figure 1). The largest of these was reported in cated. Consequently, the carbon is dissolved into the
February 2015 by JCK magazine and by GIA re- hot molten metal and transferred into the cooler re-
searchers the following month (Bates, 2015; Poon et gion, where it recrystallizes in the form of synthetic
al., 2015). The 44 samples were comprehensively in- diamond on the seed.
vestigated using both gemological and spectroscopic The main focus of HPHT technology develop-
techniques. Based on the results presented in this ment has been to create systems that can reliably
study, NDT’s lab-grown samples, though often in- produce large volumes of synthetic diamonds with
cluded, achieved excellent colors and could visually desirable properties for specific technological and in-
match high-quality natural diamonds. Nevertheless, dustrial needs. Although HPHT diamond synthesis
the samples could be readily separated from natural was first achieved in the 1950s, it was only in the
stones by characteristics that arise from their artifi- 1990s that certain manufacturers started growing
cial growth conditions. samples for the gem and jewelry trade (e.g., Shigley
LARGE COLORLESS HPHT SYNTHETIC DIAMONDS GEMS & GEMOLOGY FALL 2015 261
et al., 1997). Since color-producing dopants such as
cubic
nitrogen (yellow) or boron (blue) are so easily intro- sample cell
duced, these were typically fancy-color synthetics.
Creating colorless HPHT synthetics has been sig-
nificantly more challenging, as costly modifications
to the capsule design are necessary and additional
constraints must be placed on the chemical compo-
nents to minimize dopant uptake, particularly nitro-
gen. Eliminating contamination in the growth cell is
impossible, as trace amounts of nitrogen will likely
remain. Thus, growers have to adapt the growth
chemistry to intentionally include elements with a
strong affinity to nitrogen, termed “nitrogen getters,”
effectively trapping the nitrogen so it does not dis-
perse through the diamond lattice. Finally, growth
rates for high-purity colorless diamond (type IIa or
anvil
weak type IIb) are significantly lower than for stan-
dard type Ib (containing isolated nitrogen) synthetic
diamond, necessitating longer growth times and
greater control over the temperature and pressure
conditions. Consequently, successful near-colorless
diamond synthesis depends on the careful design of
the HPHT press and its components, the quality and
chemical composition of the solvent/catalyst melt, Figure 2. The anvil arrangement and sample cell for a
standard cubic press. Illustration by M. Toosi and
U. D’Haenens-Johansson, adapted from Sumiya
(2009).
In Brief
• The Russian company New Diamond Technology
(NDT) is producing faceted colorless and near-colorless systems designed to stabilize the thermodynamic pa-
HPHT-grown synthetic diamonds for the gem trade. rameters (pressure and temperature distribution
• eighty-nine percent of the 44 faceted samples, weigh- across the sample cell). Figure 2 illustrates the main
ing up to 5.11 ct, were classified as colorless. Color components of a standard cubic press, which consists
and clarity grades ranged from D to K and IF to I2. of six anvils coupled to independent hydraulic cylin-
• The quality of NDT-produced material is on par with, ders surrounding a cubic sample cell. Although this
or surpasses, that by other synthetic diamond growers. design is complicated by the requirement for syn-
• The samples were conclusively identified as HPHT chronized motion of separate anvils, it does allow the
synthetics using a combination of gemological obser- application of hydrostatic pressure across a relatively
vations and spectroscopic analysis. large volume cell compared to other press types.
When selecting materials for the cell components
of a multi-anvil apparatus, such as a cubic HPHT
press, it is necessary to consider their ability to with-
the choice of nitrogen getters, and the ability to con- stand the high-pressure and high-temperature condi-
trol the chemical and thermodynamic conditions tions necessary to synthesize diamond with the
across the reaction layer. Further complications re- desired properties. This will depend on both the com-
sult from the need for high crystal quality and mini- pressive strength of the materials and component
mal inclusion uptake. geometries. Generally, the choice of materials is lim-
ited to either complex ceramics (Al, Zr, or Mg oxides)
HPHT GROWTH CONSIDERATIONS BY or salt-based materials (iodides and chlorides) (Strong,
NEW DIAMOND TECHNOLOGY 1977; Satoh et al., 2000; Zhu et al., 2012). Their prop-
NDT uses modified cubic HPHT presses to grow erties, coupled with the press design, determine the
gem-quality colorless synthetics, with custom-made producer’s ability to apply long-lasting hydrostatic
A B
Figure 3. A: In the multi-seed diamond growth layout on the left, the distribution of the temperature field was ide-
ally controlled across the reaction layer, resulting in even growth sizes for all the samples. B: In the growth layout
on the right, the temperature distribution was not optimized. The crystal edge sizes are less than 2 mm. Photos
courtesy of New Diamond Technology.
pressure to the sample cell. If the designs for the cell producing higher-quality crystals with more uniform
components are not optimized, they may deteriorate size.
during the HPHT run, hindering the stability of the Care must also be taken when selecting the sol-
growth parameters and in turn the synthesis quality. vent/catalyst constituents contained within the
In the worst case, there may be catastrophic failure, growth cell. The chemical properties of the interac-
with a component breaking. This may lead to costly tion boundary at the surface of the growing diamond
damage to the equipment in addition to the cost as- crystal can determine the crystal’s morphology, crys-
sociated with an unsuccessful growth run. talline quality, impurity concentrations, and propen-
Another prerequisite for successful diamond sity to trap inclusions. Iron-based solvents, which
growth is the precise control of the carbon concentra- include Ni and/or Co, are often used for HPHT dia-
tion field, which is determined mostly by the temper- mond growth (Strong and Chrenko, 1971). For color-
ature distribution field within the reaction volume of less diamond growth, it is important to suppress the
the cell (liquid solvent volume). Computer simula- introduction of boron and nitrogen impurities, which
tion methods such as FEM (the finite element generally form color-producing substitutional defects.
method) are widely used to model the carbon concen- Traces of boron can be present in either the carbon
tration field, which defines the carbon flow direction, source or the solvent/catalyst, yet careful selection of
at different thermodynamic conditions as well as for high-purity materials (with boron concentrations less
different growth cell designs (Zhan-Chang et al., than 0.1 ppm) may reduce the boron uptake to the ex-
2013). It is important to note that the correct predic- tent that it no longer produces a discernible blue color
tion of material properties inside the reaction volume, (Sumiya and Satoh, 1996). Most diamonds will readily
in particular for composites, is greatly complicated by incorporate nitrogen present within the growth cell,
the high pressures and temperatures they are sub- which can originate from the carbon source, the sol-
jected to. Correct control of the temperature distribu- vent/catalyst, and gas found in empty spaces or pores
tion (isotherms) inside the cell is also important, of the capsule. This results in a high concentration of
especially for layouts containing multiple seeds isolated nitrogen defects, which impart a yellow
and/or growth layers. Examples of two separate mul- color. To produce a colorless diamond, nitrogen get-
tiple-seed growth layouts tested by NDT are shown ters such as Al, Ti, Zr, or Hf are used (Sumiya and
in figure 3. These trials demonstrated that the Satoh, 1996). These additives will also influence the
arrangement in figure 3A resulted in a more even growth properties of the material.
temperature distribution across the reaction volume, Using an Fe-based solvent/catalyst with the addi-
Figure 4. As-grown
HPHT diamond single
crystals produced using
a simple Fe-based sol-
vent with Al as a nitro-
gen getter. The sample
on the far right is the
largest, weighing 1.40 ct.
Photo courtesy of New
Diamond Technology.
tion of only Al results in crystal growth dominated pure constituent metals for the solvent/catalyst may
by (100) and (111) faces, as shown in figure 4. While be challenging, necessitating additional purification
these are simple to facet into gems, NDT says the steps by the manufacturer. The identities and incor-
color grade of material produced through these meth- poration methods for the nitrogen getters and other
ods is generally limited to G or poorer. Conversely, additives vary for different manufacturers and re-
adding Co and Ti/Zr to the growth medium stimu- search teams, and the ultimate mix of solvent/cata-
lates the formation of high-index growth planes such lyst and getter is often proprietary. NDT claims to
as (113), (110), and (115), yet this material benefits have determined the optimal balance of chemical
from being colorless (D–F color grades). Unfortu- constituents needed to routinely produce gem-qual-
nately, the nitrogen getters may react with the car- ity colorless synthetic diamonds, divulging that their
bon source to form carbides, which affect the carbon solvent is based on Fe, Co, or a combination of the
transport and diffusion properties of the system, de- two.
creasing the carbon available for crystal growth. Orientation of the seed initiation surface also sig-
These conditions may result in trapped traces of the nificantly affects the morphology of the grown crys-
metal/catalyst melt, forming visible inclusions. To tals (figure 5). Using differently oriented seeds—with
suppress the development of such inclusions, it may (110)- or (113)-oriented initiation surfaces, for exam-
be necessary to (1) introduce additional chemicals ple—may optimize the growth morphology. To grow
such as Cu or (2) decompose the carbides otherwise a crystal whose top face is (110)- or (113)-oriented, it
created by the nitrogen getters (Sumiya and Satoh, is not strictly necessary to have a seed with a simi-
1996; Sumiya et al., 2002). Obtaining sufficiently larly oriented face, only to have the [110] or [113] di-
Figure 5. The morphology of synthetic diamonds depends largely on the orientation of the seed surface. Photo A
shows a reaction layer where the sample in the center and the two outer rings of samples were grown on (100)-
oriented seeds. Meanwhile, the samples with triangular outlines (the two inner rings) were grown using (111)-ori-
ented surfaces. The samples shown here are yellow due to the presence of isolated nitrogen. The blue samples
shown in (B) and (C) are boron-doped and were grown on (100)- and (111)-oriented substrates, respectively. The
samples shown in (B) weigh 0.92 and 0.23 ct, while those in (C) weigh 0.29 and 0.88 ct. Photos courtesy of New
Diamond Technology.
A B C
rection vertical in the growth chamber. Since seed NDT-B (4.30 ct), underwent fewer tests than the rest
crystals are typically cuboctahedral, however, it is of the samples, as specified below.
easier to orient them in the chamber with either a Color and clarity grading was performed by GIA
(100) or (111) face on the top. Consequently, most using standard diamond grading nomenclature. Sam-
HPHT diamond growth is conducted on (100)- or ples and their internal features were further exam-
(111)-oriented seeds. NDT generally uses the former, ined using a standard Gem Instruments binocular
which produce crystals with higher cutting yields microscope system and a Nikon SMZ1500 research
and more easily polished surfaces. microscope under darkfield and fiber-optic illumina-
The growth rate of individual diamond crystals tion. The presence of anomalous birefringence pat-
depends on a combination of parameters: the identi- terns, attributed to strain in the diamond lattice, was
ties and ratios of the chemical constituents used for evaluated by viewing each sample between crossed
growth; the thermodynamic conditions; the seed polarizing filters using brightfield illumination.
quantity, size, and orientation; the carbon flow rate; The specimens’ fluorescence and phosphorescence
and temperature gradient. The quantity of the seeds response to ultraviolet (UV) light from a conventional
in a run can greatly impact the growth rate of indi- four-watt combination long-wave (365 nm) and short-
vidual crystals if no additional changes are made to wave (254 nm) lamp was tested in a dark room. The
the cell, with an increase in the number of crystals emission intensities were visually compared to the
leading to a slower rate. The latter parameter is ad- fluorescence of a GIA fluorescence master set. Fluo-
justed by modifying the growth cell layout. When rescence and phosphorescence imaging was con-
evaluating growth rates quoted by manufacturers, it ducted using a DTC DiamondView instrument
is important to note that the rate increases nonlin- (illumination wavelengths <230 nm), revealing the
early with crystal mass. As the crystal grows, the sur- samples’ internal growth structures. An in-house cus-
face area (and hence the deposition area) increases. tom-built phosphorescence spectrometer consisting
Yet the deposition rate per unit area (mm/hour) is ap- of an Ocean Optics HR4000 spectrometer coupled by
proximately constant. Consequently, the crystal fiber optics to an Avantes-DH-S deuterium-halogen
growth rate (mg/hour) for a given area will rise (Sumiya light source (operating with only the deuterium light,
et al., 2005). Colorless type IIa diamond growth rates 215–400 nm) was used to further investigate the sam-
as high as 8–10 mg/hour can be achieved using ples’ phosphorescence at room temperature (see
HPHT technology, though such material is generally Eaton-Magaña et al., 2007, 2008). Consecutive spec-
plagued by metallic inclusions (Sumiya et al., 2005). tra, each with a one-second integration time, were
Limiting the maximum stable growth rate to 6–7 collected for up to five minutes following 30 seconds
mg/hour has been reported to produce higher-quality of illumination. Samples NDT-A and NDT-B were in-
growth (Sumiya et al., 2005). vestigated with the UV lamp and the DiamondView,
NDT systems can support growth for 200–250 but phosphorescence spectra were not recorded for
hours, simultaneously yielding up to 16 crystals them.
weighing up to 10 ct each. Further modifications To gain an understanding of the impurities pres-
have enabled growth rates up to 30–50 mg/hour—up ent in the samples, Fourier-transform infrared (FTIR)
to five times higher than commonly reported— absorption spectra covering the 400–6000 cm–1 range
though maintaining this rate for extended periods is were taken using a Thermo Nicolet Nexus 6700
challenging due to the higher pressure and tempera- spectrometer furnished with KBr and quartz beam
ture conditions required. splitters and a diffuse-reflectance infrared Fourier
transform (DRIFT) accessory. The instrument and
MATERIALS AND METHODS sample chambers were purged with dried air to min-
For this investigation, NDT loaned GIA 44 colorless imize absorption features stemming from atmos-
to faintly colored HPHT-grown specimens. These pheric water. Spectra were normalized based on the
samples had been faceted into round brilliant (26) or height of the two-phonon absorption in diamond
fancy shapes (18) and weighed 0.20–5.11 ct (table 1; (Palik, 1985), enabling us to calculate the absorption
see again figure 1). According to NDT, these samples coefficient for comparison and quantitative impurity
were grown separately using cubic presses in single- concentration analysis without having to know the
layer runs, and are representative of their 2014 pro- light path length through the samples.
duction methods. Because of limited access times, Supplementary analysis of impurity content was
the two largest samples, labeled NDT-A (5.11 ct) and provided by photoluminescence (PL) spectra acquired
TABLE 1. Gemological properties and calculated bulk concentrations of neutral boron (B0) impurities for 44
HPHT synthetic diamonds produced by New Diamond Technology.
Sample Carat Color Clarity Cut Dimensions (mm) Type B0 bulk concentration
weight (ct) (ppb)
NDT-A 5.11 K I1 Cut-corner rectangu- 10.05 × 8.74 × 7.01 IIb 1.2 ± 0.2
lar modified brilliant
NDT-B 4.30 D SI1 Cushion 10.29 × 9.76 × 5.39 IIb 7±1
NDT01 2.30 e SI1 Cushion 7.30 × 8.48 × 4.85 IIa nd a
NDT02 2.20 e VVS2 Heart 8.53 × 7.55 × 5.41 IIb 20 ± 3
NDT03 2.03 F SI2 Heart 7.73 × 8.41 × 5.07 IIb 5±1
NDT04 2.02 D I1 Pear 7.14 × 10.80 × 4.45 IIb 6±1
NDT05 1.90 F VS1 octagonal 7.65 × 7.68 × 4.58 IIa nd
NDT06 1.53 D SI1 Round 7.18 × 7.26 × 4.60 IIb 2.0 ± 0.5
NDT07 1.53 D I1 Round 6.49 × 9.33 × 4.08 IIb 13 ± 2
NDT08 1.50 e SI1 Round 7.14 × 7.17 × 4.44 IIb 2.6 ± 0.5
NDT09 1.41 D VS2 Square emerald 6.25 × 6.49 × 4.26 IIa nd
NDT10 1.20 e SI2 Round 6.73 × 6.75 × 4.23 IIb 2.5 ± 0.5
NDT11 1.13 D IF Round 6.57 × 6.62 × 4.10 IIb 0.8 ± 0.2
NDT12 1.04 e SI2 oval 5.58 × 7.92 × 3.40 IIb 13 ± 2
NDT13 1.03 D SI2 Round 6.39 × 6.47 × 3.99 IIb 9±1
NDT14 1.02 F SI1 Round 6.45 × 6.49 × 4.06 IIb 58 ± 9
NDT15 1.01 D VS2 oval 5.64 × 7.79 × 3.35 IIb 0.5 ± 0.3
NDT16 1.00 G I2 Round 6.50 × 6.53 × 3.79 IIb 6±1
NDT17 0.92 e VS2 Round 6.04 × 6.07 × 3.96 IIb 1.6 ± 0.5
NDT18 0.90 D VS1 Pear 5.38 × 7.78 × 3.38 IIa nd
NDT19 0.90 D SI1 Round 6.15 × 6.18 × 3.78 IIb 5±1
NDT20 0.81 D VVS2 Round 5.92 × 5.96 × 3.79 IIb 2.5 ± 0.5
NDT21 0.80 D VVS2 Round 6.07 × 6.09 × 3.57 IIa nd
NDT22 0.80 D SI2 Pear 5.34 × 7.78 × 3.17 IIb 9±1
NDT23 0.71 D VS1 Round 5.75 × 5.80 × 3.42 IIb 8±1
NDT24 0.71 D VVS1 Pear 5.08 × 7.29 × 2.99 IIb 6±1
NDT25 0.64 e SI2 Round 5.43 × 5.46 × 3.53 IIa nd
NDT26 0.58 H I1 Round 5.43 × 5.46 × 3.53 IIb 2.5 ± 0.5
NDT27 0.58 e SI2 oval 4.89 × 6.49 × 2.75 IIa nd
NDT28 0.54 e SI1 Pear 4.62 × 6.63 × 2.92 IIb 13 ± 2
NDT29 0.52 e VS2 Round 5.09 × 5.11 × 3.31 IIb 9±1
NDT30 0.51 D VVS1 Round 5.15 × 5.17 × 3.90 IIb 6±1
NDT31 0.50 e SI2 Round 5.08 × 5.12 × 3.10 IIa nd
NDT32 0.48 D VVS2 Round 5.11 × 5.14 × 3.06 IIa nd
NDT33 0.47 H IF emerald 3.93 × 4.71 × 2.67 IIb 33 ± 5
NDT34 0.43 F I1 Round 4.78 × 4.80 × 2.99 IIb 3.7 ± 0.5
NDT35 0.43 D VS1 Pear 4.38 × 6.22 × 2.66 IIb 20 ± 3
NDT36 0.38 F I1 Round 4.70 × 4.73 × 2.83 IIb 1.8 ± 0.5
NDT37 0.37 D I1 marquise 3.90 × 7.64 × 2.24 IIb 25 ± 4
NDT38 0.31 e VS2 Round 4.35 × 4.37 × 2.63 IIb 18 ± 3
NDT39 0.30 e I2 Round 4.34 × 4.36 × 2.59 IIb 1.6 ± 0.5
NDT40 0.28 F VVS2 Round 4.15 × 4.17 × 2.60 IIb 13 ± 2
NDT41 0.24 D SI1 Round 4.09 × 4.11 × 2.41 IIb 3.3 ± 0.5
NDT42 0.20 G VS1 Round 3.70 × 3.69 × 2.33 IIb 33 ± 5
a
“nd” indicates that boron was not detected.
with the samples submerged in liquid nitrogen (77 K, ing six different excitation wavelengths: He-Cd metal
or –196°C) using a Renishaw InVia Raman confocal vapor (324.8 nm), Ar-ion (457.0, 488.0 and 514.5 nm),
microspectrometer (Hall et al., 2010). To maximize He-Ne (632.8 nm), and a diode laser (830.0 nm). Spec-
the effectiveness of stimulating different luminescent tra with the 324.8 and 488.0 nm excitations were not
defect centers across a wide range of wavelengths, the collected for samples NDT-A and NDT-B, because of
system was equipped with four laser sources produc- limited access to the instrument.
RESULTS AND ANALYSIS clarity grades, marking a major breakthrough in lab-
Visual Appearance. The grading characteristics of oratory growth.
NDT’s faceted HPHT-grown diamonds are presented Poorer clarity grades generally stemmed from the
in table 1. The samples had carat weights ranging presence of inclusions, which appeared dark and
from 0.20 to 5.11 ct, with average and median opaque in transmitted light and gray and metallic in
weights of 1.08 ct and 0.81 ct. There was no clear reflected light. These inclusions took a variety of
trend between the cut shapes and sample weights, shapes (as demonstrated in figure 6), sometimes
though it was noted that the six stones weighing within the same sample. Rod-like inclusions were
over 2 ct were all faceted into fancy shapes. Thirty- noted to have lengths of up to approximately 1.5
nine (89%) of the samples were found to be colorless mm, and could be observed singly or in clusters with
(D–F), while four attained near-colorless (G or H) the inclusions aligned in one direction. Three of the
grades and the remaining (and largest) sample was a synthetics contained flat plate inclusions with ap-
faintly colored K grade. Remarkably, five of the six proximately triangular geometry, with thicknesses
2+ ct stones were colorless, and the color zoning of about 30 µm and edge dimensions up to approxi-
often seen in HPHT-grown diamonds was not appar- mately 600 µm. Some of the metallic inclusions were
ent. There was a wide clarity range for this set of irregular in shape. Sufficiently included samples
samples: internally flawless (2, or 5%); very, very (VS2–I2) were attracted to a strong magnet, depending
slightly included (7, or 16%); very slightly included on the magnet’s placement relative to the position of
(10, or 23%); slightly included (16, or 36%); and in- the inclusion. The presence of feathers, occasionally
cluded (9, or 20%). No correlation was observed be- combined with metallic inclusions, also reduced the
tween sample weight, color, and clarity grade. clarity grades of some samples. The inclusions likely
NDT11 weighed 1.13 ct and achieved D color and IF result from entrapped traces of the metallic
Figure 6. Inclusions observed in some of New Diamond Technology’s HPHT synthetic diamonds had a dark metal-
lic appearance, suggesting they were remnants of the metallic solvent/catalyst melt used for growth. The inclu-
sions took a variety of morphologies, including rods (A–D), thin plates (E), and irregular shapes (F). Images A, B, C,
D, E, and F were taken for samples NDT16, NDT25, NDT13, NDT12, NDT26, and NDT37, respectively. Photomi-
crographs by Ulrika D’Haenens-Johansson.
A 500 μm B 200 μm C 200 μm
D 200 μm E 200 μm F 200 μm
1 mm 1 mm 1 mm
A B C
Figure 7. A: Unincluded sections of the HPHT synthetic samples did not show birefringence patterns when viewed
under crossed-polarized light, instead showing smooth gradations in subdued low-order interference colors (grays,
purples, and blues), indicating low levels of strain. B and C: Localized distortions in the diamond lattice were only
apparent adjacent to inclusions or feathers. These images were taken for samples NDT27, NDT13, and NDT25, re-
spectively. Photomicrographs by Ulrika D’Haenens-Johansson.
solvent/catalyst melt. Graining, an optical phenom- UV illumination, with 41 of them (93%) fluorescing
enon whereby a series of transparent, tightly spaced yellow, orangy yellow, or green-yellow, with intensi-
lines can be observed in the sample when viewed ties that were very weak (22), weak (15), medium (2),
from certain directions, has been reported in fancy- or strong (1). The remaining three samples (7%) did
color HPHT-grown synthetic diamonds (Shigley et not fluoresce. After short-wave UV illumination, 43
al., 2004) but was not detected in any of the samples (98%) of the samples, including two that did not
in this study. show fluorescence, displayed varying degrees of per-
sistent green, yellowish green, greenish yellow, yel-
Birefringence. When viewed between crossed polar-
low, or orangy yellow phosphorescence, which could
izers, the samples only showed subdued low-inter-
be observed for up to approximately six minutes.
ference colors (grays, blues, and purples) without
Only one sample, NDT30, showed neither fluores-
forming any patterns. This was in stark contrast with
cence nor phosphorescence when exposed to either
natural diamonds in general, which show mottled,
UV emission wavelength from the gem testing lamp.
banded, or cross-hatched “tatami” patterns in a
The stronger fluorescence following short-wave UV
wider range of colors. These colors, and the absence
illumination is characteristic of colorless and near-
of a clear pattern, indicate that the synthetic material
colorless HPHT synthetic diamonds, the opposite of
was characterized by low strain levels, in agreement
the behavior seen for natural stones (Crowningshield,
with previously reported observations (Crowning-
1971; Rooney et al., 1993; Shigley et al., 1997; D’Hae-
shield, 1971; Shigley et al., 1997; D’Haenens-Johans-
nens-Johansson et al., 2014).
son et al., 2014). Localized areas of higher strain were
The above band-gap energy UV source (<230 nm)
only observed adjacent to inclusions or feathers,
from a DiamondView was able to induce blue fluo-
where the diamond lattice was distorted to accom-
rescence and phosphorescence in all the samples, as
modate the clarity feature. Characteristic birefrin-
illustrated in figure 8. It was crucial to view the
gence effects are shown in figure 7.
samples along several different directions in order
Fluorescence and Phosphorescence Behavior. The to perceive the cuboctahedral growth patterns
samples’ fluorescence and phosphorescence re- characteristic of HPHT synthetic diamonds (Wel-
sponses to long-wave (365 nm) and short-wave (254 bourn et al., 1996; D’Haenens-Johansson et al.,
nm) UV illumination from a gem-testing lamp were 2014). The patterns were usually most clear when
investigated, with the results summarized in table 2. viewed along the pavilion, rather than with the
The majority (38, or 86%) were inert to long-wave sample oriented face-up. The patterns, which arose
UV light, with the remaining samples only emitting from the different impurity uptake efficiencies for
very faint yellow-orange fluorescence. The sample the separate growth sectors, showed poor contrast
set had a stronger response to short-wave (254 nm) due to the high purity levels of the samples.
TABLE 2. Fluorescence and phosphorescence behavior of the HPHT synthetics under different
excitation sources.
Phosphorescent
long-wave (365 nm) UV gem lamp Short-wave (254 nm) UV gem lamp emission following
illumination response illumination response broadband
illumination
Sample Fluorescence Phosphorescence Fluorescence Phosphorescence 500 nm 575 nm
NDT-A None None Strong yellow-green medium green-yellow N/A* N/A
✓ ×
NDT-B None None Strong yellow-green Strong green-yellow N/A N/A
✓ ×
NDT01 None None Weak green Weak yellow
✓ ✓
✓ ×
NDT03 Very weak yellow-orange Very weak yellowish orange Weak green Weak yellow
✓ ×
NDT04 None None Weak green Very weak yellow
✓ ✓
✓ ×
NDT06 Very weak yellow-orange Very weak yellowish orange Weak yellowish green Weak yellow
✓ ✓
NDT07 None Weak yellowish orange Weak green Weak yellow
✓ ×
NDT08 Very weak yellow-orange Weak yellowish orange Weak yellowish green Weak yellow
✓ ✓
✓ ×
NDT10 Very weak yellow-orange Weak green Weak green Weak yellow
✓ ✓
✓ ×
✓ ✓
✓ ×
NDT14 None Very weak yellowish orange Weak green Very weak yellow
✓ ✓
NDT15 None None Very weak yellowish green Very weak yellow
✓ ✓
NDT16 Very weak yellow-orange Very weak yellowish orange Weak yellowish green Weak yellow
✓ ×
✓ ×
✓ ✓
NDT19 Very weak yellow-orange None Weak green Very weak yellow
✓ ×
NDT20 None Very weak yellowish orange Very weak green Very weak yellow
✓ ×
✓ ×
✓ ×
NDT23 None None Very weak green Very weak yellow
✓ ×
NDT24 None Very weak yellowish orange Weak green Very weak yellow
✓ ×
✓ ×
NDT26 None None Weak yellowish green Weak yellow
✓ ✓
✓ ✓
NDT28 None Very weak yellowish orange Weak green Weak yellow
✓ ×
NDT29 None None Very weak greenish yellow Very weak orangy yellow
✓ ×
NDT30 None None Very weak yellow None
✓ ×
NDT31 None None Weak orangy yellow Very weak yellow
✓ ×
✓ ✓
NDT33 None None medium green medium yellow
✓ ×
NDT34 None None None None
✓ ×
✓ ×
✓ ×
NDT37 None Very weak yellowish orange Very weak green Very weak yellow
✓ ×
✓ ✓
NDT39 None None Very weak orangy yellow Very weak yellow
✓ ×
NDT40 None None Weak yellow Weak yellow
✓ ×
NDT41 None None Very weak greenish yellow None
NDT42 None None medium yellowish green medium yellow
*N/A: Not analyzed.
The phosphorescence behavior of samples light centered at approximately 500 nm (2.5 eV), with
NDT01–NDT42 (totaling 42) was further investi- 13 (31%) of the specimens also showing a yellow
gated by collecting phosphorescence spectra at room phosphorescence band centered at about 575 nm (2.1
temperature following broadband illumination (215– eV), as listed in table 2. Figures 9A and B demonstrate
400 nm), collecting spectra every second with 1 sec- representative spectra for samples showing phospho-
ond integration times. All the samples emitted blue rescence only from the 500 nm band (NDT37) or
A B
Figure 8. Blue fluores-
cence (A–B) and phos-
phorescence (C–D) were
detected for all the New
Diamond Technology
synthetics observed
with the DiamondView.
Most of the samples
showed cuboctahedral
growth patterns charac-
teristic for HPHT syn-
thetics, though they
C D were not always easily
observed. The fluores-
cence and phosphores-
cence (following a 0.10
second delay) images
found here were taken
for sample NDT06 with
0.12 and 1.00 second ex-
posures, respectively.
Images by Ulrika
D’Haenens-Johansson.
from both the 500 nm and 575 nm bands (NDT08). Watanabe et al., 1997). The donor for the 575 nm
Phosphorescent emission at 500 nm is commonly band is much less certain (Watanabe et al., 1997). If
observed in natural type IIb or weakly boron-contain- one assumes the same donor for both bands, the en-
ing type IIa diamonds, as well as similarly typed ergy separation between the 500 nm and 575 nm
HPHT synthetic diamonds grown using a variety of bands cannot be simply explained by a distance dis-
solvent/catalyst melts (Watanabe et al., 1997; Eaton- tribution between the donors and acceptors. Instead,
Magaña et al., 2008; Eaton-Magaña and Lu, 2011; it is possible that the two bands involve distinct deep
D’Haenens-Johansson et al., 2014), whereas the 575 donors. The donor for the 575 nm band may be a dif-
nm band has been detected for HPHT synthetic type ferent nitrogen-related center, or even a center con-
IIb diamonds grown using Co- and Ti-containing sol- sisting of an impurity atom with or without nitrogen.
vent/catalysts (Watanabe et al., 1997; Eaton-Magaña Watanabe et al. (1997) noted that the 575 nm band
et al., 2008). A band at 580 nm has been observed in was mainly emitted from regions close to the seed
some natural type IIb diamonds, yet this band is gen- crystal, where faster growth rates were suspected,
erally very weak and rapidly decaying, making its ob- suggesting a higher probability of incorporating im-
servation rare. Neither of these bands have been purities from the solvent/catalyst.
reported for phosphorescent CVD synthetics (Wang Comparison of the decay of the 500 nm and 575
et al., 2012). The 500 and 575 nm bands are induced nm bands for the NDT samples (e.g., figure 9) re-
following illumination with wavelengths less than vealed that the latter band is longer-lived, in agree-
approximately 400 nm (>3.1 eV) and 540 nm (>2.3 ment with published results (Watanabe et al., 1997).
eV), respectively, and are thought to originate from Consequently, the samples’ phosphorescence was
donor-acceptor pair recombination between boron noted to shift from blue to yellow following illumi-
acceptors and donors, where the identity of the nation with the spectrometer’s light source. Revisit-
donor(s) is still being questioned (Watanabe et al., ing the samples’ phosphorescence behavior under the
1997). The main candidate for the 500 nm band is a UV gem lamp (table 2), those that displayed the most
nitrogen-related center, possibly isolated nitrogen, as intense 575 nm bands (relative to the 500 nm band)
it is likely present in all the types of natural and syn- often exhibited yellowish orange phosphorescence
thetic diamonds for which this phosphorescence is following long-wave (365 nm) excitation. Both the
observed (Dean, 1965, 1973; Klein et al., 1995; long- and short-wave UV lamps generated suffi-
PHOSPHORESCENCE SPECTRA
A ENERGY (eV)
3.5 3 2.5 2 1.5
Figure 9. Room-temper-
5500 ature phosphorescence
5000
PHOSPHORESCENCE INTENSITY (COUNTS)
500
spectra were collected
4500
4000 at room temperature
3500 for HPHT synthetic
3000
samples NDT01–
2500
2000 NDT42. All 42 showed
1500 phosphorescence fol-
1000 lowing illumination
500
00
with broadband UV
light (215–400 nm);
40
here the spectra for the
60 representative samples
80 NDT37 (A) and NDT08
100
s)
(B) are shown. For clar-
E(
120
M
ity, data is only pre-
TI
140
160
sented for five-second
intervals, even though
400 500 600 700 800 900 1000 data was collected
WAVELENGTH (nm) every second. A phos-
phorescent emission
band was detected at
500 nm for all speci-
B ENERGY (eV) mens (A and B), with
3.5 3 2.5 2 1.5
an additional band
2500 centered about 575 nm
500 being observed for 13

2000
(31%) of the samples
1500
(B). For the cases where
both bands were de-
1000 tected, it was noted
that directly following
500
illumination, the 500
0 nm band dominated,
0
with the 575 nm band
displayed as a shoul-
der. The 575 nm band
80 had a slower decay,
100
s)
and its emission could

E(
M
be observed even after

TI
the 500 nm emission

160
had decayed beyond
400 500 600 700 800 900 1000 detection.
WAVELENGTH (nm)
ciently high energy to induce phosphorescence from be influenced by the size of the stone, and the rate of
both bands (Watanabe et al., 1997). decay for the emission band(s). Since phosphores-
The phosphorescence duration is affected by both cence spectra were collected, it was possible to de-
the maximum phosphorescence intensity, which can termine the half-life (τ) of the phosphorescence, or
PHOSPHORESCENCE DECAY
4000 500 nm band (NDT37), τ = 9.16 ± 0.06 s

Figure 10. Phosphores-
500 nm band (NDT08), τ = 8.55 ± 0.07 s cence intensity as a
575 nm band (NDT08), τ = 32.5 ± 0.3 s function of time for the

500 and 575 nm bands
3000
for samples NDT37 and
NDT08. The circle,
square, and triangle
data points were deter-
mined from the experi-
mental data presented
2000
in figure 9. The lines il-
lustrate the fits ob-
tained to these data
using least-squares fit-
ting based on the hy-
1000
perbolic function
expressed by equation
2, from which the half-
life values τ were calcu-
lated using equation 3.
0
0 20 40 60 80 100 120 140 160
TIME (s)
the time it took for the bands’ intensities to halve. where t is the time following the extinction of the il-
Importantly, the half-life is a measure that is inde- lumination, I(0) is phosphorescence intensity at t =
pendent of the initial phosphorescence intensity. The 0, and k parameterizes the phosphorescence recom-
shape of the decay curve (the maximum band inten- bination (Watanabe et al., 1997). The observation
sity as a function of time) can also provide informa- that a hyperbolic “bimolecular” curve fit better than
tion regarding the mechanism responsible for the an exponential decay is consistent with the donor-
phosphorescence. Figure 10 presents the phosphores- acceptor pair recombination mechanism that has
cence decay data for the 500 and 575 nm bands for been used to explain the phosphorescence of dia-
samples NDT08 and NDT37, selected for their in- mond (Watanabe et al., 1997). The least-squares fits
tense phosphorescence. The bands’ maximum inten- to the data, as illustrated in figure 10, provided esti-
sities were estimated by fitting Voigt functions to the mates for k. Setting I(t)/I(0) = ½ and rearranging equa-
intensity data plotted as a function of energy, where tion 2, the half-life is then defined by
energy E (in eV) and the wavelength λ (in meters) are
related by 2–1
(3)
(1)
Thus the half-life for the 500 nm phosphores-
where h is Planck’s constant (4.13566733 × 10–15 eVs) cence band for sample NDT37 (one-band emission)
and c is the speed of light (2.99792458 × 108 m/s). The was calculated to be 9.16 ± 0.06 s. In sample NDT08
phosphorescence intensity as a function of time I(t) (two-band emission), the half-life was 8.55 ± 0.07 s
was best modeled using a hyperbolic function for the 500 nm band and 32.5 ± 0.3 s for the 575 nm
band.
FTIR Absorption Spectroscopy. FTIR spectroscopy

(2)
was used to evaluate the samples’ boron- and nitro-
FTIR SPECTRA Figure 11. FTIR absorp-
tion spectroscopy re-
vealed that 35 (80%) of
35
the 44 synthetic dia-
monds contained de-
tectable concentrations
30
of neutral substitutional
boron (B0), with a char-
ABSORPTION COEFFICIENT (cm–1)
25
acteristic absorption
B0 band centered at about
2800 cm–1, and were
20 thus classified as type
IIb. Bulk B0 concentra-
tions for these synthet-
15 ics are tabulated in
table 1. The remaining
samples did not show
10 any boron- or nitrogen-
related IR absorption
and were classified as
5
type IIa. The spectra are
translated vertically for
clarity. Traces for sam-
0
ples NDT-A and NDT-B
are not shown due to
4500 4000 3500 3000 2500 2000 1500 1000
poorer signal-to-noise
WAVENUMBER (cm–1) ratio.
gen-related impurity content, with the resulting spec- show any defect-related absorption features. As these
tra shown in figure 11. For 35 (80%) of the HPHT samples also exhibited identical phosphorescence, it
synthetics, the only defect-related IR absorption seen is suggested that they contain boron centers at con-
was from neutral substitutional boron defects, which centrations below the FTIR detection limit. It is im-
create an asymmetric absorption band at approxi- portant to note that these spectra are not unique to
mately 2800 cm–1, resulting in a type IIb classifica- these samples, as colorless natural diamonds and
tion. Boron incorporation has been found to be HPHT- and CVD-grown synthetics can also be type
growth sector–dependent in HPHT synthetic dia- IIa or type IIb.
monds, with higher concentrations observed in {111}
growth sectors (Burns et al., 1990). Consequently, the Photoluminescence Spectroscopy. Photolumines-
boron distribution across these multi-sector samples cence spectra were collected for the samples im-
(as seen using DiamondView imaging) was also in- mersed in liquid nitrogen (77 K) using excitation
homogeneous. Their faceting made it impossible to wavelengths of 324.8, 457.0, 488.0, 514.5, 632.5, and
conduct FTIR absorption experiments through single 830.0 nm, enabling the detection of defects that emit
growth sectors. Nevertheless, to enable semi-quan- light from within the ultraviolet to infrared range.
titative comparison of the boron content between Data using the 324.8 and 488.0 nm lasers were not
samples, bulk boron concentrations across complete collected for samples NDT-A and NDT-B. Overall,
volumes were calculated using the integrated inten- the peaks reported here were generally weak, often
sities of the 2800 cm–1 band (Collins and Williams, detected only if the laser power was high enough to
1971; Fisher et al., 2009). These results are tabulated saturate the diamond Raman peak. Samples NDT21,
in table 1, with bulk neutral boron concentrations up NDT26, NDT33, NDT34, and NDT42, accounting
to 58 ± 9 ppb detected (NDT14, F color). The detec- for 11% of the suite, did not reveal any defect-related
tion of boron is consistent with the phosphorescent photoluminescence features.
behavior previously discussed. The remaining sam- The dominant photoluminescence features ob-
ples were assigned as type IIa because they did not served were nickel-related, with the 830.0 nm laser
PHOTOLUMINESCENCE SPECTRA
NORMALIZED PL INTENSITY (ARB. UNITS)

NORMALIZED PL INTENSITY (ARB. UNITS)
1.4
0.6
882.1 883.8 1.2 483.8
0.5
NDT30 1.0 484.1
484.4
0.4 483.6
NDT18 0.8
0.3 NDT17
NDT17 0.6
NDT11
0.2 0.4 NDT12
NDT03
NDT10 0.2
NDT40
0.1
NDT02
NDT06
NDT31
0.0 NDT40 0.0
-0.2
878 879 880 881 882 883 884 885 886 887 888
480 482 484 486 488 490 492 494
WAVELENGTH (nm) WAVELENGTH (nm)
Figure 12. Representative photoluminescence spectra Figure 13. For 27 of the 42 samples, photolumines-
taken at liquid-nitrogen temperatures using 830.0 nm cence spectra collected at liquid-nitrogen tempera-
laser excitation revealed a Ni-related defect that pro- tures using excitation from a 324.8 nm laser showed a
duces a doublet with peaks at 882.1 and 883.8 nm in multiplet with peaks at 483.6/483.8/484.1/484.4 nm.
32 (73%) of the New Diamond Technology samples. This feature is commonly thought to originate at a
This doublet is thought to originate from the same Ni-related center. For clarity, the Raman-normalized
Ni-related center that produces a doublet commonly spectra for only five representative samples are
reported at 1.4035/1.4008 eV (883.15/884.85 nm) for shown, translated vertically to avoid overlap.
synthetic diamonds grown using Ni-containing sol-
vent/catalyst melts. It is possible that the apparent
energy shift resulted from an instrument calibration
error. The 830.0 nm laser did not excite any other de-
fect-related features. Once normalized to the unsatu- tion (Nazaré et al., 1991). The separation of the peaks
rated diamond Raman peak height, the spectra have has been found to increase with increasing
been translated vertically for clarity. linewidths, suggesting that it is affected by strain lev-
els. The 1.40 eV center is active in both photolumi-
nescence and absorption, and is found exclusively in
the {111} growth sectors in HPHT synthetic diamonds
exciting a doublet at 882.1/883.8 nm (1.405/1.402 eV) grown in the presence of Ni (Collins et al., 1990;
in 32 samples (73%) and the 324.8 nm laser exciting Nazaré et al., 1991; Collins, 1992; Lawson and Kanda,
a multiplet with lines at 483.6/483.8/484.1/484.4 nm 1993). Similarly, the 483.6/483.8/484.1/484.4 nm
(2.563/2.562/2.560/2.559 eV) in 27 samples. No com- multiplet, also known as the “484 nm” or “2.56 eV”
ment can be made about whether samples NDT-A center, is commonly observed with photolumines-
and NDT-B would show the latter multiplet, as no cence in the {111} growth sectors of HPHT synthetic
data was collected with the 324.8 nm excitation. Rep- diamonds grown using Ni-based solvent/catalysts.
resentative spectra, with the peak intensities normal- Hence, it is also thought to be Ni-related, though its
ized to the intensity of the unsaturated Raman peaks, structure has yet to be determined (Dean, 1965;
are shown in figures 12 and 13. The 882.1/883.8 nm Collins et al., 1990; Collins, 1992; Nazaré et al., 1995).
doublet is thought to be the one previously reported These nickel-related features indicate the presence of
at 1.4035/1.4008 eV (883.15/884.85 nm), which is Ni in the solvent/catalyst melt used by NDT, inten-
often referred to as the “1.40 eV center” (e.g., Nazaré tional or otherwise.
et al., 1991; Yelisseyev and Kanda, 2007) where the Nitrogen-related features were only observed in
shift in wavelength may be related to an instrument the form of nitrogen-vacancy centers in the neutral
calibration error. The model for the 1.40 eV center is (NV0) and negative (NV–) charge states with zero-
an interstitial Ni+ atom distorted along a 〈111〉 direc- phonon lines (ZPLs) at 575 (2.156 eV, not to be con-
fused with the phosphorescence band) and 637 nm
(1.945 eV), respectively. These could be detected
using either the 488.0 nm (not shown) or 514.5 nm
lasers; figure 14 shows representative data using the
PL INTENSITY (ARB. UNITS)

736.9
latter excitation source. The features were weak, gen- 736.6
erally detectable only when the laser power was high NDT18
enough to saturate the diamond Raman peak. The
NV0 ZPL, often more intense than that for NV–, was NDT17
detected in 26 (59%) of the samples. The NV– ZPL

NDT03
was only seen in 11 (25%) samples. These features
are common in both synthetic (type IIa and IIb) and NDT05
natural (type IIa) diamonds. A feature at 503.2 nm NDT08
(2.463 eV), possibly the H3 (N-V-N0) defect, was not

detected in any of the NDT synthetics using either
457.0 or 488.0 nm lasers. This feature was previously 734 735 736 737 738 739
observed in near-colorless HPHT synthetics pro- WAVELENGTH (nm)

duced by AOTC (D’Haenens-Johansson et al., 2014).
The only remaining defect-related photolumines-
cence feature observed was the negative charge state Figure 15. The negatively charged silicon-vacancy
center, SiV –, with emission at 736.6 and 736.9 nm,
was weakly detected in 18 (41%) of the New Dia-
Figure 14. Twenty-six of the samples showed photo- mond Technology samples when using either 488.0,
luminescence features attributed to nitrogen-vacancy 514.5, or 633.0 nm laser excitation at liquid-nitrogen
centers, with the neutral (575 nm) charge state more temperatures. The selected spectra presented here
intense than the negative (637 nm) charge state. have been translated vertically for clarity.
These could be detected using either 488.0 or 514.5
nm laser excitation. These selected photolumines-
cence spectra, taken at liquid-nitrogen temperatures
using 514.5 nm excitation, have been translated verti-
cally for clarity. R is the first-order Raman peak for
diamond, the broad structure spanning 580–596 nm
of the silicon-vacancy center, SiV–, which introduced
is the second-order Raman spectrum for diamond, a weak doublet ZPL at 736.6/736.9 nm (1.683/1.682
and the sharp peak at 584.5 is the first-order Raman eV). SiV–, which could be excited by the 488.0, 514.5,
peak for liquid nitrogen, which is used to cool the or 632.5 nm lasers (figure 15), was detected in 18
samples. (41%) of the samples. Its presence is rare in HPHT
synthetic diamonds, though it has been reported in
intentionally doped samples studied by Clark et al.
PHOTOLUMINESCENCE SPECTRA (1995) and Sittas et al. (1996), as well as in gem-qual-
R ity HPHT synthetic samples of unknown origin (Moe
575
and Wang, 2010; Wang and Moe, 2012) and those pro-
637 duced by AOTC (D’Haenens-Johansson et al., 2014).
PL INTENSITY (ARB. UNITS)
NDT31
The low concentrations suggest that its presence
NDT05
may be unintentional.
NDT17
NDT10
It is noteworthy that some of the PL features that
NDT11 were weakly detected in several of the colorless and
NDT18
NDT22
near-colorless HPHT synthetic diamonds grown by
AOTC—with peaks at 658 (1.884 eV), 675.2 (1.836
NDT07 NDT35
NDT29
eV), 706.9 (1.753 eV), 709.1 (1.748 eV), and 712.2 nm
NDT30
(1.740 eV)—were not observed in the samples pro-
NDT02
duced by NDT (D’Haenens-Johansson et al., 2014).
For the AOTC samples, these peaks were seen for di-
520 540 560 580 600 620 640 660
amonds grown in either BARS or toroid presses,
WAVELENGTH (nm)
which used different solvent/catalyst melts and ni-
trogen getters.
DISCUSSION colorless samples weighed over 2 ct, with clarities
Comparison with Other Gem-Quality Synthetics. ranging from VVS2 to I2 (though this improved up to
Near-colorless and colorless synthetic diamonds of IF for smaller sizes). The 4.30 ct D-color, SI1 clarity
gem quality are produced using HPHT or CVD specimen (NDT-B) is the largest faceted laboratory-
growth methods, with CVD material the most com- grown diamond of this color grade available to date
monly encountered “white” synthetic diamond seen (also reported by Poon et al., 2015), surpassing the
at GIA’s laboratory. In recent years, advances in CVD largest CVD synthetic in both size and color known
growth methods and the introduction of post-growth to the authors (while sharing the same clarity grade)
decolorizing HPHT treatments have led to the rapid (D.NEA, 2015). Although not studied for this inves-
evolution of synthetic diamond quality. Production tigation, a 10.02 ct E-color, VS1 square-cut emerald
of colorless synthetics in modest sizes (approxi- HPHT-grown synthetic (graded by IGI), also by NDT,
mately 0.30 ct) began around 2007 (Wang et al., is so far the largest reported polished synthetic dia-
2007); by 2011, high-clarity samples weighing ap- mond grown using either HPHT or CVD technolo-
proximately 1 ct were widely available (Wang et al., gies, of any color grade (International Gemological
2012). While the size of CVD synthetics has contin- Institute, 2015; Wang and Moses, 2010; Poon et al.,
ued to increase, the colors have been generally lim- 2015; N.DEA, 2015). These results suggest that
ited to near-colorless or faintly colored, and colorless HPHT synthetic colorless gem-quality diamonds are
grades are rare (Wang et al., 2013; Pure Grown Dia- now on a par with, and could potentially surpass,
monds, 2015; Washington Diamonds Corp., 2013; those produced by CVD methods.
D.NEA, 2015). The largest near-colorless CVD syn- Interestingly, AOTC uses both BARS and toroid
thetic reported to date is a 3.16 ct sample with G presses, while NDT uses cubic presses (D’Haenens-Jo-
color and SI1 clarity, produced by Washington Dia- hansson et al., 2014). This demonstrates that different
monds Corp. (D.NEA, 2015), which was graded by a HPHT methodologies can yield comparable products,
laboratory other than GIA. The record had previously with NDT’s cubic press methods currently yielding
been held by a 3.04 ct, I color, and SI1 clarity CVD larger specimens. Like the AOTC synthetics, most of
synthetic by Pure Grown Diamonds (formerly the NDT samples (80%) contained trace amounts of
Gemesis), also graded by a non-GIA laboratory (Pure boron, peaking at a bulk concentration of 58 ± 9 ppb
Grown Diamonds, 2014). (NDT14). Meanwhile, the previously published bulk
Conversely, HPHT growth for the gem trade was concentrations for representative AOTC samples
geared mainly toward fancy-color synthetic diamonds. were often higher, and in general the samples with
Colorless gem-quality material has only been com- higher boron concentrations had poorer color grades
mercially available since 2012 (D’Haenens-Johansson (further from colorless), introducing a blue hue that
et al., 2012, 2014). The main HPHT synthetic dia- could produce grades up to Fancy Light blue, depend-
mond producer in this sector is AOTC, whose coloring on the sample’s size and faceting arrangement
less synthetics are polished to sizes up to 1 ct (AOTC, (D’Haenens-Johansson et al., 2014). DiamondView flu-
2015). Although the D-Z diamond colors achieved by orescence images revealed that the intensity contrast
HPHT producers are typically closer to colorless than between the different growth sectors was significantly
those by CVD producers, their material is generally weaker for the NDT specimens. These results may
smaller and may be visibly included. Outside of the imply that the HPHT samples grown by NDT have a
gem industry, HPHT synthetic diamond producers lower impurity content than those produced by
such as Sumitomo Electric have reported growing 7– AOTC, explaining the higher percentage of D–F color
8 ct “rough” crystals described as “colorless” type IIa, grades achieved (89% compared to 33%).
though their material is intended for technological ap-
plications (Sumiya et al., 2002). If one further expands Identification Features. Although NDT sells its
the search field to include unpolished colored HPHT faceted synthetics with full disclosure of their origin,
synthetics, the largest reported in the literature is a it is prudent to be aware of the relevant identification
34.80 ct “yellow” specimen, grown solely for research features in case these goods are at some point reintro-
purposes by De Beers (Koivula et al., 2000). duced into the gem trade without disclosure. NDT’s
NDT’s HPHT-grown synthetic gem diamonds in gem-quality HPHT synthetics can be conclusively
this investigation mark a dramatic improvement in identified, though this often relies on a combination
the combination of colors and sizes attainable. Five of gemological and spectroscopic observations. Sev-
eral of these characteristics are similar to those for or aggregated nitrogen defects. Natural, treated, and
HPHT synthetics from other sources. CVD or HPHT synthetic colorless diamonds can all
Magnification may reveal the presence of metallic belong to these types. Since some 98% of natural di-
inclusions, which are not observed in CVD synthetic amonds contain A-aggregates (nitrogen pairs, which
diamonds and are extremely rare, and with differing absorb at approximately 1280 cm–1), their absence
geometries, in natural diamonds (Sobolev et al., 1981). from a sample’s IR spectrum would indicate a need
If sufficiently large or plentiful, the metallic inclu- for detailed analysis. PL, like FTIR, cannot alone con-
sions can be confirmed by attraction to a strong mag- clusively identify an HPHT synthetic, yet it was ef-
net. Examination through crossed polarizers showed fective at revealing suspicious impurities. The
that the samples had extremely low strain levels, as Ni-related features observed at 882.1/883.8 nm
indicated by low-order interference colors (blue and (1.405/1.402 eV) and at 483.6/483.8/484.1/484.4 nm
grays) and the inability to resolve a clear birefringence (2.563/2.562/2.560/2.559 eV) are seldom detected for
pattern. This is common for HPHT synthetics natural type IIa and IIb diamonds and have not been
(Crowningshield, 1971) but in stark contrast to both reported for CVD synthetics (Nobel et al., 1998; Cha-
CVD synthetic and natural stones, which show cross- lain, 2003). The 736.6/736.9 nm (1.683/1.682 eV) SiV–
hatched, mottled, or banded birefringence patterns, feature, often used in the identification of CVD-
often in a variety of colors. The fluorescence and grown samples, has been detected only rarely in nat-
phosphorescence response of these HPHT synthetics ural diamonds or in other HPHT synthetics (Breeding
was stronger to short-wave than long-wave UV, the and Wang, 2008; Moe and Wang, 2010; Wang and
opposite of the behavior seen in natural stones Moe, 2012; D’Haenens-Johansson et al., 2014). A
(Crowningshield, 1971; Shigley et al., 1997). Long-last- weak SiV– feature was detected in a subset of the
ing phosphorescence, a characteristic rarely detected specimens in this study. Its presence remains a
in natural diamonds, should be regarded with caution. source of concern, though a stone with this feature
If any of the above gemological observations are noted, could be either natural or CVD synthetic.
further investigation using advanced testing tech-
niques is crucial for conclusive origin determination. CONCLUSIONS
The high-energy UV illumination from a Diamond- New Diamond Technology’s colorless HPHT-grown
View instrument will readily induce fluorescence in synthetic diamonds are being faceted into gems for
all diamonds, natural or synthetic. Inspection of the the jewelry trade. This study of 44 representative
table, crown, and pavilion facets of the NDT samples samples revealed that the company can create
revealed cuboctahedral growth sectors, which are faceted colorless and faintly colored synthetics
characteristic of HPHT synthetics, though the pat- weighing up to 4.30 ct and 5.11 ct, respectively. Al-
tern was very weak in some. Additionally, all the though the samples spanned the full range of clarity,
samples exhibited strong phosphorescence. A phos- NDT has demonstrated the capacity to produce high
phorescence spectrometer could aid in their identifi- clarities. One such specimen, a 1.13 ct round brilliant
cation, as 13 of the 42 samples tested in this manner with a very good cut grade, characterized by D color
(31%) showed a phosphorescence band at approxi- and IF clarity, was remarkable for its quality. All in-
mately 575 nm, which has only been reported in cer- dications point to high-color and high-clarity HPHT
tain HPHT-grown synthetic diamonds (Watanabe et synthetics becoming more prominent in the gem
al., 1997; Eaton-Magaña et al., 2008). Despite the sim- trade. The colorless stones in the study are compara-
ilar emission wavelengths, this band should not be ble to top-quality natural diamonds and surpass (for
confused with the weak 580 nm band that has been this color range) the sizes achieved by alternative pro-
infrequently detected for some natural type IIb dia- ducers. NDT’s continued focus on expanding produc-
monds. The 500 nm band, also observed, cannot be tion and increasing the sizes of their colorless
considered indicative of a synthetic origin, as it is samples emphasizes the need for awareness by grad-
often displayed by natural type IIb diamonds (Eaton- ing laboratories, members of the diamond trade, and
Magaña et al., 2007, 2008). consumers. Through careful analysis using gemolog-
Further spectral analysis using FTIR absorption ical, fluorescence, and spectroscopic methods, these
and PL spectroscopy may also be helpful. All the products can be confidently identified as synthetic,
NDT samples were either type IIb or type IIa—i.e., thus posing no threat to a well-informed, responsible,
they did not contain detectable amounts of isolated and transparent diamond trade.
ABOUT THE AUTHORS ACKNOWLEDGMENTS
Dr. D’Haenens-Johansson (ujohansson@gia.edu) is a research The authors are grateful to Tamazi Khikhinashvili and Aleko Arens
scientist, Mr. Moe is a research associate, Mr. Johnson is the su- from New Diamond Technology, in St. Petersburg, Russia, for
pervisor of diamond advanced testing, and Dr. Wang is the direc- providing HPHT-grown synthetic diamonds for this investigation.
tor of research and development at GIA’s New York laboratory. Dr. Brian Bujarski, formerly a technician at GIA’s New York laboratory,
Katrusha is a scientific partner of New Diamond Technology, Ltd. is thanked for his help with PL data acquisition.
REFERENCES
AOTC (2015) Synthetic white diamonds, http://aotc.com/unique/ D’Haenens-Johansson U.F.S., Moe K.S., Johnson P., Wong S.Y., Lu
white-diamonds [date accessed: Feb. 23, 2015]. R., Wang W. (2014) Near-colorless HPHT synthetic diamonds
Bates R. (2015) Company claims to have produced 5 carat syn- from AOTC Group. G&G, Vol. 50, No. 1, pp. 30–45,
thetic diamond, JCK Magazine, Feb. 17, http://www.jckonline. http://dx.doi.org/10.5741/GEMS.50.1.30.
com/2015/02/17/company-claims-to-have-produced-5-carat- D.NEA (2015) 3.16 ct G color round cut created diamond,
synthetic-diamond. http://d.neadiamonds.com/lab-created-diamonds/WB1225.
Bovenkerk H.P., Bundy F.P., Hall H.T., Strong H.M., Wentorf R.H. Eaton-Magaña S., Post J.E., Heaney P.J., Walters R.A., Breeding
(1959) Preparation of diamond. Nature, Vol. 194, No. 4693, pp. C.M., Butler J.E. (2007) Fluorescence spectra of colored dia-
1094–1098, http://dx.doi.org/10.1038/1841094a0. monds using a rapid, mobile spectrometer. G&G, Vol. 43, No.
Breeding C.M., Wang W. (2008) Occurrence of the Si-V defect cen- 4, pp. 332–351, http://dx.doi.org/10.5741/GEMS.43.4.332.
ter in natural colorless gem diamonds. Diamond and Related Eaton-Magaña S., Post J., Heaney P.J., Freitas J., Klein P., Walters
Materials, Vol. 17, Nos. 7–10, pp. 1335–1344, http://dx.doi.org/ R., Butler J.E. (2008) Using phosphorescence as a fingerprint for
10.1016/j.diamond.2008.01.075. the Hope and other blue diamonds. Geology, Vol. 36, No. 1, pp.
Burns R.C., Cvetkovic V., Dodge C. N., Evans D.J.F., Rooney M.- 83–86, http://dx.doi.org/10.1130/G24170A.1.
L.T., Spear P.M., Welbourn C.M. (1990) Growth-sector depend- Eaton-Magaña S., Lu R. (2011) Phosphorescence in type IIb dia-
ence of optical features in large synthetic diamonds. Journal of monds. Diamond and Related Materials, Vol. 20, No. 7, pp.
Crystal Growth, Vol. 104, No. 2, pp. 257–279, http://dx.doi.org/ 983–989, http://dx.doi.org/10.1016/j.diamond.2011.05.007.
10.1016/0022-0248(90)90126-6. Fisher D., Sibley S.J., Kelly C.J. (2009) Brown colour in natural di-
Chalain J.P. (2003) Gem News International: A natural yellow diamond and interaction between the brown related and other
amond with nickel-related optical centers. G&G, Vol. 39, No. colour-inducing defects. Journal of Physics: Condensed Matter,
4, pp. 325–326. Vol. 21, No. 36, 364213, 10 pp., http://dx.doi.org/10.1088/0953-
Clark C.D., Kanda H., Kiflawi I., Sittas G. (1995) Silicon defects 8984/21/36/364213.
in diamond. Physical Review B, Vol. 51, No. 23, pp. 16681– Hall M., Lu R., Wang W. (2010) U.S. Patent No. US
16688, http://dx.doi.org/10.1103/PhysRevB.51.16681. 2010/0220311A1. Issued September 2.
Collins A.T., Williams A.W.S. (1971) The nature of the acceptor International Gemological Institute (2015) International Gemo-
centre in semiconducting diamond. Journal of Physics C: Solid logical Institute (IGI) Hong Kong certifies record-breaking,
State Physics, Vol. 4, pp. 1789–1800, http://dx.doi.org/10.1088/ world’s largest colorless grown diamond, May 26, http://igi-
0022-3719/4/13/030. worldwide.com/igi-certifies-worlds-largest-colorless-grown-di-
Collins A.T., Kanda H., Burns R.C. (1990) The segregation of nickel- amond.html.
related optical centers in the octahedral growth sectors of syn- Klein P.B., Crossfield M.D., Freitas J.A., Collins A.T. (1995) Donor-
thetic diamond. Philosophical Magazine Part B, Vol. 61, No. 5, acceptor pair recombination in synthetic type-IIb semiconduct-
pp. 797–810, http://dx.doi.org/10.1080/13642819008207562. ing diamond. Physical Review B, Vol. 51, No. 15, pp.
Collins A.T. (1992) The characterisation of point defects in diamond 9634–9642, http://dx.doi.org/10.1103/PhysRevB.51.9634.
by luminescence spectroscopy. Diamond and Related Materials, Koivula J.I., Tannous M. Schmetzer K. (2000) Synthetic gem ma-
Vol. 1, Nos. 5–6, pp. 457–469, http://dx.doi.org/10.1016/ terials and simulants in the 1990s. G&G, Vol. 36, No. 4, pp.
0925-9635(92)90146-F. 360–379, http://dx.doi.org/10.5741/GEMS.36.4.360.
Crowningshield R. (1971) General Electric’s cuttable synthetic di- Lawson S.C., Kanda H. (1993) An annealing study of nickel point
amonds. G&G, Vol. 13, No. 10, pp. 302–314. defects in high-pressure synthetic diamond. Journal of Applied
Dean P.J. (1965) Bound excitons and donor-acceptor pairs in natu- Physics, Vol. 73, No. 8, pp. 3967–3973, http://dx.doi.org/10.1063/
ral and synthetic diamond. Physical Review, Vol. 139, No. 2A, 1.352861.
pp. A588–A602, http://dx.doi.org/10.1103/PhysRev.139.A588. Martineau P.M., Lawson S.C., Taylor A.J., Quinn S.J., Evans D.J.F.,
Dean P.J. (1973) Inter-impurity recombinations in semiconductors. Crowder M.J. (2004) Identification of synthetic diamond grown
Progress in Solid State Chemistry, Vol. 8, pp. 1–126, using chemical vapor deposition (CVD). G&G, Vol. 40, No. 1,
http://dx.doi.org/10.1016/0079-6786(73)90004-6. pp. 2–25, http://dx.doi.org/10.5741/GEMS.40.1.2.
D’Haenens-Johansson U.F.S., Moe K.S., Johnson P., Wong S.Y., Moe K.S., Wang W. (2010) Lab Notes: Silicon-vacancy defect found
Wang W. (2012) Lab Notes: Near-colorless HPHT-grown syn- in blue HPHT-grown synthetic diamond. G&G, Vol. 46, No. 4,
thetic diamonds from Advanced Optical Technology Co. G&G, p. 302.
Vol. 48, No. 2, p. 141. Nazaré M.H., Neves A.J., Davies G. (1991) Optical studies of the 1.40-
eV Ni center in diamond. Physical Review B, Vol. 43, No. 17, pp. Sumiya H. (2009) Recent advances in high pressure apparatus for di-
14196–14205, http://dx.doi.org/10.1103/PhysRevB.43.14196. amond synthesis. The Review of High Pressure Science and Tech-
Nazaré M.H., Mason P.W., Watkins G.D., Kanda H. (1995) Optical nology, Vol. 19, No. 4, pp. 264–269, http://dx.doi.org/10.4131/
detection of magnetic resonance of nitrogen and nickel in high- jshpreview.19.264 [in Japanese].
pressure synthetic diamond. Physical Review B, Vol. 51, No. 23, Sumiya H., Satoh S. (1996) High-pressure synthesis of high-purity
pp. 16741–16745, http://dx.doi.org/10.1103/PhysRevB.51.16741. diamond crystal. Diamond and Related Materials, Vol. 5, pp.
Nobel C.J., Pawlik T., Spaeth J.-M. (1998) Electron paramagnetic 1359–1365, http://dx.doi.org/10.1016/0925-9635(96)00559-6.
resonance investigations of nickel defects in natural diamonds. Sumiya H., Toda N., Satoh S. (2002) Growth rate of high-quality
Journal of Condensed Matter, Vol. 10, No. 50, pp. 11781– large diamond crystals. Journal of Crystal Growth, Vol. 237–
11794, http://dx.doi.org/10.1088/0953-8984/10/50/017. 239, Part 2, pp. 1281–1285, http://dx.doi.org/10.1016/S0022-
Palik E.D. (1985) Handbook of Optical Constants of Solids. Aca- 0248(01)02145-5.
demic Press, London. Sumiya H., Toda N., Satoh S. (2005) Development of high-quality
Poon P.Y., Wong S.Y., Lo C. (2015) Large HPHT-grown synthetic large-size synthetic diamond crystals. SEI Technical Review,
diamonds examined in GIA’s Hong Kong laboratory. GIA Re- Vol. 60, pp. 10–16.
search & News, Mar. 16, http://www.gia.edu/gia-news-re- Wang W., Moses T. (2010) Lab Notes: Large (4+ ct) yellow-orange
search-large-hpht-grown-synthetic-diamonds-examined-in-gia- HPHT-grown synthetic diamond. G&G, Vol. 46, No. 4, p. 301.
hong-kong-laboratory. Wang W., Moe K.S. (2012) Lab Notes: Silicon-vacancy defect in
Pure Grown Diamonds (2014) World’s largest laboratory pure grown HPHT-grown type IIb synthetic. G&G, Vol. 48, No. 4, pp. 304–
diamond unveiled. Business Wire, Dec. 30, http://www.busi- 305.
nesswire.com/news/home/20141229005491/en/adding- Wang W., Hall M.S., Moe K.S., Tower J., Moses T.M. (2007) Latest-
multimedia-worlds-largest-laboratory-pure-grown#.votnbs63raq. generation CVD-grown synthetic diamonds from Apollo Dia-
Pure Grown Diamonds (n.d.) Lab grown diamonds, http://www.pure- mond Inc. G&G, Vol. 43, No. 4, pp. 294–312, http://dx.doi.org/
growndiamonds.com/diamonds/single [date accessed: Jun. 23, 10.5741/GEMS.43.4.294.
2015]. Wang W., D’Haenens-Johansson U.F.S., Johnson P., Moe K.S.,
Rooney M.-L.T., Welbourn C.M., Shigley J.E., Fritsch E., Reinitz I. Emerson E., Newton M.E., Moses T.M. (2012) CVD synthetic
(1993) De Beers near colorless-to-blue experimental gem-qual- diamonds from Gemesis Corp. G&G, Vol. 48, No.2, pp. 80–97,
ity synthetic diamonds. G&G, Vol. 29, No. 1, pp. 38–45, http://dx.doi.org/10.5741/GEMS.43.4.294.
http://dx.doi.org/10.5741/GEMS.29.1.38. Wang W., Moe K.S., Yeung S.F., D’Haenens-Johansson, U. (2013)
Satoh S., Sumiya H., Tsuji K., Gouda Y. (2000) US Patent No. Lab Notes: Very large CVD-grown. G&G, Vol. 49, No. 1, p 50.
US006129900A. Issued October 10. Watanabe K., Lawson S.C., Isoya J., Kanda H., Sato Y. (1997) Phos-
Shigley J.E., Moses T.M., Reinitz I., Elen S., McClure S.F., Fritsch phorescence in high-pressure synthetic diamond. Diamond
E. (1997) Gemological properties of near-colorless synthetic di- and Related Materials, Vol. 6, No. 1, pp. 99–106
amonds. G&G, Vol. 33, No. 1, pp. 42–53, http://dx.doi.org/ http://dx.doi.org/10.1016/S0925-9635(96)00764-9.
10.5741/GEMS.33.1.42. Washington Diamonds Corp. (2013) Going where no diamond-produing
Shigley J.E., McClure S.F., Breeding C.M. Shen A.H.-T., Muhlmeis- lab has gone before: A true colorless, one carat, lab grown CVD
ter M. (2004) Lab-grown colored diamonds from Chatham Cre- diamond, Oct. 3, http://www.washingtondiamondscorp.com/going-
ated Gems. G&G, Vol. 40, No. 2, pp. 128–145, http://dx.doi.org/ where-no-diamond-producing-lab-has-gone-before-a-true-colorless-
10.5741/GEMS.40.2.128. one-carat-lab-grown-cvd-diamond/
Sittas G., Kanda H., Kiflawi I., Spear P.M. (1996) Growth and Welbourn C.M., Cooper M., Spear P.M. (1996) De Beers natural ver-
characterization of Si-doped diamond single crystals grown sus synthetic diamond verification instruments, G&G, Vol. 32,
by the HTHP method. Diamond and Related Materials, Vol. No. 3, pp. 156–169, http://dx.doi.org/10.5741/GEMS.32.3.156.
5, Nos. 6–8, pp. 866–869, http://dx.doi.org/10.1016/0925- Yelisseyev A., Kanda H. (2007) Optical centers related to 3d tran-
9635(95)00449-1. sition metals in diamond. New Diamond and Frontier Carbon
Sobolev N.V., Efimiva E.S., Pospelova L.N. (1981) Native iron in Technology, Vol. 17, No. 3, pp. 127–178.
diamonds of Yakutia and its paragenesis. Soviet Geology and Zhan-Chang L., Xiao-Peng J., Guo-Feng H., Mei-Hua H. Yong L.,
Geophysics, Vol. 22, No. 12, pp. 18–21. Bing-Min Y. (2013) FEM simulations and experimental studies
Strong H.M. (1977) US Patent No. US40142673. Issued August 16. of the temperature field in a large diamond crystal growth cell.
Strong H.M., Chrenko R.M. (1971) Further studies on diamond Chinese Physics B, Vol. 22, No. 1, 014701–014706,
growth rates and physical properties of laboratory-made dia- http://dx.doi.org/10.1088/1674-1056/22/1/014701.
mond. The Journal of Physical Chemistry, Vol. 75, No. 12, pp. Zhu H., Pearson K., Kim J.R. (2012) US Patent No.
1838–1843, http://dx.doi.org/10.1021/j100681a014. US2012/0192785A1. Issued August 2.
FEATURE ARTICLES
LETŠENG’S UNIQUE
DIAMOND PROPOSITION
Russell Shor, Robert Weldon, A.J.A. (Bram) Janse, Christopher M. Breeding, and Steven B. Shirey
The Letšeng-la-Terae diamond mine in Lesotho, a small nation surrounded by South Africa, is unique in
that it produces some of the world’s largest and highest-value diamonds from a relatively small volume
of kimberlite ore. The mine, operated by De Beers during the late 1970s and early 1980s, was difficult to
sustain economically because of its remoteness and very low ore grade, which made production costs
very high. Since 2000, sharply rising prices for large diamonds have permitted the mine to reopen prof-
itably as a 70/30 venture between Gem Diamonds Ltd. of South Africa and the government of Lesotho.
To improve recovery of large diamonds, Letšeng’s owners have implemented new processing technology
that provides better identification of these crystals before processing and a crushing mechanism that re-
duces their potential breakage. In recent years, several important diamonds recovered from Letšeng have
been sold to London luxury jeweler Laurence Graff.
W
hile the Premier (now Cullinan) mine in Until 2004, Letšeng had been shut down for more
South Africa is usually cited as the tradi- than 20 years because its low grade and remote loca-
tional source for very rare large diamonds, tion made mining unprofitable at existing diamond
in recent years the Letšeng-la-Terae mine in Lesotho prices. Up until the past decade, De Beers and other
has become the major producer of such stones (figure mining companies typically sacrificed large crystals
1). Although its total output is relatively small— for the sake of rapid throughput, which maximized
about 100,000 carats yearly, compared to two million efficiency and reduced mining costs. Only after
or more carats from most major diamond mines—the prices for large stones began rising sharply did the
average value of its production in 2014 was just mining economics change in favor of preserving
above US$2,500 per carat (Gem Diamonds Ltd., them. Because the economic viability of Letšeng
2014, 2015a). This is by far the world’s highest per-
carat value, more than 21 times higher than the $116
average for diamonds mined worldwide, according to Figure 1. Rising prices and demand for large dia-
monds have fueled the success of the Letšeng mine in
Kimberley Process data.
northern Lesotho. Shown here are two rough stones: a
The Letšeng mine is also the world’s most consis-
299.35 ct slightly yellowish partial octahedron and a
tent source of type IIa diamonds (those with excep- colorless 112.61 ct type IIa diamond. Photo by Robert
tionally low nitrogen content), which account for Weldon/GIA; courtesy of Gem Diamonds Ltd.
about one-fourth of its production (Gem Diamonds
Ltd., 2014). The mine claims six of the 20 largest di-
amonds ever discovered—478, 493, 527, 550, 601,
and 603 ct (table 1). The largest of these, the Lesotho
Promise, was found in 2006 and sold to London jew-
eler Laurence Graff for $12.4 million. In the years
since, Graff has also purchased the 493 ct Letšeng
Legacy, the 478 ct Light of Letšeng, and the 550 ct
Letšeng Star.

GEMS & GEMOLOGY, Vol. 51, No. 3, pp. 280–299,
http://dx.doi.org/10.5741/GEMS.51.3.280.
© 2015 Gemological Institute of America
280 LETŠENG DIAMOND MINE GEMS & GEMOLOGY FALL 2015

Figure 2. This remote village deep in the Maloti Mountains near the Letšeng mine is primarily dedicated to
agriculture and sheep herding. Villagers also form the workforce for the diamond mine. Basuto ponies such as the
one seen here are esteemed for their speed and surefootedness in the steep, rocky mountain passes. Photo by
Robert Weldon/GIA.
hinges on its ability to produce very large diamonds, for minerals, including diamonds. Large-scale agri-
the owners have employed new recovery processes culture was nearly impossible, which made it unat-
to reduce damage and breakage of such diamonds. tractive for colonial settlers. As a result, Basutoland
The mine is a major force of Lesotho’s economy, em- remained fairly isolated from the wars and mineral
ploying about 1,500 workers, 90% of them locals. rushes that shaped the southern portion of the con-
Letšeng accounts for 70% of the country’s corporate tinent during the 20th century (“Lesotho,” 2014).
tax revenue and 60% of its foreign exchange earnings By road, Letšeng is located approximately 214 km
(M. Maharasoa, pers. comm., 2014).
BACKGROUND Figure 3. This map of Lesotho, formerly Basutoland,

The Kingdom of Lesotho is a nation of rugged moun- shows how the small country is completely sur-
rounded by South Africa. The discovery of diamonds
tains (figure 2) set within the boundaries of South
in Lesotho came 90 years after the South African
Africa (figure 3). Its capital, Maseru, is located about finds of the 19th century.
250 miles south of Johannesburg. When diamonds
were discovered in South Africa in 1867, Lesotho ZIMBABWE
Orapa
(then known as Basutoland) was sparsely settled by
Windhoek
agrarian villagers and shepherds living under the rule BOTSWANA MOZAMBIQUE
of King Moshoeshoe. Venetia
Gaborone
The tiny kingdom struggled to remain independ- Premier
AT
NAMIBIA Jwaneng (Cullinan)

ent from its much larger neighbor, first resisting Zulu
LA
Pretoria
Finsch
NT
invasions from the south and then becoming em- Johannesburg

SWAZILAND
IC
broiled in the growing tensions between the British Kimberley Letšeng

Maseru
and Afrikaners in southern Africa during the 1870s.
OC
N
Durban A
The Afrikaners (descendants of early Dutch settlers)
EA
E
Jagersfontein LESOTHO C
SOUTH AFRICA O
N
had begun to claim parts of the territory. While Ba- N

IA
sutoland’s craggy mountains made for good defense, N Cape Town D
IN
with more than 80% of the nation at least 5,000 feet
above sea level, the British annexed the region in Diamond mines
0 500 km
1871. The rugged terrain also hindered exploration
LETŠENG DIAMOND MINE GEMS & GEMOLOGY FALL 2015 281

27˚E 28˚E 29˚E
Figure 4. This map of

Lesotho shows the
N Butha Buthe Mothae Approximate craton boundary country’s principal dia-
(beneath overlying sediments
Liqhobong and lavas) mond mines within the
Mothae Drakensberg basalts.
29˚S Letšeng mine 29˚S
Khaphamali Letšeng was the first di-
Kao
amond deposit discov-
Mapholaneng
Mokhotlong
ered in Lesotho and
Maseru
remains the largest one,
t
ic
producing significant
tr
n
is
o
rat
D
numbers of large dia-
-c
g
on
on
a ton monds. Like Letšeng,
ti
-cr
ho
off the other three mines in
ok
M
the area—Kao, Liqho-
bong, and Mothae—are
all located on Archean
30˚S 30˚S
craton underlain by
older mantle at depth.
There are no known pri-
mary diamond deposits
in the off-craton part of
the country. Map cour-
0 50 km
tesy of Telfer and
McKenna (2011).
27˚E 28˚E 29˚E
Basalts of Drakensberg Group – Lower Jurassic Road

Clastic sediments of the Stormberg Group Town
(Molteno, Elliot and Clarens Formations) – Upper Triassic Diamond digging or kimberlite pipe
Clastic sediments of the Beaufort Group – Lower Triassic / Upper Permian
(133 miles) northeast of Maseru. The route is paved green, and purple mudstones, previously called Red
except for the last 30 km, which is a well-graded Beds), and Clarens (cream-colored aeolian sands, pre-
gravel road with many sharp hairpin turns. The viously called Cave Sandstone) formations. These are
mine’s complete name is Letšeng-la-Terae (also overlain by the Early Jurassic (183 Ma) Drakensberg
spelled Terai), which means “the swamp at the turn,” Beds, a monotonous series of flood basalt lava flows.
as it is located in a low saddle in the mountain range The base of the lavas lies at an elevation of 1,600 me-
where the road makes an abrupt turn (Whitelock, ters, while the top is above 3,200 meters, for a strati-
1979). The diamondiferous deposit was found in kim- graphic thickness of more than 1,600 meters.
berlite under the superficial swamp sediment layer. Fine-grained basaltic dikes and sills contemporane-
ous with the lavas occur in the lowlands and moun-
GEOLOGICAL FRAMEWORK tains in a wide zone trending ESE-WNW. Hundreds
Figure 4 shows that the western lowlands and parts of kimberlite pipes and dikes intruded in the Middle
of southern Lesotho consist of horizontally bedded to Late Cretaceous (120–65 Ma) in many parts of
sediments, whereas a thick series of flat-lying southern Africa, and in Lesotho in the 95–85 Ma in-
basaltic lavas form the mountains in the east. The terval (Wilson et al., 2007). They occur mainly in
sediments belong to the upper part of the Karoo Su- northern and northeastern Lesotho (again, see figure
pergroup (late Carboniferous to mid-Jurassic, 330– 4), occupying the same ESE-WNW zone as the fine-
180 Ma), which once covered a major portion of grained basaltic dikes. Known at present are 39 pipes
southern Africa. In Lesotho, these sediments form and 23 dike enlargements, usually called blows. Of
part of the Stormberg series, which are Triassic these 62 bodies, 24 are diamondiferous. Moreover,
(roughly 250–200 Ma) in age and consist of pale mul- there are more than 300 individual dikes. The age of
ticolored sandstones with intercalated grits, mud- intrusion of the Letšeng pipes has not been measured
stones of the Molteno (fluvial sands), Elliot (red, directly but is assumed to be similar to neighboring

TABLE 1. Largest diamonds recovered from Letšeng.
Name Discovered Size (ct) Value Sale price original Stromberg sediments and volcanics remain
per carat and the upper portions of the pipes are still preserved.
Unnamed 1965 527 no data no data
Lesotho Brown May 1967 601 US$302,400 US$503 GEOLOGY AND PETROGRAPHY OF THE
Lesotho Promise Oct. 2006 603 US$12.4 M US$20,564 LETŠENG PIPES
The high proportion of very large and type IIa dia-
Letšeng Legacy Sept. 2007 493 US$10.4 M US$21,095
monds in the Letšeng kimberlite has puzzled geolo-
Light of Letšeng Sept. 2008 478 US$18.4 M US$38,493 gists. Letšeng sits near the edge of the Archean craton
Letšeng Star Aug. 2011 550 US$16.5 M US$30,000 (see box A). In the Archean, this location might have
permitted distinctive low-nitrogen, diamond-forming
fluids access to the deep lithospheric mantle where
pipes. For instance, zircon from Mothae is 87 Ma the Letšeng diamonds were grown. It is possible that
(Davis, 1977), while ages of perovskite from Kao, these diamonds did not grow in one short geologic
Liqhobong, and Mothae range from 89 to 95 Ma event but slowly, over a longer geologic interval. If so,
(Lock and Dawson, 2013). the high proportion of large diamonds could relate to
According to Clifford’s Rule (as refined by Janse, especially slow growth rates from these fluids. But
1994), the occurrence of economic diamondiferous confirmation of these ideas would depend on detailed
kimberlites is restricted to the Archean (>2,500 Ma) mineralogical studies of growth zonation and accu-
interior of the seismically stable, erosionally flattened rate age information, in particular for absolute age
area of continental crust plus mantle known as a cra- and/or core-to-rim ages that would reflect the dura-
ton (see Shirey and Shigley, 2013), which is composed tion of diamond growth. Such studies do not exist at
of a basement older than 1,800 Ma. Janse (1994) represent because of the high value of these stones—
fined Clifford’s Rule in stating that economic kimber- they cannot be sacrificed for research—and the excep-
lites occur only on archons (the Archean part of a tionally low abundance of mineral inclusions.
craton) older than 2,500 Ma. As previously noted, in The Letšeng kimberlites consist of two pipes, the
Lesotho this cratonic basement is hidden by overlying Main and the Satellite (figure 5), measuring 17.2
Triassic to Early Jurassic sediments and lavas (again, hectares (540 × 365 m) and 5.2 hectares (425 × 130
see figure 4). Archean basement must be present at m), respectively. The pipes are carrot-shaped bodies,
depth because pieces of Archean-age mantle and crust vertically intrusive and slightly tapering with verti-
occur in the Letšeng kimberlite pipes, showing that cal to slightly inclined walls, 83° at most. They are
they fall within the window of potentially economic elongated in a northerly direction, roughly perpendi-
diamond deposits. Letšeng lies close to the eastern cular to the general westerly trend of the fine-grained
edge of the Kaapvaal craton, where remnants of the basaltic dikes, which suggests that the pipe locations
Figure 5. The Main pipe (left) and Satellite pipe (right), photographed during a 2014 visit. The larger Main pipe
was originally mined by De Beers. Today both are operated by Letšeng Diamonds. The basalt raft is the circular
area in the upper right portion of the Satellite pipe. Photos by Robert Weldon/GIA.

BOX A: ARCHEAN CRATONS AND THE LOCATION OF LETŠENG
The southern African continent has a fascinating geo- than the bottom of the craton. A first result of this ex-
morphological history and geology that helps explain the tensive erosion of major parts of the country is that the
region’s diamond distribution. In addition, the various top portion of southern African kimberlites have been re-
mines across the region have taught the world’s geolo- moved and their diamond cargo has been transported
gists many of the basic geologic principles of diamond westward along the Vaal and Orange River systems to
formation and occurrence. form the alluvial and marine diamond deposits of today.
The basic geologic structure of the region is one A second result of this erosion is that diamonds in their
where geologically old (3 to 3.7 billion years) and stable host kimberlites are mined today at very different levels
central portions of the continent known as “cratons” are in the volcanic pipe. Since the kimberlites were erupted
surrounded by younger, deformed regions (figure A-1). through the craton before erosion took place, diamonds
Diamonds occur in primary, relatively young (often 90 at Letšeng occur roughly a mile higher in the pipe than
million but up to 1.2 billion years in age) kimberlite vol- they do at other localities in South Africa such as Kim-
canic pipes that have brought them from the bottom of berley or Premier/Cullinan.
the cratons to the surface. Kimberlites that erupt through In some ways, every kimberlite and diamond popu-
the deformed regions off-craton are devoid of diamonds. lation is unique. But Letšeng is highly unusual. Mine ge-
This basic southern African fact, that the cratons are ologist Debbie Bowen and her colleagues have described
where the diamonds occur, has been applied around the several features that make this deposit unique: the lack
world as the primary exploration approach. of octahedral diamonds, the high percentage of resorbed
Radiometric dating of mineral grains in many dia- and rounded dodecahedral shapes, the yield of gem-qual-
monds shows that they are old, ranging in age from about ity stones in the larger size categories, and especially the
a billion years to as old as the craton itself. Since the first abundance of type IIa (nitrogen-free) D-color stones. She
cratonic blocks were formed by massive continental col- compared Letšeng diamonds to South African alluvial di-
lision and thickening, diamond formation likely occurred amonds, whose large average stone size is a result of the
as part of this process. Diamond-forming fluids were car- destruction of the poorer, weaker diamonds during trans-
ried down into the deep regions of the mantle under the port (Bowen et al., 2009). Circumstances that complicate
continents where diamonds can grow. But given that our understanding of Letšeng diamonds include the ex-
many diamonds can be younger, later post-collision un- tensive vertical mixing that occurs during a kimberlite
derthrusting of oceanic slabs during a process known as eruption, the lack of control on such a chaotic process,
subduction was likely another cause of later formation. and the inability to sample such expensive diamonds for
After this long history of active continent building research purposes.
and modification, southern Africa was then subjected to The position of the Letšeng-la-Terae kimberlite pipe
weathering and extensive erosion. Erosion plays the at the edge of the craton may or may not be partly re-
major role in the genesis of secondary diamond de- sponsible for some of the unique diamonds found there.
posits—those of alluvial (riverine) and marine derivation. Knowing the age of diamond growth would help place
Erosion is very dependent on height above sea level, and geological constraints on the types of processes that sup-
therefore uplift largely determines how much erosion plied the diamond-forming carbon and what its sources
will occur. Johannesburg stands a mile above sea level, might have been. No definitive age studies have been car-
while parts of Lesotho are nearly two miles above sea ried out on Letšeng diamonds, and there have been few
level. This is highly unusual for a craton. The cause of geochemical studies of the inclusions or the diamonds
this uplift is not known, but it is relatively recent and re- themselves. Such studies are now underway and will
lated to hotter upwelling mantle occurring much deeper help us understand these questions.
were influenced by crosscutting structures. The con- The simple structural division into crater, dia-
tact between the kimberlite and the lava flows of the treme, and root zones has been modified during the
country rock are sharp, giving no indication of up- last decade (Scott Smith et al., 2008). Based on mi-
folding or chemical alteration. In places there is a 1 croscopic observation of core samples, petrographers
to 1.3 m marginal zone within the pipe where the now recognize volcaniclastic and hypabyssal kimber-
kimberlite is sheared, with mud seams and calcite lite rocks instead (figure 6, top). Volcaniclastic kim-
veins. The Letšeng kimberlites belong to the group 1 berlite occurring in the crater, and layered in
class of kimberlites, the generally non-micaceous successive pulses, may be resedimented due to
type. Their texture is volcaniclastic. slumping. This resedimented volcaniclastic kimber-

CC
contain easily visible centimeter-sized olivine crys-
ZC
tals set in a groundmass of olivine (mainly altered to
serpentine), phlogopite, and calcite. Crystals of pur-
O
DB LB plish to dark red pyrope, black ilmenite, and green
MB
V chrome diopside can occasionally be seen in the rock
N B
GC or among the minerals in the surface rubble.
J
P The Letšeng kimberlite is covered by one meter
RT KC
Of
of spongy soil (primarily on the Main pipe), two me-

fsh
er
KB F l Riv
o re
Vaa
L
ters of leached white gravel, five meters of brown
allu
N basaltic gravel (mainly on the Satellite pipe), one

K A
via
Orange River E
JF
l di
C meter of yellow ground (oxidized kimberlite), and 20

O
am
A
T N
meters of soft kimberlite “blue ground” and “yellow
on
L
A NNB IA
ds
N D
T
IC IN
O
CFB
C
E
A
0 500 km Figure 6. Top: Plan view of the Main and Satellite
N
kimberlite pipes at Letšeng, showing the location of
Figure A-1. In this geologic map of southern Africa, current
underground tunnels within the pipes. Bottom: A
or recently active diamond mines (O = Orapa, V = Vene- cross-sectional map of the Main and Satellite kimber-
tia, P = Premier/Cullinan, J = Jwaneng, F = Finsch, K = lite pipes. The green area in the Satellite pipe is the
Kimberley, JF = Jagersfontein, L = Letšeng) are confined to large basalt raft that dropped into the pipe some 90
areas of Archean crust (more than 2.5 billion years old) million years ago. From Telfer and McKenna (2011).
known as cratons, marked by long-term geological stabil-
ity. These Archean cratons (CC = Congo Craton, ZC = Main Pipe
Zimbabwe Craton, LB = Limpopo Belt, KC = Kaapvaal
Craton) are shown in brown, along with two metamorphic Satellite Pipe
terranes (KB = Kheis Belt, MB = Magondi Belt) suspected B
of containing Archean crust. The Vaal and Orange Rivers, Tunnels
Tunnels
which drain South Africa to the west, are the sources of al-
luvial workings along the river systems and marine de-
posits off the coasts of South Africa and Namibia. A
Relatively recent and substantial uplift of southern Africa
caused these diamonds to be weathered from their origi- Boreholes
nal on-craton kimberlite hosts and moved westward by N
the rivers to the Atlantic Ocean. Note that there are no di- Boreholes
amondiferous kimberlites in the green metamorphic mo-
bile belts (DB = Damara Belt, GCB = Ghanzi-Chobe Belt,
RT = Rehoboth Terrane, KB = Kheis Belt, MB = Magondi West - A East - B
Belt, NNB = Namaqua-Natal Belt, CFB = Cape Fold Belt)
Meters above mean sea level (mamsl)
that surround the cratons. Note the landlocked kingdoms

of Swaziland and Lesotho within South Africa and the in- 3200
Satellite Pipe Main Pipe
teresting location of Letšeng at the edge of the Kaapvaal
Craton. 3000
2800
2600
lite becomes more massive, nonlayered kimberlite
in the pipe and is often referred to as tuffisitic kim-
2400
berlite breccia. This is the general kimberlite breccia 0 200m
present in many specimen collections. Hypabyssal

kimberlite is a dense aphanitic (evenly fine-grained, Satellite Pipe Main Pipe
NVK K North Surface
non-breccia) rock occurring in the root zone. The de- SVK K4 Optimal Pit
gree of erosion in the general countryside can expose Raft (Basalt) K6 (March 2011)
K South
these rocks at different levels in the vertical zoning
(Scott Smith et al., 2008). The Letšeng kimberlites

ground”. Below this depth, the kimberlite becomes throughout and accompanied by a wide variety of
hard bluish grayish greenish rock descending at least basement rocks (gneisses and amphibolites) and ul-
700 meters (see pipe cross-section in figure 6). tramafic nodular fragments that include garnet lher-
During public digging access from 1959 to 1967 zolites (Lock and Dawson, 2013). Garnet xenocrysts
(figure 7), artisanal miners occasionally recovered are also relatively abundant (Bloomer and Nixon,
very large crystals from the gravels on top of the 1973). In the Satellite pipe two varieties are recog-
pipes and in the gravels of nearby streams that drain nized: NVK and SVK (north and south volcaniclastic
the Main and the Satellite pipes. Formal mining by kimberlite), while a raft of basaltic rock occupies the
De Beers (1977–1982), Letšeng Diamonds (Pty) Ltd. upper part of the pipe (again, see figure 6).
(2004–2006), and Gem Diamonds Ltd. (2007 to the The Main pipe accounts for 75% of the mine’s
present) has removed most of these near-surface de- output and has an average grade of 0.88 carats per
posits. Mining currently takes place in the hard kim- hundred metric tons of material, while the Satellite
berlite, which requires blasting and crushing.
Similar to kimberlites elsewhere, the pipes con-
tain several pulses of kimberlite magma that are
slightly different in age and, more importantly, in the
In Brief
size, content, and quality of the diamonds. It is there- • Letšeng has produced some of the world’s largest dia-
fore necessary to outline their horizontal and vertical monds in the past decade, with a production value av-
eraging more than $2,500 per carat.
dimensions to recover diamonds of the highest grade.
Four varieties of kimberlite are recognized in the • Letšeng’s diamonds are unusual for their dodecahedral
Main pipe: K North, K South, K6, and K4 (again, see form and high percentage of type IIa specimens.
figure 6, top). K6 is economically the most impor- • The mine was sub-economic for many years because
tant, as it contains the highest diamond content. It of its low ore grade, until prices for very large dia-
monds began rising sharply.
forms a separate sharply demarcated explosive vol-
canic breccia pipe known as a diatreme in the west- • To remain economically viable, the mine employs new
technology to reduce breakage of large stones in the re-
ern part of the kimberlite, measuring four hectares
covery process.
(245 ×185 m). This is the area previously mined by
De Beers and now by Gem Diamonds. K6 is a soft,
friable, greenish gray rock with visible crystals of
olivine and occasionally orthopyroxene, set in a fine pipe has a yield of 2.6 carats per hundred metric tons
groundmass. Xenoliths of basaltic lava are abundant (Gem Diamonds Ltd., 2013). Their average is by far
the lowest for any diamond mine in the world, which
range from 5 to 400 carats per hundred metric tons
Figure 7. The informal diggings in the Maloti but typically from 50 to 100 carats per hundred met-
Mountains in 1967, at a swampy area later known ric tons. By comparison, De Beers’s Venetia mine, lo-
as Letšeng-la-Terae. From Nixon (1973); courtesy of cated 500 miles to the north, yields about 110 carats
De Beers.
per hundred metric tons (Krawitz, 2014).
EXPLORATION
Stockley (1947) reported kimberlites in the Maloti
Mountains of Basutoland as early as 1947, though
no diamonds were found at the time (Nixon, 1973).
Several additional kimberlites reported by Col. Jack
Scott in 1955 added to the notion there might be di-
amonds, though none were discovered. Scott later
got De Beers to assist him in further prospecting op-
erations. But it was Peter H. Nixon, as a Harry Op-
penheimer PhD scholar from the Research Institute
of African Geology at the University of Leeds, who
discovered in 1957 the two diamondiferous kimber-
lite pipes at Letšeng-la-Terae (Nixon, 1960). He
noted that Lesotho’s many kimberlite pipes are

The eureka moment for Lesotho came in 1967,
when a digger, Ernestine Ramaboa, unearthed a
brown colored diamond weighing 601.26 ct in the
area of Letšeng-la-Terae. Ramaboa and her husband,
Petrus, reportedly secreted the diamond away, feign-
ing illness to avoid arousing the suspicions of fellow
diggers, and then walked for four days across the
mountains to the capital city of Maseru to sell their
find (figure 8). News of the find spread quickly, and
it was reported that the government had to step in to
guarantee the Ramaboas’ safety (Balfour, 1981).
At the time it was reported to be the world’s
eleventh-largest gem-quality diamond—and the
largest reported found by an artisanal miner
(Sotheby’s, 2008). The diamond was named the
Figure 8. Ernestine Ramaboa and husband Petrus Lesotho Brown (figure 9). The government set up a
hold up the Lesotho Brown diamond, which she dis- sealed tender to obtain the best possible price for the
covered in 1967. At the time it was the world’s sev- diamond. According to the September 1967 issue of
enth-largest diamond. The Ramaboas reportedly Diamond News and S.A. Jeweller, this colossal gem
received more than $250,000 after taxes. Photo was first sold for 216,300 rand (equivalent to
courtesy of UPI. $302,400 at the time) to South African dealer Eugene
Serafini. In 1968 it was purchased, also via auction,
by American jeweler Harry Winston. A fitting epi-
mostly barren of diamonds. Occasional reports of
logue to the story is that Winston invited the Ram-
large diamond finds began to circulate nonetheless.
aboas to New York City for the unveiling and
Lesotho Kimberlites (1973), a book Nixon later com-
cleaving of the diamond. The moment was broadcast
piled and edited as chief geologist in the country’s
on live television across the United States (Balfour,
Department of Mines and Geology, noted that some
1981). The Lesotho Brown rough yielded 18 dia-
1,200 diggers were working the area by 1965, a num-
monds, including the Lesotho I, a flawless 71.73 ct
ber that quickly swelled to some 6,000 within two
diamond, and the Lesotho III (figure 10), a 40.42 ct
years.
marquise purchased by Aristotle Onassis and given
to Jacqueline Kennedy as an engagement ring. In
Figure 9. This postage stamp, issued in 1976, shows 1996 the Lesotho III ring sold at an auction of the for-
the colossal 601.26 ct Lesotho Brown diamond. mer First Lady’s estate for $2,587,500 (“Jackie O
The image pays homage to the Ramaboas (seen in Lesotho diamond sells…”, 1996).
the background) and Lesotho’s King Moshoeshoe II
(encircled).
Figure 10. The Lesotho III, a 40.42 ct marquise fash-
ioned from the Lesotho Brown, was given by Aristotle
Onassis to Jacqueline Kennedy for their engagement.
Photo courtesy of Sotheby’s.

GOVERNMENT DIGGINGS was nearly 65% of sales. The mine’s profitability was
Ramaboa’s discovery brought a wave of local small- an open question, given the amount of investment
scale miners to the Letšeng-la-Terae area. Despite needed for its infrastructure (“Onstream, upstream,”
their rudimentary tools and mining methods, they 1977). Announcing the decision, a De Beers press re-
found a number of diamonds (Maleleka, 2007). In an lease detailed the challenges:
effort to gain control over Letšeng-la-Terae, the gov- The design of the recovery plant took into consideration the
ernment, through the Lesotho National Develop- particular situation of the mine—the difficulties of access,
ment Corporation, granted Rio Tinto Zinc (RTZ, the effect of the gale force winds and the sub-zero tempera-
now Rio Tinto) an exploration license in 1968. RTZ tures on site. For example, gale force winds experienced at
upgraded the road and airstrip in the remote area and Letšeng would pose problems for conveyor scaffolds and the
built a sampling plant. Beginning in 1969, RTZ sam- normal sized recovery plant buildings and the icy conditions
pled the Main pipe to about 55 meters deep, recover- could affect the conventional slurry and gravity circuits.
The steeply sloping ground too had to be considered in re-
ing 2,635 carats over the duration of the project. The
lation to the pumping of slurry and gravels in the treatment
company’s final report, issued in 1972, confirmed plant. The plant designers, who also have to bear in mind
what many people suspected: that the grade was very the recovery of the larger diamonds peculiar to the main
low, though it occasionally produced unusually large pipe at Letšeng, took the decision to excavate and locate the
diamonds. In the sampling carried out by RTZ, for plant within the mountainside itself secure from the ele-
example, over 90% of the value came from less than ments (De Beers Consolidated Mines Limited, 1975).
10% of the diamonds—six stones, in fact—one of Furthermore, a road costing three million rand ($3.7
them nearly 50 ct (Chadwick, 1981). There was also million at the time) would be built through the moun-
the remoteness and inaccessibility of the area, which tains, connecting the mine to both Johannesburg,
at the time made development costs uneconomically South Africa and the country capital, Maseru. The
high. After RTZ declined to mine the deposit, the mine officially opened in November 1977. At a cost of
government granted licenses to individual diggers more than 30 million rand (US$37 million), it went at
using limited mechanization who worked three- least 6 million rand ($7.4 million) over its intended
square-meter claims. For their part, the locals con- budget. De Beers geared the plant to do the primary
tinued to find diamonds (Maleleka, 2007). sorting and washing to identify larger gems before the
After RTZ’s departure, the government of Lesotho ore went to the crusher. The company also installed
asked De Beers in February 1975 to reevaluate the data an X-ray machine to further identify larger diamonds
collected by Rio Tinto. In a move that surprised some after the initial crushing to retrieve them before sec-
analysts, De Beers chairman Harry Oppenheimer ondary crushing (Chadwick, 1981).
signed an agreement with the prime minister in De- In 1978, a 130 ct colorless diamond was recovered,
cember 1975 to develop the mine. “De Beers will go which mine manager Keith Whitelock noted would
ahead with the mine in spite of the risk—an obvious help defray operating costs for a year. Diamond prices
risk judged by Rio Tinto’s findings against such a ven- peaked in 1980, with the high-quality stones leading
ture,” according to one newspaper story (Payne, 1975). the way. But prices fell sharply the following year and
Even De Beers harbored some misgivings. As Oppen- continued sliding. In 1982, De Beers decided to close
heimer noted, “We are certainly doing something that operations (figure 11) because the cost of running
is pretty risky which I don’t think other people would Letšeng had become untenable (Blauer, 2010).
be inclined to do, and while I think that this is some For its five years of operation, De Beers’s produc-
way of philanthropy, we certainly were influenced by tion totaled 272,840 carats recovered from processing
our desire to play a part in the development of 9.4 million tons of kimberlite at the Main pipe, for a
Lesotho” (“Latest reports from the Lesotho Dept. of grade of 3.03 carats per hundred metric tons (table 2).
Mines,” 1978; Madowe, 2013). The agreement fac- About 65% of the output was gem quality, while
tored in a 25% government share and a seat for the 27% was industrial and 8% was bort (suited for
government on the board of directors of a Lesotho- crushing into abrasives). Only 7.6% of the produc-
based joint venture formed by De Beers. Between 600 tion was larger than 14.7 ct, yet these accounted for
and 900 people, mostly Basotho, were to be employed. 62% of the mine’s revenue (Venter, 2003).
Diamonds would be sold through De Beers’s mar-
keting arm, then called the Central Selling Organi- SECOND LIFE FOR LETŠENG
zation (CSO). The amount owed to the state from tax After De Beers’s departure, the mine was worked spo-
and dividends, once the capital had been amortized, radically by artisanal diggers from 1983 to 2003, but

no formal mining was done. A second life for Letšeng
began in 1999 when the Lesotho government, com-
ing off several years of political instability and vio-
lent demonstrations in Maseru, sought help to
redevelop the resource to provide sorely needed na-
tional revenue (Shor and Weldon, 2014). The result
was a consortium, Letšeng Holdings SA, comprised
of Matodzi Resources, Johannesburg Consolidated
Investments (JCI), and Letšeng Guernsey, that took
a 76% share in the venture, with the government
holding the remainder. JCI, a South African gold min-
ing firm, provided the start-up funding of approxi-
mately US$20 million (Venter, 2003). After extensive
negotiations, the government and consortium agreed Figure 11. Diamonds over 3 ct, such as these two dia-
to a 7% royalty on diamonds sold, coupled with a monds found recently at the recovery plant at
Letšeng, were extracted from tailings left behind from
35% corporate income tax—much lower than the
De Beers’s operation at Letšeng-la-Terae. Photo by
15% royalty and 50% corporate tax levied on De Robert Weldon/GIA; courtesy of Letšeng Diamonds.
Beers (Venter, 2003). By the time the mine was ready
to begin producing, JCI had acquired a 50% share of
the Letšeng Holdings consortium and commissioned
a second processing plant designed to double its ca- anticipated an annual output of 68,000 carats begin-
pacity. Mining began in May 2004, starting with the ning the following year (Hill, 2006; Telfer and
ore that De Beers had left stockpiled. The company McKenna, 2011).
TABLE 2. Diamond production of Letšeng, 1955–2014.

Period Carats Ore (metric tons) Value/ct (US$) Grade (cpht) Operator
1955–1968 62,070 nd nd nd artisanal
1969–1972 2,000 80,000 nd 2.5 RTZ
1977–1982 272,840 9.4 M nd 3.03 De Beers
2004 14,000 nd nd nd Letšeng-JCI
2005 37,000 nd nd nd Letšeng-JCI
2006 54,677 2.6 M 1,100 2.11 Letšeng-JCI
2007 73,916 4.0 M 1,976 1.85 Letšeng–Gem Diamonds
Value in US$ per carat; cpht = carats per 100 metric tons; nd = no data
Sources: Bowen et al. (2009) and Gem Diamonds annual reports for 2007 to 2014

open-pit mining. Traditional open-pit diamond min-
ing involves gradual hollowing out of the entire kim-
berlite pipe in circular stages and, as the pit grows
deeper, cutting away walls in progressively wider
concentric circles. With each circle, the cut (or
“waste stripping”) becomes wider, so deeper mines
incur increasingly higher costs as the percentage of
waste to kimberlite rises. In addition, waste disposal
becomes much more costly.
With the split-shell concept, the waste stripping
and mining is done on one side of each pit in alter-
nate cycles—one cycle on the northern side of each
pit, then switching to the southern side. The waste
is removed via a common ramp between the two
pipes. The split-shell method can target higher-grade
Figure 12. This state-of-the-art processing plant at
areas of each pipe and defer a large amount of waste
Letšeng is designed to minimize damage to large dia-
monds and improve the recovery rate. The plant takes stripping, because only part of the kimberlite is being
advantage of the region’s mountainous topography by mined at any one time. As a result, mining compa-
incorporating a gravity-fed recovery process. Photo by nies can recoup their costs much faster. A study con-
Robert Weldon/GIA. ducted at De Beers’s Venetia mine in South Africa,
which converted to the split-shell method after seven
years of traditional mining, found it reduced costs by
25%. Earthmoving equipment requirements were re-
Expansion began in earnest in 2005, when duced by 30%, and yearly waste removal went from
Letšeng Holdings held an initial public offering (IPO) 32 million to 20 million metric tons, while maintain-
with the goal of raising $31.4 million to build the ing the same level of diamond production (Gallagher
new processing plant (figure 12). The sale was buoyed and Kear, 2001).
by the discovery of a 186 ct diamond, quickly fol- There are risks, however. First, concentrating
lowed by a 72 ct stone (Zhuwakinyu, 2005). mining operations in the highest-grade sections of
Despite the IPO, the company fell short of the ad- the deposit for too long can shorten the mine’s life
ditional $26 million needed to develop the mine to by rendering the lowest-grade areas of the pipe un-
full operation. As a result, JCI began negotiations to economic. Mining operations must be carefully bal-
sell the Letšeng Holdings stake to Gem Diamonds, a anced between the sections of the pipe to ensure this
deal that was finally signed in the fall of 2006 for does not happen. In addition, the costs will eventu-
$130.1 million. It resulted in a restructured owner- ally rise after it becomes necessary to remove the
ship with a new name, Letšeng Diamonds Ltd. The waste from the sections of the pipe that have been
government took a 6% larger stake, now totaling deferred (Gallagher and Kear, 2001).
30%, in exchange for a $5.54 million loan that would At Letšeng, De Beers mined a small section called
be repaid to the new company (Hill, 2006). Gem Di- K6, the highest-grade portion of the Main pipe down
amonds owns the remaining 70% interest. to 120 meters, and began working the “softer ore”
Restarting production remained a challenge be- section of the Satellite pipe (Telfer and McKenna,
cause of the difficulties noted above and because the 2011). Today, most of the mining is done in the area
depth of the deposit was never fully sampled by Rio adjacent to the section De Beers worked (K4). The
Tinto or De Beers (R. Burrell, pers. comm., 2014). Main and Satellite pipes are being worked at about
150 meters, with indicated resources (sampled with
MINING AND PROCESSING reasonable knowledge of the grade) down to 210 me-
When Gem Diamonds acquired the right to mine the ters. Gem Diamonds engineers note that the mine’s
concession in 2006, Letšeng had resumed active min- previous operators did not adequately sample deeper
ing for just over a year, with an annual production of areas of the pipe. The company has recently sampled
55,000 carats. To lower costs and increase efficiency, down to a depth of 476 meters at the Main pipe, still
Gem Diamonds turned to a relatively new mining finding diamonds but uncertain of the grade. This is
method, designed for twin pipes, called split-shell an inferred resource. The Satellite pipe has an in-

TABLE 3. Probable mining reserves as of December
31, 2010.
Complicating the mining of the Satellite pit is
Occurrence Tonnage Grade Carats Value
(cpht)a per carat that during the kimberlite eruption 90 million years
ago, an immense chunk of basalt country rock—
Satellite pipe 17.02 M 2.07 353,000 US$3,170
about 1 hectare (10,000 square meters) wide and 150
Main pipe 59.22 M 1.46 867,000 US$2,414 meters deep—was detached into the southern sec-
Stockpile 0.17 M 0.87 1,500 US$1,209 tion of the kimberlite pipe. This creates an immense
Total 76.41 M 1.60 1,221,700 US$2,631 waste expense and an obstacle to mining the pipe
(see figure 6, bottom). Current plans are to mine the
a
cpht = carats per 100 metric tons open pits until 2038 (Gem Diamonds Ltd., 2015b).
Letšeng produces approximately seven million
metric tons of ore and 20 million metric tons of
ferred resource down to 628 meters. The company waste per year and employs local contractors for
has test-bored as deep as 800 meters and is still eval- much of its mining activities, with the exception of
uating the samples (R. Burrell, pers. comm., 2014). blasting. Drilling is done with Atlas Copco Roc L8
A 2011 report noted that indicated resources as of and L6 machines. The mine employs a mixed loading
December 31, 2010, were calculated at about 77 Mt and hauling fleet of Caterpillar 385C and 390D exca-
(million metric tons), at a grade of 1.61 carats per vators, together with Caterpillar 773 (50-ton capac-
hundred metric tons (cpht), yielding 1,238,000 carats ity) rigid-body dump trucks for hauling waste and
worth $2,629 per carat at 2011 market prices, or Caterpillar 740 (40-ton capacity) articulated dump
roughly $33 billion. Inferred resources added another trucks hauling ore (figure 13). The mixed fleet is
141 Mt at a grade of 1.72 cpht, yielding 2,423,200 suited for handling different ore and waste profiles
carats worth $2,733 per carat, or $67 billion (Telfer during various stages of mine development.
and McKenna, 2011). This equates to a 30-year mine
life. A more immediate probable mining reserve is IMPROVING RECOVERY AT LETŠENG
calculated at 76.4 Mt at a grade of 1.60 cpht, yielding Letšeng’s financial model is dependent upon its abil-
1,221,700 carats worth $2,631 per carat in 2011 (see ity to recover large stones intact. Because of this,
also tables 3 and 4). standard diamond recovery processes, which crush
Letšeng is also unusual in that it yields only a the ore into small pieces for fast, efficient recovery,
small percentage of diamonds less than two millime- cannot be used without major modifications. Gem
ters. Because of this, the recovery operations do not Diamonds invested $9.9 million in new diamond
screen for diamonds below that size (Bowen et al.,
2009).
Figure 13. A 50-ton dump truck and front-end loader
From these figures, the value of one ton of ore can
remove the kimberlite ore after blasting at the bottom
be calculated as roughly $42, which is in line with of the Main diamond pipe. Photo by Robert Weldon/
other economic kimberlite mines in southern Africa. GIA; courtesy of Letšeng Diamonds.
Previous and current mining has shown that the
Satellite pipe has a slightly higher grade and a slightly
higher content of large diamonds than the Main pipe
(Telfer and McKenna, 2011).
TABLE 4. Probable, indicated, and inferred reserves

of Letšeng, as of December 31, 2014.
Probable Indicated Inferred Total
Ore (million tons) 131.9 187.1 106.8 293.9
Grade (ct/100 tons) 1.71 1.73 1.67 1.71
Carats (million) 2.26 3.23 1.78 5.01
Value per carat US$2,045 US$2,036 US$2,051 US$2,073
Source: Gem Diamonds Ltd. (2015a)

TABLE 5. Production of large diamonds at Letšeng,
2005–2014.
Size (ct) 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
processing technology to spare the large stones, all
because of changes in the diamond market since the
mine first operated in the 1970s and 1980s. As Gem >100 6 5 5 7 5 6 5 3 7 7
Diamonds chief executive officer Clifford Elphick ex- 60–100 7 14 9 16 10 10 19 13 16 21
plained, prices for very large diamonds were rela- 30–60 30 40 39 74 76 61 59 61 50 69
20–30 49 47 65 88 98 89 91 110 71 101
tively low compared to smaller stones, so it was
more cost-efficient to damage such goods during pro-
cessing in the name of speed. In recent years, how- Total >20 ct 92 106 118 185 189 166 174 187 144 198
ever, the rising number of billionaires and Source: Gem Diamonds annual reports
ultra-wealthy connoisseurs has pushed prices of
major diamonds to record levels. Now mines must
increasingly focus on recovering these high-value
stones—especially as they go deeper and mining
costs rise. lion metric tons of ore per year. An additional 1.3
After Gem Diamonds acquired a majority share million tons per year, mainly from waste dumps, are
of Letšeng Diamonds, the company built a second processed by an outside contractor, Alluvial Ven-
treatment plant that would double the mine’s capac- tures, through their two plants.
ity. The plant’s owners also introduced new technol- The primary ore crushers break the large kimber-
ogy to minimize damage to large diamonds and lite blocks into pieces about 125 mm in diameter.
reduce construction and operating costs. The design From there the ore goes through a scrubbing process
of Plant 2 took advantage of the mine’s topography, that uses vibration to remove excess clay, rock, and
with the processing plant set on a 17° slope. Material other foreign matter. In this process, the kimberlite
would be fed from the top end, letting it flow down- pieces grinding together do the scrubbing without ad-
ditional grinding agents that could damage the
ward stage by stage through the separating and crush-
stones. From there they proceed by slurry to the
ing processes, reducing the need for pumps and
coarse crushing, which sizes the material down to 45
conveyors (Bornman, 2010). Plant 2 went online in
mm diameter—a 200 ct rough is approximately 30
the fall of 2008. Together the two plants treat 5.7 mil-
mm in diameter.
Although the company had recovered a number
Figure 14. Inside Plant 2, an observation window of large diamonds by 2013, executives remained con-
into the conveyor belt taking the large pieces of kim- cerned that major stones were not surviving the
berlite to the crushers. Photo by Robert Weldon/GIA; crushing process, particularly because Letšeng type
courtesy of Letšeng Diamonds. IIa diamonds generally display excellent cleavage,
and a significant proportion of them are cleavage
fragments (Bowen et al., 2009). At the beginning of
that year, the throughput of ore was deliberately
slowed to allow closer observation of the conveyor
to spot large stones (figure 14). After several months,
though, there was still no clear correlation between
processing speed and diamond damage.
In the second quarter of 2013, the company re-
placed the Plant 2 crushers, large steel jaws that
broke the ore down to small pieces in three stages,
with a cone crushing system, configured specifically
for Letšeng’s production (Cornish, 2013). In this new
system, the ore is loaded into a receptacle and passed
over a rapidly spinning cone that also oscillates side
to side, crushing the ore against the sides of the pas-
sage without exerting the diamond-damaging pres-
sure of the jaw system (Gem Diamonds Ltd., 2014).
Gem Diamonds subsequently replaced the crushers
in Plant 1 as well. According to Letšeng Diamonds

tals by measuring their specific radiometric response
to X-rays and eject them from further processing
(Cornish, 2013). The general principle behind XRT is
that diamond shows up as a much brighter image
than other Si-bearing minerals. Minerals containing
Fe or heavier elements will show up even darker.
XRT sorting does not rely upon a diamond’s trans-
parency because it does not deal with visible light
range. Even if the sample is sufficiently coated to be
completely opaque, the XRT sorter will detect it. The
company tested these X-ray units on tailings, recov-
ering $12 million worth of diamonds over six months
(Cornish, 2013; Gem Diamonds Ltd., 2014).
Figure 15. Mazvi Maharasoa, CEO of Letšeng Dia- CHARACTERISTICS AND MARKETING OF
monds, the joint venture between Gem Diamonds LETŠENG DIAMONDS
and the Lesotho government, photographed in front of The Letšeng mine is famous for its very large dia-
the processing plant. Photo by Russell Shor/GIA; monds (tables 1 and 5). The largest, the Lesotho
courtesy of Letšeng Diamonds.
Promise (figure 16), was discovered in 2006. At 603
ct, it surpassed the 601.26 ct Lesotho Brown. Al-
though only marginally larger, the Lesotho Promise
chief executive officer Mazvi Maharasoa (figure 15), is a type IIa diamond, and its absence of color re-
the new crushers sized the kimberlite chunks into sulted in a stunning collection of 26 diamonds of D
pieces roughly 50 mm (about 2 inches) in diameter, color and Flawless clarity (figure 17). A significant
instead of 20 mm as in the past. percentage of diamonds from Letšeng have excep-
The result, beginning in May 2013, was an imme- tional clarity and transparency, no fluorescence, and
diate improvement in the number of large diamonds
recovered. From January through April 2013, Letšeng
yielded one diamond over 100 carats and three dia- Figure 16. The 603 ct Lesotho Promise is the fifteenth-
monds between 45 and 55 carats (table 5). From May largest rough diamond ever discovered. London lux-
2013 until year’s end, 17 very large diamonds were ury jeweler Laurence Graff purchased it in 2006 for
recovered, including one over 150 carats, three that $12.4 million. Photo courtesy of Laurence Graff.
were approximately 100 carats, and 13 between 60
and 80 carats (Gem Diamonds Ltd., 2014).
After crushing, the kimberlite pieces are con-
veyed to the recovery area, where the diamonds are
separated from waste rock by X-ray technology.
XRL (X-ray luminescence) and XRF (X-ray fluores-
cence) units, developed in the 1980s, take advantage
of the fact that diamonds will fluoresce when ex-
posed to intense X-rays. The crushed kimberlite
particles are dropped in a thin stream under a beam
of X-rays. When a diamond fluoresces, it activates a
photo-detector, triggering a jet of air that deflects the
diamond into a collector box (Valbom and Dellas,
2010). Older X-ray recovery units identified dia-
monds by fluorescence alone, causing nonfluorescent
diamonds (which include many type II) and some
fancy-color diamonds to slip through to the waste
piles. More recent methods adopted at Letšeng em-
ploy X-ray transmissive (XRT) technology before the
crushing process to identify large rough type II crys-

BOX B: MORPHOLOGY OF LETŠENG DIAMONDS
Most natural gem‐quality diamonds contain nitrogen
impurities (type I) and grow in the form of an octahedron.
A small percentage contain no measureable nitrogen im-
purities in their lattice (type II) and often occur as irreg-
ular, flattened, or elongate crystals. Diamonds usually
crystallize in the earth’s mantle in rocks known as peri-
dotites or eclogites. These are picked up and transported
to the surface by a type of molten rock called kimberlite.
Figure B-1. With increasing resorption, the morphology

of a diamond crystal changes from octahedral to do-
decahedral forms. From Tappert and Tappert (2011).
Volume Loss
≤1%
Figure B-2. These diamonds, each weighing approxi-

3% mately 4 ct, represent several days of mine production at
Octahedral
habit
Letšeng in September 2014. Most crystals show strong re-
sorption features and dodecahedral forms. Photo by
RE
SO
Robert Weldon/GIA.
RP
TI
5%
O
N
Diamonds are chemically unstable in a kimberlite

magma, and the outer surface of the diamond is often
20% etched and partially dissolved during transport, a process
Transitional known as resorption. This process causes an octahedron
habit
to lose volume and gradually convert to a dodecahedral
form (figure B-1).
35%
Several days of production at the Letšeng mine in Sep-
tember 2014 yielded the group of approximately 4 ct dia-
monds seen in figure B-2. Nearly all of them revealed
strong resorption marked by smooth, rounded surfaces
45->99% with indistinct features. The near-colorless and brown
(Rounded) stones showed irregular shapes, typical of type IIa dia-
Dodecahedral
habit
monds, whereas the three yellow type Ia diamonds in the
center presented obvious dodecahedral forms. Of the
larger stones in figure B-3 (approximately 8–10 ct range)
that were available during the mine visit, all were near-
are classified as D-color and Flawless (figure 18). for the year. The mine recovered a record 198 dia-
They are type IIa diamonds with roughly dodecahe- monds larger than 20 ct, 54 more than the previous
dral shapes; octahedral shapes are very rare, compris- year. Letšeng’s 2014 production was valued at an av-
ing less than 1% of the Letšeng population (Harris, erage of $2,540 per carat, buoyed during the final
1973; Bowen et al., 2009; see box B). quarter by the recovery of 13 large diamonds whose
In 2014, Letšeng reported that it found seven dia- price exceeded $1 million each, including a 299.3 ct
monds over 100 ct. By year’s end it had sold five of yellow diamond (price undisclosed), a 112.6 ct color-
them, which represented 14% of Letšeng’s total sales less diamond that sold for $5.8 million, and a 90.4 ct

Figure B-3. Larger dia-
monds from Letšeng, like
these crystals ranging
from 8 to 10 ct, were
usually type IIa with ir-
regular forms. Photo by
Robert Weldon/GIA.
colorless to slightly brownish with distinctive resorption on both counts.

features and irregular or elongate shapes, characteristics in- Type IIa diamonds from Letšeng usually have irregular
dicative of type IIa diamonds. shapes. The 22.87 ct type IIa stone in figure B-4 (left) has
Letšeng diamonds are well known for some degree of excellent color and clarity but no distinct form other than
dodecahedral form, with more than 95% of the recovered a single octahedral face with visible trigons. In contrast,
stones showing significant resorption effects (Bowen et type Ia diamonds show more distinct forms such as the
al., 2009). The mine is also known for producing a high partially resorbed 32.97 ct yellow octahedron in figure B-
percentage of large type IIa diamonds of exceptional color 4 (right). Octahedral forms like this are relatively rare at
and clarity, but with irregular crystal forms. The Letšeng Letšeng.
mine production available to the authors was consistent
Figure B-4. These two

crystals from Letšeng il-
lustrate the morphologi-
cal differences between
irregularly shaped type
IIa diamonds such as the
22+ ct stone on the left
and type Ia octahedral
stones such as the 32+ ct
stone on the right. Photos
by Robert Weldon/GIA.
colorless diamond that brought $4.2 million (Max, ernment, which receives at present a 7% royalty
2015). In the first half of 2015, the average value based on value. From the mine they are exported to
slipped to $2,264 per carat, though 13 diamonds of- Antwerp, where they undergo another KP check be-
fered during that period sold for more than $1 million fore proceeding to the company’s sales office at Lange
each (Gem Diamonds Ltd., 2015c). Herentalsestraat in the city’s diamond district. At the
The initial sorting and valuing is done at the mine sales office, the rough is cleaned by an acid wash to
site to satisfy the Kimberley Process (KP) require- remove all surface material and then revalued and
ments for export and to establish values for the gov- sorted for sale (C. Elphick, pers. comm., 2014). Dur-

Figure 19. This rough diamond dodecahedron from
Letšeng weighs 11.14 ct. Given its rare, attractive
shape and color, the stone would command a high
asking price at auction. Photo by Robert
Weldon/GIA; courtesy of Gem Diamonds Ltd.
goods grouped by weight, quality, and shape.

Letšeng rough is offered ten times per year in on-
line auctions conducted through its Antwerp office
Figure 17. Twenty-six D-Flawless diamonds of various (figure 19). Diamonds larger than 100 ct are offered
shapes cut from the Lesotho Promise were assembled individually in special sales.
by Graff into a necklace valued at $60 million. Photo Unlike other smaller diamond producers, Gem
courtesy of Laurence Graff. Diamonds maintains its own sales office and does
not employ a contractor. Selling its own goods in this
manner provides Letšeng with complete flexibility
ing the sorting process, the Antwerp sales team first over the marketing of its rough production, said CEO
sets reserve prices on individual rough diamonds Clifford Elphick (figure 20), while reducing sales and
larger than 10.8 ct and then on parcels of smaller marketing costs from 2.5% to 1.5%.
“We have about 150 customers for each sale,” said
Elphick. “They have a ten-day viewing period to re-
Figure 18. This group of type IIa rough diamonds ranges view the goods, and can bid up to the final minute.”
from almost 14 ct to 91.07 ct. Such diamonds make Unlike some rough diamond auction models
Letšeng a feasible mining proposition. Photo by Robert where high bids are averaged into across-the-board
Weldon/GIA; courtesy of Gem Diamonds Ltd. prices for a particular category, the winners buy their
lots for their high bid price. “Occasionally, if a bid
seems unrealistically high, we will ask the client to
review it,” Elphick explained. “Once in a while some
people misplace a decimal point.”
Elphick noted that in the eight years of selling di-
amonds from Letšeng, both the sales staff and clients
have developed a strong understanding of the goods
and their prices, so the process usually runs very
smoothly.
While the very large stones capture the headlines
and high price tags, most of Letšeng’s earnings come
from commercial goods (figure 21) and 5–15 ct goods,
which actually achieve the highest per-carat prices.
“We get prices over $50,000 per carat for the top qual-
ities: type II goods of high color, which we call

Figure 22.These rough diamonds, a 3.72 ct faint
pink and a 9.55 ct grayish blue stone, are from
Figure 20. Clifford Elphick, CEO of Gem Diamonds,
Letšeng’s limited production of fancies. Photo by
emphasizes the value of large type IIa diamonds but
Robert Weldon/GIA; courtesy of Gem Diamonds Ltd.
notes that a significant portion of production in-
cludes commercial goods. Photo courtesy of Gem
Diamonds Ltd.
ber of ultra-wealthy individuals over the past 15
years has sparked demand for large, high-quality di-
‘Letšeng-y,’” he said. “The cutters like these because amonds, driving prices higher and making Letšeng
they are easy to work, with high yields. Many of profitable to mine again. But Elphick added that the
them polish to D color—perhaps a little higher than per-carat price begins diminishing above 15 carats be-
a normal D, because they are type II with very little cause of the risks still inherent in diamond cutting.
graining.” “If they make a mistake on a ten-carat stone, it won’t
In addition, the mine occasionally produces some put them out of business. But when millions of dol-
fancy-color diamonds—two or three blues a year, and lars are involved, they have to hedge a bit.”
a larger number of pinks and yellows (figure 22). Lots that do not sell at the company tender auc-
Elphick explained that the explosion in the num- tions are sent to contractors for polishing. These ac-
Figure 21. During a visit

to Letšeng, the authors
were offered the oppor-
tunity to photograph
several days’ worth of
production. The mine
produces a number of
“three-grainers,” such
as this group of 88
stones weighing a total
of 66.40 carats. Photo
by Robert Weldon/GIA;
courtesy of Letšeng Dia-
monds.

Figure 23. These rough
diamonds (22.30, 91.07,
and 23.37 ct) were auc-
tioned through Letšeng’s
sales office in Antwerp.
Such exceptional type
IIa diamonds of un-
usual shape—many are
cleavage fragments like
these—comprise a sig-
nificant part of Letšeng’s
production and offer the
tantalizing possibility
that they were once
parts of larger gems.
Photo by Robert Wel-
don/GIA; courtesy of
Gem Diamonds Ltd.
count for $15 to $20 million yearly, he said, noting the challenges of moving equipment to a remote lo-
that most of the buyers for polished goods are also cation high in the mountains of Lesotho, the mine’s
the company’s rough clients. current majority owner has invested in radically
All of Letšeng’s unsold rough is polished in modified diamond recovery methods to greatly re-
Antwerp. Elphick and Maharasoa both stated that duce the damage to large diamonds, particularly
the plan is to establish a polishing operation in those above 50 ct. The company markets Letšeng’s
Maseru as part of a goal for the mine to contribute unique production through its own sales office in
15% of Lesotho’s gross national product. Elphick said Antwerp (figure 23), via regularly scheduled auctions.
that the company already has the space to begin a Despite the profitability, there is concern that the
small polishing operation in the capital but is await- mining license is finite, expiring in 2019, and that a
ing permits from the government. renewal may carry more costly terms. In a 2010
“It could never become a large operation that paper, Mazvi Maharasoa, CEO of Letšeng Diamonds,
would compete with those in India,” Elphick ac- noted that operating costs of 70 million loti per year
knowledges, “but we can start small and see if we (US$6 million) were too high given the finite nature
can expand later.” of the lease, the fluctuating exchange rates, and the
costs of building and maintaining access roads and
CONCLUSIONS transporting supplies. Demand for very large dia-
The Letšeng mine is a unique diamond resource monds such as those produced at Letšeng has re-
whose economic viability has been realized by strong mained high through turbulent economic times, and
increases in per-carat prices for large diamonds since the mine’s operators are confident that it will con-
2000. While earlier attempts to mine the deposit tinue to yield large diamonds, the sale of which will
were unsuccessful because of the low ore grade and make headlines around the world.
ACKNOWLEDGMENTS
ABOUT THE AUTHORS
The authors would like to thank the kind assistance of Mazvi
Mr. Shor is senior industry analyst, Mr. Weldon is manager of
Maharasoa, CEO of Letšeng Diamonds (Maseru, Lesotho); Clif-
photography and visual communications, and Dr. Breeding is a
ford Elphick, CEO of Gem Diamonds (Johannesburg, South
research scientist, at GIA in Carlsbad, California. Dr. Janse is a di-
Africa); Robert Burrell, manager of strategic technical services of
amond exploration consultant living in Perth, Western Australia.
Letšeng Diamonds; and Darrol van Deventer, principal mining
Dr. Shirey is a senior scientist in the Department of Terrestrial
engineer of Gem Diamonds. Prof. Steve Richardson (University
Magnetism of the Carnegie Institution in Washington, DC.
of Cape Town) provided a final review of the manuscript.

REFERENCES
Balfour I. (1981) Famous diamonds of the world (X): the Lesotho Lock N.P., Dawson J.B. (2013) Contrasting garnet lherzolite xeno-
Brown. Indiaqua, Vol. 29, No. 2, pp. 123–125. lith suites from the Letseng kimberlite pipes: inferences for the
Blauer E. (2010) De Beers sells historic mines. Rapaport Diamond northern Lesotho geotherm. Proceedings of the 10th Interna-
Report, Vol. 33, No. 9, pp. 66–67. tional Kimberlite Conference, Vol. 1, pp. 29–42. Bangalore. Ge-
Bloomer A.G., Nixon P.H. (1973) The geology of Letseng-la-terae ological Society of India, Special Issue, 350 pp.
kimberlite pipes. In P.H. Nixon, Ed., Lesotho Kimberlites, Madowe A. (2013) The mining process for an open pit diamond min-
Lesotho National Development Corporation, Maseru, pp. 20–36. ing operation — A case study on Letseng diamond mine in
Bornman F. (2010) Letšeng mine no. 2 plant project – A process en- Lesotho. Journal of the Southern African Institute of Mining and
gineering and design review. Southern African Institute of Min- Metallurgy, Vol. 113, No. 7, pp. 547–554, http://www.saimm.co.
ing and Metallurgy: Diamonds—Source to Use 2010, Vol. 110, za/Journal/v113n07p547.pdf
No. 11 (November), pp. 271–284, http://www.saimm.co.za/ Maharasoa M. (2010) A company perspective on mining in Lesotho.
Conferences/DiamondsSourceToUse2010/271-Bornman.pdf Rethinking the management of mineral resources in Lesotho.
Bowen D.C., Ferraris R.D., Palmer C.E., Ward J.D. (2009) On the Southern Africa Resource Watch, No. 6, pp. 32–33,
unusual characteristics of the diamonds from Letšeng-la-Terae http://www.archive-org-2013.com/org/s/2013-11-23_3214541_
kimberlites, Lesotho. Lithos, Vol. 112S, pp. 767–774,
28/Rethinking-the-Management-of-Mineral-Resources-in-
http://dx.doi.org/10.1016/j.lithos.2009.04.026
Lesotho-sarwatch-org
Chadwick J. (1981) Big stones offset low grade at Lesotho’s dia-
Maleleka D. (2007) Diamond mining in Lesotho. Southern Africa
mond mine. Indiaqua, Vol. 29, No. 2, pp. 27–33.
Resource Watch, No. 4, September, http://www.resource
Cornish L. (2013) Gem Diamonds’ big expansion, on a minimal
budget. Mining News, May 16, http://www.miningne.ws/2013/ governance.org/sites/default/files/LesothoFA.pdf
05/16/a-big-expansion-on-a-minimal-budget-for-gem-diamonds/ Max D. (2015) Gem Diamonds’ Q4 “encouraging end to positive
Davis G.L. (1977) The ages and uranium content of kimberlites year.” Idex, Jan. 27, http://www.idexonline.com/FullArticle?id=
and associated rocks. Carnegie Institution of Washington Year- 40358
book 1976, pp. 631–654. Nixon P.H. (1960) A mineralogical and geochemical study of kim-
De Beers Consolidated Mines Limited (1975) Letšeng-la-Terai. berlites and the associated xenoliths. Unpublished PhD thesis,
Press release, March 4. University of Leeds.
Gallagher M.S., Kear R.M. (2001) Split shell open pit design concept Nixon P.H., Ed. (1973) Lesotho Kimberlites. Lesotho National De-
applied at De Beers Venetia mine South Africa using the Whittle velopment Corporation, Maseru, 350 pp.
and Gemcom software. Journal of the Southern African Insti- Onstream, upstream (1977) Indiaqua, Vol. 15, No. 1, pp. 20–21.
tute of Mining and Metallurgy, Vol. 101, No. 8, pp. 401–409, Payne A. (1975) Lesotho signs up with De Beers. Rand Daily Mail,
http://www.saimm.co.za/Journal/v101n08p401.pdf Feb. 5.
Gem Diamonds Ltd. (2014) Annual report 2013, http://www.gem- Scott Smith B.H., Nowicki T.E., Russell J.K., Webb K.J., Hetman
diamonds.com/gem/uploads/reports/reports/Gem-Diamonds- C.M., Harder M., Mitchell R.H. (2008) Kimberlites: Descriptive
AR.pdf geological nomenclature and classification. 9th International
Gem Diamonds Ltd. (2015a) Annual report 2014, http://www.gem Kimberlite Conference, Extended Abstract A-00124.
diamonds.com/gem/uploads/reports/reports/gem-diamonds-ar- Shirey S.B., Shigley J.E. (2013) Recent advances in the understanding
2014.pdf of the geology of diamonds. G&G, Vol. 49, No. 4, pp. 188–222.
Gem Diamonds Ltd. (2015b) Letšeng Mine - Life of Mine 2015, Shor R., Weldon R. (2014) Diamonds from the roof of the world.
http://www.gemdiamonds.com/gem/uploads/reports/presen- GIA Research & News, http://www.gia.edu/gia-research-news-
tation/2015-05-22-letseng-life-of-mine-plan-update-presenta- diamonds-roof-world-shor-weldon
tion-may-2015.pdf Sotheby’s (2008) Magnificent and extremely rare diamonds to be
Gem Diamonds Ltd. (2015c) H1 2015 trading update, July 30, offered in Geneva on November 19, 2008. Press release.
http://www.gemdiamonds.com/gem/en/media/press-releases/ Stockley G.M. (1947) Report on the Geology of Basutoland. Basu-
?ref=279 toland Government, Maseru, pp. vii, 7–114.
Harris J.W. (1973) Observations on Letseng-la-terae diamonds. In Tappert R., Tappert M.C. (2011) Diamonds in Nature: A Guide to
P.H. Nixon, Ed., Lesotho Kimberlites, pp. 37–38. Rough Diamonds. Springer-Verlag, Berlin and Heidelberg, p. 21.
Hill M. (2006) Gem all set for Letseng expansion as JCI approves Telfer C.A., McKenna N. (2011) National Instrument NI43-101:
sale. Mining Weekly, September 28, http://www.mining-
Technical Report on the Letšeng Diamond Assets in Lesotho.
weekly.com/article/gem-all-set-for-letseng-expansion-as-jci-
Venmyn Rand, Johannesburg, pp. 21–96.
approves-sale-2006-09-29
Valbom D., Dellas G. (2010) State of the art recovery plant design.
Jackie O Lesotho II diamond sells for $2.6 million at Sotheby’s (1996)
Southern African Institute of Mining and Metallurgy, Vol. 110,
Rapaport, http://www.diamonds.net/News/NewsItem.aspx?
ArticleID=1203 No. 11, pp. 243–252, http://www.saimm.co.za/Conferences/
Janse A.J.A. (1994) Is Clifford’s Rule still valid? Affirmative exam- DiamondsSourceToUse2010/243-Valbom.pdf
ples from around the world. In H.O.A. Meyer and O. Leonardos, Venter I. (2003) Lesotho gem-mine targets full output in 2004. Min-
Eds., Proceedings of the Fifth International Kimberlite Confer- ing Weekly, Aug. 28, http://www.miningweekly.com/article/
ence 2, Diamonds: Characterization, Genesis and Exploration, lesotho-gemmine-targets-full-output-in-2004-2003-08-29
pp. 215–235. Departamento Nacional da Produção Mineral, Whitelock K. (1979) Letseng-la-Terai: The diamond mine at the
Brasilia, 390 pp. turn by the swamp on the roof of Africa. Optima, Vol. 28, No.
Krawitz A. (2014) De Beers value proposition. Rapaport, 2, pp. 120–136.
http://www.diamonds.net/News/NewsItem.aspx?ArticleID=4 Wilson M.G.C., McKenna N., Lynn M.D. (2007) The Occurrence
8494&ArticleTitle=De+Beers+Value+Proposition of Diamonds in South Africa. Council for Geoscience, Pretoria,
Latest reports from the Lesotho Dept. of Mines (1978) Diamond 104 pp.
News and S.A. Jeweller, Vol. 40, No. 1, pp. 35–37. Zhuwakinyu M. (2005) Lesotho gem-miner to be dual listed. Min-
Lesotho (2014) South African History Online, http://www.sahistory. ing Weekly, May 13, http://www.miningweekly.com/article/
org.za/places/lesotho lesotho-gemminer-to-be-duallisted-2005-05-13

NOTES & NEW TECHNIQUES
ORIGIN DETERMINATION OF DOLOMITE-RELATED

WHITE NEPHRITE THROUGH ITERATIVE-BINARY
LINEAR DISCRIMINANT ANALYSIS
Zemin Luo, Mingxing Yang, and Andy H. Shen
In Chinese culture, white is usually regarded as

Provenance studies on nephrite jade are of gemo- the color of purity and perfection (He, 2009). Due to
logical as well as archeological importance. Ori- its quality and well-established historical reputation,
gin determination is difficult to achieve relying white nephrite from Xinjiang has maintained the
solely on typical gemological properties and ob- highest price in the Chinese market. Xinjiang
servations. In this study, the authors present a new nephrite is usually translucent, with colors ranging
statistical analysis method, based on linear dis- from white and greenish white to yellow. The best
criminant analysis (LDA) and trace-element data material from this province is white nephrite with
from laser ablation–inductively coupled plasma– greasy luster, known in the trade as “mutton fat”
mass spectrometry (LA-ICP-MS), to achieve sig- jade. The purity, fine texture, and luster of white
nificantly improved origin determination (an nephrite from Xinjiang come together to create an
average of 94.4% accuracy) for eight major elegant cultural symbol (figure 1C).
dolomite-related nephrite deposits in eastern Asia. White nephrite from other deposits, including
As the principle is to compare one location to the Baikal (Russia), Chuncheon (South Korea), and Qing-
rest in one round and continue such binary com- hai (China), has similar characteristics and is difficult
parison until all possible localities are tested, the
method is referred to as iterative-binary LDA (IB-
LDA). Combined with trace-element composition In Brief
data from LA-ICP-MS, it could prove to be a use-
ful technique for the geographic origin determi- • There is no universally accepted method to determine
the origin of East Asian dolomite-related nephrite.
nation of other gemstones.
• IB-LDA, a statistical analysis method developed by the
authors, uses an algorithm based on trace-element data
to determine the origin of nephrite from eight major
N ephrite jade has been of great interest to East
Asian cultures dating back to the Neolithic Age
(Wen and Jing, 1996; Tsien et al., 1996; Wen, 2001; Har-
production areas in China, Russia, and South Korea.
• This method yielded greater accuracy than the tradi-
low and Sorensen, 2005). Historically, it has been con- tional LDA method, and may be useful for determining
the origins of other gem materials.
sidered a symbol of wealth and power for Chinese
emperors and nobility (figures 1A and 1B). Nephrite
from countries such as China, Russia, and South Korea
to distinguish with the unaided eye or even micro-
(Ling et al., 2013) continues to attract considerable at-
scopic observation (Wu et al., 2002; Ling et al., 2013).
tention as a gem and ornamental material, particularly
In some cases, material from other locations has been
in the Chinese market. Thus, geographic origin deter-
falsely represented as Xinjiang nephrite jade to fetch
mination is important in terms of proper classification,
higher prices. A reliable quantitative method for ge-
valuation, and archaeological implications.
ographic origin determination is urgently needed to
protect nephrite consumers.
GEMS & GEMOLOGY, Vol. 51, No. 3, pp. 300–311,
Nephrite is a gem-quality tremolite or actinolite
http://dx.doi.org/10.5741/GEMS.51.3.300. polycrystalline aggregate. It is usually classified as ei-
© 2015 Gemological Institute of America ther dolomite-related or serpentine-associated, which
300 NOTES & NEW TECHNIQUES GEMS & GEMOLOGY FALL 2015
A C
Figure 1. A: An ancient pen-

dant discovered in the
tomb of Marquis Yi, ruler of
Zeng State during the War-
ring States period (ca. 475–
221 BC). © Hubei Provincial
Museum. B: The imperial
seal made with white
nephrite for Emperor Yong-
zheng during the Qing dy-
nasty (1723–1735 AD). ©
Sotheby’s. C: “Autumn
B
Moon Over Han Palace”
(left) and “Spring Morning
Over Han Palace” (right)
placed second in the Gems
& Jewelry Trade Associa-
tion of China’s 2012 Tian-
gong Award for excellence
in jade carving. © The
Boguan Auction Company.
have different formation processes and different con- uyan (Liaoning province, China); Luodian (Guizhou
centrations of Fe, Cr, Co, and Ni, as well as oxygen province, China); Liyang (Jiangsu province, China);
and deuterium isotopes (Yui and Kwon, 2002; Har- Baikal (Russia); and Chuncheon (South Korea). Sec-
low and Sorensen, 2005; Siqin et al., 2012; Adamo ond, because of their similar standard gemological
and Bocchio, 2013). Dolomite-related nephrite is properties, such as color, transparency, luster, refrac-
usually white, greenish white, yellow, or light gray, tive index, specific gravity, and major element com-
due to the relatively low concentrations of the ponents (Liao and Zhu, 2005; Ling et al., 2013; Liu
transition metal elements listed above. Serpentine- and Cui, 2002), dolomite-related nephrite jades from
associated nephrite is usually green. A typical exam- these locations are very difficult to distinguish.
ple is Siberian green nephrite from Russia, known Geochemical research has shown that trace ele-
for its purity of color. The geographic sources of ser- ments can reflect the sources of gemstones (Breeding
pentine-associated nephrite, such as New Zealand, and Shen, 2010; Blodgett and Shen, 2011; Shen et al.,
mainland China, and Taiwan, can be distinguished 2011; Zhong et al., 2013). But as the number of pro-
by some characteristic elements in chromite inclu- ducing localities and the complexity of chemical
sions and by strontium isotopes (Adams et al., 2007; composition increase, simply evaluating one or two
Zhang and Gan, 2011; Zhang et al., 2012). elements cannot distinguish the different origins
There is no widely accepted method for distin- (Siqin et al., 2012). Multiple elements, and correla-
guishing the geographic origin of dolomite-related tions between those elements, must be taken into
nephrite. Two factors account for this. First, there are account. Identifying and optimizing the variables
many sources for dolomite-related nephrite in East that characterize differences among origins becomes
Asia. Eight major locations are listed in figure 2: the main focus.
western Xinjiang (including the famous Hetian area) Linear discriminant analysis (LDA) is a popular
and eastern Xinjiang province (China); Geermu, also statistical method that can reduce the multiple di-
known as Golmud (Qinghai province, China); Xi- mensions of variables and provide reliable classifica-
NOTES & NEW TECHNIQUES GEMS & GEMOLOGY FALL 2015 301
RUSSIA
Baikal
N
MONGOLIA
Figure 2. The eight
major dolomite-related
45°
Xiuyan
nephrite deposits in
Xinjiang-West East Asia reviewed in
Xinjiang-East Beijing
this study. Xinjiang
Chuncheon province has been di-
Geermu (Golmud) CHINA
vided into east and
west regions because of
SOUTH KOREA
their different ore-form-
Liyang
30°
ing conditions (Tang et
al., 1994). Modified
Luodian
from Ling et al. (2013).
0 500 km
75° 90° 105° 120° 135°
tion accuracy (Fisher, 1936; Yu and Yang, 2001; On each nephrite block, three to five points along a
McLachlan, 2004; Guo et al., 2007). Recently, this straight line with 5 mm intervals between each point
method has been used to identify the geographic ori- were selected for LA-ICP-MS testing; see the “LA-ICP-
gins of some single-crystal gemstones such as Paraíba MS Measurement” section below. We collected 60 test
tourmaline, ruby, sapphire, and peridot (Blodgett and points for each origin for further statistical analysis, to
Shen, 2011; Shen et al., 2013). be discussed in the “LDA Method” section. Due to the
In this work, the trace-element composition and heterogeneous chemical composition of polycrys-
distribution of 138 dolomite-related nephrite sam- talline nephrite and the relatively limited number of
ples from the major producing areas in East Asia have block samples (only 15–19 per each origin), we treated
been carefully summarized. Based on the trace-ele- each test point as an independent sample, so that each
ment data, we propose an algorithm, which we refer origin has a data set of 60 analyses.
to as iterative binary LDA (IB-LDA), to achieve
nearly complete separation of the dolomite-related LA-ICP-MS Measurement. Trace-element concentra-
nephrite deposits. This method may have wide-rang- tions of the 138 samples were measured using an LA-
ing applications for additional mineral and gemstone ICP-MS system at the State Key Laboratory of
origin research in the future. Geological Processes and Mineral Resources, China
University of Geosciences, Wuhan. The LA-ICP-MS
MATERIAL AND METHODS system consisted of a GeoLas 193 nm laser and an
Sample Preparation. All samples in this study were col- Agilent 7500 ICP-MS. The laser fluence was set as
lected directly from the mines in eight major East 10 J/cm2, and the ablation spot size was 32 µm. The
Asian dolomite-related nephrite jade deposits. A total widely used quantitative calibration standards of
of 138 samples, with 15–19 specimens from each lo- NIST synthetic glasses SRM610 (Pearce et al., 1997)
cality (see table 1), were chosen for LA-ICP-MS testing. and U.S. Geological Survey (USGS) synthetic glasses
The samples were cut into blocks measuring 3.0 of BCR-2G, BHVO-2G, and BIR-1G (Jochum et al.,
× 1.5 × 0.5 cm (length × width × height). The body- 2005) were used as reference materials. Three to five
color ranged from white to light greenish, with scat- spots on each sample, at approximately 5 mm inter-
tered colors such as yellow in some samples. Samples vals along a straight line, were collected for analysis.
29
with representative color from each deposit are Si was used as an internal standard. Detailed oper-
shown in figure 3. ating conditions for the laser ablation system and the
A B C D
E F G H
Figure 3. Representative 3.0 × 1.5 cm nephrite samples from the eight dolomite-related deposits in this study: (A)
Xinjiang-West, (B) Xinjiang-East, (C) Geermu (Golmud), (D) Baikal, (E) Chuncheon, (F) Xiuyan, (G) Luodian, and
(H) Liyang. Photos by Zemin Luo.
ICP-MS instrument and data reduction were the eral procedure for LDA origin determination entailed
same as those described by Liu et al. (2008, 2010). two steps. First, original data sets of 40 independent
test points for each origin except Liyang (which has
LDA Method. Linear discriminant analysis is the pri- a distinguishable feature, to be explored in the Dis-
mary tool in this classification of nephrite from cussion section), along with their trace-element in-
dolomite-related deposits. This method was designed formation, were treated as “training sets” to build
for group classification, which aims to maximize be- the discriminant functions (DFs) and find the best
tween-class variance while minimizing within-class separation. The validity of the separation is charac-
variance. Free software utilizing the statistical pro- terized by the eigenvalue (EV) and total cross valida-
gramming language R, version 3.1.2, was applied to tion (CV); see box A for detailed definitions of EV and
this statistical analysis. Based on the trace-element CV. Second, four additional samples from each ori-
concentrations collected using LA-ICP-MS, the gen- gin, along with their trace-element information,
TABLE 1. Dolomite-related nephrite samples used in this study.
Labeled Locations Mining area sources Quantity LA-ICP-MS Main color

(total 138) test points
(total 452)
1 Xinjiang-West West of Xinjiang province, 17 60 White to greenish

China; includes Hetian, white
Yecheng, Xueyanuote,
Xinzang 383, and Datong
2 Xinjiang-East East of Xinjiang province, 18 60 White to light

China; includes Qiemo and greenish white
Ruoqiang
3 Geermu Qinghai province, China 17 60 White to light
(Golmud) greenish white
4 Baikal Russia 17 60 White
5 Chuncheon South Korea 21 60 White
6 Xiuyan Liaoning province, China 17 60 White to caramel
7 Luodian Guizhou province, China 15 60 White
8 Liyang Jiangsu province, China 16 32 Greenish white
BOX A: LINEAR-DISCRIMINANT ANALYSIS
What Are DFs, EV, and CV? the validity of the model after it is established. These
In traditional LDA, each locality is classified as a group, two parameters have different functions and cannot sub-
and every trace element represents an independent vari- stitute for each other.
able. The combination of all these variables forms a
“property space,” with each variable an independent di- Implementing IB-LDA
mension. In such property space, statistical analysis as- To improve the CV accuracy rate, we designed the IB-LDA
signs weight to each variable based on relative method, which simplifies multiple-group comparisons
importance. The LDA designs a projection direction into pair comparisons. In each round, the LDA process be-
composed of weighted variables to achieve the best sep- comes a binary classification: Only the separation be-
aration between groups. Along this direction, the scat- tween one “chosen” group and the other “unchosen”
tered points from the same group are minimized within groups pool is maximized by the discriminant functions.
the group and maximized between all groups simultane- Meanwhile, the within-class variances of the chosen
ously, which can be formulated as an eigenvalue (EV), a group and the unchosen pool are minimized respectively.
term that will be introduced below. Such projection vec- In the first round of IB-LDA, we randomly selected
tors actually are the linear discriminant functions (DFs) one of the seven localities as the “chosen” group, and the
defined by LDA. The common mathematical expression remaining six as the “unchosen” group. LDA was carried
of DF is the following: out using the statistical package for this two-group com-
parison. The software automatically selected various
DF Yk = a1kX1 + a2kX2 + a3kX3 ... + amkXm + bk (1) numbers of the trace elements among all input trace ele-
Here Yk is the linear discriminant function score or ments, so that the CV value corresponding to each origin
value for a specific group K (K=1, 2,…N, where N is the in the chosen/unchosen binary classification can be
number of total groups). X1, X2, X3, … Xm are characteris- achieved. Each origin is chosen to compare with the un-
tic variables, corresponding to the selected trace elements chosen group, and the origin with the highest CV accu-
in our study; a1k, a2k, a3k, … amk are the discriminant coef- racy rate (versus the remaining six) is determined as the
ficients or weights of each characteristic variable for “optimal-chosen group” in this round. Hereafter, the se-
group K; bk is the constant of linear discriminant function lected trace elements and weights for the “optimal-cho-
for group K. sen group” and “unseparated, unchosen group” will be
EV is the key parameter for building DFs. It can be used to build the pair of DFs for the first round of IB-LDA.
understood as the ratio of total intergroup deviations to The DF for the optimal-chosen group acts as the chosen
the total intragroup deviations; the concept is defined sieve in this round, corresponding to the specific block
rigorously by using covariance matrices (see Welling, box and hole in figure 4.
2005). In general, the DFs resulting in higher EV are cho- This process is applied to the remaining six localities,
sen to build the classification model, since they can pro- with a new set of trace elements chosen by the R soft-
vide better separation between groups (Buyukozturk and ware to maximize the EV and CV values. These newly
Cokluk-Bokeoglu, 2008). selected trace elements with new weights will build a
The classification accuracy and reliability of LDA new pair of DFs for the second round of IB-LDA. Our
can be characterized by the cross-validation (CV) accu- study using this iterative-binary process identified seven
racy rate (see Kohavi, 1995; Cawley and Talbot, 2003). origins sequentially in six rounds in the following order:
The most basic mechanism for CV is: Luodian, Xiuyan, Chuncheon, Xinjiang-West, Baikal,
Xinjiang-East, and Geermu (Golmud). This process is
1. Constructing a LDA model using all data points shown in figure A-1.
except one. Six pair of DFs were obtained for this classification
2. Using this LDA model to test the omitted data process. The last pair of DFs separating Xinjiang-East and
point. Geermu (training set size of 40 samples for each) was ex-
pressed as:
3. Repeating these steps until every data point is
tested. DF Y61(Xinjiang-East) = –2.058Li + 1.554Be + 0.005Al +
4. The percentage of accurately classified omitted 0.012K – 9.807Nb – 0.461Ba + 6.452La – 11.891 (2)
data points is the CV value. A higher CV value in- DF Y62(Geermu (Golmud)) = 0.452Li – 0.682Be + 0.001Al +
dicates better validity of the classification model. 4.062Nb + 0.431Ba + 0.145La –2.376 (3)
Note that EV is the parameter characterizing the va- Herein, we defined CVorNi as the CV for the origin N
lidity of DF in the model when building DF on training in the IB-LDA. It is defined as the continued multiplica-
sets. On the other hand, CV is the parameter evaluating tion of CVi from the first round to the round in which
DF VALUE PLOTS FOR SIX ROUNDS OF IB-LDA Figure A-1. This plot il-
lustrates how six
A 50 Luodian sieve B Xiuyan sieve rounds of IB-LDA were
Unseparated 30 Unseparated
40 Blind test Blind test used to separate seven
Blind test Blind test different dolomite-re-
30 20
lated nephrite sources.
20 The horizontal and ver-
10
tical axes, named DF
Y22
Y12
10
Yi1 and Yi2, were the
0 0
two discriminant func-
CV1 : 99.6% CV2 : 98.3%
-10 Error rate : 1.43% -10 Error rate : 0.00% tions in the i round
-120 -80 -40 0 40 80 120 160 -20 0 20 40 (i=1–6). The gray
DF Y11 DF Y21 shaded area was the
D
C 30 unseparated group. Lu-
Chuncheon sieve Xinjiang-West sieve
40 Unseparated odian, Xiuyan, Chun-
Unseparated
Blind test Blind test
30 Blind test 20 Blind test cheon, Xinjiang-West,
Baikal, Xinjiang-East,
20 and Geermu (Golmud)
10
10 were determined as the
Y42
Y32
successive chosen
0 0 sieves. The CV value in
CV3 : 99.5% CV4 : 96.3%
-10 Error rate : 1.43% Error rate : 1.43%
the bottom right of
-10 each figure corresponds
-20 0 20 40 -20 -10 0 10 20 30
DF Y31 DF Y41 to the CVi accuracy rate
E F
in each round of IB-
30 30
Baikal sieve Xinjiang-East sieve LDA analysis. The error
Unseparated Geermu (Golmud) sieve
Blind test Blind test rate represents the
20 Blind test 20 Blind test
probability of misclas-
10
sifying the test set ob-
10 tained in the blind test
Y62
Y52
0
in each step of IB-LDA.
0 The sum of the six error
CV5 : 98.3%
-10
CV6 : 97.5% rates equals 1-AR.
Error rate : 0.00% Error rate : 0.71% Solid and hollow sym-
-10
-20 -10 0 10 20 30 -20 -10 0 10 20 30 bols represent the train-
DF Y51 DF Y61 ing set and test set,
respectively.
the particular origin N is classified; see equation (4). CVi Evaluating the Model Validity Based on EV, CV, and AR
is the CV accuracy rate, while i is the number of iterative As a conventional method for the machine-learning
rounds of IB-LDA: model, we used two-thirds of the total data set as training
CVorNi(i=1,2,…N) = CV1 × CV2 × … × CVi–1 × CVi (4) sets with the remaining one-third as a test set (Kohavi,
1995). We list the EV and CV values for the training set,
The overall CV (CVtotal) for this iterative-binary clas- and the AR values for the test sets, to evaluate the validity
sification model should be defined as the arithmetic mean and accuracy of the IB-LDA model. EV, CV, and AR values
value of the CVorNi value of each origin, as calculated in for traditional LDA are also presented for comparison
equation (5). with the IB-LDA model. In terms of both training sets (EV
and CV) and test sets (AR), IB-LDA shows greater validity
and accuracy than traditional LDA.
∑ The calculated results of IB-LDA CVorNi for the train-
CV ing sets are shown in figure A-2. The overall CV (CVtotal)
for IB-LDA (94.4 %) was higher than that of traditional
TABLE A-1. EV, CV, and AR values for IB-LDA and can impact the validity (EV) and accuracy (CV) of the
traditional LDA model, we conducted a series of tests by randomly re-
moving 10 data points from each group until each train-
EV CV total (%) AR (%) ing set reached 20 data points. We found that CV
IB-LDA 6.4 94.4 95.0(±2.0) generally increased as each training set grew from 20 to
50. Such an increase indicates that the variance can be
Traditional LDA 4.4 91.4 85.7(±2.5)
predicted more precisely with a larger set of samples. The
variation of CV and EV with size of the training set is
*Note: 40 test points from each origin were used as training sets; the
remaining 20 test points from each origin were chosen as test sets. The mainly caused by some “irregular” samples, which dis-
value in parentheses is the standard deviation of AR after five rounds turbs the variance in/between groups. We speculate that
of blind testing. both EV and CV eventually converge to a constant value,
which can only be justified with a larger total data set.
The most practical classification model needs to take
into account training set size, CV, EV, and other factors,
LDA (91.4%). Samples outside the “expected” groups (in
which is still a real challenge. AR values for blind testing
this case, the eight nephrite localities) in the training sets
are not included in table A-2. Because the training set is
will be classified into one of the “expected” groups in-
larger than 40 samples for each group, the remaining data
correctly according to its DF score. In machine learning
set is too sparse to supply a test set with viable statistics.
and LDA, the groups “expected” in the model must all
As a major difference between CV and AR in practice,
be included in the training sets to build the most com-
CV is extracted within the training set in the building
prehensive training set and DF. The solution for dealing
process of the classification model while AR is obtained
with “unexpected” wild samples is to collect enough
beyond the training set after the classification model
samples to define them as a new group within the train-
build. The AR method usually requires a larger original
ing set. This is why all machine-learning models require
data set to maintain good statistics for both training sets
larger data sets to improve their validity. We include the
and test sets; this requires further accumulation in the
discussion about the “unexpected” samples in the last
future.
paragraph of box A.
We wish to emphasize that we are speculating on the
CV and EV Values Depending on the Statistical Property accuracy and validity of both CV and AR. As the classi-
of Data Sets. Our analysis was based on 60 data points in fication model deals with real “unincluded” samples
each group. To evaluate how the size of the training set without any knowledge of their origin, CV and AR can
only estimate the accuracy or validity of the final analy-
sis result. In the worst-case scenario, a nephrite sample
that does not belong to any of the eight origins, it will
still be (incorrectly) classified as having one of the above
Figure A-2. The CVorNi for the identified origins in our
IB-LDA, and CV for traditional LDA. The detailed al- origins, according to its DF score. In this case, the CV
gorithm of CVorNi is shown in equation 4. IB-LDA iden- and AR are meaningless for such unexpected samples.
tified geographic origins with a higher CV (an average As with all machine-learning models, LDA requires that
of 94.4%) than that of traditional LDA (91.4%). all input (the samples’ trace-element information) be cor-
rectly paired with the expected output (geographic local-
ity) at the very beginning, when the training set is built.
In other words, the most practical model always requires
99.6
large data sets, along with the optimization and balanc-
97.9 97.4
100
94.4 ing of EV, CV, and AR values.
93.8
92.2 91.4
89.9 89.9
CV (%)
80
TABLE A-2. CV and EV values for different training

set sizes.
60 Training set size 20 30 40 50 60
IB-LDA EV (mean) 6.9 5.9 6.4 6.5 6.4
CV total (%) 93.2 93.6 94.4 95.8 94.2
40 Traditional EV (mean) 5.1 4.5 4.4 4.5 4.5
mean mean
LD XY CC XJ-W BK XJ-E GM
IB-LDA LDA LDA CV total (%) 90.0 88.6 91.4 94.0 93.3
TABLE 2. Trace-element concentrations for eight nephrite localities, by LA-ICP-MS.
Element conc., with Xinjiang-West Xinjiang-East Geermu (Golmud) Baikal Chuncheon Xiuyan Luodian Liyang
mean and stand. dev. (Group 1) (Group 2) (Group 3) (Group 4) (Group 5) (Group 6) (Group 7) (Group 8)
Li (ppm) 0.082–25.097 0.250–28.295 0.358–5.435 1.731–20.638 0.100–6.875 0.107–1.333 7.857–35.066 14.686–56.167
2.075 (0.487) 6.301 (0.888) 2.765 (0.130) 6.934 (0.597) 1.058 (0.115) 0.516 (0.037) 22.575 (0.813) 27.724 (1.932)
Be (ppm) 1.581–51.342 0.740–30.731 0.000–1.181 5.140–45.301 11.577–28.624 9.058–27.810 0.000–7.740 0.045–3.771
26.018 (0.931) 11.100 (1.159) 0.380 (0.035) 22.405 (1.613) 19.211 (0.658) 16.161 (0.471) 1.035 (0.255) 1.165 (0.134)
Na2O (wt.%) 0.042–0.438 0.021–0.182 0.008–0.147 0.056–0.133 0.034–0.082 0.024–0.067 0.098–0.221 0.457–0.913
0.088 (0.008) 0.106 (0.004) 0.081 (0.004) 0.099 (0.003) 0.057 (0.001) 0.044 (0.001) 0.179 (0.004) 0.658 (0.019)
Al2O3 (wt.%) 0.338–0.899 0.081–0.974 0.025–0.338 0.380–0.898 0.355–0.983 0.266–1.22 0.122–0.373 0.324–1.410
0.573 (0.021) 0.630 (0.030) 0.136 (0.009) 0.621 (0.015) 0.604 (0.015) 0.505 (0.021) 0.209 (0.007) 0.550 (0.033)
K2O (wt.%) 0.014–0.112 0.025–0.231 0.005–0.091 0.028–0.193 0.024–0.115 0.019–0.078 0.036–0.119 0.116–0.294
0.041 (0.003) 0.099 (0.007) 0.049 (0.003) 0.067 (0.003) 0.048 (0.002) 0.041 (0.002) 0.092 (0.002) 0.217 (0.007)
Sc (ppm) 1.103–79.327 0.850–4.588 0.208–3.682 0.284–3.125 0.518–4.734 1.097–4.117 0.000–9.147 0.614–2.800
4.88 (1.260) 1.893 (0.086) 1.088 (0.087) 1.773 (0.091) 1.333 (0.101) 1.960 (0.070) 1.971 (0.197) 1.670 (0.119)
Rb (ppm) 0.295–53.317 0.003–52.369 0.000–1.302 1.845–9.708 0.421–11.329 0.179–3.799 0.000–0.662 0.339–28.726
4.840 (1.075) 12.162 (1.589) 0.568 (0.040) 4.203 (0.239) 2.679 (0.275) 1.666 (0.125) 0.279 (0.022) 3.139 (1.121)
Sr (ppm) 4.705–32.795 3.763–22.434 3.297–26.933 2.677–18.908 8.515–32.997 3.920–22.541 15.157–31.291 141.267–1272.699
14.721 (0.886) 11.275 (0.477) 16.241 (0.853) 10.619 (0.489) 21.268 (0.529) 13.778 (0.591) 20.950 (0.681) 274.233 (21.62)
Y (ppm) 1.225–18.285 0.159–6.479 0.000–1.778 0.294–7.185 0.255–10.827 1.032–11.420 3.077–29.474 0.361–3.772
4.278 (0.404) 2.991 (0.196) 0.514 (0.055) 1.046 (0.147) 0.882 (0.163) 3.953 (0.206) 11.138 (0.713) 1.468 (0.143)
Zr (ppm) 0.354–15.483 0.000–8.733 0.000–11.187 0.000–16.652 0.000–6.811 0.191–9.151 0.000–22.154 0.007–15.557
4.900 (0.455) 2.739 (0.258) 0.892 (0.207) 1.544 (0.357) 0.720 (0.124) 2.420 (0.221) 2.770 (0.459) 3.220 (0.626)
Nb (ppm) 0.189–11.698 0.000–3.022 0.000–0.322 0.126–1.938 0.033–0.566 0.037–0.789 0.173–1.096 0.119–0.938
1.441 (0.270) 0.808 (0.113) 0.113 (0.011) 0.519 (0.037) 0.208 (0.016) 0.518 (0.022) 0.737 (0.023) 0.319 (0.031)
Mo (ppm) 0.000–0.171 0.000–0.266 0.000–0.625 0.000–1.034 0.000–0.779 0.000–0.345 0.000–3.191 0.000–0.232
0.027 (0.006) 0.034 (0.009) 0.078 (0.016) 0.055 (0.020) 0.118 (0.021) 0.032 (0.009) 0.408 (0.106) 0.041 (0.010)
Cs (ppm) 0.197–13.125 0.071–26.950 0.000–1.682 0.401–4.311 0.057–1.935 0.069–2.447 0.042–0.942 0.000–2.395
3.771 (30.493) 7.620 (0.951) 0.718 (0.058) 2.075 (0.139) 0.525 (0.034) 0.587 (0.056) 0.412 (0.024) 0.278 (0.095)
Ba (ppm) 1.368–26.254 0.284–18.403 0.000–9.237 0.430–30.014 0.859–9.097 0.374–9.951 3.099–195.132 0.196–7.535
9.231 (0.755) 5.887 (0.379) 4.151 (0.251) 7.745 (0.783) 2.463 (0.174) 4.159 (0.194) 15.413 (4.120) 2.611 (0.325)
La (ppm) 0.000–39.004 0.040–2.282 0.008–0.571 0.011–0.523 0.194–1.159 0.000–0.887 0.886–18.591 0.277–4.565
2.601 (0.828) 0.615 (0.067) 0.286 (0.024) 0.179 (0.017) 0.678 (0.021) 0.460 (0.030) 5.083 (0.352) 0.913 (0.138)
Ce (ppm) 0.053–39.916 0.025–4.772 0.000–1.615 0.069–1.384 0.589–2.900 0.027–1.659 0.536–28.944 0.593–7.261
3.452 (0.922) 1.400 (0.156) 0.560 (0.049) 0.462 (0.040) 1.480 (0.042) 1.001 (0.065) 4.938 (0.543) 1.678 (0.211)
Pr (ppm) 0.000–13.764 0.010–0.506 0.000–0.283 0.000–0.170 0.056–0.302 0.000–0.252 0.256–4.649 0.094–1.188
0.997 (0.307) 0.181 (0.019) 0.069 (0.007) 0.042 (0.005) 0.157 (0.006) 0.124 (0.008) 1.322 (0.087) 0.309 (0.036)
Nd (ppm) 0.000–17.535 0.000–2.460 0.000–1.443 0.000–0.677 0.015–1.405 0.015–1.051 1.552–15.303 0.365–4.672
2.098 (0.477) 0.811 (0.080) 0.297 (0.036) 0.266 (0.022) 0.661 (0.036) 0.536 (0.040) 5.486 (0.382) 1.238 (0.139)
Sm (ppm) 0.000–14.859 0.000–0.707 0.000–0.385 0.000–0.393 0.000–0.624 0.000–0.395 0.334–3.806 0.000–0.773
1.464 (0.397) 0.204 (0.019) 0.079 (0.011) 0.075 (0.009) 0.126 (0.016) 0.162 (0.011) 1.176 (0.083) 0.263 (0.029)
Eu (ppm) 0.000–0.140 0.000–0.117 0.000–0.135 0.000–0.145 0.000–0.111 0.000–0.187 0.047–0.637 0.000–0.129
0.017 (0.003) 0.019 (0.003) 0.027 (0.004) 0.019 (0.004) 0.026 (0.003) 0.076 (0.006) 0.279 (0.016) 0.044 (0.006)
Gd (ppm) 0.072–5.373 0.000–0.726 0.000–0.430 0.000–0.510 0.000–0.367 0.049–0.523 0.000–3.222 0.000–0.589
0.768 (0.132) 0.246 (0.019) 0.090 (0.013) 0.099 (0.012) 0.148 (0.014) 0.230 (0.013) 1.136 (0.076) 0.275 (0.029)
Tb (ppm) 0.009–2.413 0.000–0.136 0.000–0.090 0.000–0.065 0.000–0.094 0.005–0.107 0.043–0.462 0.000–0.124
0.202 (0.049) 0.050 (0.003) 0.015 (0.002) 0.008 (0.002) 0.017 (0.002) 0.053 (0.003) 0.180 (0.012) 0.038 (0.005)
Dy (ppm) 0.133–8.037 0.011–0.731 0.000–0.294 0.000–0.429 0.000–1.252 0.117–0.957 0.354–2.767 0.027–0.596
0.922 (0.188) 0.347 (0.023) 0.084 (0.011) 0.102 (0.012) 0.124 (0.021) 0.410 (0.021) 1.145 (0.071) 0.207 (0.023)
Ho (ppm) 0.027–5.253 0.000–0.160 0.000–0.082 0.000–0.112 0.000–0.296 0.045–0.593 0.045–0.593 0.010–0.108
0.379 (0.114) 0.083 (0.006) 0.020 (0.002) 0.016 (0.003) 0.027 (0.005) 0.023 (0.006) 0.224 (0.018) 0.040 (0.004)
Er (ppm) 0.034–22.467 0.000–0.674 0.000–0.213 0.000–0.914 0.000–0.533 0.021–1.091 0.000–2.316 0.004–0.318
1.839 (0.558) 0.243 (0.020) 0.052 (0.006) 0.095 (0.017) 0.070 (0.011) 0.296 (0.020) 0.715 (0.062) 0.114 (0.015)
Tm (ppm) 0.007–4.507 0.000–0.100 0.000–0.062 0.000–0.083 0.000–0.107 0.006–0.180 0.000–0.374 0.000–0.034
0.218 (0.081) 0.030 (0.003) 0.010 (0.002) 0.010 (0.002) 0.013 (0.002) 0.039 (0.003) 0.091 (0.010) 0.015 (0.002)
Yb (ppm) 0.063–15.424 0.017–0.802 0.000–0.316 0.000–0.363 0.004–1.416 0.004–1.416 0.000–1.850 0.000–0.308
0.818 (0.263) 0.204 (0.019) 0.064 (0.010) 0.093 (0.031) 0.068 (0.011) 0.234 (0.025) 0.571 (0.055) 0.103 (0.017)
Lu (ppm) 0.003–11.295 0.000–0.113 0.000–0.047 0.000–0.086 0.000–0.103 0.003–0.264 0.000–3.899 0.000–0.058
0.378 (0.191) 0.026 (0.003) 0.008 (0.001) 0.008 (0.002) 0.014 (0.002) 0.030 (0.004) 0.322 (0.083) 0.014 (0.002)
Hf (ppm) 0.000–4.405 0.000–0.321 0.000–0.223 0.000–0.257 0.000–0.302 0.000–0.299 0.000–0.748 0.000–0.423
0.395 (0.102) 0.073 (0.009) 0.024 (0.005) 0.025 (0.006) 0.036 (0.006) 0.063 (0.008) 0.112 (0.023) 0.096 (0.018)
Ta (ppm) 0.000–8.495 0.000–0.188 0.000–0.046 0.000–0.090 0.000–0.085 0.000–0.108 0.000–0.170 0.000–0.157
0.622 (0.225) 0.043 (0.005) 0.007 (0.001) 0.011 (0.002) 0.010 (0.002) 0.028 (0.003) 0.039 (0.004) 0.037 (0.005)
Bi (ppm) 0.000–0.784 0.000–2.758 0.000–0.434 0.000–0.068 0.000–0.210 0.000–0.021 0.000–2.496 0.000–0.015
0.084 (0.026) 0.076 (0.045) 0.045 (0.010) 0.007 (0.002) 0.022 (0.004) 0.007 (0.001) 0.229 (0.056) 0.006 (0.001)
Pb (ppm) 0.293–7.710 0.346–10.454 0.055–3.691 0.368–17.393 0.925–4.882 0.118–3.243 0.271–100.569 0.157–5.250
1.195 (0.173) 2.948 (0.228) 1.618 (0.106) 2.016 (0.280) 1.895 (0.096) 0.561 (0.059) 3.424 (1.637) 0.771 (0.149)
Th (ppm) 0.029–5.299 0.036–2.406 0.000–0.546 0.000–0.983 0.011–0.675 0.005–0.620 0.000–1.100 0.074–0.892
0.686 (0.108) 0.236 (0.042) 0.074 (0.012) 0.150 (0.025) 0.112 (0.011) 0.212 (0.017) 0.254 (0.025) 0.355 (0.034)
U (ppm) 0.146–5.967 0.081–2.427 0.000–0.946 0.028–0.660 0.070–0.947 0.530–2.479 0.378–1.345 0.069–1.942
1.182 (0.154) 1.080 (0.062) 0.487 (0.028) 0.179 (0.018) 0.345 (0.018) 1.448 (0.039) 0.882 (0.028) 0.304 (0.057)
were treated as test sets. The purpose of a test set is odian samples had an isolated distribution region,
to estimate how the classification model will deal while the other six localities still exhibited overlap.
with the data that was not included in the “training The full classification of all seven groups by “one-
set” to build the DFs. As conventional settings for pass” LDA must balance variance among all seven
machine learning models (Kohavi, 1995), two-thirds groups. This may be the reason why traditional LDA
of the total data sets (280 data points) were used as a presented a relatively low CV of 91.4 % and a rela-
training set, and one-third (the other 140 data points) tively low EV of 4.48 with a training set size of 40
were used as test set. This setting applies throughout testing points for each group. For the details of DFs
the paper unless other specific modifications are and CV, again see box A.
mentioned. Detailed descriptions of training sets and
test sets also appear in box A. IB-LDA to Optimize the Separation of Dolomite-
Related Nephrite Geographic Origins. If only aiming
RESULTS AND DISCUSSION to distinguish one group from all the others while ig-
noring the differences within the remaining unclas-
For the 138 nephrite samples from the eight loca-
sified groups, certain group identities should be more
tions, the concentration of elements from Li to U (45
accurate and distinguishable. Hence, we designed the
elements total) were obtained by LA-ICP-MS meas-
“iterative-binary” LDA (IB-LDA) to optimize the sep-
urement. To make our classification model inde-
aration. The procedure used to build the discrimi-
pendent of sample color, the transition metal
nant function and criteria database for all seven
elements (Ti, V, Cr, Mn, Fe, Co, Ni, and Cu) were ex-
origins is described in box A.
cluded from further data analysis. Only 34 trace ele-
Classifying different sources of dolomite-related
ments were involved in creating the LDA model. The
nephrite by IB-LDA is analogous to sorting blocks
mean value and standard deviation value of each
with different shapes by putting them into the corre-
trace element are calculated based on 60 detection
sponding holes. As shown in figure 4, seven blocks
points for each origin, as shown in table 2.
represented seven different dolomite-related nephrite
Trace-Element Analysis to Separate Liyang from origins. The red triangle, pink four-pointed star, purple
Other Origins. As shown in table 2, the nephrite sam- five-pointed star, yellow six-pointed star, green square,
ples from Liyang (group 8) displayed much higher con- blue pentagon, and orange hexagon represented Luo-
centrations of Sr (>140 ppm) and Na2O+K2O (> 0.57 dian, Xiuyan, Chuncheon, Xinjiang-West, Baikal, Xin-
wt.%) than the other seven localities (Sr<40 ppm, jiang-East, and Geermu (Golmud) respectively. We
Na2O+K2O < 0.56 wt.%). These results confirmed pre- then needed to build corresponding blocks with dif-
viously reported conclusions (Zhang et al., 2011; Siqin ferently shaped holes (named “chosen sieves” here-
et al., 2012; Ling et al., 2013) that the distinctive Sr, after) using the training set samples. The original 280
Na, and K concentrations could be used as a quantita- training set samples from the seven origins were used
tive discriminant to separate Liyang nephrite samples. to create seven chosen sieves in sequence after six
These obvious diagnostic chemical signatures ren- rounds of IB-LDA processes (figure 4). The different
dered further analysis of Liyang nephrite unnecessary. holes represent different DFs built for each chosen
The only real challenge was to distinguish the remain- sieve. The sequence of those chosen sieves directly
ing seven nephrite origins. Table 2 shows overlapping corresponds to the six rounds of IB-LDA, which must
trace-element distributions. Therefore, it is important be unchanged to maintain the validity of the classifi-
to derive effective discriminants from the trace ele- cation method. We found that the IB-LDA model pro-
ments to separate the origins. duces better CV and EV values than traditional LDA,
as shown in table A-1, which indicates IB-LDA can ef-
Traditional LDA in Dolomite-Related Nephrite Ori- fectively improve the accuracy and validity of the
gin Determination. From the outset, we attempted sieves (classification model).
to use the traditional LDA method to classify the Next, we evaluated the performance of the sieves
seven dolomite-related nephrite localities at once. built by IB-LDA on unincluded data in a blind test.
Each nephrite locality is classified as a group, and Four new samples from each origin were used as a
every input trace element represents an independent testing set to check the reliability and accuracy of the
variable. Seven linear discriminant functions (DFs) established chosen sieves (figure 5). All samples were
for seven independent nephrite groups were built si- treated as “unknown” blocks to be tested on the
multaneously. The results showed that only the Lu- seven sieves in sequence. When the unknown block
Fundamental Steps in IB-LDA
1st sieve: Luodian 2nd sieve: Xiuyan 3rd sieve: Chuncheon
Figure 4. These blocks

illustrate the process of
determining the
dolomite-related
nephrite origins by IB-
LDA. The first step is
building the “chosen
sieves” sequence
4th sieve: Xinjiang-West 5th sieve: Baikal through six rounds of
IB-LDA using the
“training set” samples.
The boxes had different
holes corresponding to
“chosen sieves” with
discriminant functions
?
in each round of IB-
LDA. The second step
(inset) is to use these
sieves to determine the
origins of testing set
6th sieve: Xinjiang-East 7th sieve: Geermu
samples step by step in
a blind test. Four test
set samples from each
origin were chosen ran-
domly from the total
data set, excluding the
2nd step of IB-LDA:
Use “test sets” and “sieves” training set samples.
to perform a blind test.
1st step of IB-LDA: Use “training sets” to build “sieves”

(discriminant functions) corresponding to seven origins.
fit into the correct sieve, its “shape” was identified. calculated in the first step of IB-LDA, as shown in
For example, the first chosen sieve (Luodian) allowed table A-1.
only the triangular blocks to be sorted out, which
meant only Luodian samples could be identified and CONCLUSIONS
all the other samples were still in an “unseparated” We propose that an IB-LDA model, combined with
status. The unseparated samples went through the trace-element information from LA-ICP-MS, is an ef-
second IB-LDA process, where a four-pointed star fective method for determining the origin of
sieve representing Xiuyan was used to extract appro- dolomite-related nephrite deposits. We consider the
priately shaped blocks, and the unseparated pool was application of IB-LDA to the quantitative classifica-
reduced once more. tion of nephrite origin a significant improvement
This process should be repeated until all the un- over the traditional method. The origin information
separated blocks can be identified. The number of reflected by trace-element data has been well ex-
test sets with the correct classification divided by the plored and applied in nephrite origin determination.
number of the total test sets is the accuracy rate (AR) The LDA method presents obvious statistical advan-
for such blind testing. The same blind test is per- tages in dealing with the massive quantity of
formed five times on different test sets, and the av- nephrite trace-element data. Finally, the successful
erage AR is calculated. We note that the AR value performance of IB-LDA remarkably improved dis-
obtained is generally consistent with the CV value criminant accuracy, increasing the CV accuracy rate
Blind Testing in IB-LDA
? ?
? ?
Figure 5. The schematic
diagram of the blind
test, the second step in
IB-LDA. Only when the
shape of an “unknown”
block agrees with one
of the sieve shapes can
its origin be deter-
mined.
?
Through blind testing, all “test sets” can be identified by sieves with high accuracy.
Luodian Xiuyan Chuncheon Xinjiang-West Baikal Xinjiang-East Geermu
from 91.4% in traditional LDA to 94.4% and identi- cumulating more nephrite samples for use as training
fying CVorNi values specific to each origin. The dis- sets will further improve the reliability of the dis-
criminant functions database built by IB-LDA criminant function database. The IB-LDA method
obtained a 95.0% accuracy rate in the testing of 28 should also prove useful for the origin determination
unknown samples. We believe that collecting and ac- of other gemstones.
ABOUT THE AUTHORS nancially supported by the Fundamental Research Funds for the
Dr. Zemin Luo is a member of the Center for Innovative Gem Testing Central Universities (20140139043), the Natural Science Fund of
Technology (CIGT), a lecturer at the Institute of Gems and Jewelry Hubei Province (2014CFB347), Quality Industry Research Special
Studies (IGJS), and a fellow of the Geology Postdoctoral Research Funds for Public Welfare Projects (201210228), and Research
Station, at the China University of Geosciences (Wuhan). Prof. Fund of Center for Innovative Gem Testing Technology (CIGTXM-
Mingxing Yang is the dean of the IGJS and leader of the classifica-
201401). Sincere thanks to Prof. Yongsheng Liu and Prof.
tion group at CIGT. Prof. Andy H. Shen is a Hubei 100 Talent Distin-
guished Professor at the IGJS and the director of the CIGT. Zhaochu Hu for providing LA-ICP-MS testing at the State Key
Laboratory of Geological Processes and Mineral Resources,
ACKNOWLEDGEMENTS China University of Geosciences (Wuhan). The authors also thank
The nephrite samples were provided by the jade study group at the manuscript’s three peer reviewers for their constructive com-
the China University of Geosciences (Wuhan). This work was fi- ments and suggestions.
REFERENCES
Adamo I., Bocchio R. (2013) Nephrite jade from Val Malenco, Italy: Bulletin, Vol. 55, No. 15, pp. 1535–1546.
Review and update. G&G, Vol. 49, No. 2, pp. 98–106, McLachlan G.J. (2004) Discriminant Analysis and Statistical Pat-
http://dx.doi.org/10.5741/GEMS.49.2.98 tern Recognition. John Wiley and Sons, Hoboken, New Jersey.
Adams C.J., Beck R.J., Campbell H.J. (2007) Characterisation and Pearce N.J.G., Perkins W.T., Westgate J.A., Gorton M.P., Jackson
origin of New Zealand nephrite jade using its strontium iso- S.E., Neal C.R., Chenery S.P. (1997) A compilation of new and
topic signature. Lithos, Vol. 97, Nos. 3–4, pp. 307–322, published major and trace element data for NIST SRM 610 and
http://dx.doi.org/10.1016/j.lithos.2007.01.001 NIST SRM 612 glass reference materials. Geostandards and
Blodgett T., Shen A.H. (2011). Application of discriminant analysis Geoanalytical Research, Vol. 21, No. 1, pp. 115–144,
in gemology: Country-of-origin separation in colored stones http://dx.doi.org/10.1111/j.1751-908X.1997.tb00538.x
and distinguishing HPHT-treated diamonds. Proceedings of the Shen A.H., Koivula J.I., Shigley J.E. (2011) Identification of extra-
Fifth International Gemological Symposium, G&G, Vol. 47, terrestrial peridot by trace elements. G&G, Vol. 47, No. 3, pp.
No. 2, p. 145. 208–213, http://dx.doi.org/10.5741/GEMS.47.3.208
Breeding C.M., Shen A.H. (2010) LA-ICP-MS analysis as a tool for Shen A.H., Blodgett T., Shigley J.E. (2013) Country-of-origin deter-
separating natural and synthetic malachite. News from Research, mination of modern gem peridots from LA-ICPMS trace-ele-
Oct. 11, http://www.gia.edu/gia-news-research-nr101410 ment chemistry and linear discriminant analysis (LDA).
Buyukozturk S., Cokluk-Bokeoglu O. (2008) Discriminant func- Geological Society of America Abstracts with Programs, Vol.
tion analysis: Concept and application. Eurasian Journal of Ed- 45, No. 7, p. 525.
ucational Research, Vol. 33, pp. 73–92. Siqin B., Qian R., Zhuo S.J., Gan F.X., Dong M., Hua Y.F. (2012)
Cawley G.C., Talbot N.L.C (2003) Efficient leave-one-out cross- Glow discharge mass spectrometry studies on nephrite minerals
validation of kernel fisher discriminant classifiers. Pattern formed by different metallogenic mechanisms and geological
Recognition, Vol. 36, No. 11, pp. 2585–2592, http://dx.doi.org/ environments. International Journal of Mass Spectrometry, Vol.
10.1016/S0031-3203(03)00136-5 309, pp. 206–211, http://dx.doi.org/10.1016/j.ijms.2011.10.003
Fisher R.A. (1936) The use of multiple measurements in taxo- Tang Y.L., Chen B.Z., Jiang R.H. (1994) Zhongguo Hetian yu (Het-
nomic problems. Annals of Eugenics, Vol. 7, pp. 179–188, ian Jade of China). Xinjiang People’s Press, Xinjiang, China, pp.
doi:10.1111/j.1469-1809.1936.tb02137.x 160–200 [in Chinese].
Guo Y.Q., Hastie T., Tibshirani R. (2007) Regularized linear dis- Tsien H.H., Lo H.J., Lin S.B. (1996) Crystal growth and whitening
criminant analysis and its application in microarrays. Biosta- of archaic tremolite yu. Acta Geologica Taiwanica, Vol. 32, pp.
tistics, Vol. 8, No. 1, pp. 86–100, http://dx.doi.org/10.1093/ 131–147.
biostatistics/kxj035 Welling M. (2005) Fisher linear discriminant analysis. Department
Harlow G.E., Sorensen S.S. (2005) Jade (nephrite and jadeitite) and of Computer Science, University of Toronto, http://www.cs.huji.
serpentinite: Metasomatic connections. International Geology ac.il/~csip/Fisher-LDA.pdf
Review, Vol. 47, No. 2, pp. 113–146, http://dx.doi.org/10.2747/ Wen G. (2001) Ancient Chinese jade and the jade age (translation
0020-6814.47.2.113 from Zhongguo guyu yuqi shidai by E. Childs-Johnson). In E.
He G.M. (2009) English and Chinese culture connotation. Asian So- Childs-Johnson, Ed., Enduring Art of Jade Age China: Chinese
cial Science, Vol. 5, No. 7, pp. 160–163, http://www.ccsenet.org/ Jades of Late Neolithic through Han Periods, Throckmorton
journal/index.php/ass/article/view/2985 Fine Art, New York, pp. 31–34.
Jochum K.P., Willbold M., Raczek I., Stoll B., Herwig K. (2005) Wen G., Jing Z. (1996) Mineralogical studies of Chinese archaic
Chemical characterization of the USGS reference glasses GSA- jade. Acta Geologica Taiwanica, Vol. 32, pp. 55–83.
1G, GSC-1G, GSD-1G, GSE-1G, BCR-2G, BHVO-2G and BIR- Wu R.H., Zhang X.H., Li W.W. (2002) Petrological characteristics
1G using EPMA, ID-TIMS, ID-ICP-MS and LA-ICP-MS. of Hetian jade in Xinjiang and nephrite from Baikal area in Rus-
Geostandards and Geoanalytical Research, Vol. 29, No. 3, pp. sia: Acta Petrologica et Mineralogica (Sinica), Vol. 21, supp.,
285–302, http://dx.doi.org/10.1111/j.1751-908X.2005.tb00901.x pp. 134–142 [in Chinese].
Kohavi R. (1995) A study of cross-validation and bootstrap for ac- Yu H., Yang J. (2001) A direct LDA algorithm for high-dimensional
curacy estimation and model selection. Proceedings of the data - with application to face recognition. Pattern Recognition,
Fourteenth International Joint Conference on Artificial Intel- Vol. 34, No. 10, pp. 2067–2069.
ligence, Vol. 2, No. 12, pp. 1137–1143. Morgan Kaufmann, San Yui T.F., Kwon S.T. (2002) Origin of a dolomite related jade deposit
Mateo, California. at Chuncheon, Korea. Economic Geology, Vol. 97, No. 3, pp.
Liao R.Q., Zhu Q.W. (2005) Chemical composition analysis of 593–601, http://dx.doi.org/10.2113/gsecongeo.97.3.593
nephrites from different localities in China. Journal of Gems Zhang Z.W., Gan X.F. (2011) Analysis of the chromite inclusions
and Gemmology, Vol. 7, No. 1, pp. 25–30 [in Chinese]. found in nephrite minerals obtained from different deposits
Ling X.X., Schmadicke E., Wu R.H., Wang S.Q., Gose J. (2013) Com- using SEMEDS and LRS. Journal of Raman Spectroscopy, Vol.
position and distinction of white nephrite from Asian deposits. 42, pp. 1808–1811.
Neues Jahrbuch Mineralogie – Abhandlungen, Vol. 190, No. 1, Zhang Z.W., Gan X.F., Cheng H.S. (2011) PIXE analysis of nephrite
pp. 49–65, http://dx.doi.org/10.1127/0077-7757/2013/0229 minerals from different deposits. Nuclear Instruments and
Liu J., Cui W.Y. (2002) Study on nephrite (tremolite jade) from Methods in Physic Research B, Vol. 269, No. 4, pp. 460–465,
three localities in China. Journal of Gems and Gemmology, http://dx.doi.org/10.1016/j.nimb.2010.12.038
Vol. 4, No. 2, pp. 25–29 [in Chinese]. Zhang Z.W., Xu C.Y., Cheng H.S., Gan X.F. (2012) Comparison of
Liu Y., Hu Z., Gao S., Günther D., Xu J., Gao C., Chen H. (2008) In trace elements analysis of nephrite samples from different de-
situ analysis of major and trace elements of anhydrous minerals posits by PIXE and ICP-AES. X-Ray Spectrometry, Vol. 41, pp.
by LA-ICP-MS without applying an internal standard. Chemical 367–370.
Geology, Vol. 257, Nos. 1–2, pp. 34–43, http://dx.doi.org/ Zhong Y.P.，Qiu Z.L., Li L.F., Gu X.Z., Luo H., Chen Y., Jiang Q.Y.
10.1016/j.chemgeo.2008.08.004 (2013) REE composition of nephrite jades from major mines in
Liu Y.S., Hu Z.C., Zong K.Q., Gao C.G., Gao S., Xu J., Chen H.H. China and their significance for indicating origin. Journal of
(2010) Reappraisement and refinement of zircon U-Pb isotope the Chinese Society of Rare Earth, Vol. 31, No. 6, pp. 738–748,
and trace element analyses by LA-ICP-MS. Chinese Science http://dx.doi.org/10.11785/S1000–4343.20130615 [in Chinese].
Editors
Thomas M. Moses | Shane F. McClure
Unusual Graining Structure in Pink

DIAMOND
Natural pink diamonds examined for
origin of color determination often
display visible pink coloring along
planes of stress. This graining is visi-
ble at a slight angle to the plane and
can occur as a single plane or multiple
parallel planes. The color may be ei-
ther uniform or unevenly distributed
along the graining. One pink stone re-
cently examined in the Carlsbad lab Figure 1. Left: A colorless hole within a natural pink diamond, observed
was noted to have a uniform plane of with darkfield illumination. Field of view approximately 2 mm. Right:
pink color with a colorless hole (figure The colorless hole in polarized light, showing the cone of strain. Field of
1, left). The diamond was graded Very view approximately 2 mm.
Light pink, so the colored plane had
low saturation, but the irregularity
was impossible to miss in the micro- of stress. Whether pink graining oc- changing the color of diamonds, partic-
scope. When examined under polar- curs during diamond growth or is ularly those with an undesirable
ized light, a cone of strain could be formed after by stresses or heat from brownish color. This treatment process
seen intersecting the plane and fitting the earth (or a combination of the has been practiced in the trade for more
perfectly in the colorless hole (figure two) is also poorly understood. It is than 15 years. An important feature of
1, right). therefore a mystery as to whether the this treatment is the etching often left
Pink graining is the result of an cone of strain formed first or concur- behind on the diamond surfaces. Re-
unknown defect with an unknown rently and prevented the formation of cently, one loose yellow diamond
formation mechanism (J.G. Chapman pink graining at the circle of intersec- rough, weighing 2.47 ct and measuring
and C.J. Noble, “Studies of the pink tion, or if the cone formed afterward 7.09 × 6.59 × 5.46 mm, was examined
and blue coloration in Argyle dia- and destroyed the color centers asso- in the New York lab (figure 2). Without
monds,” Summer 1999 G&G, pp. ciated with pink graining. Until dia- magnification, the rough had a vibrant
156–157). It seems to be related to the mond formation is more fully yellow color and an overall clean-look-
plastic deformation of diamond, evi- understood, this dynamic interplay of ing surface. Closer microscopic obser-
denced by the color lying along planes strain will remain a mystery. vation showed a very slightly etched
Troy Ardon surface, as well as some altered graphi-
tized crystal inclusions in the diamond.
Editors’ note: All items were written by staff
Distinctive fluorescence reactions to
members of GIA laboratories. long- and short-wave ultraviolet radia-
Yellow HPHT-Processed tion similar to those of HPHT-
GEMS & GEMOLOGY, Vol. 51, No. 3, pp. 312–322. Rough Diamond processed material were observed, with
© 2015 Gemological Institute of America High-pressure, high-temperature (HPHT) a medium chalky bluish green under
annealing is a very effective method for long-wave and medium to strong
312 LAB NOTES GEMS & GEMOLOGY FALL 2015

UV-VIS-NIR SPECTRUM

.6
.4
496 (H4)
503 (H3)
.2
986 (H2)
400 500 600 700 800 900

WAVELENGTH (nm)
Figure 2. This 2.47 ct rough dia-
mond was HPHT processed to
improve its color. Unlike conven- Figure 3. The yellow rough’s UV-Vis-NIR spectrum, showing H2 (986
tional HPHT-treated diamonds, nm), H3 (503 nm), and H4 (496 nm) peaks, indicates HPHT treatment.
this stone has a natural-looking
surface.
sophisticated, allowing for the reten- minder that buying rough does not as-
tion of carat weight during the cutting sure a lack of treatment.
greenish yellow under short-wave UV. process. This case serves as a clear re- Sally Chan
This was suspect, and further testing
was needed. UV-Vis-NIR and mid-IR
spectroscopy revealed the rough dia- Figure 4. Mid-IR spectroscopy revealed an absorption peak at 1344 cm–1,
mond to be HPHT processed. indicating the presence of isolated nitrogen. This occurrence and corre-
The UV-Vis-NIR spectrum, seen in sponding peak are characteristic of HPHT processing.
figure 3, showed a rise in absorption
from 500 nm, associated with strong
H3 (503 nm) and H4 (496 nm) peaks, MID-IR SPECTRUM
along with a sharp H2 (986 nm) peak.
.5
These are indicative of HPHT treat-
ment. Mid-IR spectroscopy (figure 4)
revealed peaks at 1170 and 1332 cm–1, .4

1170
indicating that the starting material of
this diamond appeared to be type IaB.
.3
The high temperature and high pres-
sure of the treatment created isolated 1332
nitrogen with a corresponding absorp- .2
tion peak detected at 1344 cm–1. Oc-

currence of isolated nitrogen in a pure .1
type IaB diamond is a strong indicator
of HPHT treatment.
1344
The majority of HPHT-treated 0
rough diamonds have severely burned,

etched, and pitted surfaces. This is the -.1
first time we have seen an almost eye-
clean yellow rough post-HPHT treat-
ment. With new technologies available 1500 1400 1300 1200 1100 1000
to diamond merchants, HPHT pro- WAVENUMBER (cm–1)

cessing techniques are becoming more
LAB NOTES GEMS & GEMOLOGY FALL 2015 313

which was consistent with the ob-
served color. Although similar zoning
has been observed in emerald previ-
ously (E.J. Gübelin and J.I. Koivula,
Photoatlas of Inclusions in Gem-
stones, Volume 3, Opinio Verlag,
Basel, Switzerland, 2008, pp. 433–434),
the zoning in this specimen was quite
distinct.
This emerald demonstrates the sig-
nificance of proper orientation and cut-
ting style in fashioning high-quality
gem materials.
Figure 5. Left: When table-down, the stone is mainly colorless, showing
some partially green overgrowth. Right: Face-up, the faceted stone dis- Jonathan Muyal, Nathan Renfro,
plays an even and attractive dark green emerald color. and Amy Cooper
Orange Faceted EOSPHORITE

Color-Zoned EMERALD tures were observed in the red end of The Carlsbad laboratory recently ex-
Cutting style and orientation are im- the spectrum when examined with a amined a 5.85 ct transparent orange
portant considerations when cutting handheld spectroscope. All of these oval mixed cut for identification serv-
natural crystals with strong color zon- properties were consistent with ices (figure 8). Standard gemological
ing. Careful placement of these zones emerald. During microscopic exami- testing showed an RI of 1.640 to
can distribute color by reflection, cre- nation, minute jagged fluid inclu- 1.667; the hydrostatic SG was 3.12.
ating a uniformly colored finished sions and growth tubes were The stone showed yellow and reddish
stone when viewed face-up. observed, proving the stone was of orange pleochroism. Fluorescence
Recently, the Carlsbad laboratory natural origin (figure 6). was inert to long-wave and short-
received a 33.73 ct green octagonal Interestingly, the stone showed a wave UV radiation. The most distinc-
step cut showing strong color zoning strong green color zone very close to tive internal characteristic, revealed
(figure 5). The gem cutter skillfully some of the pavilion facets, as well as by microscopic examination, was the
placed a thin green color zone in this hexagonal columnar-like growth and presence of multiphase inclusions
emerald along the pavilion facets so irregular zoning (figure 7). This type of (figure 9), suggesting a formation
that the stone would appear uni- zoning represents a late-stage influx of process with the presence of water,
formly green when in a face-up posi- chromium and vanadium into the ge- such as pegmatitic and hydrothermal
tion. Standard gemological testing netic environment, which were then processes.
revealed a refractive index (RI) of incorporated into the crystal lattice of
1.571–1.578 and a specific gravity the beryl. EDXRF revealed high levels
(SG) of 2.72. Chromium-related fea- of Cr and V in the green color zone, Figure 8. This 5.85 ct orange
mixed cut was a rare faceted
eosphorite.
Figure 6. When viewed from the
side, irregular green color zoning Figure 7. The colorless spots indi-
is observed along with minute cate parallel columnar growth of
jagged fluid inclusions. Field of white beryl overgrown by green
view 14.52 mm. emerald. Field of view 19.27 mm.

attributed to the P-O (phosphorus- references). The stone contains 55.3
oxygen) stretching vibrations (R. L. mol.% eosphorite and 40.5 mol.%
Frost et al., “Vibrational spectroscopic childrenite. The calculated formula is
characterization of the phosphate (Mn0.48Fe0.35Ca0.01Mg0.03)(Al)1.09(P0.99O4)(OH)2·
mineral series eosphorite-childrenite- 0.91(H2O), which indicates an inter-
(Mn,Fe)Al(PO4)(OH)2·(H2O),” Vibrational mediate member of the childrenite-
Spectroscopy, 2013, Vol. 67, pp. 14–21). eosphorite series with predominance of
This mineral series has a stoichiomet- the Mn phase. As a result, the stone
ric formula (Mn,Fe)Al(PO4)(OH)2·(H2O). should be classified as eosphorite.
Visible absorption spectroscopy re- Eosphorite is the manganese-rich
Figure 9. Multi-phase primary vealed that the stone’s orange color re- end member of the childrenite-
fluid inclusions are seen in the sults from the combination of Fe and eosphorite series, formed worldwide
eosphorite, indicating a peg- Mn (M.A. Hoyos et al., “New struc- in pegmatites (T. J. Campbell and W. L.
matitic or hydrothermal forma- tural and spectroscopic data for Roberts, “Phosphate minerals from
tion process. Field of view 1.26 eosphorite,” Mineralogical Magazine, the Tip Top mine, Black Hills, South
mm. 1993, Vol. 57, pp. 329–336). Dakota,” Mineralogical Record, 1986,
Energy-dispersive X-ray fluores- Vol. 17, pp. 237–254) or by hydrother-
cence (EDXRF) analysis revealed a mal phosphatization of metasedi-
Raman spectroscopy confirmed composition of 25.10 wt.% Al2O3, ments and associated with the
that the stone belongs to the children- 31.74 wt.% P2O5, 15.46 wt.% MnO, intrusion of granites (R. S. W. Braith-
ite-eosphorite mineral series (figure 11.47 wt.% FeOtot, 0.49 wt.% MgO, waite and B. V. Cooper, “Childrenite
10), with intense peaks observed at 0.25 wt.% CaO, and 15.50 wt.% H2O in South-West England,” Mineralogi-
970 and 1011 cm–1. These peaks are (water weight percent is assumed from cal Magazine, 1982, Vol. 46, pp. 119–
126). This was the first faceted
eosphorite examined by GIA’s Carls-
Figure 10. Raman analysis of the eosphorite in the 2000–200 cm–1 range bad laboratory.
showed a typical spectrum of phosphate oxyanions. The 970 and 1011 Ziyin Sun, Nathan Renfro, and
cm–1 peaks are assigned to the P-O stretching vibration, while the 1038 Amy Cooper
and 1055 cm–1 peaks are attributed to the PO43– stretching vibration. The
405, 473, 561, 595, and 618 cm–1 peaks are associated with the PO4 and
H2PO4 bending modes. Large Faceted HIBONITE
The Carlsbad laboratory recently ex-
amined an unusually large 134.43 ct
RAMAN SPECTRUM opaque, very dark brown faceted hi-
60000 bonite (figure 11). Standard gemologi-
cal testing gave spot RI values of 1.79
970 to 1.81 and a hydrostatic SG of 3.81.
50000 Hibonite, which crystallizes in the
hexagonal system, has published RI
values of 1.790 to 1.807, but we were
40000 unable to observe birefringence, prob-
INTENSITY (a.u.)
ably due to its opacity and these val-

P-O stretching ues’ proximity to the limits of the
30000 refractometer. The stone showed a
high luster and was inert to LW and
SWUV. Testing with a Geiger counter
20000 showed it was slightly radioactive. All
PO4 H2PO4 bending PO43- stretching
of these features are consistent with
hibonite, yet more advanced testing
10000 was needed for a positive identifica-
tion of this unusual material.
Raman spectroscopy gave a spec-
0 trum consistent with hibonite. Laser
200 400 600 800 1000 1200 1400 1600 1800 2000
ablation–inductively coupled plasma–
RAMAN SHIFT (cm–1) mass spectrometry (LA-ICP-MS)
analysis revealed large amounts of

schwang et al., “Hibonite: A new gem colored green jadeite. Observation
mineral,” Summer 2010 G&G, pp. under long-wave UV light revealed
135–138; Summer 2012 Lab Notes, p. strong green-yellow fluorescence
136). Although not gem quality, the from most green areas of the stone,
stone’s size makes it a rare collector’s but not from the area where Cr lines
mineral as well as the largest hibonite were seen in the spectroscope. The
GIA has identified to date. colorless portions of the stone were
Claire Ito and Ziyin Sun also inert to UV light. Strong fluores-
cence in jadeite is often indicative of
either polymer impregnation or a dye
treatment; however, fluorescence
Dyed and Natural Green JADEITE from polymers would be evenly dis-
The most coveted jadeite stones pos- tributed throughout the stone due to
sess a deep rich green color. Lower- the nature of impregnation. The fact
quality jadeite specimens are often that this stone only fluoresced from
treated with acid bleaching, polymer the green areas was strong evidence of
Figure 11. This 134.43 ct faceted impregnation, and dyes to obtain this dyeing rather than impregnation.
hibonite is the largest examined sought-after hue (E. Fritsch et al., Microscopic observation (figure
by GIA to date. “Identification of bleached and poly- 13) revealed concentrations of green
mer-impregnated jadeite,” Fall 1992 color in the boundary gaps in between
G&G, pp. 176–187). It is quite un- grains, typical of dyed jadeite. The
Mg, Ca, Al, Ti, and Fe; small amounts usual to see stones with even a small green area without fluorescence pos-
of Na, Si, Sr, La, and Ce; and traces of amount of natural green color further sessed a smoother, more consistent
numerous other elements, including altered with dyes. When this does green color, which is customary for
the radioactive elements thorium happen, the natural properties of the natural jadeite. Absorption spec-
(62.62–122.60 ppma) and uranium untreated area can make it especially troscopy in the UV, visible, and near-
(0.12–0.56 ppma). The presence of challenging to properly identify the IR range confirmed that the natural
these elements caused the mineral to stone as dyed. area contained an absorption band
react to the Geiger counter. The tho- Recently, the Carlsbad laboratory from 530 to 740 nm, punctuated by
rium and REE content is consistent examined a carved jadeite specimen natural chromium peaks at 691 and
with samples from Madagascar (figure 12). Standard gemological test- 655 nm (figure 14). The dyed green
(M.A.F. Rakotondrazafy et al., “Mode ing and Raman spectroscopy con- areas contained a double band around
of formation of hibonite (CaAl12O19) firmed that the specimen was jadeite; 627 and 665 nm, which is typical of
within the U-Th skarns from the infrared spectroscopy indicated the dyed green jadeite.
granulites of S-E Madagascar,” Con- specimen had been polymer impreg- The proper treatment identifica-
tributions to Mineralogy and Petrol- nated. Observation with a desktop tion of jadeite is essential to the con-
ogy, Vol. 123, No. 2, 1996, pp spectroscope revealed lines at 630 and sumer’s confidence in this stone.
190–201). 655 nm, which are associated with Testing only a portion of this material
Microscopic examination revealed chromium impurities in naturally might convince an observer that no
one deep green mineral inclusion
breaking the surface of the table. Its
Raman spectrum was consistent with Figure 12. Left: In natural light, the green band in the lower right section of
spinel, a known associated mineral of the jadeite specimen is caused by natural chromium, while the other green
hibonite. LA-ICP-MS analysis showed spots are the result of a dye treatment. Right: Long-wave UV light shows a
it was an iron-rich spinel within the strong fluorescence from the dyed areas, with little to no fluorescence from
hercynite(FeAl2O4)-spinel (MgAl2O4) the natural green area.
series, with almost one-third of the
magnesium substituted by iron. Al-
though difficult to see internally, other
whitish mineral inclusions were visi-
ble, as well as an extensive fracture and
cleavage network.
Orangy brown, transparent, and
well-formed hexagonal crystals of
gem-quality hibonite have been re-
ported from Myanmar (T. Hain-

Figure 13. Left: A close-up of the natural green area shows a smooth, dif-
fused color texture. Field of view 7.19 mm. Right: A close-up of a dyed
area shows color concentrations in the thin gaps between the crystal
grains of jadeite. Field of view 3.57 mm.
Figure 15. These three korner-

dye had been applied. One must al- cently submitted to the Carlsbad lab- upine beads (0.95–1.38 ct) were
ways expect the unexpected when at- oratory for identification service (fig- found to have a metallic coating.
tempting to determine treatments in ure 15). The three beads ranged from
any gemstone. 0.95 ct to 1.38 ct. Standard gemologi-
Dylan Hand cal properties revealed an RI of ap-
trapped in the drill holes. Raman
proximately 1.668–1.680, consistent
spectroscopy confirmed that the
with kornerupine. The hydrostatic
Coated KORNERUPINE Beads stones were kornerupine.
SG ranged from 3.01 to 3.19. This is
All three beads exhibited a high
Three dark brown faceted beads with slightly lower than expected for ko-
metallic luster with silver and bronze
an unusual metallic luster were re- rnerupine, but can be explained by air
coloration, which is not typical for
kornerupine. Under microscopic ex-
amination with reflected lighting, the
Figure 14. UV-Vis-NIR spectra of the jadeite in the 250–850 nm range surfaces showed spotty luster, consis-
showed chromium lines at 655 and 691 nm in the natural green area and tent with a surface coating (figure 16).
double broad bands around 627 and 665 nm in the dyed green area. Further evidence of coating was visi-
ble along the facet junctions, where
the coating was worn away.
UV-VIS-NIR SPECTRA EDXRF chemical analysis detected
3.5 large amounts of Mg, Al, and Si, con-
Natural green area sistent with kornerupine. EDXRF also
Dyed green area
revealed the presence of copper. The
authors could not find documentation
3
Figure 16. Diffused reflected light

ABSORBANCE (a.u.)
2.5 exposed spotty luster on the sur-

face of one of the tested stones,
indicating the presence of a coat-
ing. Field of view 2.17 mm.
2
Natural chromium
655 691
1.5
627
655
Dye bands
1
250 350 450 550 650 750 850
WAVELENGTH (nm)

of naturally occurring copper in ko- posed to standard short-wave UV ra-
rnerupine, so this element may have diation (254 nm) and examined in the
been used to create the metallic coat- DiamondView, this object showed a
ing. This is the first time the GIA lab- moderate white blue reaction that fol-
oratory has seen coated kornerupine. lowed the spiderweb pattern (figure
Amy Neurauter and 19, right).
Heidi Breitzmann Further investigation with Raman
spectroscopy using a 488 nm laser re-
vealed a calcite spectrum with peaks
Assembled Cultured Blister PEARL Figure 18. A microradiograph of at 159, 285, 716, 1088, and 1750 cm–1,
the intact assembled cultured similar to white coral (Corallium se-
with an Unusual Component
blister pearl prior to sawing re- cundum). A small peak related to
GIA’s Bangkok laboratory recently ex-
vealed an internal component. carotenoid pigments was present at
amined a blue-violet assembled cul-
The radio-translucency of the ob- 1020 cm–1 (e.g., J. Urmos et al., “Char-
tured blister pearl (mabe) weighing
ject, the nacre, and the shell base acterization of some biogenic carbon-
2.44 ct and measuring 10.23 × 5.86
are similar. ates with Raman spectroscopy,”
mm (figure 17). The specimen had the
American Mineralogist, Vol. 76, 1991,
appearance of a typical mabe pearl, ex-
pp. 641–646).
hibiting a boundary line between the
Based on the spiderweb structure,
nacre face and shell base. 2D microra-
been seen five times in mabe pearls radio-translucency, calcite spectrum,
diography showed a component with
submitted for testing since 2010; and general appearance, the object ap-
an unusual fluted shape, with radio-
questions about its identity and pur- peared to be a biogenic carbonate.
translucency similar to that of the
pose have been raised since the first White coral was the first material con-
CaCO3 nacre and the shell (figure 18).
observation. sidered. Further research revealed ac-
GIA has examined the character-
Out of curiosity, one client de- counts of sea urchin sections used in
istics of mabe pearls used in com-
cided to cut their sample in half in some assembled cultured blister pearls
mercial jewelry over the years (see
order to examine the interior in more from Bali, Indonesia, in “Kuta” pearls
Lab Notes: Summer 1981, pp. 104–
detail (figure 19, left). The two halves (E. Strack, Pearls, Ruhle-Diebener-
105; Summer 1991, pp. 111–112;
consisted of a thin nacre dome top, Verlag, Stuttgart, Germany, 2006, p.
Fall 1991, p. 177; Summer 1992, pp.
which was coated with a dark layer on 605). Regardless of identity, the reason
126–127; Fall 1992, pp. 195–196; Fall
the inner surface; an artificial for including this material within
1996, p. 210), but none of these
resinous material; and the white some mabe pearls remains uncertain.
notes mentioned this unusual com-
fluted object bordering the shell base. It does not appear related to either the
ponent. In Bangkok, this feature has
The fluted object was translucent, and formation of the mabe pearl compo-
a clear spiderweb structure was visi- nent or its weight and stability.
ble at 10× magnification. When ex- Artitaya Homkrajae
Figure 17. This blue-violet assem-

bled cultured blister pearl (mabe)
weighed 2.44 carats and meas-
ured 10.23 × 5.86 mm.
Figure 19. Left: The cross-section revealed the white fluted object with a
spiderweb structure bordering the shell base. Right: DiamondView im-
aging showed a moderate white blue reaction following the spiderweb
pattern.

Figure 20. Three unique abalone pearls are shown alongside a typical 12 mm golden South Sea bead-
cultured pearl for comparison.
Three Unique Large Natural Pearls yellowish green bodycolors. GIA has second piece, seen in figure 20 on the
from Haliotis (Abalone) Species reported on many Haliotis pearls in far right, was a loose abalone pearl
Natural abalone pearls are produced previous Lab Notes over the years weighing 66.83 ct and measuring ap-
(Fall 1984, p. 169; Spring 1996, pp. 47– proximately 42 × 26 × 19 mm with a
by many species of the Haliotis genus,
49; Fall 2004, pp. 259–260; Spring striking resemblance to a fish’s head.
which are ear-shaped saltwater uni-
1993, p. 51). The eye, mouth, and gill features were
valve gastropods. They may form in
Three large and unique abalone clearly outlined in the specimen,
many different sizes and shapes,
pearls from a client’s collection were which possessed a smooth surface and
though they are commonly encoun- recently submitted to the New York subtle array of iridescent greenish yel-
tered in baroque shapes, often taking laboratory (figure 20). Each piece was low colors. The third piece, weighing
the form of elongated “horns” and remarkable in its own right. A button- 113.58 ct and measuring approxi-
“teeth.” The pearls are also notable shaped pearl, measuring approximately 62 × 29 × 16 mm, was also
for their high luster and attractive mately 26 × 24 mm with impressive loose and resembled the head of a
multicolored appearance, which is vivid “peacock” blue and green hues, snake. It exhibited strong iridescence
usually dominated by blue, green, or was set in a yellow metal brooch. The with mottled patches of various colors
Figure 21. All three specimens from figure 20 showed concentric and void-related internal structures typical of
abalone pearls.

Figure 22. This photomicrograph Figure 23. These intensely colored CVD synthetic diamonds, weighing
shows the patches of color and 0.64 ct (left) and 0.43 ct (right), are color graded as Fancy Deep green-blue
botryoidal-like subsurface struc- and Fancy Deep greenish blue, respectively.
ture characteristic of abalone
pearls. Field of view 3.9 mm.
similar to the pattern observed on the by where the pearls are formed within attractive colors. Irradiated CVD syn-
scaly skin of snakes. the body of the mollusk, and often thetic diamonds without a multi-step
Microradiography revealed charac- mirror the form of the gonad (E. treatment process are rarely seen in
teristic Haliotis-related voids with Strack, Pearls, Ruhle-Diebener-Verlag the lab, though an irradiated Fancy
concentric growth structures in all GmbH & Co. KG, Stuttgart, 2006). Ex- Deep gray-blue CVD synthetic has
three pearls (figure 21). Two of the amining all three specimens at the been reported (Fall 2014 Lab Notes, pp.
pearls also revealed an iodine-rich same time allowed us to see a range of 240–241). The New York lab recently
composition, which GIA has noted is colors and shapes possible in abalone examined two such examples: a 0.64 ct
a fairly consistent property of abalone pearls. We look forward to encounter- Fancy Deep green-blue and a 0.43 ct
pearls. Such internal structures, along ing more of these unusual pearls. Fancy Deep greenish blue specimen
with their distinctive multicolored Joyce Wing Yan Ho and (figure 23).
orient, bubble-like (botryoidal) subsur- Surjit Dillon Wong Dark inclusions, faint color bands,
face structure (figure 22), interesting and chips on the girdle were ob-
chemical composition, and chalky served in both specimens under the
greenish yellow fluorescence under Irradiated Green-Blue microscope. Tatami-like strain was
long-wave UV are all characteristic of CVD SYNTHETIC DIAMONDS also seen in both diamonds under
abalone pearls. As-grown CVD synthetic diamonds cross-polarized light. DiamondView
The various shapes are determined are routinely treated in order to create images revealed parallel green linear
Figure 24. DiamondView images reveal linear striations in the pavilions of the 0.64 ct (left) and 0.43 ct (middle)
specimens. A face-up image of the 0.43 ct synthetic shows a yellowish green zone of growth interruption (right).

UV-VIS-NIR SPECTRA
anticipate that intensely colored, high-
quality CVD synthetic diamonds will
741.2
393.5 666.5 become more prevalent in the trade.
594.2
737.0 Kyaw Soe Moe, Ulrika D’Haenens-
Johansson, and Wuyi Wang
735.8
469.9
464.3
ABSORBANCE
Polished Freeform TOPAZ

Imitating Diamond Rough
Topaz is one of many near-colorless
0.64ct
diamond simulant materials that
may be faceted to bear a closer re-
semblance to diamond. Although
most diamond simulants in the mar-
ketplace are faceted, every now and
then, one comes across what appears
0.43ct to be a near-colorless octahedron di-
amond rough (see Lab Notes: Fall
400 450 500 550 600 650 700 750 800
1996, p. 205; Fall 1997, pp. 217–218;
WAVELENGTH (nm) Fall 2007, p. 250; Fall 2009, pp. 230–
231). A near-colorless 13.70 ct stone
Figure 25. Absorption spectra of the two green-blue CVD synthetics at (figure 26) was recently submitted to
liquid nitrogen temperature revealed radiation-related peaks. The peaks the New York laboratory as a rough
at 666.5 and 735.8 (both neutral self-interstitial defects) indicate that diamond from an alluvial source. At
these samples were annealed at lower temperatures or irradiated without first glance, even with a well-trained
annealing. The Si-V center defect (737.0 nm center) can also be observed eye, this stone could easily be
at the 736.5–736.9 nm doublet. Many unknown absorption bands (unla- misidentified. It bore a striking re-
beled in this spectra) are also detected. semblance to diamond rough, with a
“well-formed” octahedral shape and
trigon-like formations on the surface
striations in the pavilions (figure 24). related to the 666.5 band. Strong ab-
Face-up DiamondView images also sorption from the GR1 center, which
showed yellowish green zones, which contains neutral vacancies (V0), was Figure 26. The 13.70 ct topaz
were formed by growth interruptions detected at 741.2 nm. All these radi- under daylight-equivalent light.
(again, see figure 24). Both stones were ation-related peaks can be destroyed Note the triangular features on
pure type IIa diamonds with no other by HPHT treatment, while peaks at each of the octahedral faces.
defects in the IR spectra. High-reso- 666.5 and 735.8 nm can be annealed
lution UV-Vis-NIR absorption spec- out at temperatures ranging from 420º
tra detected Si-V center defects at to 540ºC (see A.T. Collins, “Spec-
the 736.5–736.9 doublet (737 nm troscopy of defects and transition
center, (V-Si-V)–); the spectra also re- metals in diamond,” Diamond and
vealed radiation-related peaks at 393.5, Related Materials, Vol. 9, Nos. 3–6,
464.3, 469.9, 594.2, 666.5, 735.8, and 2000, pp. 417–423). The presence of
741.2 nm (figure 25). The 393.5 band these peaks indicates that these CVD
is from the ND1 center, a negatively synthetics were either annealed at
charged vacancy (V–). An intrinsic ra- lower temperatures (<420º to 540ºC)
diation center (TR12) was detected or irradiated without annealing.
at 469.9 nm. The TR13 center at the Although microscopic features
464.3 peak is a local vibrational mode such as strain pattern can be helpful in
associated with the TR12 center, and the identification of CVD synthetics,
the 594.2 band (595 nm center) is a they are not conclusive. Diamond-
typical radiation-induced center. The View imaging is very useful for this
666.5 peak is attributed to a neutral purpose due to its ability to detect lin-
self-interstitial (I0) defect. The 735.8 ear striations. As CVD growth tech-
band, also caused by the I0 defect, is niques continue to improve, we

IR ABSORPTION SPECTRUM
(FTIR) spectroscopy, which revealed
14
the distinctive topaz absorption spec-
trum shown in figure 27, with peaks
12
at 4802 and 3650 cm–1 (K. Shinoda
and N. Aikawa, “IR active orienta-
tion of OH bending mode in topaz,”
10 Physics and Chemistry of Minerals,

Vol. 24, No. 8, 1997, pp. 551–554).
Although topaz can be easily
8
shaped and polished into an octahedral
shape, the trigon-like figures observed
6
on the surface require an additional
fabrication step and are not typically
seen in octahedral-shaped simulants.
4 Intentional material processing steps
to mask a stone’s identity remind
gemologists of the caution and care
1
that need to be taken when dealing
with gemstone identification—even
0 when the identity of the material ini-
tially seems obvious.
7000 6000 5000 4000 3000 2000 1000 Akhil Sehgal and Riccardo Befi
WAVENUMBER (cm–1)
Figure 27. Many absorption peaks related to the hydroxide ion in the crys-
tal structure of topaz are detected in this infrared absorption spectrum, in-
cluding those at 4802 and 3650 cm–1. These have never been documented PHOTO CREDITS:
in any diamond absorption spectrum. Nathan Renfro—1, 9; Jian Xin (Jae) Liao—2,
26; Robison McMurtry—8, 11, 12, 15;
Jonathan Muyal—5, 6, 7, 13; Amy Neu-
rauter—16; Lhapsin Nillapat—17; Areeya
Manustrong—18; Sasithorn Engniwat—19
of each of the faces. To further com- 3.53 (+/-0.04) for topaz. The material (left); Artitaya Homkrajae—19 (right); Sood
plicate the identification of this was doubly refractive, with a spot RI Oil (Judy) Chia—20, 23; Surjit Dillon Wong—
stone, topaz and diamond have the reading of 1.61, properties consistent 21; Joyce Wing Yan Ho—22; Kyaw Soe
same heft due to overlapping specific with topaz. Final confirmation came Moe—24.
gravities: 3.52 (+/-0.10) for diamond, from Fourier-transform infrared
For online access to all issues of GEMS & GEMOLOGY from 1934 to the present, visit:
gia.edu/gems-gemology

2015
CONGRATULATIONS
This year, hundreds of readers participated in the 2015 Gems & Gemology Challenge.
Entries arrived from around the world, as readers tested their gemological knowl-
edge by answering questions listed in the Spring 2015 issue. Those who earned a
score of 75% or better received a GIA Certificate of Completion recognizing their
achievement. The participants who scored a perfect 100% are listed below.
G&G Challenge Winners

Australia United States
Barbara Wodecki Andrea Blake
Beverly Brannan
Belgium
Sheila Sylvester Thaïs Anne Lumpp-Lamkie
Lisa Marsh-Vetter
Canada Paul Mattlin
Diane Gauthier
Daniel Novak
Italy Lydia Pringle
Chiara Piussi Jana Smith
Ric Taylor
United Kingdom
Francesca Lawley Colleen Walsh
Flora Walters
Glenn Wargo
Answers
See pages 56–57 of the Spring 2015 issue for the questions.
1 (d), 2 (b), 3 (a), 4 (b), 5 (a), 6 (d), 7 (a), 8 (c), 9 (*), 10 (a), 11 (b), 12 (d), 13 (c), 14 (c),
15 (d), 16 (d), 17 (c), 18 (b), 19 (a), 20 (b), 21 (a), 22 (a), 23 (c), 24 (a), 25 (d)
* Question withdrawn
CHALLENGE WINNERS GEMS & GEMOLOGY FALL 2015 323

Editor
Nathan Renfro
Contributing Editors
Elise A. Skalwold and John I. Koivula
Cleavage System in Pink Diamond

A 0.29 ct natural Fancy Intense purplish pink diamond
with a microscopically interesting geometric feature (figure
1) was recently examined at GIA’s Carlsbad laboratory. A
network of planar cracks, oriented in three octahedral di-
rections, was observed at the surface of a pavilion facet (fig-
ure 2). These linear cleavage cracks were caused by
localized strain on an octahedral plane.
The octahedral plane in diamond has the weakest
atomic bonding, which means that when a diamond is
strained, this is the direction that will be most affected.
This stone offers the perfect example of strain products in
Figure 2. Three directions of cleavage along octahe-

dral planes are seen in this pink diamond. Photomi-
Figure 1. This Fancy Intense purplish pink 0.29 ct dia-
crograph by Jonathan Muyal; field of view 1.22 mm.
mond was host to a complex cleavage network. Photo
by Robison McMurtry.
pink diamond. The purplish pink color results from plastic

deformation along octahedral planes. Also present along
the octahedral plane is a zone where the strain was so great
that it was relieved by cracking, creating the cleavage net-
work seen in figures 2 and 3.
The cleavage network in this diamond is the most aes-
thetic one encountered by the author to date due to its
finely textured geometric pattern. Interestingly, the clue to
About the banner: Modified Rhineberg illumination provides high contrast

to etch features on the prism face of a beryl crystal. Photomicrograph by
Nathan Renfro; field of view 2.86 mm.
Editors’ note: Interested contributors should contact Nathan Renfro at
nrenfro@gia.edu and Jennifer-Lynn Archuleta at jennifer.archuleta@gia.edu
for submission information.
GEMS & GEMOLOGY, VOL. 51, NO. 3, pp. 324–330.
324 MICRO-WORLD GEMS & GEMOLOGY FALL 2015

Conversely, the source could be a radioactive solution that
was able to penetrate surface-reaching cleavages in the di-
amond. In both cases, the source produced alpha-particle
radiation, which has a limited penetrating distance in dia-
mond. As a result, the radiation spot or stain is confined
to a zone close to the radioactive material. Such radiation
stains are not uncommon in diamonds, though they are
often removed by the faceting process.
The presence of a large green internal radiation stain in-
dicates that this diamond was irradiated in a near-surface
geological environment where temperatures high enough
to modify the GR1 were unlikely. This means that the ir-
radiation took place in the crust after the diamond was
transported up from the mantle by the kimberlite magma
Figure 3. Between crossed polars the diamond shows eruption. Although the geographic origin of this rough dia-
high-order interference colors consistent with the mond is uncertain, it is one of the more intriguing examples
strain causing the cleavage network along the octahe- of a natural “radiation-stained” diamond we have seen.
dral planes. Photomicrograph by Jonathan Muyal; James Shigley
field of view 0.62 mm. GIA, Carlsbad
the planar cleavage network lies in the bodycolor of the

Blue Gahnite Inclusions in Cat’s-Eye Heliodor
stone, which indicates significant strain. This high-order Four golden oval cabochons (figure 5) were submitted to
strain is easily revealed between crossed polars. If a cleav- the Carlsbad laboratory by L. Allen Brown (All That Glit-
age network is present in a diamond such as this, it may ters, Methuen, Massachusetts) for scientific examination.
have a strain-induced bodycolor as well. Standard gemological testing revealed that the stones were
beryl. Under incandescent fiber-optic illumination, the
Jonathan Muyal
stones showed a very sharp cat’s eye, caused by growth
GIA, Carlsbad
Rough Diamond Fragment with a Large Green Figure 4. This 2.02 ct rough diamond displays promi-
Cleavage Surface nent green coloration along an internal cleavage sur-
A colorless 2.02 ct flattened diamond crystal fragment (fig- face. Such coloration is thought to result from the
ure 4), from Rowan Beach of San Francisco, was examined presence of a natural radioactive solution within the
at GIA’s Carlsbad lab. The fragment exhibited a large bluish cleavage. Photo by James Shigley.
green internal surface that was readily visible without mag-
nification. Initially thought to be a colorful mineral inclu-
sion, it proved to be a radiation-induced coloration along the
internal surface. When a diamond is irradiated (during lab-
oratory color treatment or occasionally in nature), carbon
atoms are removed from their normal position in the dia-
mond lattice; this creates what scientists call the “GR1” op-
tical defect (i.e., a vacant atom position in the lattice). The
presence of this defect causes the diamond to selectively ab-
sorb light toward the red end of the visible spectrum, while
the remaining portions are transmitted through the dia-
mond and, when recombined, seen as a green or bluish green
color. The GR1 optical defect can be removed in the lab by
heating the diamond to over 600°C; this heating changes the
green color to yellowish brown. The added energy allows a
carbon atom to again occupy the vacant lattice position.
Green or bluish green color in diamonds is most often
the result of exposure to a source of radiation in a geologic
environment near the earth’s surface. The source of expo-
sure could be a nearby grain of a radioactive mineral that
gives rise to a small “radiation spot” on the outer surface.
MICRO-WORLD GEMS & GEMOLOGY FALL 2015 325

Figure 5. These four he-
liodor cabochons dis-
play sharp eyes under
incandescent fiber-optic
illumination. Photo by
C.D. Mengason.
tubes and fine needle-like inclusions oriented parallel to from Sri Lanka (E.J. Gübelin and J.I. Koivula, Photoatlas of
the c-axis. Under magnification, blue crystals with octahe- Inclusions in Gemstones, Vol. 3, Opinio Verlag, Basel,
dral morphology (figure 6) were the most distinctive inter- Switzerland, 2008, p. 290), detailed study of gahnite may also
nal characteristic. provide some indication of the geographic origin of heliodor.
Raman microspectroscopy was used to identify the blue Ziyin Sun, Nathan Renfro,
octahedral inclusions as gahnite, a zinc-rich end member of Jonathan Muyal, and Adam Steenbock
the spinel group with the general formula ZnAl2O4. It occurs GIA, Carlsbad
most commonly as an accessory mineral in granitic peg-
matites, usually associated with almandine-spessartine gar-
net, muscovite, beryl, tourmaline, and nigerite (D.R. Soares,
Pezzottaite Debuts as the Newest Trapiche Gem Mineral
et al., “Chemical composition of gahnite and degree of peg-
matitic fractionation in the Borborema Pegmatitic Province, Since the first enigmatic six-spoked emeralds were sent
northeastern Brazil,” Anais da Academia Brasileira de to GIA in the mid-1960s for analysis (H.L. McKague,
Ciências, 2007, Vol. 79, No. 3, pp. 395-404). According to “Trapiche emeralds from Colombia,” Fall 1964 G&G, pp.
Brown, the stones were purchased in Tucson in 2010 and 210–213, 223), the family of trapiche-type gem minerals
were represented as being from Brazil. These are the first has grown to include a variety of species and morpholo-
beryls with gahnite inclusions that GIA has examined to gies. One of the latest mineral varieties to exhibit the
date. Since gahnite is a diagnostic inclusion in sapphires rare trapiche form is a relatively new and rare mineral it-
self: pezzottaite.
Following its appearance at the 2003 gem shows in Tuc-
Figure 6. Diffused fiber-optic illumination reveals son, pezzottaite has quickly become an item coveted by
well-formed blue octahedral crystals in addition to gem and mineral collectors around the world. A member
the hollow tubes that cause the chatoyancy. The black of the beryl group and with trigonal symmetry (the beryl
needle-like inclusions may be hematite. Photomicro- species has hexagonal symmetry), it was approved by the
graph by Nathan Renfro; field of view 1.10 mm. International Mineralogical Association in September 2003
as a unique species and named after mineralogist Dr. Fred-
erico Pezzotta (see B.M. Laurs et al., “Pezzottaite from Am-
batovita, Madagascar: A new gem mineral,” Winter 2003
G&G, pp. 284–301).
In 2004, while working in the gem markets of Mae Sai
and Mae Sot along the border of Thailand and Myanmar,
Elaine Rohrbach of Gem-Fare obtained a small collection of
pale pink trapiche crystals and crystal fragments reported to
be of Burmese origin. Laboratory analysis later proved these
crystals to be the cesium-containing mineral pezzottaite
(previously confirmed localities include Madagascar and
Afghanistan).
A doubly terminated, well-developed crystal obtained
by one of the authors (JIK) exhibits the tabular prismatic
habit characteristic of the species, but with a complete six-

Figure 7. As viewed through the pinacoid face, strong Figure 9. Eye-visible dark needles were seen through-
oblique fiber-optic lighting illuminates the six-legged out the pezzottaite crystal with diffused transmitted
trapiche structure emanating from the c-axis of this illumination. Despite a clear Raman signal, the fea-
doubly terminated 2.33 ct pezzottaite crystal. Photo tures could not be conclusively identified. Photomi-
by Kevin Schumacher. crograph by Nathan Renfro; field of view 1.10 mm.
spoke trapiche structure that is clearly visible in both re- minated crystal obtained by author EAS from the same par-
flected and transmitted light (figure 7). Another doubly ter- cel shows secondary crystals growing out of each pinacoid
face, each of which are also trapiche crystals (figure 8). Sim-
ilar to trapiche crystals of other mineral species, each of the
Figure 8. Another well-developed, doubly terminated spokes radiates from the central c-axis toward the prism
pezzottaite crystal from the same locality, weighing faces and in a plane parallel to the basal pinacoid. Exploring
6.17 ct and measuring 11.19 × 10.71 × 8.61 mm, has the former crystal’s inner world revealed not only the de-
separate crystals growing at random angles on each tails of the white spokes and a spray of golden brown nee-
pinacoid face. Each of these smaller crystals also ex- dles and eye-visible black needles, but also a delightful
hibits trapiche growth. The white trapiche growth surprise: minute euhedral blue crystals (figures 9 and 10).
reaches all six prism faces in both of the tabular crys- Unfortunately, the black needles were not identifiable de-
tals, as well as those faces on the secondary growth spite their distinct Raman spectrum. Due to their depth
crystals. Photo by Elise A. Skalwold. within the host, the brown needles and blue crystals remain
mysteries as well. Let this be a lesson that nondestructive
Figure 10. Using diffused transmitted and oblique

fiber-optic illumination, a golden-brown spray of nee-
dles is shown deep in the crystal, along with a bright
blue crystal to its right. Their depth precluded analy-
sis by Raman spectroscopy. Photomicrograph by

Figure 11. In one method of applying Rheinberg-type illumination, glass slides are fitted to the end of two fiber-
optic illuminators and painted with red and blue ink (left) to create a filter that is easily positioned by moving the
colored portion of the glass slide into the light path (right). Photos by Danny J. Sanchez.
testing cannot always provide answers, but we can be con- An interesting technique that gives the microscopist an-
tent with exploring the beauty of the micro-world while other lighting tool is modified Rheinberg illumination, also
leaving some mysteries for the future. known as differential color illumination (M. Pluta, Ad-
Elise A. Skalwold vanced Light Microscopy: Specialized Methods, Vol. 2,
Ithaca, New York PWN-Polish Scientific Publishers, Warsaw, 1989, pp. 113).
This method, as modified for gemological microscopy, em-
John I. Koivula
ploys the use of a contrasting color filter between each il-
GIA, Carlsbad
lumination source and the subject (figure 11) to achieve an
“optical staining” effect (figure 12). When viewing crystal-
Modified Rheinberg Illumination lographically aligned subjects such as negative crystals or
Lighting control is one of the most important considera- inclusions with well-defined, reflective crystal faces, each
tions for maximizing the use of the gemological micro- illumination source highlights areas that have the same
scope; with greater control over illumination sources, more crystallographic orientation (figure 13). This provides dra-
information may be gathered from observing a specimen. matic false-color contrast to an otherwise low-contrast
Figure 12. This silvery metal sulfide inclusion in fluorite shows angular crystal faces, but the low contrast in the
image on the left makes it difficult to resolve the orientation of the faces from different sections of the inclusion.
Rheinberg illumination provides stark contrast between differently oriented regions by using two colored illumi-
nation sources that reflect their independent color off of crystallographically aligned crystal faces, as seen in the
center and right images. This allows the microscopist to easily observe the different orientations of the crystal
faces in the metal sulfide inclusion, as evidenced by the dramatic boundary of the two contrasting colors. Pho-
tomicrographs by Danny J. Sanchez; field of view 1.66 mm.

Figure 13. With modified Rheinberg illumination to impart contrasting color on some of the crystal faces, the neg-
ative crystals in the rock crystal quartz (left) prove to be crystallographically aligned (right). Photomicrographs by
subject. This enhanced contrast makes it easier to observe At first glance, these inclusions appeared to be ordinary
the relationship between areas in an inclusion scene with gas-liquid inclusions (figure 14, left); however, polishing
identical and also differing crystallographic orientations. through the inclusions demonstrated that the “liquid”
Nathan Renfro component was actually a glassy solid. The identification
GIA, Carlsbad of the solid as a glass rested with the phase’s optically
isotropic nature and the lack of an identifiable Raman spec-
Danny J. Sanchez
troscopic signal attributable to any crystalline phase.
Los Angeles
Glassy melt inclusions in Montana sapphires were previ-
ously documented by John Koivula (R.W. Hughes, Ruby &
Glassy Melt Inclusions in Sapphires from Montana Sapphire, RWH Publishing, Boulder, Colorado, 2007, pp.
Two-phase inclusions are common in many sapphires. 468–469). Electron microprobe analysis (EPMA) of glass in-
Physical and chemical analysis of the gas and liquid phases clusions from several samples showed them to be dacitic
present in these inclusions can shed light on the formation to trachydacitic in composition. In this sense the inclu-
conditions of alluvial sapphires, even in the absence of ev- sions are ordinary two-phase inclusions in which the orig-
idence of source rock lithology. We report here some initial inal liquid phase was a silicic magma that quenched to
observations of a different type of two-phase inclusion, form a silicate glass. The gas phase is presumed to have
seen in alluvial sapphires from the Rock Creek and Mis- originally been dissolved in the silicate magma. Later ex-
souri River deposits in the state of Montana. solution produced a separate gas phase as volatile solubility
Figure 14. Left: A two-phase glassy melt and gas inclusion and a glass-filled discoid fracture in a sapphire from
the Rock Creek deposit in Montana. Center: A two-phase glassy melt and gas inclusion and a glass-filled dis-
coid fracture in a sapphire from the Rock Creek deposit. Right: A two-phase glassy melt and gas inclusion, in a
sapphire also from the Missouri River, showing the surrounding discoid fracture. This fracture presumably filled
with a silicate liquid and partially healed, leaving an impression in some places of the corundum host’s trigonal
crystallographic orientation. Photomicrographs by Aaron Palke; field of view 1.24 mm (left) and 0.62 mm (cen-
ter and right).

coid fractures often have a “flux”-like appearance. In some
cases, the fractures seem to have partially healed, leaving
hexagonal geometric patterns as evidence of the crystallo-
graphic orientation of the sapphire host (figure 14, right).
Further work is being carried out to characterize these in-
clusions and to decipher the story they have to tell about
the genesis of alluvial Montana sapphires.
Aaron Palke and Nathan Renfro
GIA, Carlsbad
Richard B. Berg
Montana Tech of the University of Montana
Butte, Montana
Figure 15. This 4.71 ct Brazilian colorless topaz hosts Molybdenite in Topaz
a cluster of hexagonal molybdenite crystals. Photo by
A photogenic cluster of hexagonal molybdenite crystals
Kevin Schumacher.
(MoS2) was recently discovered in a 4.71 ct transparent col-
orless hexagonal step-cut topaz (figure 15). The topaz host
was identified using traditional gemological testing, and the
in the magma decreased due to falling pressure and/or tem- molybdenite cluster (figure 16) was identified through its
perature. Volatile exsolution likely also led to the discoid metallic luster, morphology, and the presence of molybde-
fractures commonly seen around melt inclusions, which num, which was detected using focused EDXRF analysis.
are filled with silicate liquid and/or the volatiles them- This gem, which came from Luciana Barbosa of the
selves (again, see figure 14, left). Figure 14 (center) captures Gemological Center in Ashville, North Carolina, is from
a snapshot of the process of volatile exsolution and the Seridó, Rio Grande do Norte, Brazil. We have never seen
“bursting” of the two-phase inclusions, in which multiple this inclusion-host association before, and it is not previ-
gas bubbles seem to have precipitated instantaneously as ously recorded in the literature. This elegant gem is a per-
the inclusion exploded out into the surrounding corun- fect micromount of molybdenite in a faceted showcase.
dum. Due to the high relief of the silicate glass, these dis- John I. Koivula
Figure 16. It was a sur-

prise to discover
molybdenite crystal in-
clusions in topaz.
Shadowed lighting re-
veals iridescence and
surface details on the
cluster, which meas-
ures 2.82 mm in length.
Photomicrograph by
Nathan Renfro.

NATURAL DIAMONDS MISTAKENLY IDENTIFIED AS lonsdaleite in the diamonds described, even if found in
NPD SYNTHETIC DIAMONDS NPD, is certainly not proof of synthetic origin.
Another point it seems the authors did not consider is
We have read with much interest and surprise the recent the presence of large quantities of inclusions, especially
G&G lab notes on two purported synthetic nano-polycrys- “graphite” crystals (more correctly nanocrystalline carbon,
talline diamonds, or NPD (Spring 2014, pp. 69–71; Winter as there are no Raman bands of highly crystalline graphite),
2014, pp. 300–301). Looking at the data presented in these and hence the samples’ low clarity/transparency. The au-
reports, it appears unlikely to us that the analyzed samples thors note that “unlike natural diamond, the stone in ques-
are indeed NPD synthetic diamonds. tion was heavily included with graphite crystals.” Yet their
In the infrared spectrum shown in the first lab note, the morphology and distribution is not unlike that seen in many
authors did not mention the presence of strong, distinct natural diamonds, and it is not documented to our knowl-
absorptions at approximately 2460 and 660 cm–1; only two edge in synthetic diamonds, and certainly not in NPD. Note
bands at 3611 and 3727 cm–1 were described. These four also the murky yellowish bodycolor, which is different from
bands are well known and correspond with some deviation
to vibrations of molecular CO2:υ2 is the angular bend at
the yellow to brown color previously described for NPD.
about 667 cm–1, υ3 is the asymmetric stretch at about 2349

But most importantly, there are natural diamonds with
the same properties as synthetic NPD. We encountered
cm–1, and the two other bands are combinations of the fun-
damentals: υ3 + 2υ2 near 3683 cm–1, and υ3 + υ1 near 3686
such stones while studying the separation of natural-color
from treated-color black diamonds around 1998–1999. The
cm–1 (Schrauder and Navon, 1993). Note that Schrauder samples we acquired as faceted stones had been taken out
and Navon describe a natural diamond, not a synthetic of the ground even earlier. Sumitomo did not officially pro-
one—the presence of CO2-like features is a relatively rare duce NPD until 2012. We have been working for many
feature of natural diamond but has never been described years on the topic of “CO2-rich diamonds” and have 20
for any type of synthetic diamond. This oversight of CO2- faceted samples of natural polycrystalline black “CO2-rich”
rich diamonds is even more puzzling as GIA scientists and “lonsdaleite-rich” diamonds in our laboratory collec-
have participated in our publications on this variety of ditions that were acquired from large parcels of natural-color,
amonds (Hainschwang et al., 2006; Hainschwang et al., natural black diamond prior to 1999. Additionally, we own
2008), or published their own work on the subject (Wang an impact diamond from the Popigai meteorite crater (see
et al., 2005), including a lab note by the authors on a natu- Yelisseyev et al., 2013). These 21 diamonds have optical and
ral black CO2-rich diamond (Johnson and Moe, 2010). spectroscopic properties similar to those of the diamonds
Then the interpretation of the dominant one-phonon described in both lab notes., and vary only by the intensity
absorption band at 1220 cm–1 is missing. Interestingly, it is of the “CO2-like” and one-phonon absorptions. They all
also missing in previous papers on NPD (Skalwold, 2012; contain distinct amounts of nanocrystalline carbon inclu-
Skalwold et al., 2012), even though the band was first doc-
umented in the 1980s for natural and synthetic polycrys-
talline diamond. This band has been attributed to
absorption of the diamond lattice distorted by inclusions Figure 1. Two large indented naturals observed in a
of lonsdaleite; EPR spectroscopy has identified a very high pair of natural lonsdaleite-rich diamonds offer strong
concentration of broken C-C bonds due to interconnected evidence of their natural origin. Photos by Thomas
–
textures of diamond (111) and lonsdaleite (1010) (Bokii et Hainschwang.
al., 1986). Lonsdaleite has been known since the early
1960s (Ergun and Alexander, 1962), but its existence as a
discrete material has been questioned recently (Németh et
al., 2014). Instead, faulted and twinned cubic diamond has
been suggested as responsible for the unusual properties of
so-called lonsdaleite-rich diamonds. Since additional re-
search is clearly required to confirm or deny the existence
of lonsdaleite, we will continue to call these diamonds
500 μm 500 μm
“lonsdaleite-rich” for simplicity’s sake. Yet the presence of
LETTERS GEMS & GEMOLOGY FALL 2015 331

PL SPECTRA
701.8
Sample FN3674 REFERENCES
720.8
590 nm 732.0 Bokii G.B., Bezrukov G.N., Klyuev Y.A., Naletov A.M., Nepsha
band 669.2 V.I. (1986) Natural and Synthetic Diamonds. Nauka Press,
Moscow (in Russian).
PL INTENSITY
Ergun S., Alexander L.E. (1962) Crystalline forms of carbon: a pos-

First-order
Raman
sible hexagonal polymorph of diamond. Nature, Vol. 195, No.
405 nm excitation
First-order 4843, pp. 765–767, http://dx.doi.org/10.1038/195765a0
Raman 473 nm excitation Hainschwang T., Notari F., Fritsch E., Massi L., Breeding C.M.,
532 nm excitation Rondeau B. (2006) Natural “CO2-rich” colored diamonds.
G&G, Vol. 42, No. 3, p. 97.
635 nm excitation
Hainschwang T., Notari F., Fritsch E., Massi L., Rondeau B., Breed-
400 500 600 700 800 900 1000 1100 ing C.M., Vollstaedt H. (2008) HPHT treatment of CO2 contain-
ing and CO2-related brown diamonds. Diamond and Related
WAVELENGTH (nm) Materials, Vol. 17, No. 3, pp. 340–351, http://dx.doi.org/
10.1016/j.diamond.2008.01.022
Figure 2. The photoluminescence spectra of a natural Johnson P., Moe K.S. (2010) Lab Notes: Black diamond with solid
CO2 micro-inclusions and phosphorescent zones. G&G, Vol.
polycrystalline lonsdaleite- and “CO2-rich” diamond 46, No. 2. pp. 140–141.
recorded using four different laser excitations. These Németh P., Garvie L.A., Aoki T., Dubrovinskaia N., Dubrovinsky
spectra differ distinctly from the spectra of the NPD L., Buseck P.R. (2014) Lonsdaleite is faulted and twinned cubic
synthetic samples. diamond and does not exist as a discrete material. Nature Com-
munications 5, article no. 5447, http://dx.doi.org/10.1038/
ncomms6447
Schrauder M., Navon O. (1993) Solid carbon dioxide in a natural di-
amond. Nature, Vol. 365, No. 6441, pp. 42–44, http://dx.doi.org/
sions and are heavily strained due to their apparent poly- 10.1038/365042a0
crystalline structure, just like NPD. We have several sam- Skalwold E.A. (2012) Spotlight: Nano-polycrystalline diamond
ples with naturals on their surface that are undoubtedly sphere: A gemologist’s perspective. G&G, Vol. 48, No. 2, pp.
remnants of the etched surface of natural crystals and not 128–131, http://dx.doi.org/10.5741/GEMS.48.2.128
Skalwold E.A., Renfro N., Shigley J.E., Breeding C.M. (2012) Charac-
of NPD blocks (figure 1).
terization of a synthetic nano-polycrystalline diamond gemstone.
Additionally, some of these faceted diamonds contain G&G, Vol. 48, No. 3, pp. 188–192, http://dx.doi.org/10.5741/
fissures lined with iron oxide, plus solid unidentified inclu- GEMS.48.3.188
sions other than graphite. These features are unambiguous Wang W., Moses T., Moe K.S., Shen A.H.-T. (2005) Lab Notes:
proof of natural origin. On top of this, the photo- Large diamond with micro-inclusions of carbonates and solid
CO2. G&G, Vol. 41, No. 2, pp. 165–167.
luminescence spectroscopy data of the natural samples Yelisseyev A., Meng G.S., Afanasyev V., Pokhilenko N., Pusto-
varies distinctly from what is published in Skalwold et al. varov V., Isakova A., Lin Z.S., Lin H.Q. (2013) Optical proper-
(2012): The spectra excited by 405, 473, 532, and 635 nm ties of impact diamonds from the Popigai astrobleme.
lasers are characterized by very broad bands, always with Diamond and Related Materials, Vol. 37, pp. 8–16,
three somewhat less broad bands superimposed at 701.8,
720.8, and 732.0 nm (figure 2), a signature very different
from the broad band with relatively sharp bands at 612 and
667 nm described by Skalwold et al. (2012). Thus, they are REPLY
natural diamonds with properties close to NPD.
We appreciate the letter from Thomas Hainschwang and his
Looking at the similarities between our reference sam-
colleagues related to our two nano-polycrystalline diamond
ples of natural, apparently polycrystalline “lonsdaleite-
(NPD) lab notes from 2014. We identified the samples as
rich” diamonds and the samples described in the two
NPD primarily using FTIR spectroscopy and DiamondView
recent G&G lab notes, the weight of the evidence strongly
images, as performed by Skalwold et al. (2012). We also con-
indicates that the specimens examined by GIA are not syn-
firmed our identification using XRD analysis, though these
thetic NPD but natural black diamonds.
results were not included in the short lab notes.
Thomas Hainschwang (thomas.hainschwang@ggtl-lab.org) A broad peak at 1220 cm–1 in the mid-IR spectrum is
GGTL Laboratories – GEMLAB (Liechtenstein) typical of NPD samples (figure 3; see also Skalwold et al.,
Balzers, Liechtenstein 2012). Solid sintered NPD (i.e., cubic diamond) can be
Franck Notari achieved by growth conditions of ≥15 GPa and ≥2,200ºC
GGTL Laboratories – Gemtechlab (Sumiya, 2012). However, this P-T condition varies depend-
Geneva, Switzerland ing on the graphite starting material’s grain size and crys-
Emmanuel Fritsch talline nature. Consequently, a hexagonal diamond
University of Nantes, CNRS, Institut des Matériaux structure can also be observed in NPD, in addition to cubic
Jean Rouxel (IMN) diamond. The broad band at 1220 cm–1 in our samples is
Nantes, France most likely attributed to the structure of 6H hexagonal di-
332 LETTERS GEMS & GEMOLOGY FALL 2015

amond (Wu, 2007). In the one-phonon region (400–1332
cm–1), 2H hexagonal diamond (lonsdaleite) and 3C cubic di-
amond are not IR active. A 6H diamond of natural origin
has never been reported. Sumiya (2012) confirmed that
NPD contains 50 to 100 ppm of nitrogen, mostly aggre-
gated as A, B, and N3 center with less than 1 ppm of single
nitrogen. Thus, the 1220 cm–1 peak, which is located well
within nitrogen spectral region, can probably be enhanced
by nitrogen A and B aggregates.
NPD contains oxygen from 10 to 30 ppm (Skalwold,
2012). Thus, the CO2-related features could be observed in
these samples. Although peaks at 3727 and 3611 cm–1
(stated) and 2358 and 735 cm–1 (not stated) in the Spring 2014
lab note are similar to solid CO2 features, their positions are Figure 4. This DiamondView image of the 0.90 ct
not quite matched with established peaks. The XRD pattern sample shows the typical fluorescence pattern ob-
of our samples shows two sharp peaks at 44 and 75 degrees served in every NPD sample examined by GIA. Image
(2θ), consistent with NPD grown at 15 GPa and 2500ºC by Paul Johnson.
(Sumiya, 2008). We felt that a detailed discussion of IR and
XRD analyses was beyond the scope of our lab notes.
DiamondView images of NPD usually show a distinct,
mottled fluorescence pattern (figure 4). This typical pattern comments or conclusions. The letter from Dr. Hain-
has been observed in all NPDs we have examined at GIA’s schwang and his colleagues showed photos of lonsdaleite-
New York laboratory. rich diamonds. Since we have not seen these samples,
Every stone submitted to GIA is carefully examined however, a comparison cannot be made. For a true com-
using a gemological microscope and tested using standard parison, we would need IR spectra and DiamondView im-
and advanced gemological techniques before we make any ages, as these are key identification tools for NPD
synthetics. Polycrystalline diamonds (e.g., carbonado and
framesite) can be found in nature (Harlow and Davies,
Figure 3. This mid-IR spectrum of a 0.90 ct sample is
2005), but these are extremely small and would not reach
figure 14 from the Spring 2014 lab note. The four peaks
the sizes of the specimens examined in the lab notes.
at 3727, 3611, 2358, and 735 cm–1 are similar to estab-
We respect Dr. Hainschwang’s concerns but feel confi-
lished solid CO2 peaks. But the peak positions are
dent in our identification.
somewhat different, and they could be of different as-
signment. The broad peak at 1220 cm–1 is most likely Kyaw Soe Moe and Paul Johnson
attributed to the structure of 6H hexagonal diamond, GIA, New York
which has never been reported in natural diamond.
REFERENCES
Harlow G.E., Davies R.M. (2005) Diamonds. Elements, Vol. 1, No.
MID-IR SPECTRUM 2, pp. 67–70, http://dx.doi.org/10.2113/gselements.1.2.67
Skalwold E.A. (2012) Nano-polycrystalline diamond sphere: A
1220 gemologist’s perspective. G&G, Vol. 48, No. 2, pp. 128–131,
http://dx.doi.org/10.5741/GEMS.48.2.128.
Skalwold E.A., Renfro N., Breeding C.M., Shigley J. (2012) Transpar-
ent, faceted nano-polycrystalline synthetic diamond. News from
2358
Research, http://www.gia.edu/ongoing-research/transparent-
faceted-nano-polycrystalline-synthetic-diamond
ABSORBANCE
3611 Sumiya H. (2012) Novel development of high-pressure synthetic

3727 diamonds “ultra-hard nano-polycrystalline diamonds.” SEI
735 Technical Review, No. 74, pp. 15–23, http://global-sei.com/
technology/tr/bn74/pdf/74-03.pdf
Sumiya H., Irifune T. (2008) Microstructure and mechanical prop-
erties of high-hardness nano-polycrystalline diamonds. SEI
Technical Review, No. 66, pp. 85–92, http://global-sei.com/
technology/tr/bn66/pdf/66-11.pdf
4500 4000 3500 3000 2500 2000 1500 1000 Wu B.R. (2007) Structural and vibrational properties of the 6H
diamond: First-principles study. Diamond and Related Ma-
WAVENUMBER (cm-1)
terials, Vol. 16, No. 1, pp. 21–28, http://dx.doi.org/10.1016/
j.diamond.2006.03.013
LETTERS GEMS & GEMOLOGY FALL 2015 333

Contributing Editors
Emmanuel Fritsch, University of Nantes, CNRS, Team 6502, Institut des
Matériaux Jean Rouxel (IMN), Nantes, France (fritsch@cnrs-imn.fr)
Gagan Choudhary Gem Testing Laboratory, Jaipur, India (gagan@gjepcindia.com)
Christopher M. Breeding, GIA, Carlsbad (christopher.breeding@gia.edu)
COLORED STONES AND ORGANIC MATERIALS
Large star emerald. Star emeralds are rare specimens, and

most are found in either Brazil or Madagascar. In 1999,
Manuel Marcial de Gomar (Key West, Florida) purchased a
1,000-carat parcel of rough emeralds from Diallo Mah-
moud, an independent emerald dealer based in Africa.
Mahmoud noted that the parcel came from Madagascar.
Marcial de Gomar later noticed that some crystals showed
indications of chatoyancy. He chose this particular piece
thinking it could be cut to produce a cat’s-eye emerald.
Miami-based cutter Jone Ribeiro cut the stone into a 25.86
ct double cabochon (figure 1). Marcial de Gomar put the
stone into his personal collection, where it stayed for 13
years, until independent gemologist and appraiser Martin
Figure 2. Minute fluid inclusions oriented parallel to the
Fuller examined Marcial de Gomar’s collection in
basal plane of the crystal are the cause of this emerald’s
2013. Both experts recognized that this emerald was a star
asterism. The inclusions exhibit thin-film interference,
and not a cat’s eye. It was subsequently sent to GIA’s Carls-
causing rainbow colors that are visible at high magnifi-
bad laboratory for an identification report.
cation. Photo by Nathan Renfro; field of view 1.8 mm.
The stone is described as an oval double cabochon nat- Courtesy of Manuel Marcial de Gomar.
ural star emerald. The asterism was caused by minute
hexagonal fluid inclusions oriented parallel to the basal

Figure 1. A weak, six-rayed star is visible in this 25.86 ct plane of the crystal (e.g., Spring 2004 GNI, p. 104; K.
emerald cabochon. Photo by Robert Weldon/GIA, cour- Schmetzer, et al., “Asterism in beryl, aquamarine, and
tesy of Manuel Marcial de Gomar. emerald—an update,” Journal of Gemmology, Vol. 29, No.
2, 2004, pp. 66–71). They exhibit thin film interference
with direct, pinpointed light that caused the asterism, as
well as rainbow colors seen at high magnification (figure
2). Prior to this report, very few asteriated emeralds were
recorded (e.g. Fall 1995 GNI, p. 206; Spring 1995 GNI, pp.
Editors’ note: Interested contributors should send information and illustra-

tions to Stuart Overlin at soverlin@gia.edu or GIA, The Robert Mouawad
Campus, 5345 Armada Drive, Carlsbad, CA 92008.
GEMS & GEMOLOGY, VOL. 51, NO. 3, pp. 334–345.
334 GEM NEWS INTERNATIONAL GEMS & GEMOLOGY FALL 2015

Figure 3. The 4.40 ct natural red rutile on the left was Figure 5. Near the girdle, rectangular tabular inclusions
compared with a 4.47 ct orange flame-fusion synthetic are observed under fiber-optic illumination. Photomi-
rutile. Photo by C.D. Mengason. crograph by Ziyin Sun; field of view 2.43 mm.
(figure 4). Near the girdle, some unknown rectangular tab-

60–61; Spring 2002 GNI, p. 104; Summer 2006 GNI, pp.
ular inclusions were also observed, which were not identi-
171–172). This is the eleventh, and largest, star emerald
fied by Raman spectroscopy (figure 5). Both internal
known to date.
features indicated the specimen was natural rutile. Straight
Carl Chilstrom and angular growth banding was also observed (figure 6).
GIA, Carlsbad Raman spectroscopy for both the red sample and the or-
ange synthetic revealed major peaks at 610, 446, and 242
Natural faceted red rutile. The Carlsbad laboratory recently cm–1 and minor peaks at 818, 707, and 319 cm–1, confirm-
received a 4.40 ct dark red marquise brilliant (figure 3, left) ing their identity as rutile (see the RRUFF database at
for examination. Standard gemological testing revealed a rruff.info).
refractive index that was over the limit of the RI liquid, as Laser ablation–inductively coupled plasma–mass spec-
well as a hydrostatic specific gravity (SG) of 4.27. The spectrometry (LA-ICP-MS) revealed that the dark red rutile had
imen was inert to both long-wave (LW) and short-wave a composition of 98.2% TiO2 and 1.5% Fe2O3 by weight,
(SW) UV. Under reflected light, it showed a metallic luster. with traces of Mg, Al, Si, Ca, Sc, V, Cr, Cu, Zn, Zr, Nb, Cd,
Raman testing identified the material as rutile. To distin- In, Sn, Sb, Hf, Ta, W, Pb, and U. The presence of Nb
guish it as natural or synthetic rutile, we compared it with (176.81–296.93 ppmw), Ta (15.13–25.81 ppmw), Hf (0.15–
a known flame-fusion orange synthetic rutile sample (fig- 0.40 ppmw), Zr (3.29–4.84 ppmw), Pb (0.01–0.56 ppmw),
ure 3, right). and U (0.07–0.27 ppmw) suggested a natural origin. The or-
Under magnification, the most distinctive internal ange synthetic rutile contained 99.9% TiO2, plus traces of
characteristic of the dark red specimen was an octagonal Mg, Si, Cr, Co, Cu, and Zn. Traces of iron (1.66 ppmw,
blockage with a long growth tube extending to the surface 0.0238% Fe2O3 by weight) were also detected. Synthetic
Figure 4. In this natural red rutile, an octagonal block- Figure 6. The natural rutile shows straight and angu-
age is capped at the end of the growth tube. Photomi- lar growth banding. Photomicrograph by Ziyin Sun;
crograph by Ziyin Sun; field of view 1.40 mm. field of view 4.92 mm.
GEM NEWS INTERNATIONAL GEMS & GEMOLOGY FALL 2015 335

Figure 7. Face-up view of two quartz cabochons from Figure 8. The stars in the quartz samples, seen from the
Brazil, weighing 116.95 ct (left) and 44.12 ct (right) with base, are eye-visible in sunlight. Photo by Jaroslav
cross-like quartz inclusions. Photo by Jaroslav Hyršl. Hyršl.
rutile boules created by the Verneuil process have an or- sions is uncertain, but they are strong evidence of nature’s
ange color with 0.04% Fe2O3 and a clear reddish color with ability to produce rarities.
0.2% Fe2O3 (C.H. Moore, Jr. and R. Dahlstrom, “Synthetic Jaroslav Hyršl (hyrsl@hotmail.com)
rutile crystal and method for making same,” U.S. Patent Prague
2792287). The red specimen’s lower chemical purity indi-
cates natural origin. Serpentine cabochon with unusual olive green color. Ser-
Visible spectra of both stones were taken. The red body- pentine is a common ornamental stone in the jewelry
color of the natural rutile matched the transmission win- trade. With colors ranging from yellowish green to deep
dow between 650 and 700 nm in the red portion of the green, it is commonly used as an imitation of jadeite or
spectrum, which was likely caused by iron. nephrite.
Although rutile is a common inclusion in many min- Recently, the Gem Testing Laboratory at the Shiji-
erals, the few crystals that are faceted usually weigh less azhuang University of Economics in Hebei, China, exam-
than 1 ct. This faceted stone was very rare because of its ined an olive green cabochon (figure 9) weighing 4.05 ct and
large size. It was the first faceted natural rutile examined measuring approximately 12 × 8 × 6 mm, reportedly pur-
by GIA’s Carlsbad laboratory. chased from China’s Shandong province. The sample was
Ziyin Sun (zsun@gia.edu), Amy Cooper, and relatively clean, with a vitreous luster and without any
Adam Steenbock eye-visible inclusions. Without magnification, it resembled
GIA, Carlsbad high-quality jadeite jade. Standard gemological testing gave
Rare “star and cross” quartz from Brazil. The author pur-
chased two very unusual quartz samples (figure 7) in Brazil Figure 9. This 4.05 ct translucent serpentine is unusual
in 2015. Both were cut as cabochons with a slightly curved for its olive green color and lack of impurities. Photo by
bottom; one weighed 116.95 ct (35.0 × 30.3 × 15.6 mm), Yanjun Song.
while the other weighed 44.12 ct (28.3 × 22.3 × 10.0 mm).
The larger cabochon was slightly milky, while the smaller
one was almost clear. Both showed a strong six-rayed star
when illuminated (figure 8). The stars were centered on the
top of the cabochons, which indicates that both were cut
perpendicular to the c-axis. The most unusual feature was a
cross-like inclusion in both stones. It was white and coarse-
grained, only slightly translucent. The longer part of both
crosses measured about 4 mm wide. The shorter part had a
thickness 1.5 to 2 mm and the angle of the intersection was
about 85 degrees in both cabochons, which meant that the
two specimens were probably cut from the same crystal. RI
results and Raman analysis showed the cross to be quartz,
rather than a foreign mineral. The origin of the cross inclu-

FTIR SPECTRUM
50
1046
tine from Liaoning, indicating that the specimen may have
482
formed from the serpentinization of mafic-ultramafic
REFLECTANCE (%)
40
rocks. We suspect it may have originated in Taian City in
Shandong province, which is famous for “Taishan jade,” a
30
558 newfound variety of serpentine.
20
672 Although serpentine is common in the Chinese mar-
ket, this was the first time our researchers have encoun-
10
tered a sample of such high quality.
Yanjun Song and Lu Zhang
0 Shijiazhuang University of Economics, Hebei, China
4000 3000 2000 1000
WAVENUMBER (cm–1) Thai-Myanmar petrified woods. Petrified woods (figure 11,

left) have long been used as a gem and ornamental material
Figure 10. The FTIR spectrum of the 4.05 ct emerald- due to their beauty, luster, durability, and relative rarity.
green cabochon sample exhibits characteristic peaks at Certain minerals (mainly quartz, opal, or moganite) crys-
1046, 672 and 482 cm–1, all consistent with serpentine. tallize or precipitate in the hollow cavities and cells of the
woods in a fossilization process. Fossil trading is govern-
ment regulated in some countries, and gemologists should
a spot RI of 1.56 and a hydrostatic SG of approximately know the useful identifying characteristics of petrified
2.56; the sample was inert to both LW and SWUV. These woods, which can be misidentified as jasper and agate.
properties are consistent with serpentine. Four main peaks With this in mind, the author examined Thai-Myanmar
were observed in the 1100–400 cm–1 region of the infrared petrified wood samples, fashioned as cabochons (figure 11,
reflectance spectrum (figure 10). The two peaks at about right), for gemological characterization.
1046 and 482 cm–1 were related to the fundamental Si-O More than 100 Thai samples were collected from the
vibrations. The remaining two peaks at 672 and 558 cm–1 four localities shown in figure 12: Tak, in Ban Tak-Oke
are attributed to the OH– and Mg-O vibrations, respec- (TAK); Khon Kaen, in Ban Hin Khao (BHK); Chaiyaphum,
tively, confirming the sample’s identity as antigorite, a in Noen Sa Nga (CYP); and Nakhonratchasima, in Kroke
species in the serpentine family. Duan Ha (KDH). About 100 Burmese samples were do-
EDXRF was used to differentiate the sample from the nated by carvers from Mae Sai in Chiang Rai province
common “Xiuyu” variety of serpentine, from China’s (MYR), near the border of Thailand and Myanmar. Thai
Liaoning province. Testing revealed that both Cr and Ni specimens were collected from either Quaternary gravel
were significantly higher in this material than in serpen- beds of fluvial deposits or the high-terrace sands and grav-
Figure 11. Left: Petrified

woods are carved and
sold as ornamental gem
materials, such as these
specimens from the
Mae Sai market near
the Thai-Myanmar bor-
der. Right: A few of the
cabochons from this
study. Photos by Seri-
wat Saminpanya.
2 cm

100o 105o 110o
CHINA
MYANMAR
Mae Sai
(MYR)
Hanoi
20o
LAOS GULF
OF
TONKIN
Tak (TAK)
Khon Kaen (BHK)
Rangoon
THAILAND SOUTH
Figure 12. A map show-
Chaiyaphum (CYP) CHINA ing the localities in
SEA Thailand and Myanmar
Nakhonratchasima (KDH) 15o
where the petrified
Bangkok wood samples origi-
nated. Adapted from
TNAM
Saminpanya and
Sutherland, 2013.
CAMBODIA
VIE
Ho Chi
Phnom Minh City
Penh
GULF
10o
OF
THAILAND
N
Collection sites
0 200 km
Major cities
5o
els of paleorivers. Tektite grains found in the sediments The rough surfaces may have been altered and stained
date back 700,000 to 900,000 years (P. W. Haines et al., by soil or iron oxides, resulting in black, brown, cream,
“Flood deposits penecontemporaneous with ~0.8 Ma tek- white, and red coloration. Colorless areas and different
tite fall in NE Thailand: Impact-induced environmental ef- shades of brown, black, gray, white, cream, yellow, orange,
fects?” Earth and Planetary Science Letters, Vol. 225, 2004, and red appeared below the surface. Some samples exhib-
pp. 19–28). ited color banding. Most Thai samples were opaque or
Thirty-eight samples (all but seven from Thailand) were nearly opaque; those from Myanmar were translucent and
tested by standard gemological methods, Raman spec- lighter in tone. The surfaces showed varying degrees of lus-
troscopy, and X-ray diffraction (XRD). Thirty-seven of these ter and fracturing. RI ranged from 1.425 to 1.543 (tested at
came from dicotyledonous plants, indicated by vascular the flat polished base of the cabochon), and SG ranged from
bundles concentrated in an outer ring of active growth. Pe- 1.963 to 2.616. Most samples fluoresced weak to moderate
riods of slow (winter) and fast (summer) growth create the chalky white under LWUV, but some were inert or showed
growth rings in the trunk. One sample came from a palm very weak white fluorescence.
tree, and had vascular bundles dispersed throughout the A gemological microscope can often reveal a well-pre-
trunk. served wood grain structure in petrified wood samples.

R AMAN SPECTRA
0.288
Studies of wood structure generally involve the transverse,
tangential, and radial sections of the trunk (R.F. Evert,
Esau’s Plant Anatomy: Meristems, Cells, and Tissues of 6
0.286 CYP-A
the Plant Body: Their Structure, Function, and Develop-
3 A
ment, 3rd ed., John Wiley & Sons, Hoboken, New Jersey, TAK-
2006, pp. 203, 292–296). A transverse section of one of the 0.284 CYP-
A6
samples (figure 13) showed several tiny oval-shaped pores
QM
INTENSITY (COUNTS)
filled with quartz or opal (e.g., S. Saminpanya and F.L.
Sutherland, “Silica phase-transformations during diagenesis 0.282 Q
within petrified woods found in fluvial deposits from Thai-

-1A
M BHK
land-Myanmar,” Sedimentary Geology, Vol. 290, 2013, pp. Q
Q
0.280
15–26). The samples also showed thin parallel rays, groups
of cells oriented perpendicular to the trunk’s main axis. In
some samples, the parenchyma (the most common plant
0.278 Q
tissue cells) consisted of pale lines or bands running across
the rays. Others showed parenchyma surrounding a pore
on the long and parallel wood grains of the tangential sec- 0.276
tion, or dark-colored pores extending the length of the
trunk. On the radial section, the pores ran the length of the
trunk, with wood grains perpendicular to the pores. 0.274
Raman spectra were characteristic of opal (according to 250 450 650 850 1050 1250 1450 1650 1850
the RRUFF database, rruff.info), quartz, and moganite (K.J. RAMAN SHIFT (cm–1)
Kingma and R.J. Hemley, “Raman spectroscopic study of
microcrystalline silica,” American Mineralogist, Vol. 79,
Figure 14. Raman spectra of petrified wood samples
1994, pp. 269–273). Figure 14 shows the peaks for the TAK,
from Chyaphum (CYP), Tak (TAK), and Khon Kaen
BHK, and CYP samples; peaks for KDH (which included
(BHK) with quartz (Q) peaks at 465 cm–1. Samples from
opal) and MYR samples were previously presented (Samin-
Chyaphum show a moganite (M) peak at 503–506 cm–1.
panya and Sutherland, 2013). Most of the quartz material
displayed peaks at 354, 395–397, and 464–466 cm–1, with
different colors within a sample exhibiting different peak
intensities. Some samples showed only major peaks of
quartz at 464–466 cm–1. Moganite was mixed with quartz
in one sample apiece from CYP, KDH, and MYR, indicated by peaks at 503–506 cm–1, close to the reference peak of 501
cm–1 (Kingma and Hemley, 1994). Two KDH samples
showed bands centered at 1587 and 1607 cm–1, characteris-
Figure 13. Oval pores and rays (wavy white lines) are tic of opal (Saminpanya and Sutherland, 2013), which was
seen on a transverse section of this petrified wood sam- probably responsible for the weak to moderate chalky white
ple, and the growth rings are visible as two lines cross- fluorescence in some of the samples.
ing the image. Photomicrograph by Seriwat XRD showed that the quartz peaks had an average 2Θ
Saminpanya; field of view 10 mm. of 21.0, 26.7, 36.6, 39.6, 40.4, 42.5, 45.9, 50.2, and 55.0 de-
grees; the opal-CT peaks had an average 2Θ of 20.7, 21.7,
36.0, 44.0, and 56.9 degrees (again, see rruff.info). Quartz
appeared in the XRD peaks of samples from all localities,
but the opal-CT was found only in samples from KDH and
MYR. The diffractogram of opal-CT in one sample from
KDH appeared noisy. This may be due to its amorphous
nature, though extending the XRD scan times may im-
prove the resolution of the peaks. At this stage, XRD can
only detect quartz and opal from these samples.
Seriwat Saminpanya
Srinakharinwirot University, Bangkok
SYNTHETICS AND SIMULANTS
Coated rock crystal imitation of ruby. Recently the Dubai

Central Laboratory received for identification a rough stone

Figure 16. The rock crystal shows dark red under trans-
mitted light to imitate ruby. Photo by Nazar Ahmed
Ambalathveettil, Dubai Central Laboratory.
Figure 15. This coated rock crystal imitation of ruby has

a brown skin. Photo by Nazar Ahmed Ambalathveettil, light and no coating was observed there. The stone’s SG
Dubai Central Laboratory. was 2.60. The coated area gave a patchy chalky blue and
purple reaction under LWUV and was inert to SWUV. Spe-
cific gravity and inclusions indicated a rock crystal; this
was confirmed by Raman spectroscopy.
weighing 616.93 g and measuring approximately 114.3 × This was the first time the Dubai Central Laboratory
65.8 × 53.0 mm (figure 15). Most of it was covered by a had identified such a large coated rough stone imitating
brown skin, while other parts were transparent. Viewed ruby.
through one of the transparent areas, the stone showed a Nazar Ahmed Ambalathveettil (nanezar@dm.gov.ae)
red color. Under oblique light, the stone appeared dark red, and Mohamed Karam
with some areas that were colorless (figure 16). Gemstone Unit, Dubai Central Laboratory
Microscopic examination with transmitted light re- Dubai, United Arab Emirates
vealed negative crystals, fingerprints, and two-phase inclu-
sions. Observation of the transparent and opaque areas on Color-change synthetic cubic zirconia as peridot imitation.
the surface showed red, green, and blue areas of coating Peridot is a common gemstone in the jewelry industry; how-
that could be removed using sharp needles. These coated ever, it is rarely simulated using substances other than glass
areas gave the stone a purplish red color overall, but some (Summer 2004 Lab Notes, p. 165). The State Gemological
transparent areas appeared colorless under transmitted Centre of Ukraine recently received for identification an
Figure 17. This syn-

thetic cubic zirconia
imitation of peridot,
with an estimated
weight of 4.50 ct, is
shown in daylight (left)
and incandescent light
(right). Photos by Igor
Iemelianov; field of
view 13.20 mm.

oval-cut gemstone, with an estimated weight of 4.50 ct, set
in a ring. The stone was greenish yellow in daylight (figure
17, left) and orangy yellow in incandescent light (figure 17,
right). The client who submitted the ring specifically wished
to know if the center stone was peridot. The sample was
transparent and isotropic, with an RI of 2.150 (measured by
a refractive index meter) and was inert to both SW and
LWUV radiation. Microscopic examination revealed no in-
clusions. The measured relative reflectivity (RR 65–67), ob-
tained using a Presidium DuoTester, clearly matched that
of cubic zirconia, a result in line with the other gemological
observations. This allowed us to identify the stone as cubic
zirconia. Chemical analysis with EDXRF showed Zr and Y,
with minor Pr and Nd. These impurities can cause color
change in this stone. Using fiber-optic light with the spec-
troscope, we observed a complex set of absorption lines in
both the orange and green-blue regions (the same measure-
Figure 18. This 37.50 ct plastic bead contained visible
ment can be seen at http://www.gemlab.co.uk/peridot
fragments of carbonate-type coral (the lighter spots).
.html). This study shows the importance of studying a vari-
Photo by Gagan Choudhary.
ety of simulants, even those for less-expensive gems such as
peridot.
Iurii Gaievskyi (gaevsky@hotmail.com)
and Igor Iemelianov enough to rule out that possibility. A few spots of lighter
State Gemological Centre of Ukraine, Kiev color (again, see figure 18), visible to the unaided eye, made
it more interesting. Under strong fiber-optic light, minute
Coral inclusions in plastic. Composites assembled from transparency was observed, while microscopic examination
opaque-to-translucent and transparent gem materials such revealed numerous gas bubbles scattered throughout the
as turquoise, chalcedony, opal, chrysocolla, tourmaline, bead. Sharp boundaries between the lighter spots and the
and peridot have become popular over the past few years, host bead suggested that these spots of lighter color were
as evidenced by the number of samples received for iden- actually fragments of some other material. These randomly
tification at the Gem Testing Laboratory in Jaipur, India. oriented fragments also displayed some concentric radiating
Recently, we examined an orangy red bead of plastic (figure structures (figure 19, left), typically seen in carbonate-type
18) with embedded fragments of coral. corals. Some chips also showed elongated sections (figure
The round bead weighed 37.50 ct and measured 20.62 × 19, right), along with some whitish to faintly colored areas.
20.57 × 14.02 mm. The bead was represented as coral, but Although the bead was obviously an artificial product,
its smooth texture along with the luster was sufficient its major component was still to be identified. A spot RI of
Figure 19. Some of the fragments embedded in the plastic bead displayed concentric radiating structures (left), typ-
ically seen in carbonate-type coral, while others show elongated sections (right). Also note the sharp edges and
whitish to faintly colored areas of the fragments. Photomicrographs by Gagan Choudhary; field of view 5.08 mm.

R AMAN SPECTRA
1340
1140 Plastic
Coral
1005 1109 1315
1391
1427 Figure 20. Raman spec-
1200 1451
15000 1580 tra using 785 nm laser
INTENSITY (COUNTS)
1606
confirm that the major
component of the bead
is plastic (blue trace),
10000
1093 while the grains are a
carbonate such as cal-
cite (red trace). The
lack of carotenoid-re-
5000
lated peaks suggests
310 the use of white coral
717 fragments.
200 400 600 800 1000 1200 1400 1600 1800
RAMAN SHIFT (cm–1)
approximately 1.53 and hydrostatic SG of 1.97 were ob- tics have also been known as coral imitations for decades.
tained. Under UV light, the bead fluoresced orange, with a This specimen turned out to be notable because of the use
stronger reaction to SW than LW; a desk-model spectro- of coral fragments as inclusions; however, we could not
scope revealed a broad absorption band from the violet to clearly understand its purpose.
yellow-orange region. Raman spectroscopy (figure 20) con-
Gagan Choudhary (gagan@gjepcindia.com)
firmed the major component of the bead as plastic, while
Gem Testing Laboratory, Jaipur, India
the grains were identified as carbonate. We did not detect
any carotenoid-related peaks, which have been suggested
as the cause of color in pink-to red-coral (C.P. Smith et al., A jadeite bangle simulant: Hydrogrossular garnet. Jadeite,
“Pink-to-red coral: A guide to determining origin of a material significant within Chinese culture, is often sub-
color,”Spring 2007 G&G, pp. 4–15). On the basis of Raman mitted to the Lai Tai-An Gem Laboratory for identification.
spectra and concentric radiating structure, the sharp frag- Historically, the name “jade” has been applied to either
ments were identified as white coral. nephrite or jadeite. Nephrite has been used for adornment
Coral-plastic composites have been reported previously and ornamentation for thousands of years in China, and
(e.g., Fall 2008 Lab Notes, p. 253). Further, orange-red plas- nephrite objects from many dynasties are found in muse-
Figure 21. The photo on the left shows a jadeite jade bangle. The photo on the right shows two hydrogrossular gar-
net bangles that were claimed to be jadeite. Photos by Lai Tai-An Gem Lab.

R AMAN SPECTRUM
877
1600
1400
INTENSITY (COUNTS)
367
1200
1543
1000 823 1489
408
800 544
600
400
200
0
500 1000 1500 2000
RAMAN SHIFT (cm–1)

Figure 22. These hydrogrossular garnet bangles exhibit
the black pepper-like inclusions that are sometimes
found in this material. Photo by Lai Tai-An Gem Lab,
Figure 24. Raman analysis of one of the client’s ban-
field of view 8.20 mm.
gles, using a near-infrared wavelength of 785 nm, con-
firms its hydrogrossular garnet identity.
ums, auctions, and antique shops around the world. While
newer to the Chinese market, jadeite jade is becoming in- quire more rough than cabochons or fashioned stones, re-
creasingly important. Jadeite bangles (figure 21, left) the resulting in a larger amount of waste. Consequently, the
value placed on good-quality pieces is higher, and jadeite
bangles with fine textures and brilliant colors continue to
Figure 23. FTIR spectra reveal the difference between set new records in auction houses.
jadeite jade (red line, peaks at 1175, 1075, 1051, 957, We recently received two similar-looking bangles (fig-
855, 745, 666, 589, 534, 476, and 435 cm–1) and hy- ure 21, right) for identification. The client claimed the
drogrossular garnet (blue line, peaks at 954, 866, 843, pieces were jadeite, but we determined them to be hy-
616, 560, 488, and 458 cm–1). drogrossular garnet. This jadeite simulant is rarely seen in
bangle form; it is more likely encountered as carvings or
cabochons. In this particular case, the similarity to jadeite
FTIR SPECTRA
was extremely convincing.
0.4
1075 The two translucent, light to strongly saturated green and
Hydrogrossular garnet
Jadeite jade white hydrogrossular garnet bangles weighed 339.55 ct and
589
0.35 400.03 ct, and measured 75 × 11 mm and 77 × 12 mm, re-
spectively. Both exhibited a spot RI of 1.72, SG of 3.42/3.45,
inert reaction to both SW and LWUV, and a light pink reac-
476
0.3 tion through the Chelsea color filter. The absorption spec-
534
trum showed a cutoff below 460 nm, and the material
exhibited black pepper-like inclusions (figure 22); these char-
TRANSMITTANCE (%)
0.25
957 acteristics suggest that the color of the garnet is natural.
The identification was confirmed by FTIR and Raman
0.2 analysis. Peaks at 954, 866, 843, 616, 560, 488, and 458 cm–
954 1
in the infrared spectra (figure 23), and peaks at 367, 408,
544, 823, 877, 1489, and 1543 cm–1 (figure 24) in the Raman
866
0.15
560
843 spectrum are indicative of hydrogrossular garnet. The sim-
855 ilarity to jadeite could lead to a costly error, so those in the
trade must be aware of such imitations.
0.1
Larry Tai-An Lai (service@laitaian.com.tw)
745 Lai Tai-An Gem Laboratory, Taipei
0.05 666
616 TREATMENTS
0
Impregnated and dyed turquoise. Turquoise has a cryp-
500 1000 1500
tocrystalline structure, which gives rise to the gem’s poros-
WAVENUMBER (cm–1)
ity. Its vulnerability to body oils, ordinary solvents, and dirt
can induce variation of color. This porous gemstone’s ap-

Figure 25. Spectroscopic analysis of this 15.21 gram Figure 26. Some lines, along with the cavity, fluoresce
turquoise reveals that the material has been both im- white when exposed to long-wave UV radiation, indicat-
pregnated and dyed. Photo by Wen Han. ing the presence of filling material. Photo by Wen Han.
pearance and durability can be enhanced by various treat- and SWUV radiation (figure 26), suggesting impregnation.
ment techniques, such as dyeing, impregnation (with poly- Magnification showed some filling material were also pres-
mer, wax, or plastic), and the proprietary Zachery process ent in the cavities. Additionally, green color concentration
(E. Fritsch et al., “The identification of Zachery-treated was observed in the material’s cavities and cracks.
turquoise,” Spring 1999 G&G, pp. 4–16). Infrared reflectance spectroscopy revealed complicated
Turquoise, often featured in vintage jewelry pieces, is peaks containing turquoise and impregnated materials. The
becoming increasingly popular in the Chinese market. Re- bands at 1118 and 1050 cm–1 were assigned to the asym-
cently, the NGTC Beijing lab received for identification a metric stretching vibrations of phosphate units, while the
15.21 gram turquoise sphere with green color (figure 25). 835 cm–1 was caused by the bending vibration of OH units.
The specimen measured approximately 3.5 cm in diameter; Other features from approximately 647 to 482 cm–1 were
its spot RI value was 1.59 and its hydrostatic SG value was due to the phosphate bending modes of turquoise. The 2918
2.58. The test sample was inert, except for several lines and and 2850 cm–1 peaks were attributed to wax; additionally, a
a cavity with moderate white fluorescence under both LW series of small peaks from 1800 to 1350 cm–1 indicated the
IR SPECTRUM
97
1410
95
1611 Figure 27. The infrared
TRANSMITTANCE (%)
1731 1510 transmittance spectrum

1260
93 of small amounts of
powder scraped from
2850
the sample shows that
91
2918 the turquoise was im-
pregnated with three
89 components: epoxy
resin, acrylic polymer,
87 and silicone.
797
1013
85
3500 3000 2500 2000 1500 1000 500
WAVENUMBER (cm–1)

UV-VIS SPECTRA
0.3
677
respectively, and the 797 cm–1 was assigned to the Si-C
0.28 stretching vibrations of the silicones.
This sample
EDXRF chemical analysis detected mainly P, Al, Cu, Fe,
0.26 and Si. The first four are consistent with the chemical com-
0.24
Natural turquoise position of turquoise, while Si was suspected to be a byprod-
ABSORBANCE
uct of silicone impregnation. The UV-Vis spectrum showed

0.22 two absorption bands: a band centered at about 428 nm
caused by Fe3+ d-d electronic transition, and a sharp band
0.2
428
centered at approximately 677 nm (figure 28). Natural-color
0.18 turquoise does not show the 677 nm band; many Chinese
labs consider this feature characteristic of dyed material.
0.16
Turquoise is usually impregnated with acrylic polymer
0.14 or epoxy resin for stabilization in addition to the dyeing
process to change its color. This is the first time we have
0.12
encountered turquoise that was dyed and impregnated with
350 450 550 650 750
three components. Our investigation reinforces the need to
WAVELENGTH (nm)
identify any and all treatments.
Wen Han (winnerzx@126.com), Taijin Lu, Huiru Dai,
Figure 28. When compared with the spectrum of a and Jun Su
known natural turquoise sample, the 677 nm band of National Gems & Jewelry Technology
the UV-VIS spectrum indicates that the sample in this Administrative Center (NGTC)
study was dyed. Beijing
Hui Dai
Institute of Geological Experiment of Anhui Province
presence of organic matter. To identify this organic mate- Hefei, China
rial, we scraped off some powder from the edge of one of the
cavities with a blade. The resulting infrared transmittance ERRATA
spectrum indicated the presence of three kinds of polymer
(figure 27). The 1731 and 1013 cm–1 peaks were attributed 1. In the Summer 2015 article by E. Sorokina et al.,
to acrylic polymer, and the 1611 and 1510 cm–1 peaks were “Rubies and Sapphires from Snezhnoe, Tajikistan,”
due to epoxy resin; both are commonly applied to various coauthor Dr. Andrey K. Litvinenko was incorrectly
gem materials for impregnation. Peaks at 1410, 1260, and listed as director of GIA Moscow.
797 cm–1 were characteristic absorption features of silicone.
Silicones are polymers consisting of repeating units of silox- 2. In the Summer 2015 GNI entry “Crimson Prince
ane (a functional group composed of two silicon atoms and ruby from Namya” (pp. 214–215), the location of
one oxygen atom, often combined with carbon and hydro- Namya was incorrectly shown on the map. The lo-
gen). The 1410 and 1260 cm–1 peaks were caused by asym- cation has been corrected in the online versions
metric and symmetric deformation vibrations of Si-CH3 , (HTML and PDF) of the Summer 2015 issue.
IN MEMORIAM
TINO HAMMID (1952–2015)
Gems & Gemology mourns the loss of acclaimed gem and issue of Modern Jeweler. Their 25-year collaboration yielded
jewelry photographer Tino Hammid, who died July 11 at two Jesse H. Neal awards for business journalism and a pair
the age of 63 after a two-year battle with cancer. Mr. Ham- of Gem Profile hardcover volumes: The First 60 (1988) and
mid is widely remembered throughout the industry for his The Second 60 (1992). Mr. Hammid’s work appeared in
kindness and his passion for the craft. countless other books, and starting in 1987 he photographed
The son of Academy Award–winning filmmaker more than 100 Christie’s jewelry auction catalogs.
Alexander Hammid, the New York native established his G&G has been privileged to showcase Mr. Hammid’s
career in Los Angeles. From 1980 to 1982, he served as a talents for over 30 years. Longtime readers of the journal
staff photographer at GIA’s Santa Monica campus. Upon will recognize his distinctive backgrounds and use of light
leaving GIA, he became a successful freelance photogra- and shadow. Through his style and technique, he exerted a
pher, capturing some of the world’s most remarkable gems. lasting influence on gem photography. Tino Hammid is
In 1983, he began photographing gemstones for David survived by his wife, Petra, and three children. We extend
Federman’s popular Gem Profile column in each monthly our deepest condolences to his family and friends.

FA15 Gems Gemology PDF

Uploaded by

Copyright:

Available Formats

FA15 Gems Gemology PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

FA15 Gems Gemology PDF

Uploaded by

Copyright:

Available Formats

FALL 2015

THE QUARTERLY JOURNAL OF THE GEMOLOGICAL INSTITUTE OF AMERICA

Colombian Trapiche Emeralds

280 Letšeng’s Unique Diamond Proposition

NOTES AND NEW TECHNIQUES

Editorial Review Board Copyright and Reprint Permission

About the Cover

Duncan Pay | Editor-in-Chief | dpay@gia.edu

EDITORIAL GEMS & GEMOLOGY FALL 2015 221

COLOMBIAN TRAPICHE EMERALDS:

222 COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015

COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015 223

Figure A-1. Trapiche vs. trapiche-like

224 COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015

al ern Cordillera Basin in

6º N Peñas Blancas deposits are hosted by

Lower Cretaceous sedi-

mentary rocks forming

Muzo two mineralized zones

the basin, respectively.

Muzo, Coscuez, La Pita,

and Peñas Blancas,

while Gachalá, Chivor,

and Macanal lie on the

COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015 225

226 COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015

Tequendama mine (Muzo deposit)

Tequendama siliceous black shale, and

COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015 227

20 m ure 6) is located under the administration building of

the Muzo mine and corresponds to a klippe of the

Los Chulos calcareous black shale that overthrust the siliceous

228 COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015

whitish albitite), cemented by carbonates and pyrite.

event linked to evaporite dissolution and driven by Administration

Trapiche and non-trapiche emeralds are associated Thrust

COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015 229

230 COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015

Hexagonal prismatic Pinacoidal growth

Figure 9. Different mor-

COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015 231

232 COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015

COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015 233

Figure B-1. This diagram shows

nature of the interfaces is classi-

234 COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015

COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015 235

236 COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015

Coscuez T3 Center, arms 6 ‖ AbBS Ab, Dol, Brl, Cal, 3 3

Peñas T5 Hexagonal green 4 ‖ AbBS c: 0.68–0.72 0.30–0.40 Kln, Fe(O,OH), Ab, 3 3 3 3

Muzo T6 Center, Arm up to ‖ None 3 3

Muzo T7 Small core, arms, 10 ‖ None Mca, Kln, Tur, Dol, 3 3

Muzo T9 Hexagonal green 5 ‖ AbBS Brl, Mnz, Qtz, Pt, 3 3 3

Muzo T12 Small hexagonal 12 ‖ AbBS

Muzo T13 Core, arms 5 ‖ Ab

COLOMBIAN TRAPICHE EMERALDS GEMS & GEMOLOGY FALL 2015 237

arm core arm en

pyramid’s vertex the pyramid’s vertex (figure 13B). This is confirmed

Figure 15. X-ray computed tomography images of a