INFLUENCE OF ADSORPTION PHENOMENA ON OXIDATION OF METALS AT HIGH TEMPERATURES

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP80-00809A000700240058-0
• Release Decision: RIPPUB
• Original Classification: U
• Document Page Count: 15
• Document Creation Date: December 22, 2016
• Document Release Date: July 11, 2011
• Sequence Number: 58
• Publication Date: September 22, 1955
• Content Type: REPORT

Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
STAT
Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
 Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
INFLUENCE OF ADSORPTION PAENCMENA ON OXIDATION
OF METALS AT HIGH TEMPERATURES
[Comment: This report presents the full text of an article by
s.- Di.il. f"br M64auRle, submf{ted from
the -Institute of Physical Chemistry of the University of Greifswald,
and published in Metall, Volume 6, 1952, No 11/12, pages 285-291,
Berlin.
Numbers in parentheses refer to appended authors' bibliography.)
Statement of the problem
In spite of the simple equation for the chemical reaction
metal + oxygen= metal-oxide
the oxidation of metals has a rather complicated reaction mechanism since the
reaction product, the metal-oxide, separates the two reactants from one an
other. Thus, the total oxidation process comprises the two phase boundary
reactions at the phase boundaries metal/metal-oxide and metal-oxide/oxygen and
the diffusion of ions and electrons through the metal-oxide layer as depicted
in Figure 1.
inlerna~ stratum
Pl a se
Figure 1. Schematic Representation of the Density of the Quasi-Free Electrons
in the-Oxide Layer of a Scale System Metal/Metal-Oxide/Oxygen
In the last few decades, a large amount of theoretical and experimental
research has been published to explain these component processes and to clarify
their effect on the total course of the oxidation. As a result, today we are
in a position to decide clearly whether diffusion phenomena in the oxide layer
or phase boundary reactions are rate-determining. Although in the case of the
formation of solid porefree surface layers the parabolic oxidation law of
Taman (1) is valid, in the case of porous surface layers the phase boundary
reactions are slowest and are therefore the rate-determining reactions because
here the oxygen passes quickly enough through the pores of the surface layer
to the metal.
I -~
Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
 Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
For the following discussion, however, we wish to consider the formation
of porefree oxide layers only. Under these conditions the diffusion phenomena
of the starting materials, either the metal or metalloid, are usually rate-
determining. According to a hypothesis by Wagner (2), atoms do not diffuse
through the oxide or scale layer; only ions and electrons do so. Furthermore,
Wagner Vas able to show that a migration of ions or electrons can take place
in the scale layers only through the lattice vacancies. (There are ions in
the interstices and ionic lattice vacancies or quasi-free electrons and elec-
tron-defect places.) By extending bhe Wagner-Schottky Lattice Vacancy Theory
(3), Wagner was able to develop, by means of the general formula, a theory of
the oxidation of a metal of which the oxidation rate can be calculated from
the free energy of the oxidation, from the electrical capacity and the trans-
fer numbers of the ions and electrons, or from the free energy of the oxidation
and the autodiffusion coefficients of the ions participating in the diffusion
through the scale layer. By extension of these relationshils and application
of the lattice-vacancy theory extended to heterotypical mixed phases (5), Wagner
(6), as well as Hauffe and his collaborator (7), applied the Theory of Oxida-
tion Phenomena to metal alloys.
While more has become recently known about oxidation reactions with rate-
determining diffusion phenomena, the participation of phase boundary reactions
has had only limited consideration. Judging from available experimental obser-
vations, in most cases with fast diffusion phenomena, the phase boundary is the
rate-determining component of the oxidation. Thus, the dissociation rate of
the oxygen molecule or an established adsorption equilibrium can be considered
of importance. However, this by no means exhausts the possibilities of the
phase boundary reaction.
In analogy to the surface layer theory of the crystall rectifier of
Schottky and Spenke (8), Hauffe and Engell (9) have been able to show that,
for the chemisorption of oxygen on a metal-oxide, a surface layer of the
oxide of up to 1,000 angstroms is added to the phase boundary; this will be
discussed in more detail below. The phase boundary metal-oxide/oxygen is,
first, the place at which the adsorption of gaseous oxygen and the subsequent
formation of oxygen ions take place, i.e., the chemisorption. As indicated
elsewhere, the surface stratum of the oxide plays a critical role. (9, 10)
Because of the importance of the participation of the surface s'.-atum to oxida-
tion phenomena, which can lead to experimental findings that cannot be inter-
preted directly in terms of present knowledge, we wish to deal in this paper
especially with such phase boundary reactions.
Moreover, in dealing with reactions on the phase boundary metal/metal
oxide, the existence of a surface stratum will have to be considered, and
in this case assumptions based on the Theory of Crystal Rectifiers will be
promising.
Adsorption and Phase Boundary Reactions
Recently, Hauffe and Pfeiffer (11) referred to the fact that the oxidation
rate of iron at fixed temperatures in CO-CO2 mixtures is determined by phase
boundary reactions. To some degree the experiments can be interpreted by means
of plausible assumptions concerning the chemisorption of the C02 on the FeO
scale layer.
Earlier, Moore and Lee (12) attempted to interpret their oxidation research
on zinc in a similar manner. Their experiments showed a clear relationship of
the oxygen pressure to the oxidation rate that is inexplainable in the Wagner
Scale Theory. (2, 4) Also, Moore and Lee assumed, in the interpretation of
their experiments, that the chemisorption of the oxygen on the Zn0 of the sur-
face layer was the rate-determining factor. Of course, they found a parabolic
Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
 Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0 I
function that was not consistent with the first assumption. How both circum-
stances (oxygen pressure relation .p to the oxidation rate and the parabolic
time functions) can be interpreted ZnO at the same time from the adsorption proper-
own below after discussion of the general con-
cepts. Furthermore, Vernon and coworkers (13) were able to show that t9e oxida-
tion rate of zinc, especially at lower temperatures between 190 and 300 C, can
be described by a logarithmic time function, which can also indicate phase
boundary phenomena.
It is thus understandable that the problem of adsorption on scale layers
has been of great interest. Also the following observation is of interest
in this connection: If an iron sheet is heated in high vacuum (p0, about
10-6 to 10-1} Torr) at 1,000?C, there is no measurable weight incroase of the
sample even after several hours. If, on the other hand, the sample is placed
in a CO-CO2 mixture that has a very low oxygen partial pressure (per a 10-10
Torr), then at the same temperature after several hours the sample will be
completely oxidized. (11) The oxidation rate under the experimental conditions
is proportional to the quotient (pC(~2/pC0)2/3. As the cause of this phenomena,
we might assume a specific adsorption of C02i which in this case acts as the
oxidizing agent, on the iron oxide surface layer. Molecular oxygen, on the
other hand, is evidently adsorbed much less strongly and therefore effects an
oxidation rate that is much slower at these low pressures.
In the following discussion, the "chemisorption" (activated adsorption
activated by chemical forces between the surface layer of the adsorbent and
the adsorbate) of oxygen on oxide layers will be considered. Essentially, we
shall bring the same concepts into use that we developed i- zr lecture at
the meeting of the Deutsche Bunsengesellschaft in Berlin in January 1952 (9)
and in another place with W. Schottky (10).
The influence of chemisorption on scale phenomena can be quite variable.
A possibility of this influence has already been briefly indicated for the
example of the oxidation of iron in a CO-CO2 mixture. After discussing the
general mathematical interrelationships of chemisorption and then several spe-
cial aspects of the relationship between lattice vacancy phenomena in oxide
surface layers and chemisorption, we will consider the influence of chemisorp-
tion on the mechanism governin; the formation of "thick" scale layers (layer
thickness X10-1% cm). According to current theoretical developments (9), an
influence of chemisorption is to be expected on the growth of such layers if
the chemisorption itself or subsequent reactions at the phase boundary oxide/
oxygen are rate-determining. Furthermore, it is evident what influence the
alloying materials can exert in this case.
Next, we will consider the relationship between chemisorption and the
formation of "thin" surface layers (10-6 to 10-5 cm). In this case, the
chemisorption controls the kinetics of the tarnish phenomena if the trans-
fer processes in the surface layer are rate-determining. The question whether
phase boundary reactions or transfer processes are rate-determining can be
answered from the form of the tarnish function according to the following as-
pects:
1. A parabolic time function points in all cases to the fact that trans-
fer processes are rste_determ.ining (Taimrann Tarnish Law). If an electron-defect
conducting surface gayer is formed, the parabolic tarnish constant is completely
dependent on the oxygen pressure. (2, !,,) If the surface layer is an excess
conductor, then the tarnish constant is a function of the oxygen pressure only
with "thin" surface layers.
Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
 Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
2. A linear time function always means that the phase boundary reactions
are rate-determining. In this case, we are dealing with a reaction at the
oxide/oxygen boundary; thus the tarnish constant depends on the oxygen pressure.
Of course, with defect-conducting oxide layers, the rate of the phase boundary
reactions at the metal/metal-oxide boundary can also be dependent on the oxygen
pressure so that a clear conclusion concerning the reaction mechanism in the
case of a linear time function is possible only for metals whose oxides are
electron-excess conductors.
Finally, these considerations will be applied to the oxidation of zinc
(according to the measurements of Moore and Lee), and it will be shown that
theory and experimental data are in good agreement.
Adsorption and outer layer Formation
For the following discussion, we assume that an oxide layer on a metal
or alloy is more than 1,000 angstroms thick and that the growth of this layer
continues. In the case of an electron-excess conducting surface layer, the
mass transfer from the metal to the boundary oxide/oxygen is effected by ca-
tions in the interstitial positions or oxygen ion lattice vacancies, which are
formed continually at the metal/metal-oxide boundary and combine with the
chemisorbed oxygen to form the metal-oxide at the oxide/oxygen boundary. The
entire process might take place in the following steps:
1. Solution of the metal In the oxide with dissociation of the dissolved
metal atones into interstitial position cations Moo and quasi-free electrons
Me = Meo?? + 20(H) (Ia)
The index H signifies here and in the following discussion a particle
in the semiconductor interior; therefore, in the internal phase of the surface
layer.
2. Adsorption of oxygen on the oxide:
oZ(G) = o2(adsorbed)
(ib )
3. The chemisorption of the oxygen with the consumption of the quasi-
free electrons of the oxide layer-.
1/2 o2(adsorbed) t 20(H) = 202-(0-) (Ic)
The symbol 02-(a') will represent a chemisorbed oxygen ion to which
two electrons of the ZnO are attached, either because a surface bonding (cova-
lent or ionic) is formed between oxygen and oxide, or in the manner suggested
by Weyl (14) by polarization of the conductivity bond of the oxide through
the oxygen of the surface layer. An 02-(a-) forms, therefore, two "surface
charges." Figure 2 illustrates the energy relationship on the oxide/oxygen
boundary which initiates process (Ic).
4. Formation of a new lattice plane of metal-oxide by reaction between
chemisorbed oxygen and interstitial cations which migrate by diffusion or
other transfer means to the cxidc/oxygen boundary:
Me0? ? + O2-(0')r"_ i.TeO (Id)
Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
STAT
 Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
ancL that
have been formed in the vicinity oftheumetal/met
al-oxidetboundary byndis ocia-
tion of the, metal atoms dissolved in the oxide, at the oxygen/metal-oxide boun-
dary the electrons are attracted to the cations on the basis of their greater
mobility; however, the spatial separation of the positive and negative charges
in the interior of the oxide layer ("internal phase") with adequate electron
concentration and layer thickness is so unimportant that a quasi-neutrality can
be assumed.
The charge distribution in the vicinity of the oxide/oxygen boundary be-
haves otherwise. By the transfer of quasi-free electrons in surface charges
through the chemisorption of oxygen at the surface, an electron-repelling sur-
face field results and thereby there is a decrease in negative charges within
the surface stratum of the oxide and hence the formation of a positive space
charge is sponsored (9) (see Figure 1). The process is analogous to the forma-
tion of surface strata of crystal rectifiers and permits us to deal formally
in the same manner as Schottky and Spenke (8) have dealt with these surface
strata (Figure 3),
bond
donors /- o_
bonet 01171,5_31711,
e(ectreirl IeveI of fIV
bsorkmd oyyyti before
Fa elec't'p o tra,,51 k
Figure 2. Simplified Energy Diagram for the Adsorption of oxygen on ZnO Before
the Electron Transfer From the Conductivity Bond
Sanitized Copy Approved for Release 2011/07/12 CIA-RDP80-00809A000700240058-0
 Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
oxide Iayer
no
l l
OZ-
k ~
/ u
-11
bo,,,,da, - ----s
su;.6 a La/e/iry7rn
Figure 3. Simplified Representation of the Charge in Concentration of Quasi-
Free Electrons n9 and the Metal Ions in the Interstitial Position
n, in the Internal Phase (1/2n,, = me = nR) and in the Surface
Stratum
The surface stratwc formation and chemisorption are controlled reci,ro-
cally and both are affected by the type and concentration of the lattice va-
cancies in the oxide. It is therefore possible to make predictions from the
lattice vacancy situation in the internal phase of the oxide layer concerning,
the adsorption behavior of the layer that is of treat importance in the clay-..
fication and effect of the tarnish phenomena if the phase boundary reactions
are rate-determining.
Influence of the S ace Charge in the Surface Stratum on the Transfer oP cations
Through the Oxide Laver
As mentioned at the outset, the proGress of the oxidation is made
according to Wagner (2), by the fact that the metal ions and the electrons :'.:?c
transferred through the oxide layer to the metal-oxide/oxygen boundary. ii:;
transfer can result both from diffusion .ad ;:._f-,ration in the electrical ?ie).d
so that in
general the following function is valid.
se = De p - I- (E)?ne. Be
sK= DKW i'9K(E(~)?nK. BK
These relationships must be fulfilled at each position of the se: eouc:.r-
tor. In these equations, S desiGnates the stream (i.e., the number c" c"_ cir_ r;
or cations which arrive per second on a sq c;,: of the outer surface) o,f cation:
or electrons toward thb outer surfaces; D is the diffusion coef_icicnt `
sec; B is the mobility in cm2/volt?sec; 4 is the position coordinate dircc-,,r1
from the gas/oxide boundary toward the interior; and ('(E) i., the field st .,;.h
in volt/cm at position Furthermore, n1. and ne mean the concentration o.'-metal ions in the interstitial positions and the quasi-free electrons: and y K the charge of the cations in the interstices. The concentration dictril,uiion n'
the positive and negative charges in the surface stratum is defined these con-
ditions, as well as by the following equations from the Boltzmann 'unction.:
vh n
su jgce m
1, / /;a se
Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
 Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
VD)
nR = exn(- (III)
Here nR is the concentration of the quasi-free electrons in the surface
stratum; nH is the same in the internal phase of the surface layer; VD
is the diffusion potential between the internal phase and the surface based
on the space charge in the surface stratum of the surface layer; and
A = kT/e.
Under the action of the space charge in the surface stratum, the migration
of the lattice interstitial cations to the oxide/oxygen boundary is accelerated.
To the diffusion, according to the first term of the sum of equation II, is
added a field current represented in equation II by the second term. Thus
there is a reduction in the concentration of positive charges of the surface
stratum and consequently a reduction of the field strength C(?) in the surface
stratum and of the diffusion potential VD and thereby again a reduction in
field current. A stationary position is reached when the concentration dis-
tribution of the interstitial cations in the surface stratum are so oriented
that a divergence-free stream of cations flows through the entire surface layer
(surface stratum and internal phase). In the internal phase, the forward flow
of the electrons ahead of the cations brings about a diffusion energy that is
in addition to the drop in concentration. This type of mass transfer is called
ambipolar diffusion. For the stream of the interstitial cations, it is true
that in this case of ambipolar diffusion in the internal phase we have, using
DK = BKB,
SK - (l+)K)?V?BK. fore ~.1 (IIe)
In the surface stratum, the diffusion component compared with the field
component is to be disregarded, so that from equations IIb and IIc we can write
for the stationary divergence-free cation current through the total surCace
layer the postulate
1+1/K)T1?BKXr5f>_1=~V{?nK( )?BK?E (6~) no22) we can write:
( 1)= 4wo. = K1.Pa/,22n2.exp(-2"(tl)?~l) (IX)
with K' = K=A~.N
and finally after modification
G (tl) 0n(pO2--"'K"2)-2AF( 413 . (X)
In this expression LnC(41) is negligible compared with ?(~1) because e(41) is
of the order of 105 V/cm.
The transfer of cations from the metal to the oxide/oxygen boundary takes
place according to equation IIb. If the field strength is great, then the dif-
fusion current is negligible compared with the field current, and one obtains:
Sk = nk?Bk?(B(~,).YK
From equations X and XI there follows as the final equation from the cation flow
(i.e., the number of cations which arrive at the surface per second per sq cm)
SK = nK.BK.y . .'en(po2.n4.K'2) (XII)
If the transfer of the cations to the surface is rate-determining for the
oxidation of the metal, then equation XII leads to a time function of the form
dim A
it
where A = NL.BK.Ve?~Gr 2 +Jn(n4 K'2)J
(Vm = mol volume of the oxide; N L= Loschmidt number.)
a parabolic tarnish function.
The field strength of the scale layer amounts to about 105 V/cm for layer
thicknesses of about 1,000 angstroms. If the layers are too thick then the
reaction comes to a halt if the temperature is not high enough for the diffusion
to cause an adequate transfer of cations to the surface. This stoppage of the
growth or transition to another time function will result from thinner scale
layers the lower the oxygen pressure and the lower the temperature because accord-
ing to equation X in both of these cases (F(,Xl) and thereby the field current of
the cations decreases.
The oxygen pressure dependence of the rate constants of the parabolic tarnish
function XIII follows from equation XII
Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
 Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
The slope of the curves in Figure 4 with small layer thickness may be
such that first equations XII and XIII determine the true time function and
after a transition region a formula becomes valid that corresponds to a trans-
fer of cations by ambipolar diffusion. If the scale layer reaches the thick-
ness of the surface strati= and exceeds it, then the surface stratum is detached
from the metal and moves, without changing its thickness, forward with the oxide
oxygen boundary. Since the space charge of the surface stratum is not effective
in the interval phase, the internal phase is also free of appreciable electric
fields. The surface stratum shields the internal phase, so to speak. Thus, if
the scale layer has become appreciably thicker than 2, the transfer of the in-
terstitial cations can occur only by diffusion and a new -- of course also
parabolic -- time function with smaller rate constants and a different tempera-
ture dependence of the rate constants is valid.
Since the thickness of the surface stratum is, according to equation V,
proportional to VD and consequently inversely proportional to the concentration
of the quasi-free electrons in the internal phase nH, this break in the curve
should lie at the minimum thickness of the surface layer the greater nH is.
Accordingly, the shift to the other time function must occur earlier with Zn-Al
alloys as with Zn-Li alloys. No experimental evidence in this direction is yet
available.
With the oxidation of zirconium and titanium, Gylbransen and coworker (19)
found a deviation from the parabolic tarnish function with small oxide layer
thickness that is analogous to the phenomena discussed above concerning the
findings with zinc.
Discussion of the Findings of Moore and Lee
As already mentioned, Moore and Lee (12) found a parabolic tarnish func-
tion for the oxidation of zinc and simultaneously an oxygen pressure dependence
of the oxidation rate constants that was incompatible with the Wagner Oxidation
Theory (see Figures 5 and 6). However, these counterstatements are cleared up
if it is kept in mind that the ZnO scales studies by Moore and Lee had a thick-
ness of only 100-1,200 angstroms. In the case of these thin scales, the trans-
fer of zinc ions through the ZnO layer is c ..used principally by electric fields.
As pointed out above, however, the field strength in the surface stratum at
the ZnO/02 phase boundary is a function of the surface concentration of the
chemisorbed oxygen and thereby a function of the oxygen pressure itselr. Since,
however, in the case of the Moore-Lee studies the ZnO layer is identical with
the surface stratum (g< 1), equations XIII and XIV are valid; i.e., thereby
the findings of Moore and Lee which were noteworthy at first become reasonable.
To check the oxygen pressure dependence of the rate constants, we plotted
the rate constants calculated by Moore and Lee from their measurements against
log pO . Figure 7 shows the straight lines obtained. As is known, the experi-
mental2data will be described satisfactorily by equation XIV. On the basis of our
graph, the Langmuir function introduced by Moore and Lee for time -detcr;nini-_%r
adsorption appears to us to be of slight value.
The measurements of Moore and Lee may be clearly interpreted, on the other
hand, as we have shown in explanations given above. It is, therefore, not to
be assumed that in this case the active adsorption of the oxygen on the ZnO is
itself rate-determining for the oxidation of the zinc. Bather it is the trans-
fer of cations through the surface stratum, which here is identical with the
scale layer, to the oxide/oxygen boundary which is considered as the slowest
process since a phase boundary reaction as the slowest reaction must give a
linear time function. Of course, the transfer rate depends on the diffusion
potentials and thereby the oxygen pressure, so that the simultaneous occurrence
of an oxygen pressure dependence of the reaction rate and a parabolic time
function is compatible.
We are especially obliged to Dr W. Schottky for numerous suggestions.
-13-
-1
Sanitized Copy Approved for Release 2011/07/12: CIA-RDP80-00809A000700240058-0
 Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0
1. G. Tersnann, Z. anorC, Chem. 111, 78 (1920)
2. C. Wagner, Z. phys. Chem. (B) 21, 25 (1933)
3. C. Wagner and ::. Schottky, Z. Phys. Chem. (B) 11, 163 (1930);
C. Wagner, Z. phys. Chem., Bodenstein-Festband, 17 (1931)
4. C. Wagner, Z. phys. Chem. (B) 32, 447 (1936); "Diffusion and High
Temperature Oxidation of Metals" in Atom Movements [sic], Ohio, 1951, 153
ff..
5. E. Koch and C. Wagner, Z. phys. Chem. (B) 38, 295 (1937); C. Wagner J.
chem. phys., 18, 62 (19?0); K. Hauffe, Ann. phys. (6) 8, 201 (1950);
Fehlordnungserscheinungen and Leitungs-vorgange in ionen- and elek-
tronenleitenden fasten Stoffen" in Ergebn. exakt. rraturwiss, 25, 193 (1951)
6. C. Wagner, J. Coor. Mat. Protection 5 No. 5 (1948);
C. Wagner and K. E. Zimens, Acta Chem. Scand. 1, 547 (1947)
7. K. Hauffe, Metallberfl. (a) 5, i (1951); K. Hauffe and Ch. Gensch, Z.
phys. Chem. 195, 116 (1950); 195, 386 (1950); 196, 427 (1950)
8. W. Schottky, Z. phys. 113, 367 (1939); 118, 539 (1942); W. Schottky and
E. Spenke, Wiss. Veroff. Siemens-We_rken, 18, 225 (1939)
9. K. Hauffe and H. J. Engell, Z. Elektrochen, 56, 336 (1952)
10. H. Engell, K. Hauffe, W. Schottky, Z. phys. Chem. (published in abstract)
11. K. Hauffe and H. Pfeiffer, Z. Elektrochon, 56, 390 (1952)
12. W. J. Moore and K. K. Lee, Trans. Faraday Soc., 47, 501 (1951)
13. W. H. Vernon, E. J. Akeroyd, E. G. Stroud, Inst. Metals, 65, 301
(1939)
14. W. A. Weyl, Trans. New York Acad. Sci.II, 12, 245 (1950)
15. C. Wagner, J. Chem. Phys. 18, 62 (1950)
16. K. Hauffe and A. L. Vierk, Z. phys. Chem., 176, 160 (1950)
17. Ch. Gensch and K. Hauffe, Z. phys. Chem. 196, 427 (1950)
18. N. Cabrera, N. F. Mott, Reports on Progress in Physics, Vol XII, 163 (1949)
19. E. A. Gulbrancen, K. F. Andrew, Trans. A. J. M. E., Metals Trans. 185, 515,
741 (1949)
Sanitized Copy Approved for Release 2011/07/12 : CIA-RDP80-00809A000700240058-0