The adsorption of CO gas by metals supported on silica

SURFACE SCIENCE 7 (1967) 229-249 0 North-Holland Publishing Co., Amsterdam

THE

ADSORPTION

OF CO GAS BY METALS

SUPPORTED

ON SILICA

C. R. GUERRA and JACK H. SCHULMAN Stanley-Thompson

Laboratory,

Krumb School of Mines, Columbia New York 10027. U.S.A.

University,

New York,

Received 1 November 1966; revised manuscript received 13 February 1967 The adsorption of CO gas by the metals Rh, Ir, OS, Re, Ru and Au supported on silica was studied. From electron micrographs and measurements of the surface area of metal covered silica, the metal particles appear as thin clusters on the silica surface ranging from about 20 8, in diameter and probably approaching monolayer thickness. The infrared spectra shows adsorbed CO species of the type M-CO, Mz-CO and M-(CO)z. At the higher pressures used with Rh, Ir and OS, a band at 2080 cm-l is attributed to a weakly held CO in sites already containing one chemisorbed CO. The possible occupancy of corner and edge sites in Re and Ru by more than two CO molecules is discussed. A correlation is

presented between the force constant of M-CO species and the number of d holes of the metal indicating a decrease of the bond order of adsorbed CO on account of the formation of metal-carbon bonds. The correlation is additional evidence of the role of d electrons in adsorption.

1. Introduction Infrared spectra have been used by investigatorsl-5) to characterize the species resulting from the adsorption of CO by transition metals. The colloidal metal particles required for suitable infrared transmission are supported on refractory oxide powdersI) or obtained by evaporation of the metal in CO atmospheres6) and by exploding wires in innert gas atmospheres’). Reflectance spectra of CO adsorbed on metal films has also been obtaineds). The spectra have been found to vary with the mode ofpreparation of the metal particles and the type of support. The disagreement in the transmittance and reflectance spectra of CO on Ni and Rh has been attributed to the difference in size of metallic aggregates in films relative to colloidal particlesc). Various structural effects observed in condensed filmsg), such as preferential development of certain crystal faces, may also affect the spectra of adsorbed CO. The spectral variations of CO on nickel on silica, alumina or titania have been attributed to differences in the electronic properties of the supporting oxides affecting the meta15). Spectral variations may also result from changes in the activity of the adsorbent caused by high temperatureslo), sintering of non-supported metal samples6) and residual gases6,ll). 229

230

C. R. CiUERRA

AND

J. H. SCHULMAN

The available spectra of adsorbed CO indicate certain trends in the position and intensity of bands independent of the methodof sample preparation. Some of thesevariations of the adsorbent

in the spectra are thought metals. This investigation

to depend on the electronic deals with the adsorption

nature of CO

by the metals iridium, osmium, rhodium, ruthenium, and rhenium supported on silica. The present results are compared with published spectra to examine the effect of metals with various electronic configurations on the spectra of adsorbed CO. 2. Experimental The technique for the preparation of the supported metal samples is similar to that used by previous investigatorss). High-area silica (Cabosil HS-5) was used as support. The silica powder was mixed with aqueous solutions of one of the following salts (NH,),IrCl,, (NH,),OsCl,, (NH,),RuCl,, NH,ReO,, 2 (NH,),RhCl, .3H,O and AuCl,. After drying the powder, a sample was compressed into a pellet at 3OOOpsi and placed in a glass cell with MgO windows connected to a high vacuum system”). The pellet was reduced with hydrogen at 375°C for about 20 hours and evacuated while hot at about 10e6 Torr for at least 1 hour. After cooling in vacuum to room temperature the sample was further evacuated to about low7 Torr and then allowed to react with CO. The silica (G. L. Cabot Co., Boston, Mass.) and the metal salts (K&K Laboratories, Plainview, N.Y.) are over 99% pure. Some of these salts which are readily soluble in water are used in the industrial production of these metalsla) by reduction with hydrogen. The gases (Matheson Co., East Rutherford, N. J.) are 99+% pure. Hydrogen was passed by a catalytic purifier (Baker Deoxo from Engelhard Industries, E. Newark, N.J). All the gases were dried by passage through traps cooled with liquid nitrogen. The infrared spectra were obtained with a modified Perkin Elmer 21 spectrophotometer with CaF, optics5). To minimize scattering, very thin samples were used containing 38.5mg of silica and 3.9 x 10-4cm3 of metal per 1 inch-diameter pellet. The slit width at 2000 cm- ’ was set at 0.204 mm. the metals The 2400-1200 cm- ’ spectra were recorded after equilibrating with CO for 5 minutes at pressures from IO- ’ to 15 cm Hg at room temperature, For the higher CO pressures, the spectra were also obtained after several hours of exposure. The spectra of only the gaseous phase, CO on pure silica and CO on the metal salts after partial reductions were also obtained. The samples used for the infrared measurements were also dispersed in collodium and formvar films and examined with the electron microscope (Siemens Elmiskop I). The gas adsorption by the samples was measured using a fixed volume gas adsorption apparatus.

ADSORPTION

OF CO GAS

BY METALS

231

3. Results 3.1. PHYSICAL CHARACTERISTICS OF THE SAMPLES basis ? ‘he metal-to-silicate ratio was standardized on a volume-to-weight to 2tpproximate the amounts used by other investigatorss). Thus each pellet con tained 38.5 mg of silica and 3.9 x 10m4 cm3 of metal. For these re :lative

Fig. 1.

Silica powder before the addition of metals in a collodium film.

Fig. 2. Silica powder supporting Re metal (54 % metal by weight). The small blat k dots ititute the metal phase. The large black areas result from an inhogomogeneous tIispersion of sample in the collodium film.

232

amounts

C. R. GUERRA

a calculation

assuming

AND

J. H. SCHULMAN

a homogeneous

dispersion

of metal atoms

on the silica surface indicates that the metal atoms would be separated by distances equivalent to 2 to 3 atomic diameters. The appearance of the metal particles dispersed on the silica powder was observed with the electron microscope. Fig. 1 shows the silica powder before the addition of a metal, the smaller silica particles appear to be 200 to 300A in diameter. Fig. 2 shows the silica powder containing 54 weight y0 Re metal which is 6 times the amount used for the infrared measurements. Using 9 weight y0 metal, the smaller particles remain unchanged but the large agglomerates become less numerous. The smaller metal particles are about 2OA in diameter and appear to form larger clusters by agglomeration. Moreover, at higher magnifications the smaller particles do not appear homogeneous but show regions of higher density about 8 A in diameter which would correspond to clusters of at least 9 atoms. Similar results were obtained with the other metals. The size of the clusters appears to depend on the concentration of metal salt per surface area of the silica dispersed in the solution for impregnation and other preparative conditions. Drying and heating with H, produced no appreciable sintering with the metals studied as evidenced by similar appearance under the microscope. However, heating of gold supported on silica was observed to produce particles approaching macroscopic dimensions which is attributed to sintering of the smaller original particles13). Other investigators have observed clusters ranging from about 65 A for Ni evaporated in CO atmospheresr”) and from 40 to 1008, for Ni particles obtained by reduction of Ni(OH), on silica powderIs). The surface area of the powders was measured using the adsorption of argon at liquid nitrogen temperaturesls). The total area of the powders before and after the addition of the metals was about 245m2/g. A small decrease in area was noticed after several reductions, as shown in fig. 3 for Rh where after two reductions the area decreased by about 2%, suggesting some sintering of the silica particlesr5). The equal areas of the powders with or without metals indicate that the metallic clusters observed with the microscope are not separated from the silica phase, what would tend to increase the surface area, but rather that they are on the surface of silica particles. From the relative amounts of metal and silica per sample, calculations indicate that a monoatomic layer of metal would cover 14% of the silica surface. CO

3.2. SPECTRAO~A~SORB~D 3.2. I. Silica and CO gas No infrared bands pressures. At higher

of CO adsorbed on pure Cabosil appear CO pressures for several hours, bands

at low gas appear at

ADSORPTION

3

Q Pure

OF CO GAS

Si 02

I A

0

Rh on SiOn,

1

lS+

Reduction

3

2

P/P, Fig. 3.

233

BY METALS

4

x IO

BET plot of the adsorption of argon at -195°C by samples used for infrared measurements. The weight of silica is the same in all samples.

2050-2000 cm-’ which must be attributed to adsorbed CO because they are absent when other gases are used instead of CO. Heat treating the silica with or without H, affects the position and intensity of these bands. Also, a gaseous species, absorbing weakly at 2060 cm- ‘, appeared when silica heat treated with H, was exposed to CO for several hours. The species was absent when silica was heated without H,. The infrared bands observed with CO on Cabosil without metals do not appear when the Cabosil is covered with metal particles, particularly when ammonium-chloro salts are used as the source of metals. Cabosil covered with Au obtained by reduction of AuCl, produced bands resembling those of CO on pure Cabosil but of weaker intensity. Since calculations show that even a monoatomic layer of metal covers only 14% of the silica surface in the samples, the absence of SiO,-CO bands must be attributed to surface effects caused by the decomposition of the ammonium in the salts. No bands resulted from CO on the metal salts before reduction. The SiO,-CO bands may be easily differentiated from metal-CO bands, because the latter are usually more intense at any pressure. 3.2.2. Rhodium Fig. 4 gives the spectra obtained for the adsorption of CO by Rh, the band frequencies are shown in table 1 of the appendix. The 2080 cm- ’ band is usually not clearly resolved from the 2065-2040 cm-’ band. Weaker bands attributed to SiO,-CO species and carbon-oxygen groups resulting from some decomposition of COr,t’) are not shown. Increasing the CO pressure

234

C. R. GUERRA

shifts the 2040-2065cm-’

J. H. SCHULMAN

band to higher frequencies.

ascribed to the heterogeneity among adsorbed moleculesr*) with adsorptionlg).

AND

This effect has been

of the metal surfacesr), lateral interactions and the change in the metal work function

The assignment

of the bands

by previous

investigators

using Rh films6) and alumina as support3) (see table 1) did not consider that for Rh-(CO), the 2080cm-’ band increases with time more than the 20201990cm-’ band. This effect possibly invalidates the assignment of this band to only Rh-(CO), species where both CO molecules are identically attached

0.15

w 0,1

T L

,

/ RHODIUM

---.14.575 cm 20.5 Hrs. 3 Hrs.

_-

z” ;;:

5 Min. 14.575 cm

4.365 0.4 cm 4X lO*cm

g 2 0.05 a

I

i

OL----2200

2000 FREQUENCY,

I800 cm-’

Fig. 4. Spectra of CO adsorbed by Rh metal. The full lines correspond to adsorption for 5 minutes at gas pressures given in cm Hg. The dashed lines show the variation with time.

to a Rh site. Species like Rh-(CO), have CzV symmetryzo) and the CO stretching fundamentals would result in two bands corresponding to the symmetric and the antisymmetric modes. Thus the change in intensity of one band relative to the other is difficult to explain. Measurements of the adsorption of H, and CO by Rh on silica may be used to determine the number of CO molecules per Rh site. Fig. 5 shows the volume of H, adsorbed by pure silica and Rh on silica at 0°C. Considering that the weight ratio of silica to Rh in the samples is about 8, pure sihca shows very little adsorption of H, as noted by other investigatorszr). Rhodium exposed to Ar and CO, reduced with H, at 375°C and evacuated, adsorbed less H, than during the jnitial exposure, probably because of sintering of the silica and annealing of the metal particles. Rhodium, treated with H,S for several hours reduced and evacuated, adsorbed less H, than before the treatment. Presumably, as observed with H,S on Niez), the H,S leaves sulfided sites where H, does not adsorb. The slopes of the lines in fig. 5 sug-

ADSORPTION

OF CO GAS

235

BY METALS

gest that the increase in adsorption with pressure is affected by the permeability of the pellet to the gas at the various pressures. Fig. 6 shows plots of CO adsorption by silica and Rh on silica at 0°C. Again considering the weight ratio of silica to Rh, silica adsorbs very little CO. Rhodium adsorbed more CO after several reductions with H, than during the initial exposure. This effect is less noticeable in the spectra but the 2080cm-’ band increases more than the 2065-2040cm-” after more than one reduction. The treatment of Rh with H,S decreased the volume adsorbed of CO although the spectra shows an increase of the 2080cm-’ band. INITIAL

1 I

AFTER

EXPOSURE

~~~

:

- ppC-’ AFTER

z-

H2S TREATMENT , xp __o/ d --

_o_ / - a().a< 0

/--

I 2

I

__-a-SiO2

J_--_l 4

1 6

PRESSURE,

Fig. 5.

WITHOUT I

I 8

Rh I

/ IO

I

I> 12

cm Hg

Volume of HZ adsorbed

at 0°C by Rh supported on silica. The weight of silica in all samples (3.8 g) is 8 times the weight of Rh.

The ratio of adsorbed CO to H may be obtained from figs. 5 and 6 by subtracting the values corresponding to pure silica from those of silica with Rh. For the initial exposure to the gases the CO/H ratio varies with pressure from 1.3 to 0.9. After several exposures the CO/H ratio is about 1.2 to 1.6. These values may be compared with CO/H = 0.5 obtained for films of Rh “3) at residual pressures of about 0.1 mmHg. The assu~nption that a residual pressure of 0.1 mm Hg of H, is enough for a monolayer coverage 24)has been questioned by some investigators 1%25). In this investigation the higher pressures were necessary to test the infrared results, although it was noticed that most of the adsorption occurred at residual pressures of a few mm of Hg. Assuming that at these pressures each metal site is occupied by one H atom from the dissociation of H,, the CO/H ratios indicate that the main infrared band is caused by the Iinear Rh-CO species and that Rh-(CO), sites are

236

C. R. GUERRA

AND J. H. SCHULMAN

more common than Rh,-CO sites. The relative intensities of the infrared bands corroborate these results since the band attributed to Rh,-CO species is much weaker than the Rh-CO AFTER

and Rh(CO),

bands.

SEVERAL

TREATMENT

IAL

0

4

’

/ 8

PRESSURE,

Fig. 6.

I

EXPOSURE

,

I2

I

I

,

1

16

cm Hg

Volume of CO adsorbed at 0°C by Rh supported on silica. The weight of silica in all samples (3.8 g) is 8 times the weight of Rh.

X-ray diffraction studies of metal carbonyls2s) indicate that the C-O bond length is about 1 to 1.2 A. The M-C bond length corresponds to half the metal atom diameter (2.7 A for Rh”7)) and the diameter of C (1.2 A2*)). Thus two CO molecules may fit on a surface site and edge and corner sites could accommodate more CO molecules. The variations in intensity of the 2080 and 2020_1990cm-’ bands may be explained, assuming that at some surface sites containing two CO molecules both CO molecules are not identically attached to the metal atom and may be designated M-(CO)(CO) species. The variations of the spectra lead to the speculation that the first CO molecule reaching a site is more strongly adsorbed (band at 20652040 cm- ‘) than the second CO molecule reaching this site (band at 2080 cm-‘). The fact that the 2080 cm- ’ band increases greatly with time, disappears in less than one minute of evacuation and reappears readily upon readsorption, corroborates this explanation. Also M-(CO)(CO) sites may held, change into M-{C0)2 sites, where the CO molecuIes are identically producing two infrared bands corresponding to the symmetric and antisymmetric modes. The relative position of these bands is comparable to the CO bands at 2051 and 1987cm- ’ of C,H,Rh(CO),2g) where the CO molecules have localized CZV symmetry.

ADSORPTION

OF CO GAS

237

BY METALS

3.2.3. Iridium

Fig. 7 gives the spectra for the adsorption of CO by Ir, the band frequencies are shown in table 1. The 203&2035cm- ’ band of Ir-CO shifts to higher frequencies with CO pressure. The 2080cmM1 band of Ir-(CO)(CO) increases with time and is readily eliminated by evacuation. Although Ir(CO),Cl with C,, symmetry has CO bands at 2079 and 2033cm- ’ 2% the evacuation tests of CO from Ir indicate that Ir-(CO), species are not appreciable. The 1993cm-1 band probably is the antisymmetric mode of the

I

I

I

1

I

IRIDIUM

---- 18.7 cm 20 Hrs. 2 Hrs.

i -5

Min 18.7 cm

:\

: ’ 1\

0.62 cm

0.05

0

l_

f 2200

2oDD FREOUENCY,

1800 cm-l

Fig. 7. Spectra of CO adsorbed by Ir metal. The full lines correspond to adsorption for 5 minutes at gas pressures given in cm Hg. The dashed lines show the variation with time.

Ir-(CO), species. For comparison, C,K,Ir(CO), with CO molecules of localized CZVsymmetry has CO bands at 2037 and 1957cm-r as). The band at 1910-1890cm-1 may be assigned to Ir,-CO species. 3.2.4. Osmium Fig. 8 gives the spectra for t.he CO adsorption by OS, the frequencies are shown in tabfe 1. The OS-CO band at 2012-2008cm-’ shifts to higher frequencies with CO pressure. The OS-(CO)(CO) band at 2080cm-’ increases with time, but evacuation leaves a weak band at 2060cm-’ which is probably due to residual OS-(CO),. The band at 1880-1840cm-’ may be attributed to OS,-CO species. The bands observed for CO on OS do not correspond with any of the bands reported for 0s,(CO),,a9).

238

C. R. GUERRA I

AND

3. H.

I

SCHULMAN

I

I

I

OSMIUM 0.15

---- 13.3 cm 22 t-h.

! lz g

o.l

-

0.05

-

g u

‘5 Min 13.3 cm 6.62 cm 0.5 cm 5X lO+cm

1

0

I

2200

I 2000 FREQUENCY,

I

1800 cm-’

Fig. 8. Spectra of CO adsorbed by OS metal. The full lines correspond to adsorption for 5 minutes at gas pressures given in cm Hg. The dashed line shows the variation with time.

Rlzenium Fig. gives the for the adsorption by Re. spectra were as reproducible as of the previous the band frequencies shown in 1. Evacuation indicate that 1950cmW1 band from Re,-CO and the and 2010-1992cm-r result from Re-(CO), and species. The of band heads be attributed surface heterogeneity 1). It might also be specuI

0.15

! /I -

5 Min.

“u 0.1 2 E :: : 0.05

0

I

2200

,

,

I

FREQUENCY,

I

1800

2000 cm-l

Fig, 9. Spectra of CO adsorbed by Re metal. The full lines correspond to adsorption for 5 minutes at gas pressures given in cm Hg. The dashed lines show the variation with time.

ADSORPTION

OF CO GAS

BY METALS

239

lated, on the basis of the structure of C,H,Re(CO), 29) and Re,(CO),,sQ), that isolated Re atoms and edge and corner sites can be shared by more than two CO molecules to account for some of the bands.

Fig. 10 gives the spectra for the CO adsorption by Ru. CO pressures of a few mm Hg moved the whole 2100-i200cm-1 to regions of lower transmittance resulting in several band heads. Evacuation tests indicate that the main bands at 2010-1990 and 1910-1870cm-’ result mostly from Ru-CO and Ru,-CO species. The other band heads may be attributed to surface heter-

__.‘-.) -

m a

5 Min 14.4 cm 5.48 cm OBcm 0.26 cm 4 x 10-5

0.05

I 2200 FREQUENCY, Fig.

cm-l

Spectra of CO adsorbed by Ru metal. The futi lines correspond to adsorption for 5 minutes at pressures given in cm Hg. The dashed line shows variation with time. 10.

ogeneityl) and to Ru sites with more than one CO molecule. For comparison ~C,H,RU(CO)~ with localized CzVsymmetry has bands at 1950, 1938, 1767 and 1764cm-’ apparently resulting from CO vibrations 2”). 3.2.7. Variatiom in the spectra The infrared bands of the M-CO species shift if the metals are subjected to repeated adsorption~vacuatio~ cycles with or without reduction with Hz. The shift is to higher frequencies for Rh and Re (dv<5cm-‘) and to lower frequencies for OS (dv
240

C. R. GUERRA

AND

J. H. SCHULMAN

tors1~s,r4~31) and are attributed to changes in the electronic structure of the metals. Structural changes in the metal particles affect also the spectra as noted with Au (see below). The variation

in intensity

of the M-CO

band is

shown in fig. 11. Most of the adsorption of CO occurs at pressures of a few mmHg and the band intensities become constant at pressures of a few cmHg. For Re and Ru, the M-CO band intensity is affected by contributions from other bands. Data for CO adsorbed on silica supported Au are also shown. CO on Au produces a single weak band at 2080-2070cm-’ assigned to Au-CO species. The weak band intensity is attributed to smaller 0.15

____-_ __ __~. ,._,_sh_._. -Tag:.-:‘-.‘: I .1’1171_ $,,_---I_-~ _ - - -& - -

I-I /’

F-

w 0.1

I’II 2 a i!, Lit? 0

__--

____--

lr

R”

_--

____-----

‘,

za 0.05

0

t

1

I

I

/

I 5

0 CO

Fig. 11.

-

I

-t---1-7

PRESSURE,

_---I IO

Au I

I

I

I

I I5

I

cm. Hg

Variation of the intensity of the M-CO band as a function of gas pressure for similar samples of metals on silica.

total surface

from the coalescence

of small metal particles

into larger ones

by heating. The Au-CO band has been observed more intense and at higher frequencies if the original metal particles are prevented from coalescingls). At high CO pressures and long times 4u covered silica produces also bands attributed to SiO,-CO species. 4. Correlation of the spectra with properties of the metals The vibrational analysisss) of the spectra is hampered because, for most metal-CO species, the necessary force constants are not known even for simple three-body (metal-carbon-oxygen) systems. However, for preliminary correlations, variations in the spectra of CO upon adsorption may be

ADSORPTION

compared

with metal properties.

OF CO GAS

241

BY METALS

Neglecting

vibration

interactions

adsorbent and adsorbate, either the frequency or the force constant M-CO may be used in correlations according to the equation w = (l/n> (F/u>+

between of CO in

(1)

where u is reduced F the calculated force constant. Variations in the force constant result from variations in the bonding of CO caused by bonding interactions with the adsorbent metal. Correlations of force constant versus the % d character of the metallic bondssb), % d character times the valencys5), work function of the metals6), and initial heat of CO adsorptions’) show no trends on account of the scattering of the values. However, as shown in fig. 12, a plot of force constant of CO calculated from the bands of M-CO species versus the number of d electrons required to fill the d orbitals of the metal atoms results in a linear pattern. The line slopes down from about the force constant corresponding to gaseous COss) to the force constant corresponding to a double bonded CO group33). I

/

I

I

I

I

CIO _-__-_

!

I8 -

v7-..$y;

E 5 0 m 0 16 5 2

cu Au Pd

_

z 0 ‘4 0 v c !z

I2

P

1st Long Row 2nd Long Row

_C:Q_.

0 3rd Long Row

I 0 NUMBER

Fig. 12.

. . . $---. _ -_--_ ._ ‘_Q‘; .____ u Pt .Y@..; a.‘,-? Rh Ir -- .--.. Ni co ps Re RU

I I

1

I

I

2

3

4

5

OF ELECTRONS REQUIRED d ORBITAL

TO FILL

of the force constant of adsorbed CO as a function of the number of electrons required to fill the d orbitals of the metal atoms.

Variation

The values shown for Rh (2065-2040cm-‘), Ir (2035-2030cm-1), OS (2012-2008cm-1), Ru (2010-1990cm-1), Re (2050-2030 and 2010-1992 cm-l) and Au (2080-2070cm-‘) are from this work. With Re there is the possibility that the higher frequency band results mostly from sites containing more than one CO molecule. The values obtained from published

242

spectra

C. R. GUERRA

for silica

supported

metals

AND

J. H. SCHULMAN

are for Cu (2100cm-1)1),

Pd (2060

cm-‘)I), Pt (2075cm-‘)I), Ni (2050cm-‘)5), Fe (2005cm-‘)a) and Co (2025cm-1)4). Other bands reported for Ni (2075cm-‘)l) and Co (2070)*) are attributed to weakly held CO 5, and probably correspond to M-(CO)(CO) of M-(CO), species. Other band (1960cm-‘)l) has been reported for Fe although some difficulties in the reduction process were experienced. Among the metals with more than 5 electrons missing from the d shells, data for MO have been reported’). Unfortunately only the high frequency side of the band is given (2070cm-‘) but it appears to follow the trend of the other metals if electron pairing is taken into consideration. The dependance of force constant on bond order may be obtained from the empirical relation between force constant and bond length (Badger’s rule) as) and Pauling’s equation for bond order 39) : R=a-blogF,

(2)

R=R,-clogn,

(3)

In F = k + (c/b) In n ,

(4)

where R is the bond length, R, the single bond length, n the bond order and a, 6, c and k constants. Accordingly a log-log plot of F versus n should give a straight line of slope (c/b). A log-log plot of the data in fig. 12 diminishes the scattering and produces a smooth line. No additional information may be presently derived, because of disagreement about the bond order of CO and the slopes obtained from force constants (of CO and H,CO)sa) or from tabulations of the constant b in eq. (2)40). Variations in the force constant represent variations in the strength or bond order of CO. The correlation in fig. 12 suggests that the bond order of CO varies upon adsorption in a manner approximately proportional to the number of missing d electrons in the metal atoms. Some of the scattering of the data appears to decrease, when approximate force constants for metal carbides are used in 3-atom models, i.e. M-C-O. Perturbation of an atom by the surrounding ligands would also affect the electronic population in the d orbitals. Thus, a better correlation would be of force constant of adsorbed CO versus the d holes in the metal phase. Unfortunately, there are very little experimental data on the d holes in transition metals. Fig. 13 is a plot of force constant corresponding to the M-CO band versus the number of d holes in Fe, Co, Ni, Cu, Pt and Pd in Bohr magnetons41, 42). Fig. 13 may be approximately compared with fig. 12 where each electron corresponds to a magnetic moment of + Bohr magneton”“). The plot in fig. 13 again indicates a correlation between bond order of CO and number of d holes in the metal.

ADSORPTION

OF CO GAS

The bond order of CO decreases d holes of the metal on account

in a manner

Fig. 13.

proportional

of the formation

cu Q

243

BY METALS

to the number

of metal-carbon

of

bonds44).

Pt

Variation of the force constant of adsorbed CO as a function of the number of d holes in the metal.

Acknowledgements The authors wish to financial support of this discussions with Dr. D. also indebted to Mr. L.

thank work. J. C. Schulz

the International Nickel Company for the The suggestions of Dr. C. E. O’Neill and the Yates are gratefully acknowledged. We are for the electron micrographs. Appendix

THE EFFECTOF O,, AND NH, AND H,S ON THEADSORPTIONOF CO BY METALS

Eflect of 0, Treatment

I consisted

of removing

CO from a metal by evacuation

and

admitting 0, to react at 25°C with the residual CO on the surface. The residual Rh-CO, Ir-CO and OS-CO bands were quickly eliminated by O,, but the Re-CO and Ru-CO bands remained constant even under high 0, pressures. Treatment II consisted of reacting with 0, at 25°C residual CO on a metal surface evacuating and readmitting CO to observe its adsorption. The main CO bands then appeared at 2098-2094,2064-2060 and 2027-2020 cm-l for Rh, at 2070-2055,2042-2040 and 2020cm-1 for Ir and at 2028-

244

C.R.GUERRA

ANDJ.H.SCHULMAN TABLET

Infrared bands of CO adsorbed by metals on silica

(a)

(b)

Metal

Admittance of CO Frequency (cm-r)

Evacuation of CO Frequency (cm-l)

Rh

2080 2065-2040 2020-1990 1900-l 890

Ir

OS

Re

Ru

2080 2035-2030 1993 1910-1890

2040-2035 (s)cc) 2018-1990(w) 1890 (w)

2020 1990 1910

(s) (w) (w)

2080 2012-2008 1880-1840

2060 (w) 2006-1990 (s) 1890 (w)

2100-2090 2050-2030 2010-1992 1950 1890

2100-2090(w) 2040 (s)

2010-1990 1910

2010-1990(w) 1910-1870 (s)

1950 1890

(s) (w)

Assignment

Rh-(CO)z, Rh-(CO)(CO) Rh-CO Rh-(C0)2(“) Rhz-CO Ir-(CO)a, Ir-(CO)(CO) Ir-CO lr-CO2(“) Ira-CO OS-(CO)z,

OS-(CO)(CO)

OS-CO(d) osz-co

Re-(CO)z? Re-CO? RepCO

Ru-CO? Ruz-CO?

(a) (b) (c) (d)

CO Pressures of 0.1 Evacuation for over Strong (s) and weak Partial contribution

to 15 cm Hg. 1 hour at about 10-j Torr. (w) intensities. of SiOa-CO species.

2018 for OS. Treatment III consisted surface with 0, at 25°C evacuating

of oxidizing a freshly prepared metal and admitting CO to observe its ad-

sorption. Treatment III produced bands similar to those after Treatment II for CO on Rh, Ir (2050-2045cm-‘) and OS (2020-2014cm-‘). The results are shown in tables 2a, 3a and 4a, where the dashed lines A and B and C or M-(CO)(CO) species of untreated correspond to M-CO and M-(CO), metals. It is evident that 0, tends to suppress the band resulting from two CO molecules per metal site and to shift the M-CO band to higher frequencies. Treatment 11 of Re caused the 2010-1992cm-’ band to split and all the bands to decrease in intensity as also noticed with Ru. Treatment III of Ru decreased the intensity of bands and completely eliminated the CO bands of Re.

ADSORPTION

245

OF CO GAS BY METALS

TABLE

2

Efffect of surface treatments on the bands of CO on Rh. The height and width of symbols A and _L indicate the intensity and position of the bands. The superpostion of bands is indicated by A. The dashed lines show the position of the bands before treatments. n._-

I

02

Treat. D: --__

A

i

I

_L-

J_

h

;

I

A.

8

Ci

Ii:’

i

/

Press.

J_ Treotz

CO

A

<

A

> IO cm

10M6cm

-___Time > 5Min.

_-(evocuotion)

/ /

>

I Hr.

/

;;a

AL;_+ >IOcm

5 Min.

/ A

ii,

/x1.;&

A':

J.3

i

6.-d3

-.I_

1

A Treat.

m

;

~b

A

/ I

4

I Treot. .-II /h

lcm

5 Min. 5 Min.

< toe5 cm Press.

Time > 5 Min.

LrL--!-_.l0

> IOcm

>

/ ’ /L-L--e

> tOcm

5 Min.

I - IOcm

5 Min.

,5-J___-

O.l- Icm

5 Min

t

< iO-5cm

5 Min.

/

h

CO

< Id6cm

0

&,V

,a,

:

‘Z%-A-kz

AL:, /-. >~ ‘I I\ Treat:

I

A,

B

i I

Trea’t. II

Press.

CO

I Hr.

Time

C.

< Id”cm

> 5 Min.

LIIETL_

>lOcm

> I Hr

_-

iI!% ’ >’

:

1

:

/ /

>IOcm

5 Min.

-

I-IOcm

5 Min.

-

O.l-lcm

5 Min.

c fOe5cm I

5 Min.

d

‘/

it I

(evacuation)

c;

Lii,

:.- Hz S

o.l-

I

I

I

5 Min.

I -iOcm

j/

I

\._

2 100 2050 FREQUENCY,

I

A

/5 I 2000

cm.-’

(evacuation)

C. R. CUERRA

246

AND

TABLE

J. H. SCHULMAN

3

of surface treatments on the bands of CO on Ir. The height and width of symbols A and 1 indicate the intensity and position of the bands. The superposition of bands is indicated by A. The dashed lines show the position of the bands before treatments.

Effect

a.-

0:

I !2s_s~cO~_~_ _?%

I

,y

Treot.lI

_L

Treot.‘m __-__

b-L/.

.,;i

B

A

CT”

-c 10-6cnl

>

> I

5Min

)A

Lh :

_

:*

>!Ocm

IA

/ 68

,

‘P

>iOcm

5

%

I

5 Min

(A

I-1

..

-IOcm

(evacuation)

Hr

Min.

/ h

‘\

A Treat. ~ ~__

~___~

0

O.l-

b

<10w5cm

h !

I

A

~~
Press.

Al

I

A

*:

> IOcm

> I

A :

>

:L

4

I - IOcm

5 Min.

G I

A.

O.l-

5 Min.

Ax-2 :.-

H;S

,-, Treat.

I I

Time

^c

Al

I 1

CO

> 5 Min.

,-L

/I

u

5 Min. 5 Min.

< IO+cm

*

8

Icm

I

Treat.

I

II

m

m

e

A I”’

b

:d_

m

I Z%

,=

4

m’

:-._L

4

IOcm

Hr.

5 Min.

Icm

< 10b5cm T--Press. CO

5 Min. Time

< 10-6cm

> 5 Min.

> IO cm

> I Hr.

C /

>

5 Min.

IOcm

I-IOcm

5 Min.

O.l-lcm

5tvlin.

/ \h

A+

\_

I

1

2100

I zobo

2050 FREQUENCY,

A

cm-’

<

(evacuation)

IO-“cm

I

5 Min.

~evocuotion)

ADSORPTION OF CO GAS BY METALS

247

TABLE 4 Effect of surface treatments A and 1 indicate indicated I I.-

02

by 4.

on the bands of CO on OS. The height and width of symbols

the intensity

and position

of the bands. The superposition

The dashed lines show the position 1

A Treat.

II

A ,’

Press.

.+L ,

I

II

I

7 5 Min

A

>

IO cm

>

*

> I-

.- NH3

A Treat.

I

I

5 Min

I cm

5 Min

,?A

/CL

/

Press. CO

Time

<

IO&m

> 5 Min

>

IO cm

>

>

IO cm

II

&_

:_I_

LiL

I-IO

LfYl

\~ LL

I

A Treot. I

1

Treot.

I

k

0.1-l cm

5 Min

IO-Sm

m

I IPA

Time

lO3m

7 5 Min

IO cm

7

<

7

‘>

Jl-,

\/ &

I 2050

2000 cm-’

IOcm

I Hr 5 Min

I-IO

cm

5 Min

0.1-l

cm

5 Min

<

I FREQUENCY,

5 Min

Press. CO

A7

jtb

5 Min

5 hlin

II A

L’\

I Hr

cm

<

(evacuation)

I

A,,

8

5 Min

I

Treat. A,,’

II

2100

IO cm

I

B

I .- HZS

5 Min

G5cm

(evacuation)

IHr

IO cm

O.l-

Q,

‘\ ----<

/

Time

IO-?m

L

N’i

CO

<

,*

171:

I

of bands is

treatments.

I

I

Tre0t.m

I

of the bands before

IO-lm

1

5 Min

(evacuation)

C. R. GUERRA

248

AND J. H. SCHULMAN

Effect of NH, Treatment I consisted of exposing a fresh metal surface to NH3 at 25 “C, evacuating, admitting CO and observing its adsorption. The main CO bands appeared at 2084-2080, 2070-2060 and 2020-2018cm-’ for Rh, 2048-2037 and 2000cm- ’ for Ir and about 2012cm- ’ for OS. Treatment 11consisted of reducing with H, at 375°C a fresh metal surface exposed to NH, at 25”C, evacuating, admitting CO and observing its adsorption. Treatment II produced bands similar to those after Treatment I for CO on Rh (2080, 20702040, 2020-1990cm-‘), Ir (2080, 2035-2030, 2000cm-‘) and OS (2080, 2008cm-‘). The results are shown in tables 26, 3b and 4b. Without a subsequent reduction step, NH, affects the metals as O,, but a reduction with H, eliminates the effect of NH, and causes band shifts to lower frequencies. Treatment I eliminated the CO bands on Re while Treatment I1 decreased the intensity slightly, Both treatments decreased the band intensities of CO on Ru. Effect of H,S

Treatments 1 and II with H,S were similar to those with NH,. Both treatments produced similar bands with Rh (2086-2080, 2070-2058, 20201990cm-l) and eliminated the Rh,-CO band (1900-1890cm-‘). Treatment II produced bands of greater intensity than before the treatment. The bands of CO on Ir (2080, 2030-2026, 1993cm-‘) and OS (2080, 2008-1995cm-‘) showed lesser intensities than before the treatments. The results are shown in tables 2c, 3c and 4c. Treatment I almost eliminated the CO bands on Re while Treatment II decreased the intensity slightly. Both treatments decreased the band intensities of CO on Ru. References 1) R. P. E&hens and W. A. Pliskin, Advan. Catalysis 10 (1958) 1.

2) 3) 4) 5)

N. N. Kavtaradze, Kinetica i Kataliz 3 (1961) 378. A. C. Yang and C. W. Garland, J. Phys. Chem. 61 (1957) 1504. J. S. Cho and J. H. Schulman, Surface Sci. 2 (1964) 245. C. E. O’Neill and D. J. C. Yates, J. Phys. Chem. 65 (1961) 901.

6) 7) 8) 9)

C. W. Garland et al., J. Phys. Chem. 69 (1965) 1188. C. P. Nash and R. P. De Sieno, J. Phys. Chem. 69 (1965) 2139. H. L. Pickering and H. C. Eckstrom, J. Phys. Chem. 63 (1959) 512. D. W. Pashley, in: Thin Films, Papers presented at the Seminar of the ASM, Oct. 19-20 (1963). G. C. Bond, Catalysis by Merals (Academic Press, New York, 1962) p. 471. R. Suhrmann, Advan. Catalysis 7 (1955) 303. C. A. Hempel, Rare Metals Handbook, 2nd Ed. (Reinhold Pub. Co., London, 1961). D. J. C. Yates, Esso Research and Engineering, private communication. C. W. Garland et al., J. Phys. Chem. 69 (1965) 1195.

10) 11) 12) 13) 14)

ADSORPTION

OF CO GAS

BY METALS

249

15) G. C. A. Schuit and L. L. Van Reijen, Advan. Catalysis 10 (1958) 242. 16) S. Brunauer, P. H. Emmet and E. Teller, J. Am. Chem. Sot. 60 (1938) 309. 17) L. J. Bellamy, The InfraredSpectra of Complex Molecules, 2nd Ed. (Methuen, London, 1958). 18) D. 0. Hayward and B. M. W. Trapnell, Chemisorption, 2nd Ed. (Butterworths, Washington, 1964) p. 221. 19) J. H. De Boer, in: Chemisorption, Ed. W. E. Garner (Academic Press, New York, 1957) p. 27. 20) G. Herzberg, Molecular Spectra and Molecular Structure, Vol. II (D. Van Nostrand, New York 1964) p. 106. 21) D. A. Dowden, Discussion of paper by G. C. A. Schuit et al., in: Chemisorption, Ed. W. E. Garner (Academic Press, New York, 1957) p. 55. 22) C. E. O’Neill, Ph. D. Thesis, Columbia Univ., New York (1961). 23) M. A. H. Lanyon and B. M. W. Trapnell, Proc. Roy. Sot. (London) A 227 (1955) 387. 24) 0. Beeck, Advan. Catalysis 2 (1950) 151. 25) D. F. Klemperer and F. S. Stone, Proc. Roy. Sot. (London) A 243 (1957) 375. 26) E. R. Corey and L. F. Dahl, Inorg. Chem. 1 (1962) 521. 27) E. A. Moelwyn-Hughes, Physical Chemistry, 2nd Ed. (Pergamon Press, London, 1961) p. 25. 28) L. Pauling, The Nature of the Chemical Bond, 3rd Ed. (Cornell Univ. Press, New York, 1960) p. 224. 29) H. P. Fritz and E. F. Paulus, Z. Naturforsch. 18b (1963) 435. 30) N. Flitcroft et al., Inorg. Chem. 3 (1964) 1123. 31) C. E. O’Neill and D. J. C. Yates, Spectrochim. Acta 17 (1961) 953. 32) R. F. Wallis, Surface Sci. 2 (1964) 146. 33) G. R. Somayajulu, J. Chem. Phys. 28 (1958) 814. 34) 0. Beeck, Discussions Faraday Sot. 8 (1950) 118. 35) M. Baker and G. I. Jenkins, Advan. Catalysis 7 (1955) 1. 36) Handbook of Chemistry andPhysics, 43rd Ed. (Chemical Rubber Publ. Co., Cleveland, 1961) p. 2594. 37) G. C. Bond, Catalysis by Metals (Academic Press, New York, 1962) p. 77. 38) R. M. Badger, J. Chem. Phys. 2 (1943) 128; 3 (1935) 710. 39) L. Pauling, The Nature of the Chemical Bond, 3rd Ed. (Cornell Univ. Press, New York, 1960) pp. 239, 255. 40) D. R. Herschbach and V. W. Laurie, J. Chem. Phys. 35 (1961) 458; Lawrence Radiation Lab. Report 9694, Univ of California (1961). 41) L. Pauling, The Nature of the ChemicalBond, 3rd Ed. (Cornell Univ. Press, New York, 1960) p. 397. 42) E. 0. Wollan, Oak Ridge Nat. Lab., private communication. 43) C. A. Coulson, Valence, 2nd Ed. (Oxford Univ. Press, 1961) p. 283. 44) D. 0. Hayward and B. M. W. Trapnell, Chemisorption, 2nd Ed. (Butterworths, Washington, 1964) p. 210.