JOURNAL OF C4TALYSIS 11, 220-227 (1968)

Chemisorption of Hydrogen Sulfide on a Siupported

Nickel Catalyst

JENS R. ROSTRUP-NIELSEN
From the Haldor Topsee Research LaboratomJ, P. 0. Box 49, Vedbek, Denmark

Received February 22, 1968

The chemisorption of hydrogen sulfide on some nickel catalysts has been studied
in the temperature range 550645°C. The sulfur uptake is a function of the
ratio ~H~,Y/~H~. A saturation layer has been observed at ratios above 5 X W,
approximately, and at ratios above lo-’ bulk sulfide (NiSJ was formed. The sulfur
content of the saturation layer is correlated with the nickel area determined by volu-
metric measurements of hydrogen chemisorption at -72°C. The mechanism of the
chemisorption of hydrogen sulfide is discussed.

INTRODUCTION At high coverage the dissociated hydro-

Hydrogen sulfide is known to poison gen will gradually be desorbed as the sulfur
nickel catalysts, the hydrogen sulfide being atoms take its place. Thus it would be rea-
retained on the catalysts. sonable to believe that all hydrogen orig-
The poisoning of a nickel catalyst may inating from the chemisorbed hydrogen
occur even when the concentration of hy- sulfide might be desorbed as molecular hy-
drogen sulfide ought not to cause formation drogen. This is, however, not the case, as
of a bulk compound of sulfur and nickel. only about two-thirds of the hydrogen can
As an example, it appears from Rosenqvist’s be removed from a surface saturated with
investigation of the thermodynamics of hydrogen sulfide-even by evacuation at
nickel sulfide phases (1) that formation of 400°C. Measurements by Kemball et al. (3)
bulk-phase sulfide at temperatures around and by Muller and Gibert (4) have sup-
500”-700°C requires a p&p,, ratio of the ported Selwood’s observation concerning the
order of RY3. This ratio is about lO&lOOO dissociative chemisorption. However, from
times above what would normally cause their measurements on nickel films in the
poisoning at those temperatures. Therefore, temperature range -80’ to 0°C Kemball
a priori, hydrogen sulfide may be assumed et al. suggest that each molecule of hydro-
to be retained 1,; a chemisorption process. gen sulfide occupies three nickel atoms and
Literature on ihe chemisorption of hy- that the breakdown of the molecule as an
drogen sulfide on nickel is very scanty. Den intermediate stage may involve an HS
Besten and Selwood (2) have found by radical occupying one site and blocking a
magnetochemical measurements on nickel second with an H atom on the third site.
catalysts in the temperature range 0’ to The results of Muller and Giber%, who
115°C that hydrogen sulfide is dissocia- repeated the study of Den Besten and
tively chemisorbed, forming four chemi- Selwood, show identical rates of hydrogen
sorptive bonds with the nickel surface. The sulfide chemisorption and hydrogen de-
following mechanism is suggested : sorption, when the surface has been com-
pletely covered with dissociated hydrogen
I I I I sulfide. This, too, is hardly in accordance
H&3 (gas) + -Ni-Ni-Ni-Ni- = with the simple mechanism including four
nickel atoms.
-~-N&Ni-~& (1) All the studies mentioned above have
220
been performed at low temperatures, but fide. Swift and Butler (5) have determined
there seem to be no publications on chemi- the kinetics of this hydrolysis in the tem-
sorption of hydrogen sulfide at higher tem- perature range of 6&9O”C to be of first
peratures. The present work is part of a order with respect to the concentrations of
study of chemisorption of hydrogen sulfide both hydrogen ions and thioacetamide. The
on supported nickel at temperatures above hydrolysis has been successfully used at
400°C. The study of the saturation layer 25°C as a hydrogen sulfide generator in the
has been performed by correlating the sul- range of 0.01-100 mg S/hr.
fur content of the saturated catalyst with The apparatus was made mainly of glass,
the nickel area measured by hydrogen apart from the reactor, which was made of
chemisorption. 18.8 stainless steel. It is true that this made
the reactor adsorb hydrogen sulfide, but
METHODS this feature appeared to be of no influence
on the investigation. The reactor (internal
1. Catalysts diameter 15 mm) was placed in an elec-
In order to avoid sintering effects, four trically heated oven, and the temperature
catalysts called A, B, C, and D, which were in the catalyst layer was regulated with a
made for high-temperature employment, thyristor controller.
were used for the experiments. The cata- Analytical. The gas was analyzed for
lysts contained about 10 wt % Ni and were hydrogen sulfide by passing it for a certain
supported on different carriers consisting of period through a bubble flask containing an
magnesium aluminum spinel. Catalyst A acidic solution of cadmium acetate, and the
contained a few percent of potassium, which collected sulfur was then determined colori-
may be of no importance to this investi- metrically by the methylene blue reaction.
gation. The sulfur content of the catalysts By means of the flow of hydrogen sulfide
prior to chemisorption has been determined thus determined and the hydrogen flow
to 5 - 20 wt ppm. measured by the flow meters, the molar
ratio H, S/H, was calculated.
2. Hydrogen Sulfide Chemisorption The sulfur content of the catalysts was
Apparatus. The measurements of the determined by driving off hydrogen sulfide
chemisorption of hydrogen sulfide were with hydroiodic acid in an apparatus de-
performed at atmospheric pressure by pass- scribed by Gustavson (6). The hydrogen
ing until equilibrium was established a mix- sulfide was then analyzed as mentioned
ture of hydrogen sulfide, hydrogen and, in above. This method has a relative standard
some experiments, argon over the catalyst, deviation of about 2%, and the results were
which was placed in a tubular reactor. in accordance with those obtained from
A hydrogen flow, purified in a copper radiochemical analysis.
oven, was passed through a hydrogen sul- Chemisorption conditions. After reduc-
fide generator with a constant production of tion of the catalyst for 1 hr in dry hydrogen
hydrogen sulfide per unit of time. The at 850°C the chemisorption of hydrogen
partial pressure of the hydrogen sulfide was sulfide was performed by passing a hydro-
changed by varying the hydrogen flow. It gen sulfide containing hydrogen flow over
was also possible to send hydrogen and the catalyst at 55O’C (in some experiments
argon, free from hydrogen sulfide, to the 645°C). After having passed to the catalyst
reactor. The hydrogen and argon flows were the amount of sulfur that was expected to
adjusted by needle valves and were meas- be retained by it, the hydrogen sulfide flow
sured with flow meters. Sampling for anal- was measured before and after the reactor.
ysis could be made before and after the Then, normally, these flows were close to
reactor. each other, but the reactor was given at
The hydrogen sulfide generator was based least twice the amount of sulfur that was
on hydrolysis in an acidic solution of thio- expected to be retained before closing the
acetamide to acetamide and hydrogen sul- experiment. In some experiments a num-
222 JENS R. ROSTRUP-NIELSEN

ber of catalyst types were treated tribution of (loo), (llO), and (111) planes
simultaneously. in the surface yields a mean area of a
nickel site of 6.5 A”.
3. Nickel Area and Hydrogen However, not the area but rather the ad-
Chemisorp tion sorbed amounts of hydrogen are of interest
Principle. Chemisorption of hydrogen and for the present work.
carbon monoxide has been used for deter- Apparatus and procedure. An ordinary
mination of nickel areas; still, owing to the apparatus for determination of BET areas
fact that carbon monoxide may be bound was used. The sample container was a
to the surface in different ways (7) and be- quartz flask equipped with a tube at the
cause there is also a risk of formation of bottom so that hydrogen could flow through
nickel carbonyl (8), it seems reasonable to the catalyst layer. During reduction and
prefer hydrogen, which is assumed to be evacuation, the flask was surrounded by an
bound in a well-defined way. electrically heated oven. The evacuation
However, the hydrogen-nickel system is was carried out by means of an oil diffusion
also complicated, as a fast chemisorption pump. The hydrogen was purified in a
is followed by slow, activated effects. This copper oven, and a gold foil trap was used
feature has resulted in a great many differ- to protect the sample against mercury
ent conceptions (7, 9, IO), which are not vapor. After some few measurements, a
to be discussed here. Maybe this uncertainty palladium alloy diffusion cell was also used
is the reason for the existence of several for purification but, apparently, this had
different procedures for determination of no influence on the results.
the nickel area of catalysts by means of The measurements were mainly per-
chemisorption of hydrogen (11-15). formed under the following conditions:
The method used in the present case is
Amount of catalyst: 10-20 g.
based upon the assumption that the hydro-
Reduction: 2 hr at 850°C in a stream of
gen uptake observed by Beeck (16) result-
dry hydrogen (2 moles/hr).
ing from the fast chemisorption on un-
oriented nickel films at -196°C and pHZ = Evacuation: 850°C until p < 1O-4 mm
0.1 mm Hg, reflects a coverage with one Jfk.
Chemisorption: - 72°C (freezing mixture
hydrogen atom per nickel atom of the sur-
of ethanol and solid CO,). First measuring
face. However, in order to avoid physical
point was read after a period of half an
adsorption on the carrier, the temperature
had to be increased to -72”C, which could hour, whereas the following three to four
points were read at intervals of 10 min.
be done without any influence from the
activated hydrogen uptake. When the adsorbed hydrogen volume is
Knor and Ponec (17) have criticized the plotted against the hydrogen pressure, one
BET measurements (among which were gets an isotherm appearing as a straight
those with krypton) that were the basis of line with a slightly positive slope. The first
the results of Beeck and Ritchie (16). point will normally be located on the or-
Beeck as well as Knor and Ponec found dinate axis. The intersection of the straight
that about 0.8 mole of krypton per mole of line with the ordinate axis is taken as a
hydrogen is adsorbed on a randomly measure for a monolayer of hydrogen on
oriented nickel film. However, Beeck used the nickel surface corresponding to the
a calculated value of the area of a krypton volume adsorbed at low pressure (~0.1
molecule (14.6 AZ) whereas it appears rea- mm Hg) .
sonable to use an experimental value
(21 K2), as was done by Knor and Ponec. RESULTS
The latter value results in an area of a hy-
drogen site on the nickel surface of 8.9 IL2, 1. Chemisorption of Hydrogen Sulfide
whereas Beeck used 6.18A2. From crys- Preliminary experiments indicated that
tallographic considerations an equal dis- the chemisorpt,ion of hydrogen sulfide is re-
 CHEMISORPTION OF HzS ON NI CATALYST 223

I I fur uptake of the saturation layers (called

6w I swtppms
I
s,,) of the four catalysts. It should be noted
that the measurements on Catalyst D are
subject to great uncertainty. From the data
in Table 2 it may be concluded that the
sulfur uptake was independent of the
amount of sulfur added, which amounted to
1.3 to 30 times that which was retained. In

TABLE 1
ISOTHERMS AT 550°C AND 6&j%-CATALYST Aa

Mg S/lx hlole Hz/hr Mole A/hr r x lo'* s WPPm)

Isotherm at 550°C
I I
1 2 3 6 5 6 7 0.067 4.0 0 0.53 410
0.110 4.1 0 0.84 520
FIG. 1. H,S chemisorption. Isotherms at 550°C
0.016 4.3 0 0.12 150
and 645”C, Catalyst A. Estimated Langmuir iso- 0.100 2.0 0 1.56 515
therms are shown. 96
0.004 4.5 0 0.03
0.016 1.2 1.9 0.43 395
versible and that the sulfur uptake depends 0.011 2.5 1.9 0.14 222
on PHZS as well as on PHZ. Results from 0.374 3.8 0 3.08 543
experiments at 550” and 645°C performed 0.045 2.2 0 0.64 515
0.034 2.0 0 0.53 470
with different values of p&S and pH2 are
shown in Fig. 1 and Table 1. From Fig. 1 it Isotherm at 645°C
appears that- the amount of sulfur retained 0.144 3 .9 0 1.16 440
by the catalyst (called s) may be considered 0.100 2. 0 0.9 1.56 455
a fucction of the ratio pHzS/pHz (called 0.104 1.9 2.4 1.73 520
T) and, furthermore, that a saturation layer 0.110 1.5 1.9 2.34 559
is obtained at a ratio above 2-5 X 10mG. 0.07s 1.4 0 1.74 492
0.020 4.9 0 0.13 140
Equal amounts of sulfur per unit area are
0.311 1.3 0 7.84 545
taken up at the two temperatures.
0.130 2.0 1.9 2.0” 545
Table 2 and Fig. 2 show results from sul- 1.9 0.69 437
0.044 2.0
0.02s 4.5 0 0.20 227
0.061 4.2 0 0.45 4’20
0.054 3 .9 0 0.48 420

a 0.5 g of catalyst used in each experimrnt. Dura-

.
D tion more than 50 hr.
1CUQ. .
. ’ f = PEPS/PEP
.

ml. addition to this, the sulfur uptake is not

correlated with the H, S/H, ratios applied,
which were varying from 2 X 1O-Gto 27 X
10-F. Hence, the measured amounts of sulfur
that have been retained are assumed to
represent the saturation layers.
Scanning experiments showed that an
Irn/ 7 E.,06
: abrupt increase in the sulfur uptake occurs
when the H, S/H, ratio exceeds the value
10 20 30
for formation of bulk sulfide predicted from
FIG. 2. HzS chemisorption. Saturation layers of the thermodynamics (1). Thus, depending
various catalysts. on the duration of the experiment, Catalyst
224 JENS R. ROSTRUP-NIELSEN

TABLE 2
SATURATION LAYERS OF VARIOUS CATALYSTS”

H&3 addition
D”EtP
r
HaS flow
(mg S/W (mp s/g cat) r x lo‘* (~&Q)

Catalyst B
10 1st 0.97 1.8 15.1 705
10 21 0.63 1.3 9.8 675
10 45 1.04 4.7 16.2 685
5” 42 1.70 7.2 26.6 723
10 66 0.87 5.7 13.5 738
Cat4dgst C
10 44 0.77 3.4 12.0 235
5 43) 0.80 6.9 12.5 267
5 66 0.43 5.7 6.7 288
5 42 0.91 7.6 14.2 263
5 42; 0.86 7.3 13.4 310
Catalyst D
10 88 0.74 6.5 11.6 1045
10 44 0.59 2.6 9.2 1125
10 47 0.36 1.7 5.6 1270
10 45 0.81 3.7 12.7 923
5” 42 1.70 7.2 26.6 983
0 Tempedure: 550°C; H, flow, 2 mole/hr.
br = %S/PIl*
c Same experimwt

B contained about 0.1 to 2.7 wt % sulfur feet of the activated hydrogen adsorption
after treatment at 55O’C with Hz S/H, in is observed (Fig. 4). However, the rate of
a ratio of about 10-3. Ni,S, was identified hydrogen uptake has been too small to in-
by X-ray analysis, as expected from the fluence the determinations of the areas that
data of Rosenqvist (1). are shown in Table 3. Again the measure-
Finally, it should be mentioned that the ments on Catalyst D are subject to rela-
carrier impregnated with a potassium com- tively great uncertainty.
pound did not show any sulfur uptake at
550°C and with H, S/H, = 2.5 X 1O-s. This
means that potassium apparently does not V Ncm3 Ii2
6. .
influence the results from Catalyst A.

5.
2. Hydrogen Chemisorption
Typical isotherms are shown in Fig. 3, L..

one of which shows incidentally a measure-

ment for p = 0 mm Hg placed just where 3;;

the linear extrapolation intersects the or-

dinate axis. This may support the validity 2..

of the procedure of calculation of the area

on the basis of this extrapolation to low 1..

pressures. It should be noted that this re- P mHg

sult is not in accordance with the extrap- 10 20 30 .co 50 m 70
olation suggested by Sinfelt (11).
By leaving the sample overnight, the ef- FIQ. 3. H, chemisorption. Isothermu rt -72°C.
 CHEMISORPTION OF HzS ON NI CATALYST 225

r is shown in Fig. 5. The value of so has

been determined as the mean of the meas-
06. v Ncm”H2/g cat.
ured s values for T above 2 X 10”. By
linear regression analysis the slope n is es-
timated to be 1.1.

2.0

1.5
0.11
time Lhoursl

10 20 30 LO 50 60 70

Fm. 4. I& chemisorption. Influence of the acti-

1.0
vated chemisorption. The hydrogen uptake has
been followed by a decrease in pressure from 32 to
8 mm Hg.

RESULTS FROM MEASUREMENTS

TABLE
OF NICKEL
3
o.si/ 1.0
:
1.5 2.0
AREAS BY MEANS OF HYDROGEN
CHEWSORPTION AT -72°C Fro. 5. Estimation of n in s = ar”/l+ br” by
plotting log (s/s”-s) = n log T + log b, where so
HP uptake Ni area N”2ber
meas-
Est*of
variance
mean
=a/b. Data from isotherm at 645°C.
(Ncm’/g w/g
Catalyst red. cat.) red. cat.) urements x 104
Assuming n = 1 linear regression of r/s
A 0.224 0.78 2 16 on r fitting the well known linear form of
B 0.348 1.21 8 4 the isotherm
C 0.143 0.50 3 11
D 0.454 1.58 6 88

yields at 550°C
DISCTJSSION
Corresponding values of sulfur uptake 3.1 X 10gr
wt ppm S (52
(s) and r = pHzs/pHZ, which are plotted ‘~ = 1 + 5.3 X 109
in Fig. 1, show a correlation resembling a and at 645%
Langmuir isotherm. It has therefore been
attempted to correlate the results by an ex- 2.5 X 10gr
wt ppm S P3)
pression of the form ’ = 1 + 4.4 X 10%

UP By weighted regression analysis of the

S=l+ (2) form

where a, b, and n are constants. 1-=

In order to estimate n, the expression is s ;+;
rearranged to where the variables are separated, the ob-
tained results do not deviate significantly
logso
A-- s
= n log r + log b (3) from these expressions.
Calculated values of the expressions have
where so =cz/b. formed the basis of the curves shown in Fig.
The plot of log (s/s,, -5) against log 1, and it is obvious that the estimated ex-
226 JENS R. ROSTRUP-NIELSEN

pressions can be looked upon only as a gression analysis fitting a straight line
guidance. through the origin, the slope is determined
It is well known that it is questionable to be 0.74 for hydrogen capacity as inde-
whether it is possible to conclude from an pendent variable. Moreover, statistical
analytical expression, which covers an ex- analysis indicates a high correlation factor.
perimentally obtained isotherm, anything In other words, the surface may have
about the mechanism of the adsorption. The nearly three-fourths of a “sulfur sit,e” per
fact that a certain theory turns out to be “hydrogen site.”
in agreement with experimental data, need Now, it should be possible to estimate, in
not be a proof of its assumptions. principle, the atomic ratio of sulfur and
However, it should be mentioned that n nickel in the saturation layer, which in
= 1 after Langmuir’s assumptions repre- itself may give an idea of the atomic ar-
sents the following mechanism including oc- rangement. This implies knowledge of the
cupation of one nickel site: atomic ratio of hydrogen and nickel at the
surface after the fast chemisorption at
H,S (gas) + Ni = Ni-S + HZ (gas) (8)
-72’C; it also implies that the surfaces
By similar analysis Muller and Gibert may be considered identical under the dif-
found from their data from low tempera- ferent conditions of the two measuring
tures (4) that n = 1/, which was related to methods and that the sulfur and hydrogen
the dissociation of hydrogen sulfide in- atoms are occupying the same type of
volving occupation of two sites. “sites.”
Another method of illustrating how sulfur The latter assumption is made probable
may be adsorbed at the nickel surface is to by the observation that no measurable
compare the respective saturation layers chemisorption of hydrogen took place on a
with respect to their contents of sulfur and catalyst which was saturated beforehand
hydrogen. This has been done for the four with sulfur. Since in the two measuring
catalysts in Fig. 6 where the number of methods the catalysts have very similar
sulfur atoms retained per gram of catalyst pretreatments, and as no correlation has
has been plotted as a function of the num- been observed between sulfur capacity and
ber of hydrogen atoms chemisorbed per time of sulfidation (i.e., no sintering effects),
gram of catalyst. There appears to be a it would be reasonable to assume that the
linear dependency and by weighted re- same amount of “sites” have been involved
in the two measuring methods.
According to the general assumption that
one hydrogen atom per nickel atom is
chemisorbed at the conditions used here,
there ought to be chemisorbed 0.8 sulfur
atom per nickel atom, in the saturation
layer. However, as a consequence of what
was previously mentioned about the criti-
cism of this assumption proposed by Knor
and Ponec (4), one should rat.her use a hy-
drogen capacity of 6.5/8.9 = 0.73 hydrogen
atom per nickel atom, which will result in a
sulfur capacity of 0.54 sulfur atom per
nickel atom.
These calculations are of course very un-
certain, but the result seems reasonable in
relation to the size of the sulfide ion (about
10 A”), which hardly allows a coverage
FIO. 6. Correlation between hydrogen and sulfur above 0.5 sulfur atom per nickel atom.
uptake. Stable surface layers containing free metal
 CHEMISORPTION OF H2S ON NI CATALYST 227

atoms have also been observed in the chem- sorption complex. This requires further
isorption of hydrogen sulfide on silver by studies by means of another technique.
BBnard et al. (18)) an observation probably
corresponding to the observations made by ACKNOWLEDGMENTS
means of the LEED technique on the chem- I wish to express my thanks to Mr. Haldor
isorption of oxygen on nickel (19). As no Topsoe for permission to publish the results and
hydrogen uptake was observed on a cata- to Mrs. G. Parthum, Miss B. Johansen, Mr. C.
lyst saturated with sulfur, it may be pre- Thomsen, and many of my colleagues for as-
sistance in performing the experiments. I am very
sumed that the sulfur atoms of the satu- grateful to Dr. Hans Bohlbro for many helpful
ration layer are concealing the free nickel and stimulating discussions.
atoms.
The idea of the composition of the sat- REFERENCES
uration layer which has been suggested 1. ROSENQVIST, T., J. Iron Steel Inst. (London)
above, is somewhat different concerning the 176, 37 (1954).
role of hydrogen, from what is inferred in DEN BESTEN, I. E., AND SELWOOD, P. W., J.
the literature (6,3,4) from results obtained Catalysis 1, 93 (1962).
at low temperatures. At low temperatures SALEH, J. M., KEMBALL, C., AND ROBERTS, M.
the hydrogen originating from hydrogen W., Trans. Faraday Sot. 57, 1771 (1961).
sulfide is at least partly retained at the sur- MULLER, J. Cl., AND GIBERT, R., Bull Sot.
face, whereas the mechanism discussed Chim. France, p. 2129 (1967).
SWIFT, E. H., AND BUTLER, E. A., Anal. Chem.
above involves no chemisorption of hydro-
28, 146 (1956).
gen. This may well be ascribed to the dif-
6. GUSTAVSON, L., Talanta 4, 236 (1960).
ferent temperatures applied, the sulfur dis- 7. GUNDRY, P. M., AND TOMPKINS, F. C., Trans.
placing the hydrogen from the surface when Faraday Sot. 53, 218 (1957).
the temperature is increased. The discrep- 8. HUGHES, T. R., HOUSTON, R. J., AND SIEQ, R.
ancies concerning the number of bonds P., Ind. Eng. Chsm. Proc. Design Develop.
formed when a sulfur atom is chemisorbed, 1, 96 (1962).
will not be discussed because it must be con- 9. BEXCK, O., Advan. Catalysis 2, 151 (1950).
ceded, again, that the argument on the basis 10. SCHUIT, G. C. A., AND DE BOER, N. H., Rec.
of the isotherm cannot be conclusive. Trav. Chim. 70, 1067 (1951).
11. YATES, J. C., SINFELT, J. H., AND TAYLOR, W. F.,
CONCLUSION J. Am. Chem. Sot. 86, 2996 (1964).
1.2. CARTER, J. L., CUSUMANO, J. A., AND SINFELT,
The sulfur capacity of the saturation J. H., J. Phys. Chem. 70, 2257 (1966).
layer observed in the chemisorption of hy- lb. NIKOLAJENKO, V., BOSEEK, V., AND DANE&
drogen sulfide on nickel at temperatures V., J. Catalysis 2, 127 (1963).
about 550°C may be related to the nickel 14. GEUS, J. W., AND NOBEL, A. P. P., J. Catalysis
area as measured by hydrogen chemisorp- 6, 108 (1966).
16. SWIFP, H. E., LUTINSKI, F. E., AND TOBIN, H.
tion. A mechanism involving a single bound
H., J. Catalysis 5,285 (1966).
sulfur atom and no chemisorption of hy- i6. BEECK, O., AND RITCHIE, A. W., Discussions
drogen, seems possible but, probably owing Faraday Sot. 8, 159 (1950).
to the size of the sulfide ion, the sulfur ca- 17. KNOR, Z., AND PONEC, V., Coil. Czech. Chem.
pacity is less than one sulfur atom per Commun. 26, 961 (1961).
nickel atom. It may be reasonable then to 18. B~NARD, J., OUDAR, J., AND CABAN&BROUTY, F.,
consider the atomic arrangement of the sur- Surface Sci. 3,359 (1965).
face phase instead of considering the ad- 19. M.4c RAE, A. U., Surface Sci. 1, 319 (1964).