Nature of Adherence of Porcelain Enamels to Metals

by B. W. KINO, H. P. TRIPP, and W. H. DUCKWORTH

B a M e M d a l Institute, Columbus, Ohio

An investigation of the fundamentals of the ad- research program at BatteUe Memorial Institute aimed at
herence of porcelain enamels to metals indicated acquiring greater knowledge of the fundamentals of enamel
that good adherence is the result of metal-to- adherence. This paper describes that research and presents
metal bonds between the atoms in the base metal the findings.
and ‘the proper metallic ions in the enamel. To For many years, the accepted method of evaluating ad-
accomplish this type of bond, the enamel must herence has been to strike the enameled object a blow hard
meet certain chemical and thermodynamic re- enough to deform the metal. The enamel will shatter. If
quirements: (1) The enamel at the inteuface it flies off, exposing bare metal, the adherence is said to be
must be saturated with an oxide of the metal p r . If the metal remains completely covered with flakes
and (2) this oxide must be one which, when in of enamel, the adherence is said to be excellent. The broad
solution in the glass, will not be reduced by the objective of this research was to learn the conditions essential
metal. In the case of iron, the oxide is FeO. for obtaining so-called “excellent” adherence, and to de-
Many of the phenomena observed in commercial termine the nature of the bond when excellent adherence is
enameling were investigated and found to be achieved.
related to adherence, but not essential for its
development. An example is the precipitation pj Tli.ori.rtromkLthmhrn
of metallic particles in the enamel. Much of the An analysis of the many articles concerning the mechanism
complexity in commercial enameling arises of the adherence of porcelain enamels reveals two basic
from the limitations imposed by practical con- theories. One attributes adherence to chemical bonds, and
siderations. For example, because enamels the other attributes it to mechanical keying of the enamel
usually are &ed in air, the heavy scale devel- to a roughened base metal.
oped during the early stages of firing must be Most versions of the chemical theory rely on the presence
removed before adherence can be developed. of an oxide film as a transition zone between the steel and
Likewise, as the conditions of the enamel-metal the glass, with the oxide firmly attached to each.‘ A main
interface change rapidly during flring, “adher- objection to this theory is that the adherence between steel
ence-promoting oxides” are used to help main- and FeO,the oxide form to be expected a t the interface, is
tain the necessary conditions for the time re- not particularly good. There is considerable evidenceto show
quired in commercial enameling. Surface rough- that the heavy oxide scale present when a normally adherent
ness, although not nec- for excellent ad- enamel is underfired is detrimental to adherence.* It is hard
herence, was found to improve the apparent to see why a decrease in scale thickness alone would promote
adherence when the bond between the enamel adherence.
and the metal was relatively weak. All the mechanical theories of adherence require the ex-
istence of a rough glass-metal interface. The theories vary
mainly as to the manner in which the roughness is produced.

1. Introduction
GREAT deal has been said and written on the subject of

A the adherence of porcelain enamels (glass) to steel.

A review of the literature pertinent to enamel adherence
by O’Bannon* in 1950 contained more than 300 articles.
This interest is evidence of a need for industry to know
considerable about the subject. As further evidence, a group
of companies in the porcelain enamel fieldst sponsored a I ( u ) Karl Kautz. ‘‘Discussion on Lord’s Paper.” J . Am.
Ccrum. SOL, 20 [4]11!j-20(1937).
(b) A. Dietzel and K. Meures, “Reactions Important for
Adherence When Firing Ground Coats Contaimn No Adher-
ence-Promoting Oxides,” ibid., 18 [Z] 35-37 (19357 (translated
by R. M. King).
(c) A. W. Hull and E. E. Burger, “Glass-to-Metal Seals,”
Ph sics. 5 [12]384-405(1934); Ccram. Abstr.. 14 [6]136 (1936).
&) A. J. Monack, “Theory and F’ractice of Glass-MetalSeals.”
Gkrss I d . , 27 [8]389-94, 408-14, 420; [el 446-49. 464-70,
476; [lo] 502-604. 52228; [ll] 566-69, 676-78, 582 (1946);
Ccram. Abstr., 1947, November, p. 2233.
Presented in abbreviated form at the Fifty-Eighth Annual (c) L. Vielhaber. “Formation of Oxides Affects Adherence of
Meeting. The American Ceramic Society. New York, N. Y.. Enamels.” EmaiZwarm-Ind., 15 [49] -21 (1938); Ccrum.
April 25,1956(Enamel Division. No. 23). Received July 3,1958; Abstr.. 18 [S]149(1939).
revised copy received July 23,1959. cf) E. E. Howe. “Adherence Phenomena of Sheet Iron Ground
At the time this work was done, the writers were, respectively, Coats,” Bet& Enamding, 9 [O] 13-19 (1938);Ccrum. Abstr., 19
consultant, project leader, and chief, Ceramic Division, Battelle [8]183(1940).
Memorial Institute. Mr. Tripp is now manager. Ceramic Bond- (g) L. A. Johnson and E. E. Howe. “Factors Governing Ad-
ing Department, Gulton Industries, Incorporated, Metuchen. herence of Enamels Applied to Sheet Iron,” J . Am. Ccrum. SOG..
New Jersey. - _rioi
29 m-mi mm.
L. S. O’Bannon, Battelle Memorial Institute; private com- L(aj --%l-&u~,‘~“Further Data on Enamel Adherence,”
munication. J . Am. Cerum. SOC.,19 [4]93-108(1936).
t See Acknowledgment for a list of the sponsoring organizs- &I
(b) B. W. Kim and C. W. Stull. “One-Fire White Enamels.”
32.[I]
; 344071949).
tiOnS.
so4
 November 1959 Nature of Adherewe of Porcelain Enamels to Metals 505
The “dendrite” theoryafavors the precipitation of a metallic
phase from the glass. The “electrolytic” theory4 proposes
that electrolytic attack roughens the metal base. In either
case,the result is an interlocking glass-metal structure. The
proof of these theories requires a roundabout approach be-
cause wetting is required, and wetting cannot occur without
oxygen being available a t the interface. Therefore, under
strict analysis, the mechanical theories cannot be totally
divorced from the chemical theories. A great deal of cir-
cumstantial evidence has been amassed by the supporters
of the mechanical theories, but these data have not been Iron

conclusive. flg. 1. Sdmmatk prosentotton of galvank corrodon

The National Bureau of Standards attempted a rather in cobalt oxide-bearing grand coat as Wlearizod by
direct approach to establishingwhether adherence is chemical Di*el.
or mechanical. They investigated the character of interfacial
roughness, in Werent specimens, and correlated adherence
with both the density of anchor points (keys or mechanical
bonds) and the specific surface area observed. They reasoned
that the efiective area of bonding would be determined by
anchor-point density if adherence were mechanical, and by
specific-surface! area if adherence were chemical. Hence, Fe& + Co between the iron oxide from the scale and cobalt
by knowing which correlated better with adherence, specific- oxide. He claims this reaction has no relation to adherence.8
surface area or anchor-point density, a conclusion could be An investigationof Dietzel‘s theory by the National Bureau
drawn as to the basic nature of the bond. of Standardso substantiated the possibility of such reactions.
Although the Bureau’s initial conclusions favored mechani- A previous Bureau paper’ bad shown that cobalt was
cal adherence,’ close inspection of their data raises doubt. deposited directly on iron, thereby establishing the contact
On further investigation, the Bureau reported the followin@: necessary for the formation of a short-circuited cell. The
“As mentioned earlier, a good correlation has been shown to Bureau’s conclusionsare less positive than those of Dietzd.“
exist between the roughness of the enamel-ironinterface and They suggest that galvanic corrosion can occur under normal
the adherenceindex of specimensfmd in an air atmosphere. . . firing conditions, but since metals, which exhibit a greater
Other evidence indicates, however, that roughness of interface potential Merence in contact with iron than does cobalt,
does not completely 8ccouIlt for adherence. Specificslly, do not caw a more extensive attack, the theory has weak-
p r adherence has been observed on very rough interfaces nesses.
with high anchor point counts, and, for coatings on alloys,
fairly good adherence has been obtained without surface (2) Tln N d o n o f ~
roughening.” A consideration of the possible general nature of adherence
Dendrites of irona(0 and of cobalt’(a)l7 have been observed shows that, in the ultimate analysis, there are only three
in adherent enamels. No one, however, has shown that the types of bonds possible between solid materials, although more
dendrites cause adherence, and their presence could be in-
cidental. This fact, of course., makes the dendrite theory
of mechanical adherence presently inadequate.
The other major theory interpreted as favoring mechanical
adherence is the electrolytic or galvaniccorrosion theory.
Actually, this theory does not eliminate the possibility of * (0) J. H Healy and A. I. Andrews, “Cobalt-Reduction
chemical adherence. Even its chief exponent, Dietzel, recog- Theory for Adherence of Sheet-Iron Ground Coats,” J . Am.
nizes the necessity for an oxidation-reduction reaction (it Cerum. SOC.,34 p] 207-14 (1951).
may be called “electrolytic”) to provide the surface roughness ( b ) R. M. King, “Mechanics of Enamel Adherence: VIII,
( A ) Apparatus for Firing Enamels Under Accurate Control of
to which adherence is a s ~ r i b e d . ~ ~ ~ Temperature, Pressure, and Atmosphere; ( B ) Studies in Firing
In the hypothesis, originally presented by D i e t ~ e l , ~cobalt
(~) Enamels Under Reduced Pressures,” ibid., 16 [5]232-38 (1933).
’ (or nickel) precipitated from the glass in contact with the y u ) A. Dietzel, “Explanation of Adhering Problem in Sh’eet
iron base forms a short-circuited local cell in which iron is the Iron Enamels,” Emuilwuren-Id., 11 (191 161-66 (1934);Cerom.
Absfr., 13 [lo] 250 (1934).
anode. The current (positive electriaty) flows from the iron (b) € F.IStaley,
. “ElectrolyticReactions in Vitreous Enamels
through the melt to the cobalt and back to the iron. These and Their Relation to Adherence of Enamels to Steel,” J. Am.
local cells are not exhaustedduring firing because there is an Gym. Soc., 17 [S] 163-67 (1934).
abundance of anodic iron and M u s i n g atmospheric oxygen J. C. Richmond, D. G. Moore, H.B. Kirkpatrick, and W. N.
Harrison, “Relation Between Roughness of Interface and Adher-
has a depolarizing action on the cathode side. The result is ence of Porcelain Enamel to Steel,” J. Am. Cerum. Soc., 36 [12]
that the iron goes continuously into solution, the surface 410-16 (1953);Nufl.Advisory Comm. Aeronauf. Tech. Nok, No.
becomes roughened, and the glass anchors itself into the holes. 2934,29pp. (April 1953).
, 6 A. G. Eubanks and D. G. Moore, “Effect of Oxygen Content
This theory, therefore, is based on anchor points. of Furnace Atmosphere on Adherence of Vitreous Coatings to
Schematically, the galvanic action would appear as shown Iron,” J. Am. Cerum. Soc.,38 [7]226-30 (1955);Nufl.Advisory
in Fig. 1. Comm. Aeronaut. Tcch. Note, No. 3297, 17 pp. (May 1955).
The reactions that would be required to support such a 7 W. N. Harrison, J. C. Richmond, J. W. Pitts, and S. G.

system are as follows for a cobalt-containingenamel: Benner, “Radioisotope Study of Cobalt in Porcelah Enamel.”
J. Am. Cerum. Soc., 35 [5] 113-20 (1952);published 89 Nutl.
++ ++ Advisory Comm. Aeronaut. Tech. Note, No. 2895, 28 pp. (June
1 1) FeO Coo -c FeO Coo (to establish cell)
2) 2 w or -c 2Co’+
+
3) Co’+ 2.3-+ C d
(4) FeO - 2e- + Fe*+
20’- 1952) under the title “Migration of Cobalt During Firing of
Ground-Coat Enamels on Iron.”
a Adolf Dietzel, “Adherence in Sheet-Iron Enamelware.”
Sprechsuul, 68 [I] 3-6; [2] W23; (31 34-36; (41 53-66; [a]
For these reactionsto occur,local deviations from electrid 67-69;[6]84-86 (1935);Ceram. Absfr., 14 [5]107 (1936).
neutrality in the enamel must be assumed. This assumption 0 D.G. Moore. J. W. Pitts, J. C. Richmond, and W.N. Harri-

probably is in agreement with the presentconcepts of glass son, “Galvanic Corrkon Theory for Adherence of Porcelain
Enamel Ground Coats to Steel,” J. Am. Cerum. Soc., 37 111 1-6
structure. In addition to reactions (1) and (2), Dietzel (1954);Nutl. Advisory Comm. A e r m u t . Tech. Note. No. 2936,lQ
recognizes the probability of the reaction 2FeO + Coo-, pp. (June 1953).
506 Journal of The American Ceramic Society-King, Tripp, and Duckworth Vol. 42, No. 11
Table 1. Oxide and Batch Compositions of Basis Glass
Glass
Amount
Oxide Amount (wt. %) Batch material (parts by weight)

NarO 19.8 Sodium silicate’ 49.4

Metal CaO 5.9 CaO 5.9
-41101 5.5 A1101 5.5
Glass BIG 18.8 Dehydrated borax 27.2
so¶ 50.0 SioI 13.2
* SS65 PWD, Philadelphia Quartz Company, Philadelphia,
Pennsylvania, contained 2.5 weight % total of CoI and HlO.

Metal
Fig. 2. Metal-glass hterfacer that might promote me- during smelting. The batch components used were generally
chonical bandhg. (a) Ideal dovetail iaint and Ib) gen-
eralized metol-glass dovetail. oxides. Decomposable compounds, such as hydrates, car-
bonates, and nitrates, were avoided. Smelting was done in a
gas-fired furnace, in an electric furnace using an air atmds-
phere, or, when necessary to maintain the metal-oxide ratio
of the raw batch, in an electric furnace in an argon atmosphere.
than one may be important in a given situation. The three For most experiments, the glasses were degassed by remelting
types are as follows: (1) van der Waals’forces, (2) chemical at 1900OF. under a reduced pressure of about Sop of mercury
bonds, and (3) mechanical bonds. until gas evolution ceased. This usually required to
It can be stated that, of the first two bond types, van der hour. To avoid introducing water during fritting, the molten
Waals’ forces are much the weaker. They are the forces enamels were quenched on a cooled aluminum sheet.
which come into play when two or more atoms are brought The enamels were pulverized in a steel mortar to pass
close together without a chemical interaction taking place. through a No. 200 sieve, and then the iron was removed with
Reported bond energies range from 0.5 to 4 kcal. per mole. a magnet. The enamels were applied to the metal panels by
I n contrast, bonds of the chemical type are about 100 times dusting. The panels then were fired in an electrically heated
stronger than van der Wads’ forces, with some variation porcelain-tubefurnace.
depending on the nature of the chemical bond. Chemical In the instances where an argon atmosphere was used, the
bonds are &died into three types: electrostatic,*covalent,t argon was purified by passing it over titanium chips at 1500O
and metallic. Theoretical strengths of elements or com- to 1800’F.
pounds having chemical bonds have been calculated to be The oxide and batch compositions of the enamel used as
more than 1,OOO,OOO lb. per sq. in. Thus, with any type of the basis in nearly all the experimental enamels are given in
chemical bonding, the force that might exist at an enamel- Table I.
metal interface potentially is very strong. Adherence measurements were made using a falling-weight
If, in a purely mechanical bond, the metal is visualized as type of test. The equipment used consisted of a vertically
keying the glass to its surface in a sokt of dovetail arrange- mounted tube and a cylindrical hammer weighing li/, lb.
ment, as shown in Fig. 2, the narrowest aos section of the with a hemispherical shape a t one end. The hemisphere was
keys will be the controlling area. The strength of iron is approximately in. in diameter. The equipment was de-
fairly predictable, but the strength of a glass can vary con- signed so that enameled specimens approximately 1 by 1 in.
siderably. If the glass at the interface behaves like annealed could be used. The blow delivered by the falling weight was
glass in bulk with a tensile strength of 8OOO lb. per sq. in., such that the depression produced had a smaller radius than
the maximum strength of a mechanical bond will be about that obtained in the standard P.E.I. adherence test.* As a
5OOO lb. per sq. in. However, the glass a t the interface may result, comparison of the two tests showed that good ad-
be stronger, approaching that of the chemical bonds in the herence as judged by the P.E.I. test was only moderate as
glass, or weaker, depending on whether the conditions there judged by the falling-weight test.
increase or decrease local flaws and resultant stress mcentra-
tions. The nature of the bond developed in porcelain enamels 111. Essential Conditions for Adherence
on steel would be better understood if the relative importance Obviously, one of the first conditions for adherence is that
of chemical and mechanical bonding could be established. the enamel wet the metal base.
This was a major aim of the experimental work to be de-
scribed. m w*--
To explain how an enamel wets a metal, it is well to review
II. Materials and Procedures first what takes place during the early stages of firing. When,
Unless otherwise specified, Armco enameling iron was used for example, a piece of iron coated with dried, but unfired,
in this research for the base stock, and the iron was prepared enamel is placed in a furnace in an air atmosphere, the metal
for enameling by electropolishing$ to remove the Beilby first oxidizes. Phelps, Gulbransen, and Hidmanlo have
layer. This layer consists of the surface metal disturbed by shown that when iron is heated in air, a submicroscopiclayer
fabrication, and has properties slightly different from those of oxide is formed below 600OF. As the dry bisque enamel
of the bulk material. Standard specimens were 1.25- by d o r d s little protection from the air atmosphere, an oxide
1.25-by 0.035-in. panels. layer forms readily on the metal. The thickness of this layer
In preparing experimental enamels, care was taken to increases rapidly as the temperature rises. Under normal
eliminate gases which are normally present in frit and, when
necessary, to prevent reduction of readily reducible oxides

* “Test for Adherence. of Porcelain Enamel to Sheet Metal,”

* Also known as ionic type. Bulletin T-17. Porcelain Enamel Institute, July 1953.
t Also known as valence or homopolar type. lo R. T. Phelps. E. A. Gulbra-n, and J. W. Hickman, “Elec-
1 The general technique was that described in Chapter 3D tron Di&sction and Electron Microscope Study of Oxide F i l y
of the Electroplating Engineering Handbook. edited by A. Formed on Metals and Alloys at Moderate Temperatures,
Kenneth Graham. Reinhold Publishing Corp., New York, 1955. I d .Eng. Chetn., A m l . Ed.,18 IS] 391400 (1946).
November 1959 Nature of Adherence of Porcelain Enamels to Metals 507
Table II. Free Energies ‘of Formation of Metal Oxides Table 111. Phases Precipitated from Glass-Metal Mixtures
at 1292OF.

-
Precipitated phaae
Oxide Energy (kcat./mole Or) Oxide addition to +
basis glass Metal additive Expected Observed
2cao -25f3 Chromium None None
~

?++aQ- zr(k -219 ,

++
4/3A1 Q + 2/3A1& -217 Iron ,I

I
2Ti Q . T F d O -203 &PW. ,‘

+ -
-182 Chromium I‘

2 $2- SiG
4/3B Q 2/3&OI
-166
-162
Iron
Copper.
,,
,,
,

+
++ -
2Mn Q 2Mn0 -148 Chromum Cobalt
,‘ Cobalt

+ -
4Na a -c 2NaO -139 Iron “

+ oa -
4/3Cr Q 2/3Cra -138 &Ppe’. None None
2Zn 2ZnO -119 Chromium Aynic Aynic
4/5v 2/!5v205 -110 Iron
+
4/5P Q + 2/5P2os
2wF’+QQl-2E0 -102
-94
COPper. None
Chromum Lead
None*
Lead
4

-94 Iron ‘1

++ +
Mo Q-MoQ -94 Copper. None None

++-
Sn OI -c SnO, -94 Chromum &!per
4/3Fe Q + 2/3Fe2a -90 Iron
Cuprous oxide Cuprous oxide
++ --
2C0 0, 2coo , -80 Copper
2/3Mo O, + 2/3MoG -79 See text.
-76
+ -
2Ni Q 2NiO
4/3As a 2/3As& -73
+ -
4/3 Sb Q 2/3 SbsOa -72
+
+ -
2Pb Q -c 2PbO -63
2Bi

+ -
4/3Bi
+ Q
-2Bi0
2/3Bi&
4cu ot 2CutO
2cu 01 2cuo
-60
-50
-48
-35
the attainment of equilibrium. Under such conditions, ac-
curate determination of the amount of iron involved in the
reaction or its state of oxidation during periods of reaction
is dif6cult, if not impossible. For these reasons, metal powder
was mixed with pulverized glass and the mixture was fired
in an argon atmosphere.
firing conditions, the scale attains its maximum thickness The second variable was the solution of gases in the glass.
of to 2 mils in about 1 minute of firing,i* corresponding Gases could react in an unknown way and confuse the results.
to a gain in weight of about to 21/3 gm.per sq. ft. The To minimize these gases, the glasses were formulated, smelted
thickness attained depends on the composition of the metal, in air, and degassedas describedin Section 11.
the firing schedule, the compositionof the atmosphere, and the Six glasses were prepared. The first was the basis glass
fusibility of the enamel. Once the enamel fuses and seals given in Table I with no additions, and the second contained
the metal surface completely from the air, the oxidation of the an addition of ZroZ. These glasses contained only oxides
metal is stopped temporarily. above ferrous oxide in free energy of formation. The four
During the first stage of firing, several things happen at other glasses contained additions of Coo, Pbo, As& or
once, but the most important is the apparent wetting of the CuO, each of which is below FeO in free energy of forma-
metal by the enamel. Actually, as the glass sinters and then tion. Each oxide addition introduced 0.04 mole of 0 1 per
fuses, it wets the oxide already formed on the metal. In 100 gm.of basis glass.
commercial enameling, most of the oxygen to form this Metallic chromium, copper, or iron was added to portions
initial oxide layer comes from the air, with only a small, of these glasses in sufficient quantities to combine with two-
perhaps negligible, amount supplied by the enamel. To thirds of the oxygen contributed by the oxide addition.
understand why this is true, it is necessary to consider the In determining the amount of metal to add, it was assumed
driving force which causes scale formation, as well as many that Cr, Cu, and Fe would oxidize to Cr&, CuO, and FeO,
of the other reactions observed during enameling. This respectively, if a reaction took place. Stoichiometric ratios
force is the free energy of formation of oxide or the diiTerence of metal-to-oxideadditions were not used to minimize alloying
in free energies of the compounds present. The data for between any precipitate and excess metal addition.
glasses are very limited, but considerable information is These mixtures were fired for 2 hours at 1500’F. in an
available for materials in their standard states.’* A summary argon atmosphere. This soaking period was used to allow
of the free energy of formation for a number of oxides, at near any precipitate formed to grow enough to permit easy and
enamelingtemperature, is given in Table 11. accurate identification. It was established, however, that
To obtain a better idea of the validity of applying free- the reactions between Fe and the Coo glass and between
energy-of-formation data obtained for oxides in their standard Cr, Cu, or Fe and the As& glass were completed in less than
states to reactions in glassy systems, a number of special 10 minutes.
glass-metal mixtures were fired under controlled conditions. Tbe results, given in Table 111, can be summarized as
There were two variables wbich received special considera- follows. From the free-energy-of-formation data, none of
tion. The first had to do with the glass-metal ratio. When - these metal additions would be expected to reduce any of
enamels are fired on iron, the extreme excess of iron prevents the components of the basis glass or the zirconia-containing
glass. No reductions were observed. The free-energy-of-
formation data indicate that both Cr and Fe should reduce
COO, As&, PbO,and CuO under these conditions. These
reductions occurred, although Cr appeared more sluggish
11(a) L.C. Athy and P. C. S t d t , “Enameling Iron Oxidation than Fe as a reducing agent in these glasses, as minute
Phenomena.” Proc. Porcelain Enamel Inst. Forum. Fifth Fcnum. amounts of unreacted chromium were sometimes detected.
Oct., 1940. pp. 89-99; Cerum. Abstr., 20 141 85 (1941).
(b) A. I. Andrews and H. R. Swift, “Iron Oxide in Enamel Only one possible deviation from the results predicted from
Glass as Dissolved from the Metal Base.” J . Am. Ccram. SOC.. the free-enerm-of-formation data occurred; the oxidation
25 [Ql 217-22 (May 1, 1942). of Cu by an-ksenic glass. This may be explained if it is
1*H. P. Tripp and B. W. King, “Thennodynamic Data on
Oxides at Elevated Temperatures,” J . Am. *am. Soc., 38 [12] assumed that the added to the glass batch
43237 (1955). oxidized to A s a s when being smelted in an air atmosphere.
 508 Jo?cmal of The American Ceramic Society-King, TriQQ, and Duckworth Vol. 42,No. 11
8 m r -locking
-Fused qwrtz mechanism

\
r.Nichrome-*ouna I

Rg. 3. J c h d oullh of
centact-anglo apparatus.

lhermocoupbto

enamel

lroin cement

The A & resulting from such a change would have made

availablemore oxygen than was calculated on the assumption
that Asto,alone was present.
By these experiments, it was demonstrated that the free-
energy-of-formation data for materials in their standard
states can be applied in estimating the reactions which can
be expecte.d in complex glassy systems. These data can be
coasidered to be acceptable as a guide for predicting the
relative stability of the various components of the system to
each other and, especially, the oxides to the base metal.
Having established a basis for predicting the reactions taking
place in metal-oxide systems, the study of the wetting of
metals by enamels could be undertaken with a better under-
standing of what reactions might be expected at the enamel-
metalinterface.
To measure the angle of wetting between metals and glasses,
'/'-in. chips of glass were placed on 1-in.-square metal sheets
in a small tube furnace having a window at one end. The
atmosphere in the furnace was controlled and the chips could
be moved about in the furnace 90 that they could be both
heated and cooled in the controlled atmosphere. Wetting
angles d d be observed and measured directly. The general
arrangement is shown in Fig. 3.
On the basis of the data available, it appeared to be possible Oxygen Content of Atmosphere (%I
to prepare glasses which s h d d be inert in the presence of
Fig. 4. Effect of oxygen cont.nt of ohnosphere an contad angle
metallic iron. The glass whose compositionis given in Table I between iran and glass heated at 13OO0F.
was such a glass. Degassed samples of this glass were heated
at 1300OF. in contact 6 t h iron in a purified-argon atmosphere
for periods of 4 to 5 hours. The glass samples formed beads
making angles of from 100 to 130 degrees with the base, in-
dicating a very slight degree of wetting. Both electrolytically
pure iron (sS.sS9 weight % of Fe) and Armco enameling
iron were used in this experiment. The quality of the iron Armco iron. T w o schedules were followed. In one, the
did not produce a signilicant difference in the wetting angle plaques were heated in air as before and then the furnace
observed. However, if the glasses were not degassed, wetting atmosphere suddenly was changed to purified argon and heat-
was greatly improved; the angle of wetting dropped to about ing was continued for 10 minutes. For the other, plaques
5Odegrees. were fired in a purified-argonatmosphere for only 10 minutes
When oxygen was introduced, the wetting increased rapidly at 1500OF.
as is shown in Fig. 4. It is interesting to note that the best The results of some of these experiments are shown in
wetting was obtained in an atmosphere containing about Figs. 5 to 9. Figure 5 shows only glasses containing oxides
l/& of oxygen. Wetting actually was better than when the above FeO in their energies of formation. Although these
heating was done in air (21% of oxygen). glasses readily wetted the metal base in an air atmosphere,
Thus far, consideration has been given to a relatively none continued to wet when the atmosphere was changed to
simple glass containing oxides which are thermodynamically neutral. Figure 6shows the results with glasses each of which
stable in the presence of iron. The next phase of the work was contained an added oxide below FeO in its free energy of
concerned with the wetting charactexistics of glasses contain- formation. Here, with two exceptions, the glasses continued
ing both stable and unstable oxides. to wet the metal after the atmosphere was changed. These
For this purpose, a series of glasses was prepared by adding exceptions were the PbO- and BisO,-bearing glasses. In each
to the basis glass 0.04 mole of MnO, NiO, PbO, ZnO, A&:, of these instances, the melting points of the metals which
BirO:, CrrOa, S W s , Tiol, WOI, 21% Pas, or CaFt. Glasses would be produced by the reduction of the oxides were con-
also were prepared which contained additions of 10 to 40 siderably below 1600'F. and several hundred degrees Fahren-
weight % of FeO, 0.33 to 13.30 weight % of-CoO, or 0.72 to heit lower than that of any of the other metals which could
28.7 weight % of CuO. The enamels were lired on. dean have been produced in the other glasses in the Series.
November I959 Nature of Adherence of Po*ukrdn Enawls lo Melds 509
Pired in air Fired la air Fired in r i r
5 minut- m t 1500' P. S mluutes m t 1500' P..
tdlowtd by w o n
ror 10 .JOU~+. m t imo* P.

6.3%
WOI
4 .@%
ZlQl

a.a%
Ti&

17.8%
PbO

Figure T shows a series of FeO-bearing glasses. The only

glass that remained flat after intmduang argon was the one
that contained 40% of P a . Other data showed that the
glass at the interface needed to contain above 30% of iron
ng.7. w m O # h d i r bods olou 20 and a% ad& O( oxide in order to sustain wetting by virtue of the presence
f.rrour orid.. of iron oxide done.
 510 Journal of The American Ceramic Society- Aping, Tripp, and Duckworlli Vol. 42, No. 1 I
Fired in air
Cobalt oxide 5 minutes at 1500°F.. or less, but an addition of -I.Oyoor more
addition to Fired in air followed by argon Fired in argon
basis glass 5 minutes at for 10 minutes at 10 minutes at lROOOf:. of this oxide produced a glass which tlitl
IBOOOF. 15OOOF.
not wet readily.
The appearance of the glasses them-
selves after firing was interesting.
Many of the glasses containirig readily
reducible oxides lost much o f their
characteristic color, with the formation
of a new layer at the metal interface.
This change in color was due to the
removal of these oxides from the glass
by reduction to a metallic phase and
the sulistitution of equivalent amounts
of iron oxide. The metal produced
would be expected to separate as a
layer of dendrites or as a plate that
could impede the diffusion o f the iron
oxide into the glass by physically re-
ducing the total area through which
oxide d:ffusion could take place. In
addition, the deposition of a new
metallic phase o n the metal base could
affect adversely the wetting of the
metal Ijy the glass. The series of glasses
containing colxdt oxide (Fig. 8 ) showed
this latter effect. After a sufficient de-
posit of metallic cobalt had covered the
surface of tl<e iron, the glass terided
to draw u p into beads.
I n contrast to the overlapping den-
dritic structure o f the cobalt, as could
be observed by exaniining the speci-
mens in vertical illumination at a
magnification o f SOOX, the copper
oxide-containing glasses (Fig. !I), when
fired on iron, deposited a layer of
metallic copper that appeared to start
as minute spherulites and to grow into
sheets which contained many sinall
openings through which the base metal
could be observed. This may partly
explain the differences in degree of
wetting exhibited by glasses of high
cobalt oxide and high copper oxide con-
tents.
1 I 30' ;
The glasses containing only oxides
above FeO in their free energies o f for-
mation appeared to act merely as sol-
vents for the iron oxide fomied during
Fig. 8. Wetting studies of basis glass with additions of cobalt oxide. the initial firing in air. The glasses
containing oxides below FeO in their
free energies of formation supplied
oxygen for the formation o f additional
iron oxide and also developed metallic
Figures 8 and !) indicate the results obtained with com- deposits near the glass-metal interface, thus physically im-
positions containing cobalt ox'de or copper oxide, respectively, peding the solution of iron oxide.
as additions to the basis glass In no instaice was the wetting Glasses which wetted less readily with a change from an
affected by changing to an argon atm7sphere when the glasses air to a n argon atmosphere reversed the process to a limited
contained copper oxide. The glasses containing c J b i l t oxide extent if air replaced the argon. The glass slumped and spread
remained flat as long as the cobalt oxide addition WIS l.:jJyo out again in air to cover much of the original area, and went
 N o v e m b e r 1059 iVnture of Adherence of Porcelain Enamels to M e h l s 511
through another reversal if the air Fired in air
Copperoxide Fired i n air 5 minutes at 1500°F.,
was replaced by argon once more. addition t o 5 minutes a t followed by argon Fired in argon
basis glass 15OO'F. for 10 minutes a t 10 minutes a t 1500°F.
To investigate the effect of the IROO'F.
I
t
amount of iron oxide scale on wet-
ting by the basis glass, a number of
specimens of pickled or electro-
polished iron were preoxidized in a 0.721%

hot aqueous NaOH solution for

various times to produce several
thicknesses of oxide layer. T h e
specimens next were covered with
the inert glass and fired in argon.
Attack o f the scale began as soon as
the glass sintered, and wetting fol-
4.3ya
lowed as soon as the glass began to
flow. IIowever. good wetting could
be maintained only as long as the
iron oxide layer was present.

(2) Conditions at Glass-Metal 28.77a

Interface

Once a glass has wetted a metal,

the development of adherence can
take place. In the previous section,
it was shown t h a t the wetting of iron Fig. 9. Wetting studies of basis glass with additions of copper oxlde.
by a glass could be maintained pro-
vided two general requirements were
met:
( I ) A quantity of iron oxide be maintained in solution
in the glass which was in direct contact with the iron.
( 3 ) No continuous metallic phase be precipitated on the
iron and thus provide a parting layer between the glass and
the iron.
The importance o f the iron oxide concentration in the
glass in maintaining wetting suggested t h a t the iron oxide
Fig. 10. Interface of gloss saturated with FeO and fired on polished iron
concentration also might he important in adherence. I t in an argon atmosphere ( X 250).
therefore was postulated that excellent adherence could he
developed on iron by saturating the glass-metal interface
with ferrous oxide, and research experiments were designed
to prove or disprove this hypothesis.
I t was recognized that the proper concentration of ferrous
oxide might he acquired by ( I ) smelting the ferrous oxide in
the glass, ( 2 ) admixing iron oxide with the glass, (3) pre-
oxidizing the iron, or ( 4 ) firing in a n oxidizing atmosphere.
(A) kflect of FeO Coiiten! of Glass: If adherence were due
to the presence of a large quantity of FeO in solution in the
glass, it should he possible to demonstrate this experimentally.
For example, it should be possible t o obtain good adherence
on a smooth iron surface hy saturating the glass with ferrous
oxide. A glass was prcpared containing 43 weight yo of F e u ,
the amount, determined experimentally, required t o saturate
the hasis glass. This glass was fired on electropolished iron
in an argon atmosphere at l(i00'F. for 5 minutes. The
glass-metal interface showed only a slight grain-boundary
attack, as is illustrated in Fig. 10. This slight amount o f
roughening should cause only a small amount of mechanical
adherence. Actually the adherence was excellent, as is shown
in Fig. 1 I .
To dc ternline whether this slight roughness caused appreci-
able adherence, several previous specimens, which had been
roughened to a much greater degree than t h a t observed above,
were enameled with the hasis glass under conditions just
oxidizing enough to promote wetting without excessive oxida- F ; ~ .1 1. Results of impact test on enamel illustrated in ~ i 1 0~ showing
.
tion of the base metal. Adherence was so poor t h a t none o f excellent adherence (X3).
512 Journal of The American Ceramic Society-King, Tripp, and Duckworth Vol. 42, No. 11
Refiring and alter the surface configuration as well as introduce a
time, Commercial ground Ground coat and
(minutes) coat only inert cover coat variable quantity of oxygen into the system. I t is also doubt-
ful if the typical configuration at the interface of a fired enamel
could be duplicated exactly by other means, and any method
0 of completely removing the fired enamel would probably
oxidize the base metal to some extent.
T o circumvent these objections, iron plaques on which a
commercial ground coat had been applied and fired were
used as the starting material. This treatment produced a
typically roughened glass-metal interface. Three parallel
I
series of experiments were run. In the first, the ground-
coated plaques were refired for periods up to 20 minutes.
I n the second, the ground-coated plaques were coated with
the iron-oxide-free basis glass and refired. In the third, the
ground-coated plaques were coated with the basis glass satu-
rated with FeO and then refired. All refiring was done in a n
2 argon atmosphere. As the enamels in the second and third
series were applied to the same thickness, the adherence
obtained could be compared on the basis of an impact test.
According to the hypothesis stated, adherence should be
maintained in the specimens of the first and third series but
the specimens in the second series should gradually lose
4 adherence as the FeO diffused away from the iron-enamel
interface and into the FeO-free glass. This proved to be the
case. There was a gradual increase in roughness a t the iron-
enamel interface in all three series.
Figure 12 shows the change in the surface configuration
that occurred owing to refiring o f the first and second series.
10
The third series is not shown, but was similar. The major
change observed was due to firing time. Little difference
was observed between the series of specimens refired for the
same length of time. Impact tests, illustrated in Fig. 13,
showed an appreciable drop in adherence for the specimens
20 cover-coated with the iron-oxide-free cover coat between the
first and the fourth minutes. After that time the adherence
decreased gradually. Practically no change in adherence
was detected in 20 minutes for the specimens cover-coated
Fig. 12. Interfaces of commercially ground-coated specimens of CoO-
bearing enamels reflred in an orgon atmosphere at 1 60OoF. The speci-
with the FeO glass or those not cover-coated at all.
mens cover coated with FeO-beoring enamel were sectioned and were In addition to impact tests, data on adherence were ob-
similar to the specimens shown. ( X 500.) tained from direct tensile tests. The assembly for this test
is shown in Fig. 14. The enameled specimens were prepared
separately and then bonded to the adapter blocks with an
epoxy resin. The assembly was held in a simple jig during
the setting of the plastic to maintain proper alignment.
Tensile measurements were made in a Baldwin universal
the specimens could be sectioned for microscopic examina- testing machine. There were undoubtedly a number of in-
tion. herent errors in these tests, but they all should have tended
The foregoing experiments showed that good adherence to effect low values, not high ones.
was produced with a glass saturated with FeO and that almost The limit of this test method is the strength of the resin
no adherence was produced with a glass containing only a (about GO00 Ib. per sq. in.).
small amount of FeO. If saturation is essential for adherence, Tensile tests were made on five sets of refired specimens.
it should be possible to destroy adherence by decreasing the The results are given in Table IV.
concentration of FeO at the glass-metal interface. A series These data correlate quite well with the impact data.
o f glasses was prepared having various FeO contents. These They do not, however, establish a definite tensile value for
glasses then were applied to electropolished iron and fired excellent adherence ; they merely indicate that, when the
in an argon atmosphere. Decreasing the FeO content from adherence is good, the strength of the bond is greater than
43, the saturation limit, to 40 weight % destroyed the about GOO0 Ib. per sq. in. These tests confirmed the impact-
adherence completely as determined in the adherence test. test results; namely, that diffusion of iron oxide away from
( B ) Efect of Siirjace Roziglrness: In the study of conditions the interface had a more pronounced effect on the adherence
at the glass-metal interface, previously described, only electro- than did the interfacial configuration. No close correlation
polished iron had been used and it was demonstrated that between anchor points and adherence appeared to exist.
surface roughness was not essential for adherence. However, The strength of the bond between high-FeO glasses and
there remains the possibility that excellent adherence could smooth iron also was tested and was found to be greater
be produced by surface roughness alone, providing the enamel than 59’70Ib. per sq. in.
remained in intimate contact with the metal. Nearly all the failures in the general region of GO00 lb. per
One method of investigating this would be to prepare an sq. in. were through the enamel or the plastic. Some of these
iron surface having a definite configuration and then enamel failures removed large areas of the glass above the interface.
it in a neutral atmosphere with an inert glass. The difficulty It appeared that the tensile strength of the glass was close
with this approach is the fact that satisfactory wetting cannot to this value. No clean breaks at the base metal-enamel
be accomplished without the presence of available oxygen a t interface occurred in the adherent specimens, even though
the interface unless the glass has been saturated with FeO. several tensile loads of about 6500 Ib. per sq. in. were attained.
The available oxygen, of course, could react with the iron These results indicate, as has been suggested by many in-
November 1959 Nature of Adkcnee of Porceluin Emmeis to Metals 513
Ground coat and 0-d coat and vestigators, that the strength of
Ground-conted FeO cover coat inert Cov- coat
adherence is greater than the
strength of the glass itself.
As a matter of interest, a
typical dry-process cast-iron
enamel was m o v e d from casG
iron atatmsileloadof860lb-
per sq. in.
under conditions deviatingt
from ideal (which is always the
casf2 in owrmercial enameling),
surface r o u g h , and the ac-'
comp&ng greater specific 8ut-
face, is desirable to in- ad-
herence. Greaterspecificdace
allows the development of satis-
foctary adherence with a lower
unit bond strength. However,a
high degree of surface roughness
winnot markedly alter or replace
4 minuta the CBemiCrrl condition8 required
itaeMmalsforgoodbond. Asan
txnmpk,@weu at the saturation
limit of FCO developed exdent
adhereaceonelectropoli€ihedh.
Glasses containing d y 3 weight
yo less than the limit of satura-
tion of FeO developed ILO m e a ~ -
urable adhereace on electso-
polished iron, and developed only
about 25% adherence on sand-
blasted iron. In all instances,
firing wan in argon at 1600'F. for
5 minutee and adherence was
determined by an impact test,

(3) HypolJmkOf-
It was postulated that three iequirements were essential
for the adherenceof enamelstometals:
(1) The enamel wet the metal.
(2) The enamel at the g b m e t a l inwace be saturated
with an oxide of the metal.
(3) This oxide be one which, when in solution in the
glass, will not be reducedby the metal:
The experiments described next were designed to verify
or disprove this general hypothesis or to deline more exp3,ly
the conditions existing at the glass-metal interface when
excellent adherence is developed.

0 >6010
1 >6780
4 >6#)0
7 >m 2880
10 >m aeM)
20 >6BM) as10
514 Journal of The American Ceramic Society-King, Tripp, and Duckworth Vol. 42,No. 11
Table V. Solubility Factors for FeO
Oxide Foctor

Na¶O -0.83
&O -0.46
CaO -0.40
-0.11
-0.92
-0.00
-1.21
-1.34
-0.76
+ l .70
+1.m
+2.23

Specimens of glasses by l/4 by in. were prepared for

measurements of resistivity and Hall voltage vents tempera-
ture to determine the manner of conductance and the number
of electron carriers. Glasses containing ZnO, MnO, or 14
or 28% of FeO conductedionically.
In the case of the glass saturated with FeO (43y0 FeO),
the conductance was electronic up to about 350'C. Above
this temperature, ionic conductance became predominant.
Upon cooling, the electronic conductance did not recur,in-
dicating either a rearrangement of the structure of the glass
or a change in the state of oxidation during measurement of
the resistivity.
Ag. 15. s.cHon (hrough a platod s p d m n Bright crystab are pro- These experiments show that a glass saturated with FeO
dpitatod imn. (A) Iron plate, (81 glass, and IC) b a u metal (X250.1
has unusual properties and that a bond can be developed
betweea such a glass and iron under conditions considerably
Merent (electrodepostion) from those used in commercial
enSmeting, thus further substantiating the hypothesis con-
( A ) Direct DcvcCopmcnt of a CkGmicrrt Bond: If the bond cerningconditions for adherence.
between the metal and the glass is essentially chemical, it (B) Effectof Enamel C o u s w on S o l W y of FCO in
should be possible to produce good bond between a glass of Baris Glass: The hypothesis being investigated assumes
the proper composition and a metal simply by briuging the that good adherence between enamels and iron depends on
two together under conditions such that chemical bonding near-saturation of the glass interfacewith FeO. By changing
develops. One way of doing this would be to reverse the the enamel composition, it should be possible to change the
usual enameling process and apply the metal to the glass. solubility of FeO. However, if this hypothesis is correct,
This might be done electrolytically with a suitable plating adhereace should be developed when the glass is saturated
bath at mom temperature. The electroplating of iron would irrespective of the amount of FeO required. In this program,
have to be done on a glass saturated with FeO. most of the oxides commonly used in porcelain enamels for
Spedmens were prepared by enameling one end of a mild sheet iron were investigated both as to their d e c t on FeO
steel rod, using the FeO-saturated basis glass €iredin an argon solubility and on adherence.
atmosphere. The enameled end then was immersed in an The data were obtained by adding FeO to glass batches
iron-plating bath. The plate developed on the glass was which were then smelted in an argon atmosphere. The point
characteristic of those generally obtained on iron and had of saturation was that at which either metallic iron or wtistite
e x d e n t adherence as judged by an impact test. Regardless first appeared. In most cases metallic iron appeared h t ,
of the conductivity of the glass, such adherence d d not be and wtistite followed on the addition of a few more per cent
expected unless there were sites on the surface of the glass of FeO.
that could accept an intimate metallcbond. As might be expected, the glassformers boric oxide, silica,
A photomicrograph of a section through the iron plate, and phosphorus pentoxide increased the solubility of ferrous
glass, and the base metal is shown in Fig. 15. The interface oxide, and the glass modifiers decreased its solubility. Alu-
between the glass and the iron plate is not considered to be mina appeared to act as a modifier and not as a glassformer
rough enough to contribute much mechanical adherence. inthisrespect.
The glass contains many crystals of both oxides and iron, but In each solubility experiment, with one exception, iron or
no wtistite (ferrous oxide) is apparent. crystsls of metallic wiistite apparently was the primary phase that precipitated.
iron generally are present in glasses nearly saturated with 3% high-NarO (>W% NSP) glasses had a primary phase
FeO. Iron usually is precipitated just before wiistite begins that ~ 8 prohblyNarO-FeO*SQ-
9
to precipitate. This is a normal consequence of the fact that E x d e n t adherence on iron was attained when the primary
FeO cannot exist in the crystalline form and thus precipitates phase was iron or wtistite. Poor adherence was attained with
from the saturated glass as iron and wtistite; i.e., xFe
Fe(l-,)O, where x = 0.02 to 0.11. In Certsin areas of the
+ all compositionsmore than a few per cent below the saturation
limits. The high-NatO glasses (>20%) that precipitated a
system F e O - S i ~ F e & ~iron
, (re") is a stable primary phase, primary phase other than iron or d t e did not adhere.
and it is quite likely that iron (Fe") may also be a primary Table V gives factors by which the approximate solubility
phase in glasses of the type used in this program. Despite of FeO in glasses can be calculated. These factors were
the quantity of crystals present, the crystals did not appear derived from the solubilitycurves,and can be used to calculate
to act as plating centers or nuclei in electroplating. the solubility of FeO in glasses by following these steps:
The unusually low electrical resistivity of the high-iron (1) Multiply the weight per cent of each oxide in the glass
oxide glasses suggested an electron type of conductance. by its factor.
Noveriiber 1!)50 Nature of Adherence of Porcelain Enamels to hfetals 515
(‘2) Add products from ( 1 ) algebraically. Tungsten was not easily coated. Tungsten sheet was fired
(3) If this sum is a positive value, it represents the parts in an argon atmosphere at l500’F. using the basis glass
of FeO soluble in 100 parts of glass. A negative value means containing an additional 6.5% of W03. This was not suf-
that the glass will dissolve no FeO. ficient to promote adherence. A rather heavy preoxidatim
The use of these factors applies in all instances except when or an admixture of the W 0 3 glass and WO,, however, pro-
the glass contains hlnO. The manner in which MnO is moted excellent adherence.
associated with the FeO and SiOz in the glass makes prediction In every instance, adherence was good when sufficient
of its effect difficult. Calculations on glasses containing oxide of the base metal was present in the glass. In the cases
small percentages (<5 ) of MnO will result in underestimations of copper and iron, both the higher and lower oxides wcre
of the FeO solubility which are almost equivalent to the added to determine the relative effectiveness of sufficient
amount of MnO. additions of the higher oxide. No appreciable difference was
With the exception of the high-NazO glasses, the high-iron apparent. This was to be expected, because the higher
oxide glasses studied precipitated wustite when an excess of oxide would be reduced a t the interface if sufficient time
FeO was added. To study the effects of a different ferrous elapsed at firing temperatures.
oxide phase, fayalite (2FeO.SiO.J was prepared and fired on Ti, Ta, Cr, Mo, Mn, Fe, and Ni also were coated with
iron for S minutes a t 2400’F. in an argon atmosphere. At glasses containing none of the corresponding oxides, under
this teniperature, al)out 200’F. above its melting point, the conditions that would maintain a clean metal surface. In
fayalite was very Auid and wetted the metal well, but no these instances, the glasses did not wet well and developed
adherence resulted. This lack of adherence from a coating only very poor adherence.
containing i O % of FeO probably was caused by the differences I t is not meant to imply that none of the metals inves-
in the nature of crystalline and glass coatings. Glasses lack tigated has been enameled before, or even that these par-
the directional properties that might make different crystal- ticular metals are difficult t o enamel. These data indicate,
line phases incompatible. The arrangement of the atoms in however, that if the conditions a t the glass-metal interface are
fayalite might not fit those of the base iron; many of the iron Zuch as to favor the saturation of the glass with a stable
ions occupy interior positions in the fayalite crystal and there- oxide of the metal, virtually any metal in the periodic table
fore would not be available for bonding. probably can be enameled, provided its vapor pressure is not
(C) Coating of Various Metals: Any hypothesis that pur- too high at enameling temperatures. A representative metal
ports to explain the adherence of enamels to iron should be of each group has been coated and good adherence developed,
general enough to apply to virtually all metals or a t least to in accordance with the hypothesis.
the metals in the same group in the periodic table as iron, There are, of course, a number of physical restrictions to
since iron is not set apart markedly from all other elements. the practical application of this hypothesis. The physical
To test the hypothesis, various metals were coated with an state of the oxide a t firing temperatures, the tendency of some
adherent glass, using one or more of the following methods: oxides to reduce to metals a t relatively low temperaturcs,
(1) .4n admixture of pulverized basis glass and powdered and other properties might preclude its application in some
metal oxide was applied to a sheet of the metal corresponding instances.
to the oxide and the specimen was fired in an argon atmosphere Of considerable interest, in connection with all these en-
at 1500’ F. for 8 to 5 minutes. ameled metals, is the character of the interfaces as well as
(2) The metal oxide was smelted into the basis glass, and the glasses themselves. Figure 1 G shows several typical
the pulverized glass was applied and fired as in (1). sections of the enameled metals. I n every instance, except
(3) In those cases where the melting point of the metal on rhodium (Fig. l G ( D ) ) ,the glass has a second phase, known
was lower than that of the glass, the metal oxide was smelted to be, or presumed to be, an oxide of the base metal. In the
into a glass, and this glass was fired on top of the molten case of rhodium, a characteristic discoloration of the glass
base metal which was contained in a crucible. at the interface was noted indicating the presence of rhodium
(4) A glass containing reducible oxides was fired on a ions in the glass.
highly active metal in an argon atmosphere. The surface configurations generally appear smooth com-
( 5 ) The metal was preoxidized and then coated with an pared with commercial ground-coated interfaces. Chromium
inert glass and fired in an argon atmosphere. (Fig. lG(C)) appears to have an irregular interface with many
Using the foregoing techniques, the following metals were small particles adhering to the base metal or floating very
studied: Cu, Ag, Au, Be, Mg, Ca, Zn, Al, Si, Ti, Zr, Sn, Pb, near the base metal without actually being attached to it.
Sb, Ta, Bi, Cr, Mo, W,Mn, Fe, Co, Rh, Ni, Pd, Pt, and U The unusual appearance of the surface of the palladium (Fig.
The term “coatings” is used to refer to the enamel on these lG(E)) was probably due to the repeated firings to which it
specimens t o avoid the inference that completely satisfactory was subjected to remove the hydrogen. In each cycle, the
porcelain enamels were developed in all instances. Al- palladium oxidized readily a t low temperatures during cooling,
though several good enamels apparently were obtained, the but was reduced by the hydrogen evolution on reheating.
main objective of this investigation was to obtain Palladium has a much lower solubility for hydrogen at
appreciable adherence between the glass and the base elevated temperatures than a t low temperatures.
metal, and no concern was given to general appearance or Figure l G ( F ) shows a section through a specimen of molyb-
physical properties. Adherence was tested by the falling- denum enameled with an adherent layer of glass high in
weight test where practical. Otherwise, a bend or similar FeO. In this instance, a heavy iron plate was deposited on
test was employed. or near the molybdenum during firing. The precipitation of
Only palladium was difficult to coat. This metal evolved metallic iron would be expected as a direct consequence of the
considerable gas for an appreciable time during firing. As law of mass action. This system represents one of the rare
palladium is capable of absorbing hydrogen to the approxi- instances in which an oxide (FeO) in the enamel has the same
mate ratio of one hydrogen atom to one palladium atom, free energy of formation as one of the stable oxides of the
several hundred times the volume of the palladium can be base metal (Moo2) at enameling temperatures (88 kcal. per
absorbed. IIydrogen was probably the gas evolved. Once mole of oxygen a t about 1.530’F. in this case). This would
the gas evolution ceased, adherence was developed easily. present conditions under which the A F of the reaction (2Fe0
Beryllium was enameled by coating a lump of the metal
with the basis glass in an argon atmosphere. The high free
+ Mo * MOO*-I- 2Fe) would equal zero. The reaction
would be driven to the right by the preponderance of reac-
energy of formation of Be0 resulted in an extreme amount of tants on the left until the mole ratios shown by the equation
reduction in the glass and even in the porcelain crucible were attained. A variation of as much as 100’F. in enamel-
container. Adherence was good. ing temperatures would alter this only slightly. Generally,
516 Journal of The American Ceramic Society-King, Tripp, and Duckworth Vol. 42, No. 11

Fig. 16. Typical sections of enameled metals. ( A ) Inert glass on preoxidized nickel, (6) silver enameled in an oxidizing atmasphere, (C)
chromium enameled with an admixture of CrpOI and an inert glass, ( D ) inert glass on preoxidized rhodium, (€1 palladium enameled by using
an inert glass in an oxidizing atmosphere, and ( F ) molybdenum enameled with a glass saturated with FeO. (XSOO.)

the law of mass action has not been of major concern, as most The studies of the solubility of FeO i n enamel-type glasses
free-energy relations in enamel systems result in equilibrium showed that, a t the saturation limit, the glasses probably
constants such that the reaction proceeds virtually not a t all have an unusual structure because the glassforming ions and
or almost in its entirety. the oxygen ions which can be assigned to them are in such
This particular specimen of enameled molybdenum prob- proportions that on the average only continuous chains can
ably has many of the characteristics of enameled iron, as the be formed, as illustrated in Fig. IT. The result would be
glass is more generally exposed to the iron plate than to the that iron rather than oxygen ions may be present o n the glass
molybdenum. I t represents the unusual situation in which surface.
the metal to which the glass adheres was a stable component
of the glass until its state of oxidation was changed. The IV. Phenomena Related to Adherence
presence of wiistite crystals, also precipitated from the glass, In the study of adherence, many phenomena are observed
indicates an equilibrium system between a crystalline oxide, which must be explained if a complete investigation of ad-
crystalline iron, and a glass. herence is to be presented. The removal of oxide scale, the
An extensive study was not made of the solubility of metal precipitation of metallic phases, the attack o f the bare metal,
oxides in the basis glass, but it was determined experimentally and the use of nickel to promote adherence are important
that, to obtain good adherence on Fe, Cr, Mn, Mo, Ni, and in commercial enameling.
Co, the glass had to be saturated with an oxide of the metal.
Less than saturation amounts produced only poor adherence. (I) Affmk on Scale
The firing was done in an argon atmosphere. As most commercial enameling requires the heating of iron
(D) Study of Glass Structures: In view of the importance in air, a scale is formed on the iron. As soon as the enamel
of saturation in the development of adherence, a limited glass sinters during firing, the attack on the scale begins.
study was made of the structure of glasses containing FeO. As the enamel becomes more fluid on continued firing, the
X-ray spectrometer curves showed that certain rounded attack on the scale becomes more rapid. However, excellent
peaks shifted and also increased in intensity when the FeO adherence is not developed until all the scale is removed.
content was near saturation. This result suggests that the
characteristic Fez+-Si4+ and Fe2+-02- distances change as
the glass approaches saturation with Fez+ ions. If such a
glass is fused on a metal, a preferred orientation or some
degree of polymerization may occur at the glass-metal inter-
face. This orientation may be necessary to produce adher.
ence. The structure of the surface of silicates has been dis-
cussed by Weyl,I3 who, as reported by G r u ~ e r , ~also
‘ postu-
lated a glass-metal oxide-metal bond to explain the adher-
ence of evaporated metal films on glass.

l3 W. A. Weyl, “Wetting of Solids as Influenced by Polariz-

ability of Surface Ions”; Chapter I V (pp. 147-84) of Structure
and Properties of Solid Surfaces (Robert Gomer and C. S.
Smith, editors). University of Chicago Press, Chicago, 1953.
491 pp.; Ceram. Abstr., 1954, August, p. 153j.
R. M. Gruver, “Atomistic Approach to Adhesion to Glass,” Fig. 17. Hypothetical structure of a ferrous
Glass Znd., 37 [Z]77-80,94, 100-101 (1956). rilicote glass applied to iron.
November 1959 Nature of Adherence of Porcelain Enamels to Metals 517

Fig. 18. Iron oxide pellets in contact with glass. ( A ) Wustite pellet in basis glass, (6) hematite pellet in basis glass, (C) magnetite pellet
%
in basis glass, ( D ) wustite pellet in 1 %
COOglass, (€1 hematite pellet in 1 COOglass, and ( F ) magnetite pellet in 1% COOglass. ( X 500.)

Published data regarding the nature of the scales formed Experiments were next made to determine the relative
on sheet iron heated in air have shown them to be largely rates of solutions of the oxides of iron in glass. To investigate
ferrous oxide. I‘ The solubility of ferrous oxide is considered this, rods of FeO, FesO4, and F e a l about ‘/2 in. in diameter were
to be greater than that of the higher oxides.*(o) prepared by pressing hydrostatically at 100,000Ib. per sq. in.
In general, the rate of solution would be expected to follow These rods then were sintered at 2000°F. for 2 hours in an at-
the general accepted expression’e relating the rate of solution mosphere of argon or air as required. The rods were made in
of an ideal solute in ideal solvent to the degree of saturation this manner to produce, as nearly as possible, a nonporous sur-
and having the form face and thus to prevent disintegration by intergranular attack
of the glass. The apparent porosities of the rods were FeO
dx/dl = K ( C I - Cz) (1) 1.35%, Fe304 1.25%, and Fez08 O.i2%. An X-ray analysis
d x / d f = rate of solution of a solute x at any given time 1. of the sintered rods showed only minor impurities. The
K = a constant depending on agitation, diffusion, etc. FeO rod contained wiistite and about 5% of magnetite.
CI = concentration of solute at the limit of saturation.
C. = concentration of solute at time f . The FQOc rod contained magnetite and about 5% of hematite.
The FeZO3rod contained only hematite. Although no at-
For systems in which the differences in the values of K are tempt was made to quench these rods, wiistite was obtained
insignificant, the initial rate of solution should be greatest even though it is reportedL7to be metastable below 1060°F.
for materials having high total solubilities, because this rate (570°C.) and to decompose to form Fe and Fe,04.
is controlled by the differences between C1and C., The rods were cut into pellets. A pellet of each oxide was
As the precise limits of solubility of FeO, Fez03,and Fe304 placed in a crucible covered with pulverized basis glass and
in the basis glass were lacking, the next step was to inves- the glass was fused by heating for 20 minutes a t 1600°F. in
tigate the limits of solubility of the iron oxides. The basis an argon atmosphere.
glass, containing different amounts of the particular iron Figure 18 illustrates the nature of the attack of the glass
oxide, or its equivalent in Fe +
Fes03,were smelted in argon on the iron oxide pellets. Figure 18(A) shows the pellet of
at 1900°F. The temperature of the melt then was decreased wiisite (FeO) in the basis glass. The surface was altered only
to 1600°F. in about 10 minutes and held at this temperature slightly from the original configuration.
for 1 hour. The glass then was quenched to room tempera- Figure 18(B) shows the hematite (FezO3) pellet in the basis
ture. The resultant glasses or glass-crystalline mixtures were glass. The hematite formed smooth pellets that appeared
examined microscopically a t a magnification of 500 X to to be highly resistant to solution. The lack of iron discolora-
determine the presence of iron oxide crystals. No difficulty tion in this glass also indicated that very little solution of
was encountered with completing reactions between the Fe Fez03 occurred.
and Fe203. Using this technique, the limits of solubility of Figure 18(C) shows the magnetite (Fe304) pellet in the
FeO, Fe304,and Fep03 in the basis glass were found to be 43, basis glass. The original pellet was compact and the surface
23, and 19 weight 70, respectively. configuration resembled that of the wiistite pellet in Fig.
Although FeO was by far the most soluble, it was experi- 18(A). A rapid rate of attack on the Fe304 is indicated.
mentally observed that the scale formed under a commercial The glass was able to take this oxide into solution readily,
ground coat is dissolved more rapidly than the scale formed penetrating deep into the pellet, and rapidly producing a
under a glass of the same composition but lacking the reduc- much greater surface area for subsequent attack.
ible oxides of cobalt and nickel. The presence of such oxides These results indicate that an Fe304 scale was attacked
could only increase and not decrease the state of oxidation of more rapidly than an Fez03or even an FeO scale in spite of
the scale. the higher total solubility of the latter.

ISM. H. Davies, M. T. Simnad, and C. E. Birchenall, “Dis-

cussion of ‘The Mechanism and Kinetics of Scaling of Iron,’ ”
J . Metals, 4 [5]536-40 (1952). James White, “Physical Chemistry of Open-Hearth Slags,”
18 J. Eggert, Physical Chemistry, D. Van Nostrand Co., Inc., J . Iron Steel Insf. (London), 148 [2]579-712P (1943);p. 615;
S e w York, 1933. 632 pp. Translated by S. J. Gregg. Ceram. Abstr., 23 [9]159 (1944).
 518 Journal of The American Ceramu Society-King, Tripp, and Duckworth Vol. 42, No. 11
There remains the problem of why a scale, which is largely These data help substautiate the assumption that one of
FeO, is attacked more readily by a standard blue ground the main functions of adherence-promotingoxides ia to alkr
coat than a glass of the same composition but lacking cobalt FeO and paaibly any FerO, in the scale to F a d , thereby
and nickel oxides. facilitating the solution of the scale by the glass. The con-
From thermodynamic considerations, either Coo or NiO version of FeO or FerO, to F a 4 also would result in volume
in their standard states should react with crystalline FeO to changes that could aid solution by breaking up the scale,
form F a 4 . If the Coo or NiO in an enamel were to react the ratios of the molar volumes of these oxides being FeO

--
in a similar manner when in contact with the predominantly 1.00, I/s(FeSOd 1.18, and l/r(Fer0,) 1.21.
FeO scale such as forms on sheet iron, then FeSO4would be These experiments suggest that the reason certain oxides
produced. F a 4 would Ile dissolved more readily by the act as adherence promoters, in spite of obvious difEerences in
enamel, thus explaining the more rapid attack. The reaction crystal structure between their metals and iron,lies partly
+
taking place would be 3FeO Coo 6Fe04 Co. If the + in the fact that they help clean the metal surface.
metallic Co were to come in contact with Fe& in the scale,
more F a 4 w o u l d b eproduced according to thereaction Co + (2)
3Fer0, + Coo + 2 F a 4 . No reaction would be erpected Metallic particles, often as dendrites, have been observed
in enamels on sheet steel usually at, or near, the glass-metal
hetween the Coo and the F a 4 already present. The re-
actions should apply equally to NiO as follows: interface. Their presence has led to considerable specula-
tion as to the relation of these particles to adherence.' To
3FeO + NiO -L Fa04 + Ni investigate various conditions under which metal might pre-
cipitate, fifteen glasses were prepared, each of which contained
a Merent metal oxide added to the basis glass. These
glasses then were fired on ten Merent metals for 15 minutes
The basis glass previously used was prepared with the in an argon atmosphere.
addition of 1 weight % of Coo. This glass was placed in Aftex firing, metallic phases were observed in many of the
contact with the e t e , hematite, and magnetite pellets specimens. An analysis of the d t s showed that, in general,
and heated as before. The results are illustrated in Fig. metallic particles were produced whenever an oxide was
18 ( D ) ,(El. and (F). present in the glass which could be reduced by the metal
Figure 18(D) shows the pellet of wtistite (FeO) in the Coo- base. The presence of reducing gases might be expected to
containing glass. This pellet has assumed an altogether cause further reduction. However, most of the particles
merent appearance from that of the specimen in Fig. 18(A). observed could be explained by assuming that the base metal
Figure 18(D) shows rounded surfaces and considerable glass was the sole reducing agent. The results are summsrized
penetration, apparently the result of oxidation of the wiistite in Table VI, where the glasses have been arranged in the order
to F a 4 and subsequent solution of tbe F a d . predpi- of the free energy of formafjon of the metal oxide additions at
tated metallic cobalt, a product of this reaction, can be seen l!NO°F., and the metals have been arranged in the order of
near the pellet, further evidence that crystalline ferrous oxide the free energy of formation of the lowest oxide on which data
can react with cobalt oxidein a glass under the proper con- were available. Those oxides above and to the right of the
ditions. solid line in Table VI should have been reduced if only crys-
Figure 18(E) shows the hematite (Feral) pellet in the Coo- talline oxides had been present. Actually, the oxides were
containing glass. The pellet appears to be essentially un- in solution in the glass and there was the further possibility
changed; no oxidationreduction was observed. that alloys or complex oxides might have been formed as
Figure 18(F) shows the magnetite ( F a d pellet in the reaction products. In spite of these obvious possible dif-
Coo-containing glass. The attack was very severe as it was ferences, the general agreement ~ 8 good.
s
in the CB(K of the Coo-free glass. If 6ring conditions were vaned from oxidizing to neutral
From these results, it appears that Coo in a glass reacts to reducing, the quantity and size of t h e metallic particles
with the c r y s e e oxides of iron much as would be expected or dendrites increased. However, unless other conditions
if all the &rides were in their standard states. were met, the degree of wetting and the adherence both de-
What happens to the FeO dissolved in the glass? If this creased as the firing conditions changed in the manner in-
oxide were to react with the Coo in the glass, little scale could dicated. Repeated experiments were conducted in which
be dissolved, because the Coo near the scale as well as that adherence was almost nonexistent in spite of the presence of
in direct contact with it would be reduced. Glasses were a multitude of dendrites.
prepared containing 30 or 40% of FeO and 0.7, 1.3, 6.7, or The presence of metallic particles also provides a further
13.3% of Coo. Samples of the FeOcontaining and Coo- insight into the conditions a t the glass-metal interface. An
containing glasses were placed in contact and fired on Vycor- inert glass provides an excellent seal, practically eliminating
brand glass plates for 5 minutes a t 1500OF. in a purified- the diffusion of atmospheric oxygen for considerable periods.
argon atmosphere. No reaction was observed along the line To investigate this, a NsrO-B&-SiOs glass was smelted,
of contact. Apparently FeO dissolved in the glass did not outgassed, and then fired on electropolished iron a t 1600OF.
react with Coo dissolved in the glass. Therefore, Coo can in air for periods ranging from 1to 85 minutes.
oxidize crystalline FeO to F 4 4 and thus aid in dissolving The type of overfiting observed was the gradual burn-off of
scale, but once the iron oxide is in solution, no further reac- the enamel from the edges of the panels toward the centers.
tion will take place. The central areas were only slightly discolored even after
As a further check, glasses were prepared containing MnO 80 minutes of 6ring. This experiment indicated that vir-
or CuO. Thermodynamically, MnO is more stable and CuO tually no iron oxide was being formed after the initial scale
less stable than FeO. When these glasses were heated in was dissolved and therefore no attack of the base metal was
contact with the iron oxide pellets for 20 minutes at 1600OF. occurring owing to direct oxygen penetration through the
in an argon atmosphere, the results were qssentially as ex-
pected from thermodynamicansiderations. According to
free-energy-of-formation data, 'the MnO-bearing glass was
incapable of acting as either oxidizing or reducing agent with
respect to the iron oxides, so it was as inert as the basis glass
itself, whereas the CuO-bearing glass reacted with the FeO G. H.Spencer-Strong,J. 0. Lord, and R. M. King, "Me-
to form Fe04, thus eEectivdy increasing the rate of solution
WCS af Ensmel Adherence: VII, Further studies Of Enamel-
Metal Contact Zones by. M' k and Metallogrsphi~Meth-
of the FeO pellets. - ods," J . Am. Ceram. Sac., I S y m (1922).
November 1959 Nature of Adherence of Porcelain Enamels to Metals 519
Table VI. Occurrence of Metal Particles in Glasses Fired on Various Metals for 15 Minutes at 1500°F. in Argon

--- ~~
Probable oxide in base glass
PIOI WOa M O O ~ COO
-
Assumed ZrOl TiO? \'*OL CrzOa SnOr ZnO AsrOr N iO I'M) BitOa CuO
oxide Actual additions to base glass
Base from --- ~-
metal metal Zr02 Ti0r CrKh ZnO VZOS PtO6 SnOz WOa MoOz CoKh NiO AszOa PbO Ri:Oa CuO

Ti no + I + + * + + + +* 0 +* +* +* + +* +*
Mn MnO +* + * I +* + * +* +* +* +* 0 +* +* +* +* +* +*
Cr + +* + + +* *
Fe FcO + 0 +* +* + * + *
IL-----1 +*
W WO. I +* +* + +*
MO MOO, 7 +* + + + L--
+*
CO coo 0 I +* +* +*
Si SiO + 0 + I +* + L--l

L----
+
cu CU?O * *

KEY: + Metal particles observed near interface.

* Metal particles observed in glass proper.
0 Hazy zone near interface which may have contained submicroscopic metallic particles.

Fig. 19. Section through a Na20-B?03-Si0. glass Ared on electropolished

iron at 160OOF. in air for 40 minutes. The dendritic structures are the re-
sult of metallic iron being transported from the interface to a point where
it was precipitated in this particular habit. (X500.)

Fig. 20. Glass of 50% FeO showing wurtite dendrites and alpha-iron
crystals. ( X 500).
enamel. The absence o f appreciable attack also was es-
tablished by platinum reference plugs.
Figure I!) shows a section through a specimen fired for 40
minutes. The most important occurrence observed in this
study was the growth o f metallic iron dendrites from the base
metal in the absence o f any reducible oxides such as Coo or
NiO. This possibility has been a matter of controversy for FeO per I00 parts of basis glass. The light-gray network has
many years. The dendrites were readily apparent after 15 been identified by X-ray diffraction analysis to be wiistite
minutes of firing and, after (i0 minutes, extended 3 mils into (Feo.:60). Cubes of metallic iron, which must have been
the enamel. precipitated during the cooling of the glass, occurred with
The fact that these metallic iron dendrites even exist is these wiistite dendrites and apparently were compatible with
sufficient evidence to necessitate the reappraisal of the theories them. I n some areas, the iron crystals appear to have
concerning the penetration of oxygen to the base metal. If served as nuclei for the growth o f wiistite dendrites. This
the oxygen could readily penetrate the fused glass by solution glass wetted iron very well when fired in an argon atmosphere.
or a simple diffusion process, these dendrites would have been The ability of this glass to wet was such that, when fired as
oxidized as fast as they were formed. Instead, the dendrites a pellet, i t would spread out readily over the entire surface
were observed to grow and there was no evidence of the for- of the plaque once the glass melted.
mation of iron oxide. I n general, metallic particles precipitated in a n enamel can
The precipitation of metallic iron, as just described, was be considered to be indicators of reactions which have taken
somewhat unusual, but it was observed often in glasses which place. Their appearance can be predicted with fair accuracy
were saturated with FeO. -4s previously noted, these glasses from the thermodynamic stability of the metals a n d oxides
wetted iron and, in view of the precipitated iron crystals, present. The presence of precipitated metallic particles is
were apparently in equilibrium with iron, as illustrated in the neither indicative of, nor essential to, the development of
glass shown in Fig. -30. This melt contained 100 parts of good adherence.
 520 Journal of The American Cerumic Society-King, T r i p p , and Duckworth Vol. 42, No. 11
roughening in the absence o f atmospheric oxygen or iron
oxide. This was true of the entire series. Dietzel""'
pointed out the necessity for the presence of available oxygen
to maintain galvanic action and thereby to develop roughness.
Ile maintains, however, that the effective oxygen is that which
penetrates the fused enamel from the atmosphere after solu-
tion of the scale. He also maintains that the reaction be-
tween FeO in the srale and COO does not participate in the
development of roughness. I n marked disagreement with
this is the evidence in Figs. 2 l ( C ) and 2 l ( D ) . llere it is
shown that the oxygen contained in the scale formcd o n iron.
by heating in air, can be of importance in the development
of a highly irregular interface in the absence of atmospheric
oxygen. I n Fig. 21(C) a layer of needlelike oxide crystals
separates the layer of metallic spheroids, prolxhly cobalt,
from the enamel-metal interface. Yet this interface shows
a n excellent distribution of ideal anchor points such as are
normally attributed to galvanic corrosion. For all the NiO
and/or COOcontents investigated, the specimens preoxidized
and then fired in argon exhibited a much rougher metal sur-
Fig. 21. Sections through the interfaces of Coo-bearing enamels flred on face than did those specimens which were not preoxidized
electropolished iron for 5 minutes at 1600OF. (A] '/2 weight % COO, hut fired in air. Both of these conditions produced much
argon atmosphere, clean iron; ( 6 ) 4 weight % COO,argon atmosphere, more roughness than firing in argon only.
96
clean iron; ( C ] Yz weight COO, argon atmosphere, preoxidized iron;
Platinum plugs were inserted in several specimens to mark
(D] 4 weight % COO,argon atmosphere, preoxidized iron; ( E ) '/2 weight
% COO, air atmosphere, clean iron; and ( F ) 4 weight % COO, air the original surface position. It was observed that the total
atmosphere, clean iron. ( X 500.1 amount of attack on the metal followed the same general
sequence as the degree of roughness. I t could be noted,
also, that the major portion o f precipitated metal, such as
can be seen in Fig. 2 l ( C ) , lay below the level o f the original
metal surface. These metallic particles appeared in the
volume which probably contained the oxide layer that had
formed before the enamel fused. For the range o f NiO and/
(3) Attack on he Metal or COO contents studied, the degree o f roughness developed
The previous section showed that a molten glass can act as at the glass-metal interface was about the same. However,
an effective seal to prevent atmospheric oxygen from reaching some differences were observed. When nickel was present,
a metal surface. However, i t is known that, in commercial the metal precipitate had less tendency to form massive
enameling, the metal continues to be attacked as firing pro- dendrites, and metal was precipitated to a greater extent
gresses. This attack apparently is one function of reducible throughout the glass. This was probably the result o f X i 0
oxides such as the "adherence promoters." being more easily reduced b y FeO than is Coo.
( A ) Investigation of Role of Cobalt Oxide and Nickel To obtain some idea of the roles o f hydrogen and water
Oxide: The role of adherence promoters in the attack on vapor when dissolved in adherence oxide-bearing enamels,
the metal was studied. The experimental work was limited the glass containing 1 weight yoof NiO plus COO was smelted,
to a study of the attack on iron by glasses containing COO or using batch materials t h a t contained considerable water o f
NiO, but i t is believed that the general principles evolved constitution such as AI(ON)3. and hydrated borax.
would apply to any metal in contact with a suitable glass. This batch was smelted in a gas-fired smelter, water quenched,
When a piece of sheet steel coated with dried enamel is and not outgassed. The enamel was fired on electropolishec1
fired in air, the metal oxidizes rapidly, forming an iron oxide iron in the same manner as the other glasses in this series.
scale but leaving the metal surface relatively smooth. If the No difference in the type or degree of attack on the iron
enamel is a typical ground coat containing COO and NiO, was observed with this moisture-laden enamel compared with
the scale soon disappears and the glass-metal interface then the others. T h e surface configuration was essentially the
becomes rougher as firing continues. same as for the similar enamels from which water and hydro-
For this study, the following percentages of COO and/or gen were removed. However, the hazy layer was morc
NiO were added to the basis glass: pronounced and the scale tended to persist longer when water
Added oxides (weight % ) was present. The latter was evidenced by the presence o f
NiO NiO + COO COO
isolated patches of thin layers of scale on the preoxidized
'/, + specimen which did not appear in any o f the other specimens
'/?

2
1 '/2 +
1 + 1
'/4
'/?
'/?
1
2
of the entire series. This was probably the result o f the rc-
duction of water vapor by the iron to produce additional
4 2 + 2 4 iron oxide, thereby slowing the total solution of the scale by
8 4 + 4 8
the enamel. The specimen fired in argon alone appeared no
Tliese oxides were smelted into the glasses which were then different from similar specimens in the other series
outgassed to remove the water and other dissolved gases. ( B ) Efect of Iron Oxides in the Enomel: Hecause o f the
The glasses were fired at IGOO'F. for 5 minutes in the following effectiveness of an oxide scale in promoting surface roughness
manner. Each of the glasses was fired ( I ) in air on electro- in the presence of NiO and/or COO in the absence of atmos-
polished iron, (2) in argon on electropolished iron, and ( 3 ) in pheric oxygen, the effect of iron oxide dissolved in the glass
argon on electropolished iron preoxidized by heating in air on the interface was studied. Four glasses were prepared.
for 2 minutes at IfiOO'F. Firing at I ( i 0 O O F . for 5 minutes One was saturated with FeO, one with Fe203,one with FesOr,
in air produced excellent adherence on iron with the three and one with both FetOI and FeO. They were fired on
glasses containing a total of 1 weight yoof adherence oxides. electropolished iron at l(iOO°F. in a n argon atmosphere.
Sections of some of these specimens are shown in Fig. 21. Some corrosion resulted in each instance, but the roughness
The 1)hotomicrogral)hs show that COO, or metallic cobalt was slight and resembled grain-boundary attack. I t was not
deposited from COO,was almost totally ineffective for surface the type of roughness generally experienced in commercial
November 1 %!) Na!ure of Adherence of Porcelain Enamels to Metals 52 1
Table VII. Glasses Used to Study Changes in Enamels are shown in Table VII. The iron oxide contents were less
During Firing than half t h a t required for saturation. The glasses were
applied to electropolished iron and fired at I(iOOo or 2000'F.
in air for various periods.
The results for the enamel containing Fea04 are shown in
19.0 19.0 19.0 Fig. 22. Figures ??(A) through 2 2 ( H ) show (at a magni-
18.0 18.0 18.0 fication of 5OOX) the changes that occurred at 1GOO'F.
48 0 48 0 48.0
1.5 0 I n Fig. ??(A) the scale which formed initially was still
15.0 present. In Fig. 2 2 ( 8 ) the scale had dissolved and a fine
1.5 0 roughness was developing. Figure 22( C) shows a remarkable
degree of fine roughness with an excellent distribution of
anchor points. However, no appreciable adherence had
developed. The 1 -minute interval between Figs. 22( C) and
?2(0) resulted in a tremendous change in the degree and type
enameling. The surface of the metal remained fairly smooth, of roughness. The change was less significant after this time.
and areas o f attack were not particularly deep or irregular. Good adherence was developed after about 5 minutes of
Iron oxides in the enamels did not prove t o be effective firing.
roughening agents in the absence of atmospheric oxygen. Figures 22((1) through 22(L) show the Fe30t series fired a t
( C ) Combined liflect of Atmospheric Oxygen and Iron 2000 OF. Considerable differences exist between these speci-
Oxides in the Enamel: Since the previous study had shown mens and the previous ones. The rate of attack was much
the oxides of iron to cause little roughening in the absence of greater than that in the lower-temperature series, and the
atmospheric oxygen, the effect of firing iron oxide-containing roughness was of a greater magnitude. Again, little or no
enamels in air was studied. For this investigation, three adherence was attained until after the first appearance of a
four-component glasses were prepared. Their compositions crystallhe phase, not necessarily at the interface.

Fig. 22. Sections through the interfaces of an Fe301-containing glass flred on electropolished iron in air a t temperatures
shown. ( A ) 1 minute, (6) 2 minutes, (C) 3 minutes, ( D ) 4 minutes, (€1 5 minutes, (F) 7l/$minutes, (G) 10 minutes, ( H ) 20 min-
utes, ( 1 ) 1 minute, (I) 5 minutes, ( K )7'/2 minutes, ( I ) 10 minutes, (MI 4 minutes, ( N ) 71/2minutes, and (0)
10 minutes.
 522 Journal of The American Ceramic Society-King, Tripp, and Duckworth Vol. 42, No. 11

I X 250.1
Fig. 23. Nickel-dipped electropolished iron flred in air a t 1 60OoF. using
Fe3Or-cantaining glass. ( A and 8 ) 1 minute, (C and D) 7'/2 minutes, ( E and
F) 25 minutes. Nickel weights of 0.02 gm./sq. ft. are on the left and
0.08 gm./sq. ft. are on the right. Specimens (C) and (Dl have platinum
reference plugs an the right side. (X500.1

square foot. These were enameled with the glasses contain-

ing 13 weight yo of FeO or Fe30r (compositions 1 and 3 in
Table VII). Firing was at 160O0F. in air for periods rang-
ing from 1 to 25 minutes.
Figures 22( .%I) through 22((0) show sections through the The results of this study are shown in Fig. 2 3 , which shows
whole enamel layer. These illustrate the type and distri- the manner in which the iron surface was affected by the
bution of precipitates observed once the firing time had been nickel. The presence of platinum reference plugs in Fig. 23
sufficient to produce some degree of saturation. In every ((C) and ( D ) ) establishes that the original surface was still
instance, the enamel contained a solid layer of oxide at or present over much of the specimen. Also, although no oxide
below the surface of the enamel, but well above the surface precipitate is visible in any of the fields shown. a small amount
of the metal. The appearance of this layer occurred simul- was present after 15 minutes of firing. This is in contrast
taneously with the change in the general type of surface to the result shown in Fig. 22 and indicates that, in the
roughness as can be seen by comparing Fig. 22(C) with 22(D) presence of a nickel dip, little iron is taken into solution by the
or Fig. 2 2 ( J ) with 2 2 ( K ) . It also was noted that crystalline attack of atmospheric oxygen transportcd to the base
wiistite existed in most of these enamels. metal Only slight differences appeared between the speci-
The results with the glasses containing FeO or Fez03 were mens prepared with FeO- or FerOl-karing enamels.
not markedly different from those just described. The These data support the findings of others that the coating
periods required for removal of the scale were somewhat longer deposited by a nickel-dip operation is effective in decreasing
in these studies, as were the periods at which the interfaces the oxidation of the iron.
made the sudden change from small anchor points to coarse Nickel dip has been found t o suppress the development of
configurations. As before, adherence was not developed by a rough interface but, at the same time, helps to maintain a
what appeared to be a n ideally roughened interface. It was more uniform degree of roughness over long firing times.
developed when a crystalline phase appeared in the enamel ( E ) Role of Nickel i n Development of Surface Roughness
and no sooner. The crystal habits of the precipitated phases H'hen Used as an Alloying Agent i n Steel: Much has been
appeared to be different in the glasses containing different written about the role of nickel and cohalt when incorporated
iron oxides, and possibly the compositions of the phases in the enamel as an oxide or applied in the elemental form to
also were different. iron. Apart from corrosion studies on high-temperature
This study indicates that all the oxides of iron are effective alloys, little has been done to obtain a better understanding
as corrosion accelerators when incorporated in enamels which of the functions of these metals as alloying agents in enamel-
are fired on iron in air. Fe30r appears t o act most readily ing stock. It seemed possible that cobalt or nickel in the
as a corrosive oxide. After a brief period of time, however, steel could have an influence on the rate and type of corro-
the three oxides appear to be almost equivalent. sion of the metal during the firing process
This investigation also showed that enamel-metal interfaces Five steels* were used which were reported to contain
characteristic of those generally achieved in commercial nickel in amounts of 0.01, 0.2, 0 5 5 , 1.2i. or 2.55 weight yo.
practice can be obtained in the absence of NiO or COO. These were enameled with the glasses listed in Table VII.
(D) Efferts of Nickel Dip on Change i n Enamel-.Uetal In- Firing was at l(i00OF. in an air atmosphere for periods
terface During Firing: For many years, i t has been known ranging from 1 to 20 minutes.
that a light application of nickel on an iron surface tends to Some of the results of this study are shown in Figs. 24 and
control or limit the oxidation of metal. This treatment is 2.5. Figure 24 shows the progressive changes of the enamel-
known as a nickel dip. The effect of such a coating on the
changes taking place a t the interface during firing was in-
vestigated.
A number of electropolished iron specimens were nickel- * Supplied by The Iiiternational Sickcl Coinpaiiy, Iiicor-
dipped to a pick-up of 0.02, 0.05,and 0.08 gm. of nickel per porated, Sew York, S. Y.
November 1959 Nature of Adherence of Porcelain Enamels to hfetals 523

Fig. 25. Sections through the interfaces of four additional nickel-

bearing steels enameled with Fe304-containingglass and flred in air
for 20 minutes at 1600OF. ( A ) 0.01, (6) 0.2, (C) 1.27, and ( 0 )2.55
weight % nickel. (X250.) Electron cloud

Fig. 26. Schematic representation of glass-iron interface.

metal interface of the 0.55% nickel steel as the firing time was
lengthened. Initially, the surface acquired a fine roughness. is the case because the modifier ions are much smaller.
This transformed a t between 5 and i ’ / )minutes to a much Attack on the base iron might follow the sequence
coarser roughness. Adherence a t 5 minutes and less was (1) At the enamel-air interface,
very poor; above 5 minutes it was good. At the longer firing
times, the roughness greatly exceeded that developed by a
4Fez+(Mn2+,etc.) +
0 2 + 4Fe3+(Mn3+,
etc.) + 20r-
similar enamel on nickel-free iron. (2) At the enamel-iron initerface,
Figure 25 shows the differences in surface roughness of the 4Fea 2Fe0
+ + 4 6FeZ+
various steels after 20 minutes of firing. The higher nickel or
contents appear to have favored both greater penetration
into the metal and the development of a fine roughness on the
4Mn3+ + 2Fe0 + 2Fe2+ + 4Mn2’
surfaces of the grains. These equations suggest that “oxygen transport” is a mis-
This study points out the necessity for understanding and nomer and that the true mechanism may be one of cation
controlling alloying agents in enameling stock. I t shows diffusion or even of electron transport.
nickel to be an extremely active agent in promoting surface
, ..
roughness in the presence of atmospheric oxygen which has V. Discussion and Summary
been transported to the enamel-metal interface. The results indicate that good adherence of enamels to
( F ) Conclusions and Discussion Regarding Attack on the metals results from metal-to-metal bonding. For this to
Metal: These studies demonstrate that the primary corrosive occur, ions of the metal are required in the enamel and other
agent which attacks and roughens the iron during enameling chemical and thermodynamic requirements must be met: (1)
is atmospheric oxygen. This oxygen may be present in the The enamel at the interface must be saturated with an
scale formed during the early stages in firing or it may pass oxide of the metal and (2) this oxide must be one which, when
through the fused enamel during the later stages in firing. in solution in the glass, will not be reduced by the metal. In
Atmospheric oxygen does not pass through the enamel by the case of iron, the oxide is FeO.
simple diffusion. Instead, i t must react with oxides of var- At elevated temperatures where the metal ions in the glass
iable valence which are present in the glass. These oxides and the atoms in the metal are relatively mobile, there is a
in turn provide the mechanism by which the base metal is continual exchange at the glass-metal interface. Metal ions
attacked. from the glass diffuse into the metal, gain electrons, and be-
This general mechanism, which is called “oxygen trans- come metal atoms. Metal atoms diffuse into the glass and
port,” has been proposed by others. Actually, it is unlikely become ionized. Thus, a state of dynamic equilibrium exists
that the adherence oxides are especially effective in this a t the glass-metal interface which could not be maintained
because most of them lack two oxidation states which can if the glass were not already saturated with the proper oxide of
both be present under the conditions of enameling. For this the base metal.
reason, and because the quantity of iron oxide present gen- At lower temperatures where the atoms and ions are largely
erally is greater, oxygen transport is probably much more of fixed as to position, there may exist only an exchange of
a function of the oxides of iron than of the adherence oxides. electrons. This situation is illustrated schematically in Fig.
As a further conjecture on the method of oxygen trans- 26, which is another way of illustrating Fig. 17. Since the
port, it seems likely that in glassy systems the acceptance of difference between an atom and an ion lies soley in the con-
oxygen into the glass entails a change in valence of the glass centration of electrons in the outer orbits, the atoms and ions
modifier ions such as iron. The modifier ion which has a t the interface may alternate in properties, depending on the
changed valence remains in an interstitial position. In such concentration of electrons or the degree of ionization. This is
a system, it would not be necessary for the oxygen ions to indicated in Fig. 26 by showing the electron cloud partly sur-
move if the modified ions could move more readily. Such rounding the metallic ions in the glass. These atoms and
524 Journal of The American Ceramic Society-King, Tripp, and Duckworth Vol. 42, No. 11
ions of intermediate character provide a transition between pose instead of the 21y0present in air. If the enamel is satu-
the metallic bond in the base metal and the electrostatic rated with the oxide of the base metal, atmospheric oxygen
and covalent bonds in the glass. is not required.
Starting with an enamel that fulfills the proper requirements, Some consideration must be given to the conditions
and using firing conditions that will prevent any changes in imposed by the requirements of most commercial enameling.
equilibrium or the development of new phases, good adherence Under these conditions, it is known that COO and/or NiO
can be attained readily without consideration of the multi- added to the enamel facilitate the development of adherence.
tude of factors commonly imposed. Good adherence can be The roles of these oxides are several, and the total extent of
attained, but less directly, by employing other routes to these roles is not known at present.
reach the same interfacial conditions. However, much has been learned about COO,NiO, and other
The question naturally arises as to the significance of the oxides reducible by iron. For one thing, they help to main-
basic requirements for the development of adherence. Sat- tain good wetting under adverse atmospheric conditions.
uration of the enamel-metal interface with the proper oxide The small amount of oxygen available from these sources
of the base metal appears to be necessary for several reasons. often will prevent the loss of intimate contact between the
First, it promotes wetting, and good adherence cannot be enamel and metal during a temporary deficiency of atmos-
developed in the absence of good wetting. Studies on the pheric oxygen.
characteristic wetting of metals by glasses of different com- These oxides also are effective in accelerating the removal
positions and using different atmospheres showed that glasses of the scale. This investigation showed that Fe3OI was more
did not wet metals in the absence of available oxygen, either readily soluble in enamels than either FeO or Feoa. The
as atmospheric oxygen or as reducible oxides in the glass, reducible oxides such as NiO and COO favor the formation
unless the glasses were saturated with the proper oxide of of FeaO, from FeO, which is the major constituent of scales on
the base metal. iron when they are formed under normal enameling condi-
Second, saturation may infer that all the available posi- tions.
tions within the glass are filled and additional metal ions are Reducible oxides also can contribute to the amount of
forced into positions on a surface such as the glass-metal ferrous oxide present at the interface. The reduction of
interface. It is only when the iron ions are forced to remain these oxides by the iron contributes only a relatively small
somewhat permanently in positions intermediate between the amount of FeO, but as oxygen go-betweens or electron car-
glass structure and the base iron that they are subjected riers they could make atmospheric oxygen available at the
strongly to the forces of both of these phases, thereby sup- interface. Experimental evidence has been gathered show-
plying the linkage which produces adherence. ing that appreciable quantities of atmospheric oxygen, per
Third, glasses a t saturation are beginning to acquire a more se, do not diffuse through the molten enamel. In the absence
definite structure. Such a condition is admittedly highly of metal ions easily capable of existing in at least two valence
speculative, but is supported somewhat by cursory X-ray states, little or no oxygen gains access to the interface after
measurements. The ratio of glassformers to modifiers sug- the enamel has been fused. This suggests that a transport
gests that the normal three-dimensional network is virtually mechanism exists which could be maintained by a wide
replaced by an interweaving of two-dimensional chains. The variety of oxides often added to enamels, including those of
glass has some of the properties of both glasses and crystalline manganese, iron, cobalt, nickel, and others.
solids. Under these conditions, the metal ions may assume Reducible oxides also can roughen the base-metal interface.
more definite positions with respect to the oxygen and silicon This might be by direct corrosion, by galvanic corrosion, or
ions. by some mechanism that favors preferential attack. This dis-
The requirement that the oxide in solution in the enamel cussion on adherence would not be complete without some con-
not be reduced by the base metal is in keeping with the con- sideration of the existing theories of the adherence of porcelain
cept of equilibrium as a dynamic process. The addition of enamels. Surface roughening, or “mechanical adherence” as
readily reducible oxides, such as COO or NiO, to an enamel it is known, is the basis for many of the existing theories for
simply supplies a source of oxygen that can react with the base adherence, such as the galvanic corrosion theory. The pres-
metal. Metal is precipitated when the oxide is reduced. ence of considerablesurface roughening in nearly all enameling
This process is irreversible in a closed system and cannot be has led many to believe it to be the sole or major reason for
considered in the normal sense of equilibrium, because the adherence. This investigation indicates it to be a coincidence
metallic atoms produced are not free to revert to the ionic rather than a cause. The same may be said of dendrites of
state. Any premise that in an enamel system a reducible cobalt, nickel, or iron.
oxide, such as COO or NiO, can exist in close contact with a Three major objections exist to the acceptance of the
metal base, such as iron, is not made in full cognizance of mechanical theories of adherence. First, excellent ad-
the free energy of formation of the various oxides being con- herence can be obtained on extremely smooth metal surfaces.
sidered. Second, the presence of a roughened interface exhibiting many
A t the other extreme, from a thermodynamic viewpoint, is anchor points does not insure adherence under conditions
the use of oxides considerably more stable than the oxides deviating more than slightly from .the previously described
of the base metal. Once again the free and reversible inter- criteria for chemical adherence. The;final objection is based
change between the ionic and the atomic state is prevented on measured values for the strength oflhe bond obtained with
almost entirely. good adherence. To obtain such strength would necessitate
Not much has been said up to this time about the role of the assignment of extremely high values for the strength of
gases in enamels. Not a great deal was done in this investiga- iron, because the total cross section of the anchor points is
tion to study gases except at various times to include or ex- always a relatively small proportion of the total interfacial
clude certain ones. Much has been written about water vapor area.
and hydrogen. These have been both present and absent Surface roughness, however, is not to be discounted
during the development of good adherence in cobalt oxide- totally. Like the presence of adherence oxides, it often may
bearing and cobalt oxide-freeenamels. The presence of hydro- play a beneficial role. Experimental data show that ad-
gen is not necessary for the development of good bond, but herence is developed only over a narrow range of enamel com-
it may be that, under conditions of commercial enameling, positions lying very near the limit of oxide saturation. A t
any hydrogen at the enamel-metal interface must be removed the lower extreme of this range, the bond strength per unit
before a satisfactory bond can be developed. area is low. The development of surface roughness supplies
Oxygen is known to be necessary to promote wetting in a greater total surface area, with the possibility thereby of
ordinary enameling, but or less is sufficient for this pur- increasing the total bond to a satisfactory value. Under
November 1959 Pyroplastic Index and Firing Deformation of Ceramic Bodies 525
conditions deviating from ideal, as in coinmcrcial enameling, (2) Good adherence requires that the enamel at the
surface roughness is a factor. interface be saturated with an oxide of the base metal, and
Of the theories attributing adherence to chemical bonding. this oxide must be one which, when in solution in the enamel,
the oxide-layer theories are important. They suggest that a will not be reduced by the metal.
layer of oxide which is adherent to the metal is wetted by the (3) Good adherence of enamels to metals appears to be the
glass and becomes the transition zone between the metal and result of metal-to-metal bonding between the atoms in the base
the glass. This layer is so thin as to be undetectable by most metal and metallic ions in the enamel when certain chemical
means. There are several objections to this theory. The and thermodynamic requirements are met.
most important one is that few oxides are sufficiently adherent (4) The oxides commonly known as adherence promoters
to their metals to produce excellent adherence. The theory of help to establish and maintain the conditions necessary for
adherence advanced in this paper may appear to duplicate or good adherence without contributing directly to the bond
to be closely related to the oxide-layer theory but, in fact, i t developed.
does not recognize the degree of order at the interface required Acknowledgment
to produce an oxide layer. Some intermediate degree The writers wish to thank the many staff members of Battelle
of order in the glass structure is visualized, but this exists only Memorial Institute who contributed to this investigation.
between small groups of ions such as the Si4+,01-,and Fe2+. Among the many associated with this program, the contributions
There is no orderly arrangement of these structures into a of the following were particularly significant: C. E. Sims, A. R.
three-dimensional structure as is required to meet the delinition Elsea. L. S. O’Bannon, and C. G. Harman. Also of value were
the many suggestiqns made by technical representatives of the
of a crystal lattice or layer. The existence of image forces or sponsoring companies.
the polarization of ions at a glass-metal interface has been pos- The program was sponsored by several companies. Those
tulated as an explanation for the bond between glasses and who were active through the entire program were: Bethlehem
metals. However, the importance of metal oxide saturation Steel Company; Chicago Vitreous Corporation ; Ferro Corpora-
tion ; Frigidaire Division, General Motors Corporation; The
in the development of the adherence of porcelain enamels was 0. Hommel Company; Ingram-Richardson, Incorporated; The
not indicated. The theory presented in this paper therefore International Nickel Company, Iiicorporated ; Nash-Kelvi-
supplies a more complete explanation of the known facts. nator Corporation; Pemco Corporation; The Pfaudler Com-
pany; and A. 0. Smith Corporation.
The following companies were sponsors of a portion of the
effort: American Stove Company; Benjamin Electric Manu-
VI. Conclusions facturing Company; Borg-Warner Corporation; Climax Molyb-
denum Company; Florence Stove Company; The Fusite Cor-
(1) The bond between enamels and metals can be chem- poration: General Electric Company; Tappan Stove Company;
ical in nature. and Vitreco, Incorporated.

Pyroplastic Index and Firing Deformation of Ceramic

Bodies
by D. S. ADCOCK, J. E. DRUMMOND, and 1. C. McDOWALL
The New Zealand Pottery and Ceramics Research Association, hcorporated, Lower H a , New Zealand

Methods of assessing the W g deformation of rod during firing. They were able to predict the sag of such
ceramic bodies by the pyroplastic index were de- a rod from the parameter and the rod dimensions. This
scribed in a previous publication. Equations parameter, which they term the “pyroplastic index” (P.I.),
are presented here by which the firing defor- here denoted by D, gives a measure of the tendency of the
mations of simple geometrical shapes can be er- material under investigation to deform when fired on the
pressed in terms of the pyroplastic index and the schedule specified.
dimensions of the piece.

1. Introduction Received September 15, 1958; revised copy received April 13,
ECREASE in viscosity is the main factor leading to dis- 1959.
Technical Report No. 10 of the New Zealand Pottery and
D tortion and deformation of ceramic ware during firing.
Methods are available for determining the viscosity of
ceramic bodies at elevated temperatures, but data cannot be
Ceramics Research Association. Incorporated, issued in July 1958.
At the time this work was done, D. S. Adcock and I. C. Mc-
Dowall were, respectively, Scientific Officer and Director, The
applied to estimate firing distortion unless the changes of New Zealand Pottery and Ceramics .Research Association.
viscosity with time and temperature can be integrated. The Incorporated, and J. E. Drummond was associated with the
Applied Mathematics Laboratory, Department of Scientific and
density of the material is also relevant, and a full description Industrial Research, Wellington, New Zealand. D. S. Adcock
of the deformation tendency of a ceramic body requires a is now with the Standard-Vacuum Oil Company (N. 2.). Limited,
parameter containing terms relating to viscosity, time, and Wellington, and J. E. Drummond is with Canberra University
density. College, Canberra, A. C. T., Australia.
1 I. C. McDowall and W. Vose, “Determination of Pyroplastic
Such a parameter has been derived by McDowall and Vose,‘ Deformation in Firing of Ceramic Bodies,” Trans. Brif. Ceram.
who analyzed the central sag of a simply supported circular Soc., 50 1111 506-16 (1951); Ceram. Absfr., 1953, March, p. 476,