Professional Documents
Culture Documents
Fired Porcelain Microstructures Revisited
Fired Porcelain Microstructures Revisited
The microstructure of standard commercial porcelain con- or concentration of foreign ions in the mullite crystal lattices,
sists of ␣-quartz grains held in a complex matrix. Quartz one originating from feldspar (K2O⭈Al2O3⭈6SiO2) and the
grains are surrounded by amorphous silica-rich solution other from metakaolin (Al2O3⭈2SiO2). However, it may be that
rims while the matrix contains clay relicts of small (∼200 the large, uniform aggregates of primary mullite crystals are
nm × 40 nm × 40 nm) primary (2Al2O3:1SiO2) mullite crys- intrinsically less soluble than the fine, acicular secondary mul-
tals in aluminosilicate glass and feldspar relicts of acicular lite. Schuller6 showed that at 1400°C all primary mullite trans-
(>1 µm long) secondary (3Al2O3:2SiO2) mullite in potas- forms to secondary mullite and, due to the complete dissolution
sium aluminosilicate glass. A continuous increase in mullite of quartz, a microstructure consisting simply of glass and mul-
crystal size from the clay–feldspar relict interface to the lite needles results.
feldspar relict center and their compositions indicate a Mullite is the only stable phase in the Al2O3⭈SiO2 system at
transformation from primary to secondary mullite. In alu- atmospheric pressure. Because of its potentially favorable
minous porcelain, corundum grains are observed in addi- properties it has a significant role in the development of tradi-
tion to ␣-quartz and regions of clay and feldspar relicts. tional and advanced ceramics.9 Mullite is orthorhombic (space
Small (∼50 nm × 10 nm × 10 nm) tertiary mullite crystals group Pbam) and its composition ranges from approximately
(formed by precipitation from Al2O3-rich glass) were de- 3Al2O3:2SiO2 to 2Al2O3:1SiO2, referred to as 3:2 and 2:1 mul-
tected adjacent to these corundum grains. lite, respectively.10 The nonstoichiometry is accommodated by
an oxygen vacancy defect structure. Ordering of these vacan-
cies leads to domains and twins which give additional features
I. Introduction in diffraction patterns which have been interpreted mistakenly
as tetragonality in high-alumina mullites. An important aspect
T HE microstructures and properties of porcelain bodies fab-
ricated from triaxial mixtures of clay, feldspar, and quartz
have been studied for the greater part of this century.1–3 The
of the present work was to quantify the chemical compositions
of primary and secondary mullite observed in porcelain bodies.
purpose of the current work is to use the improved resolution Austin et al.11 showed that incorporation of up to 20 wt%
and chemical analysis capability available in modern electron calcined alumina into whiteware bodies for a corresponding
microscopes to examine porcelain microstructures. Specifi- amount of quartz improves the mechanical properties without
cally, the type and amount of mullite, the affect of alumina appreciably affecting the production control of shrinkages, fir-
additions, and the role of local glass composition are examined. ing schedules, and maturing temperatures. When flint (which is
Porcelain microstructures are grain and bond type with a predominantly quartz) is replaced by alumina, the difference
large grain or filler (usually quartz) held together by a finer between the thermal expansion coefficients (␣) of the crystal-
bond or matrix.4 The gross microstructure of fired triaxial por- line and glassy phases is also reduced. As the difference in ␣
celain consists of quartz, feldspar, and clay relicts in a glassy decreases, so does the possibility of cracking and, therefore, a
matrix.1 Lundin5 and Schuller6 reported the formation of two significant increase in the strength of electrical porcelains can
types of mullite crystals in porcelain bodies. Mullite crystals be achieved. Khandelwal and Cook12 achieved a 200% in-
coming from the clay relicts have a scaly appearance and are crease in transverse strength by adding 40 wt% alumina to a
referred to as primary mullite, whereas the long needle-shaped vitreous china body, although additions >20 wt% lengthened
mullite crystals crystallizing from the feldspar melt are termed the firing process, increased overall body thermal expansion,
secondary mullite. Schuller6 reported no difference in the crys- and reduced translucency. Increasing alumina and decreasing
tal structure or cell dimensions of the two varieties of mullite feldspar content increases the total crystalline content of the
except that the primary mullite crystals were as small as 100 Å fired body.12 Analysis of the microstructures reveals a higher
in size (Lundin7 reported them to be <0.5 m) and the sec- mullite content and decreased crystal size. This provides hard
ondary mullite crystals were large enough (>1 m) to be ob- inclusions to arrest propagation of cracks and improve the bod-
served in the optical microscope. Mullite crystals forming in ies thermal shock resistance. Increased interlocking of the mul-
the presence of a silica-rich glass liquid phase (such as a feld- lite crystals is also thought to increase thermal shock resis-
spar melt) are acicular whereas those forming in the absence of tance.13 On the other hand, Khandelwal and Cook12 compared
a liquid phase (as in the clay relict) are uniform and equiaxed.8 the strength of two quartz-containing bodies with different
At higher firing temperatures (∼1400°C), quartz and secondary mullite contents and observed that mullite content was not the
mullite are believed to partly dissolve in the glass melt whereas only factor affecting porcelain strength since the body with
primary mullite crystals are believed not to dissolve and even lower mullite content was stronger. This indicates that micro-
to grow in size.6 The dissolution of one type but stability of the structure (the quantity, size, size distribution, and shape of
other has been attributed to small differences in the chemistry various constituent phases) has an important influence on por-
celain properties. Most recently, Vazquez et al.2 investigated
the effect of alumina additions on the elastic properties of
electrical porcelains. After firing 2 h at 1270°C, the highest
value of Young’s modulus (E ≈ 53 GPa) was attained for a
M. P. Harmer—contributing editor
body containing 10 wt% alumina. After 2 h at 1320°C, E was
directly proportional to the alumina content rising to >62 GPa.
After 2 h at 1350°C, E ≈ 66 GPa for samples containing 15
Manuscript No. 189543. Received February 18, 1999; approved June 17, 1999.
wt% alumina. Vazquez et al.2 used XRD to confirm that the
*Member, American Ceramic Society. quartz:mullite ratio decreased with increased alumina content.
3584
December 1999 Fired Porcelain Microstructures Revisited 3585
Thus the increase in E can be attributed to increased mullite Standard TEM specimen preparation techniques were used,
accompanied by reduced quartz content. Similarly, physical involving grinding, polishing, dimpling, Ar+ ion beam thin-
and mechanical properties of fired porcelain bodies can be ning, and carbon coating. A Philips 420 TEM operating at 120
improved by replacing quartz and feldspar with sillimanite kV was used for bright-field (BF) imaging and selected area
sand and alumina/cordierite glass-ceramic, respectively,14 diffraction (SAD). A JEOL 3010 TEM operating at 300 kV
since the replacement particles behave as fracture-resistant was used for high-resolution electron microscopy (HREM).
dispersoids. ␣-Quartz crystals present in most of the samples or a single-
Chaudhuri and Sarkar15 gave a second heat treatment (50 h crystal Si sample was used as a standard for the camera con-
at 1150°C) to prefired porcelain bodies (63 wt% clay–12 wt% stant (L) calibration, where is the electron wavelength and
quartz–25 wt% feldspar) in an attempt to crystallize the glassy L the camera length. At least three selected area electron dif-
phase which contained 10–30 wt% Al2O3 after initial firing fractions (SADPs) from single crystals were indexed and in-
(1 h at 1400°C). Bulk aluminosilicate glasses of similar Al2O3 terzonal angles calculated with an in-house computer program.
content are prone to crystallization of mullite on heat treat- To simulate diffraction patterns and confirm pattern indexing,
ment,16 but no significant crystallization of the glassy phase in the structure factor based computer package “DIFFRACT” was
porcelain could be achieved without addition of nucleating used.23
agents.15,17–20 Alumina substituted for quartz as in alumina- The chemical compositions of crystals and matrix glass were
rich porcelains may dissolve in the SiO2-rich matrix promoting analyzed using EDS (JEM 3010 eXL) in the TEM. Semiquan-
precipitation of tertiary mullite from the alumina-saturated titative data were obtained by comparing experimental traces to
glass.21 The present study also examined aluminous porcelain standards. EDS was performed on approximately 100 mullite
bodies for the presence of tertiary mullite. crystals and 50 glassy regions. To minimize error due to sili-
A large volume of a high-viscosity glass increases the ca-rich glass covering mullite crystals, mullite crystals on
strength of porcelain bodies, although the accompanying de- the edges of samples with little or no overlapping glass were
crease in body flexibility reduces thermal shock resistance. The analyzed.
dissolution of quartz upon firing influences the chemical com-
position of the glass phase. Dissolution rate is a strong function III. Results
of grain size and is much slower for coarse, low-surface-area,
quartz. Quartz crystals observed in standard porcelain (50 wt% XRD was used to compare Por-S and Por-H. Por-S con-
clay–25 wt% feldspar–25 wt% quartz) are usually rounded in tained ∼22 vol% ␣-quartz, ∼16 vol% mullite, and ∼62 vol%
shape, showing evidence of attack by the feldspar melt.22 glass. Por-H contained ∼17 vol% ␣-quartz, ∼12 vol% mullite,
Variations in local glass composition were also investigated in ∼25 vol% corundum, and ∼46 vol% glass.
the present study. Reflected-light microscopy of Por-H revealed quartz grains
“Q”, dark pores “O”, regions of clay “P”, and feldspar “S”
II. Experimental Procedure relicts (Fig. 1(a)). Note that larger (>30 m) quartz grains are
cracked but smaller ones are not. The pores and quartz grains
Standard and aluminous porcelain samples were provided by can be more clearly seen in the SEI of Por-S (Fig. 1(b)) and the
Allied Insulators (Leek New Road, Stoke-on-Trent, England). dark quartz grains “Q” (surrounded by darker silica-rich solu-
The compositions of standard and aluminous porcelain in wt% tion rims “R”), pores “O”, regions of clay “P”, and feldspar “S”
are given in Table I. relicts in a glass matrix are shown in Figs. 2(a,b). Note the
For X-ray diffraction (XRD), powder (sieved <325 mesh) of variation in mullite crystal size at the clay–feldspar relict in-
standard (Por-S) and aluminous (Por-H) porcelain were terfaces labeled “I” and columnar crystal growth from the clay
scanned from 2 ⳱ 10°→60°, at a scanning speed of 0.5°/min, relict “P” into the feldspar relict “S”. It was not possible to
using a Philips diffractometer (with CuK␣ radiation, ⳱ measure the exact size of individual crystals due to aggregation
0.1541838 nm) at 50 kV and 30 mA. The diffractometer was of primary mullite and interlocking of secondary mullite from
calibrated with an ␣-quartz standard before use. JCPDS Cards SEM micrographs. From TEM individual primary mullite crys-
5-490 (␣-quartz), 15-776 (mullite), and 10-173 (corundum) tals in clay relicts were often elongated approximately ⱕ200
were used to identify crystal phases. The volume percents of nm × 40 nm × 40 nm or cuboidal although they occurred in
␣-quartz, corundum, and mullite were determined by compar- large (several micrometers diameter) aggregated clumps
ing the areas under the XRD peaks of these phases in porcelain whereas the acicular secondary mullite crystals in feldspar
samples to the same peaks in standards (i.e., ␣-quartz, ␣-alu- relicts had similar width and thickness but were up to several
mina, and molochite) using Sietronics XRD Trace Processing micrometers long.
Software version 2.0 (1993). The volume percent of glass pres- ␣-Quartz grains could be distinguished from corundum be-
ent was estimated by subtracting the sum of the volume per- cause the former were always surrounded by amorphous silica-
cents of crystalline phases from 100. rich solution rims, whereas the corundum grains labeled “A”
Optical and scanning electron microscopy specimens were were not (Figs. 3(a,b)). Corundum grains were located adjacent
polished using 6, 3, and 1/4 m diamond pastes after grinding to the clay or feldspar relicts (Fig. 3) whereas quartz grains
with silicon carbide powders and water. The polished surfaces were completely isolated from the remaining body constituents
were etched for 3 min in 5% HF solution and then gold/carbon by the solution rim. Figure 4 is from a region containing small
coated. A Polyvar optical microscope and JEOL 6400 SEM (ⱕ0.2 m long) mullite crystals observed in the clay relict of
(operating at 20 kV) were used. Secondary electron images Por-S. Figures 5(a–c) show a cuboidal mullite crystal (∼0.2
(SEI) were used predominantly and SEM energy dispersive m) and corresponding [001] SADP observed in the clay relict
spectroscopy (EDS) using a Link eXL detector was used to of Por-S. This crystal contained 83.3 wt% Al2O3 and 16.7 wt%
distinguish corundum from quartz. SiO2 (i.e., 74.6:25.4 mol%), which is close to 2:1 mullite. The
light region ∼50 nm within the crystal may be amorphous.
Figure 6 shows a region of acicular secondary mullite crystals
Table I. Compositions of Porcelain-S and Porcelain-H in a feldspar relict of Por-S. Figure 7 is a BF-TEM image of an
Standard porcelain Aluminous porcelain elongated (∼1.2 m long) mullite crystal and a corresponding
Body recipes (Por-S) (wt%) (Por-H) (wt%)
[1 1 1] SADP from a similar feldspar relict in Por-S. This crys-
Dorset ball clay 30 30 tal contained 70.4 wt% Al2O3 and 29.6 wt% SiO2 (i.e., 58.3:
China clay 20 10 41.7 mol%), which is close to 3:2 mullite. Figure 8 is a BF-
Potash feldspar 25 20 TEM image of a mullite crystal (∼0.7 m long) and a
Silica sand 25 20 corresponding [1 1 2] SADP observed in the aluminous porce-
Calcined alumina 20
lain. This crystal had the elongated morphology of a secondary
3586 Journal of the American Ceramic Society—Iqbal and Lee Vol. 82, No. 12
feldspar relict by their size and composition. Mullite crystals Heat Treatment: I, Mineralogical Composition,” J. Eur. Ceram. Soc., 15, 1031–
forming at the edges of corundum crystals are tertiary mullite. 35 (1995).
16
J. F. McDowell and G. H. Beal, “Immiscibility and Crystallization in
In general, the glass phase close to ␣-quartz crystals and in Al2O3–SiO2 Glasses,” J. Am. Ceram. Soc. Bull., 52 [1] 17–25 (1969).
crystal-free regions is silica-rich. K2O and Al2O3 contents vary, 17
S. P. Chaudhuri, “X-ray Study of Induced Mullitization of Clay,” Trans.
indicating the local heterogeneity of glass composition in por- Indian Ceram. Soc., 27 [1] 20–30 (1969).
18
celain bodies. In clay and feldspar relicts, mullite crystals con- S. P. Chaudhuri, “Influence of Mineralizers on the Constitution of Hard
Porcelain: II, Microstructures,” Am. Ceram. Soc. Bull., 53 [3] 251–54 (1974).
tribute to EDS peaks from glass, and therefore, it is not possible 19
S. P. Chaudhuri, “Influence of Mineralizers on the Constitution of Hard
to determine the exact glass composition, particularly in re- Porcelain 1. Mineralogical Compositions,” Am. Ceram. Soc. Bull., 53 [2] 169–
gions containing primary/tertiary mullite. 71 (1974).
20
L. Hermansson and R. Carlsson, “On the Crystallization of the Glassy
Phase in Whitewares,” Trans. J. Br. Ceram. Soc., 77, 32–39 (1978).
21
W. M. Carty and U. Senapati, “Porcelain—Raw Materials, Processing,
References Phase Evolution, and Mechanical Behaviour,” J. Am. Ceram. Soc., 81 [1] 3–20
1
A. A. Klein, “Constitution and Microstructure of Porcelain,” Natl. Bur. (1998).
22
Stand. Tech. Pap., 3–38 (1916–17). P. Rado, “The Strange Case of Hard Porcelain,” Trans. J. Br. Ceram. Soc.,
2
S. B. Vazquez, J. C. M. Velazquez, and J. R. Gasga, “Alumina Additions 70 131–39 (1971).
23
Affect Elastic Properties of Electrical Porcelains,” Am. Ceram. Soc. Bull., 77 [4] J. P. Morniroli, D. Vankieken, and L. Winter, Electron Diffraction Com-
81–85 (1998). puter Program “DIFFRACT,” vers. 3.4 (1995).
3 24
Y. Kobayashi and E. Kato, “Lightening of Alumina-Strengthened Porcelain Y. Ohya and Y. Takahashi, “Acoustic Emission from a Porcelain Body
by Controlling Porosity,” J. Jpn. Ceram. Soc., 106 [9] 938–41 (1998). during Cooling,” J. Am. Ceram. Soc., 82 [2] 445–48 (1999).
4 25
W. E. Lee and W. M. Rainforth, Ceramic Microstructures. Chapman and S. I. Warshaw and R. J. Seider, “Triaxial Porcelains—Strength and Micro-
Hall, London, U.K., 1994. structural Relations”; in Ceramic Microstructures, Their Analysis, Significance
5
S. T. Lundin, “Electron Microscopy of Whiteware Bodies,” Trans. Int. Ce- and Production. Edited by R. M. Fulrath and J. A. Pask. Wiley, New York,
ram. Congr., 4, 383–90 (1954). 1966.
6 26
K. H. Schuller, “Reactions Between Mullite and Glassy Phase in Porce- D. Mitchell, “Which Way,” 40th Mellor Memorial Lecture, Br. Ceram.
lains,” Trans. Br. Ceram. Soc., 63 [2] 103–17 (1964). Trans., 96 [2] 66–73 (1997).
7 27
S. T. Lundin, “Microstructure of Porcelain,” Natl. Bur. Stand. (U.S.), Misc. J. E. Comeforo, R. B. Fischer, and W. F. Brandley, “Mullitization of Ka-
Publ., 257, 93–106 (1964). olinite,” J. Am. Ceram. Soc., 31 [9] 254–59 (1948).
8 28
B. L. Metcalfe and J. H. Sant, “The Synthesis, Microstructure and Physical G. W. Brindley and M. Nakahira, “The Kaolinite–Mullite Reaction Series:
Properties of High Purity Mullite,” Trans. J. Br. Ceram. Soc., 74, 193–201 II, Metakaolin,” J. Am. Ceram. Soc., 42 [7] 314–18 (1959).
29
(1975). K. A. Chakravorty and K. D. Ghosh, “Kaolinite–Mullite Reaction Series:
9
I. A. Aksay, D. M. Dabbs, and M. Sarikaya, “Mullite for Structural, Elec- The Development and Significance of a Binary Aluminosilicate Phase,” J. Am.
tronic, and Optical Applications,” J. Am. Ceram. Soc., 74 [10] 2343–57 (1991). Ceram. Soc., 74 [6] 401–406 (1991).
10 30
T. Epicier, “Benefits of High Resolution Electron Microscopy for the Struc- C. J. McConville, W. E. Lee, and J. H. Sharp, “Microstructural Evolution in
tural Characterisation of Mullite,” J. Am. Ceram. Soc., 74 [10] 2359–66 (1991). Fired Kaolinite,” Br. Ceram. Trans., 97 [4] 162–68 (1998).
11 31
C. R. Austin, H. Z. Schofield, and N. L. Haldy, “Alumina in Whiteware,” S. H. Risbud, “Revisiting Metastable Immiscibility and Structure in Silica-
J. Am. Ceram. Soc., 29 [12] 341–54 (1946). Alumina Glasses”; pp. 61–71 in Ceramic Transactions, Vol. 6, Mullite and
12
S. K. Khandelwar and R. L. Cook, “Effect of Alumina Additions on Crys- Mullite Matrix Composites. Edited by S. Somiya, R. F. Davis, and J. A. Pask.
talline Constituents and Fired Properties of Electrical Porcelains,” Am. Ceram. American Ceramic Society, Westerville, OH, 1990.
32
Soc. Bull., 49 [5] 522–26 (1970). G. Urbain, F. Cambier, M. Deletter, and M. R. Anseau, “Viscosity of Sili-
13
S. P. Chaudhuri, “Ceramic Properties of Hard Porcelain in Relation to cate Melts,” Trans. J. Br. Ceram. Soc., 80 139–41 (1981).
33
Mineralogical Composition and Microstructure: VI, Thermal Shock Resistance Y. Klykova, A. Mazurkevich, and V. Panteleyev, “Image Analysis and
and Thermal Expansion,” Trans. Indian Ceram. Soc., 34 [1] 30–34 (1975). SEM Studies of Ancient Ceramics and Feldspar Porcelain Structure,” Microsc.
14
S. Maity and B. K. Sarkar, “Development of High-Strength Whiteware Anal., 31–34 [Sept.] (1998).
34
Bodies,” J. Eur. Ceram. Soc., 16, 1083–88 (1996). C. R. Perry, “Optimizing an Alumina Porcelain Body”; B.S. Thesis. New
15
S. P. Chaudhuri and P. Sarkar, “Constitution of Porcelain Before and After York State College of Ceramics, Alfred University, Alfred, NY, 1997. 䊐