2772 GEOHGERLY1iOLI)ER
J. H. DE 13oen ( T h e Hague, Netherlands). I quite agree with parent from the low temperature adsorption rliat the (:overage
Dr. Cxanderna when he says that silver oxide cannot be in:& of the surface with oxygen atoins is r:tt,e-liriiited. This is most
from silver layers and oxygen by a thorrnal mct,hod. 1Iy own probably t)ccaause of a limited niirrit)cr of primary :dsorption
experience hefore World War 11, if I remember wcll, is that a sites. I t is riot possit)le ti, :wcwunt for the low c.over:igo (e =
glow tiischarge in dilute oxygen oxidixos ii silver layer to Ag,O, 0.45) :it low ternper:ttrires iinless at 1 ~ : ~ sthree
t silver iitoiiis i ~ r o
while a t rcatment. with o m i e results in thr forination o f higher bound to one oxygen :ttoni. In fact., if :ill (he silver :iloriiu of the
silver oxides, proh:ihly :I iiiixt i i r o of Ag20:iritl Ago. Yiirfm’f: are assurrieti to I)(, c.hse-packcd iref. 20 of the paper),
uriiforrii coverage wilh :itom ions a t 6 = 0.45 (I>ig.10 -12j would
R. A. VAS KORDSTRANI) (Siwlair Ilescw-oh, 1Ioust.on). 0 1 1 correspond to a t.rue 6 of 0.33 which would be the Ag.iO that. hiis
what grounds did you dec,itic thrit t h r oxygctri prders t o :dsort) bectn proposed. JVliile t.his quantit:it ivr agrccrtictrit iu:iy be
onto a tetraticdral site? fort~nitous,tlie :issurript,ions iruide for the c.:tlcul:itiori d o not :ip-
pear to t)c wildly ype(.iil:ii ive \)evriuseof the way the reprotlric~it)lc
siirf:ic~t WBY prepared :tii(l the known repulsions of srirf:wc: ions.
Finally, oxygen stonis in A g r O oc~cupy:t posii.ion rqiiivaleni. to ihci
tc?tr:ilicdr:iI hole of t h r w c,low-pai,ltcd silver :Ltoiiis, whirli niit:Lns
thiit this nirist, b e :t prcfcrred :trraiigenient. ‘I’ht “Ag.~O“ thus
could rilso rc3preseiit the siniplcst nuc*leris for the growth of hgrO
on a silver surface when oxygen prcwures a r e re:iched bhat. make
this thcrrrioti~narnic~:illyfc::tailile.
Molecular Orbital View of Chemisorbed Carbon Monoxide
by George Blyholder
Department of Chemistry, University of Arkansas, Fnwlteville, Arkansas (Received February ?2G,196‘4)
Carbon monoxide chemisorbed on metals is considered in the light of I-Iiickel rnolccular
orbitals for the met,al-carbon-oxygen bonds. The Huckel molecular orbitals predict the
existence of a partially filled wnolecular orbital which increases the metal-carbon bond
strength but decrcascs the carbon-*xygen bond strength. I n this model only effects in the
r-bonding system are considered; ie., the u-bonds are assumed constant. This model
qualitatively explains. (1) the occurrence of several carbon-oxygen stretching frequencies
in the infrared spectra of CO adsorbed on metals; ( 2 ) differences in spectra of CO adsorbed
on evaporated and supported metals; (3) infrared band positions as a function of coverage;
(4) the effect of adsorbing other gases in addition to previously chemisorbed CO; and
( 5 ) band shifts in going from adsorption on pure Xi to a Ni-Cu alloy.
Introduction are varied, the observed infrarcd bands shift their
Recently, there has been accumulating a body of positions. For example, it has been observed that
literature dealing with the infrarcd spectra of carbon (1) the carbon-oxygen stretching frequency con-
monoxide adsorbed on metals. The carbon-oxygen sistently shifts to higher wave numbers as the fractional
stretching frequency for cheniisorbed carbon monoxide surface coverage (2) the relative intensi-
has been observed from above 2100 em.-’ down to ties of bands for carbon monoxide on evaporated
1800 cm.-l. As a variety of adsorption parameters and supported films are different6.6; (3) adsorption of
ORBITALT‘IEW
MOLECULAR OF CHEMISORBED
CARBOSNONOXIUE 27i3
other gases after carbon monoxide shifts the observed this c-bond puts a large formal negative charge on the
band positions.’ Thesc and other phcnomena havc central metal atom it is usually stated that bavk dona-
receivcd scant explanation and certainly no over-all tion from a metal d-orbital to the antibonding K*-
view of the nature of the metal-carbon-oxygen bonding molecular orbital oii the carbon monoxide ligand occurs
system has been presented and used to attempt to to reinow this excess ncgativc~charge arid stabilize
explain or corrclate these phenonicna. the bond. This htatciiient about the 7-bonding be-
The principal structural conclusion drawn, first by tween the metal and the ligand serins a bit inislcadirig
Eischcns and co-workers2 and subsequently by other in that at first glance it iniplicxs thc placement of elcc-
workers, from the observed spectra in the 5000 to 1400- trons in a high energy antibonding orbital. IIo\vc’ver,
cni.-l region was that bands above 2000 em.-’ repre- formation of siniplc IIuckel molecular orbitals indicates
sentcd a linear metal-carbon-oxygen system, whilc that thc niolecular orbitals into which these elcctrons
bands bclow about 1!)60 em.-’ werc due to a bridge go are lowcr in criergy than thc metal d-orbitals.
structure in which the carbon atom is bonded to two The a-molecular 01 Litals arf’ foriiied hy combining
nickel atoms on the surface. These conclusions werc a metal d-orbital labelcd XI,carbon p-orbital labeled
based on a comparison to the spcctra of a few carbonyls X z , and an oxygen p-orbital 1at)clcd X ? . Follon~ing
which give bands abovc 2000 em.-’ for linear mctal- suggestions in St rcitwies(irli based on the clcctro-
carbon-oxygcn groups while Fez(CO)9, which contains ncgativities of the atonis involved, the integrals arc’
bridgc carbonyl groups, gives a band7 at 1828 em.-’. assigned values fill = a - 6,IZt2 = a , I l l ? = cy p, +
ITowever, Cotton and Kraihanze14pYhavc shown that HI2 = p, fZ23 = p. Solution of the secular q u a t i o n
in substituted carbonyls a linear mctal-carbon oxygen gives energies Itl = a +
1.78, IC2 = cy, and It3 = cy -
system can havc carbon-oxygen stretching frequencies 1.78 corresponding to wavc functions t,bl = 0 21X1
down to 1800 cm. -l. Thus the original correlation + 0.58X2 +
0.78X3, $2 = 0.38X1 +
0.58Xt - 0.38X3,
suggesting the bridgt. structure is not entirely valid. arid $3 = 0.7fiXL - 0.XiXP +
0.20X3. Since CY and
Recently, work in this laboratory has shown6 that for 6arc both negative nunibers, El < E, < ICd. The wavc
carbon monoxide chemisorbed on nickel the infrared function for the lowest orbital places most of thc charge
spectrum over the entire range from 5000 to 300 cm.-l on the carbon and oxygen atoiris and adds to thc bond
does not support the bridge structure. If it is assumed, strengths for both thc carbon-oxygen and metal-
as we do, that all of the carbon monoxide is adsorbed carbon bonds. IIowcver, the sccond orbital, whose
in the lincar structure, the problem of explaining the energy indicattxs that it is lower in energy than a metal
various carbon-oxygen stretching frequencies for the d-orbital, is bonding for the nictal-carbon bond but
one structure remains. antibonding for the carbon- oxygen bold, since the
It is proposed that a simple molecular orbital view wave function has a node bctwccn thc carbon and oxy-
of the nature of the ]rictal-carbon-oxygeri bond will
go a long way in at least qualitativcly explaining and (1) R. P. ISischens and R. A. l’liskin, A-lxnn. Catalysis, 10, 1
corrclating the observed band positions and the few (1958).
observations that h a w been made on intensities. hens, S. 4.Francis, and W. A. Pliskin, J . Pli?/s. Chrm.,
6 0 , 194 (11156).
Theory ( 3 ) C . E. O’Neill and D. J. C . Yates, it~id.,65,901 (1961).
(4) C. W.Garland, ibid., 63, 1423 (1959).
Thc general molecular orbital approach to bonding ( 5 ) 1,. n. Neff, S4.S. Thesis, University of Arkansas, 1902.
in carbonyls as given by Orgcl, Ballhausen, l 2 and (6) G. Rlyholder, I’iiper 1-38, I’roceedirlgs of the Third 1nt)ernational
Richardson13 is adopted and herein adapted to chemi- Congress on Catalysis. Amsterdam, 1964, X o r t l ~ - I l o l l : dI’uhlishirlg
Co., Aniuterdnm.
sorbed carbon monoxide. The bonding in an isolated
(7) R. K. Sheline and K. S. Pitxer. J . A m . C h r m . Soc., 72, 1107
carbon nionoxidc molecule is regarded14 as rcsulting ( 1950).
from an sp,-hybrid orbital of the carbon atom combin- (8) F. A. Cotton and C. S. ,Kraihanzel, ibid.. 84, 4432 (1962).
ing with thc p,-orbital of the oxygen to produce a u- (9) C. S. Kraihanzel and F. A. Cotton, Inorg. Chem., 2, 533 (1963).
bond while the pT- and p,-orbitals of the carbon and (10) L. E. Orgel. “Transition SIetal Chrinistry.” 1Iethnrn m d ( - 0 ,
Ltd., London, 1950.
oxygen atonis combine to produec two a-bonds.
(11) L. E. Orgel, I r m r g . Chrm.. 1, 25 (1962).
This leaves a lone pair of electrons on the oxygen 2s- (12) C. J. Ballhausen. “Introrluctioll to Ligand 1:ield Theor,v,”
orbital and a lone pair in a carbon sp,-hybrid orbital, McQraw-Hill Rook Co., Inc., New York. N.Y.,1962.
which can form a coordinate bond in a complex with (13) J. R. Iiirhnrdson. “Orgnnornetallic (:heniistry.” 11. Zeiss, 1,:d..
Iieinhold I’ul)lishing Corp.. New York, N . Y., 1960.
a suitable acceptor orbital such as a d-orbital on a
(14) C . A. Coulron. “Valence.” Oxford, Piew \.ark. N. Y.,1952.
metal atom. This forms a a-bond between the carbon
(15) A. Streitwieser, J r . , “SIolecular Orbital Theory ior Organic
atom and the mctal atom. Since the formation of only Chernista.” John Wile?. and Sons, I n c . , N e w Y o r k , N. IT.,l9Gl
2774 GEORGE
BLYHOLDER
gen a t o m . h count of thc electrons available indi- of the rcmaining carbon monoxide ligands is increased
cates that this second n-inolecular orbital would not with the result that the carbon-oxygen strctchirig
be filled. In making this count, for example, for frequencics are dccreased. They observed frequericics
Cr(CO)o, it is assumed that each ligand forms a low down to 1800 cm.-'.
encrgy u-bonding orbital which is filled and a u*- It might also be expccted that as the electronic
antibonding orbital which is empty. To aid in count- charge available to go into orbital jL2 is increascd, the
ing, the electrons may bc assumed distinguishable and extinction coefficient for the carbon-oxygen stretching
the electrons originally in the carbon lone-pair orbital band would change. Jones16 has shown that as the
are placed in the u-bond. The n-orbitals labeled extcnt of n-bonding increases, the extinction coefficient
$1, of which thcrc are two for each ligand, one in the greatly increases for the carbon-nitrogen stretching
z-plane and one in the y-plane if the z-axis is the bonding band in a series of cyanide coniplexes which have
axis, are all just' filled by thc electrons originally in the bonding very similar to that in carbonyl compounds.
n-orbitals of the carbon nionoxide ligands. This In applying the above considerations to chcmi-
leavcs the six electrons originally in the outer shell sorption problems, a mctal atom on the surface is re-
of thc chromium atom to be spread out in the tn7elve gardcd as the central atom in a complex with the sur-
$2-orbit'als. Thus there is coinpetition among the rounding metal atoms and the chemisorbed molecule
ligands for cilectrons to put in thcse orbitals. as ligands. Here we do not have thc high symmetry of
While this picture is greatly oversimplified it should the unsubstituted carbonyls. For the transition
be qualitatively correct and will be shown to explain metals, the metal ligand atoms partially surrounding
a number of observations. Here the picture has been a particular surface metal atom will have available
presented in terms of molecular orbitals cxtendirig only partially filled d-orbitals with appropriate syniinetry
over the metal and one ligand rather than over the to form n-bonds with thc chosen surface atom and
cntirc: niolecule as should be done for a symmetric thereby compete with the adsorbed carbon monoxide
case like Cr(C0)6 because for the case of chemisorbed for electrons from the adsorbent atom. The extent of
carbon monoxide, the octahedral symmetry is lost and the conipetition of the surrounding metal atoms for
only one carbon monoxide ligand will be dealt with. Fur- electrons which could go into a & n-molecular orbital
t'hcr the u- and n-bonds are assumed separable and all of the metal--carbon-oxygen system will determine the
effects will be blanxd entirely on the n-bonding. This frequency and extinction cocfficicnt for the carbon-
mctans the problcins to be treated will bc limited to oxygen stretching band of thc adsorbcd moleculc.
cascs whcrt! one might rcasonably expect' the u-bonding As an adsorbent atom occupies a position in diffcrent
to rcniain approxiinately constant as an effect of a crystal faces or a position at an edge, corner, or dis-
variablc is observed. location, thc number of surrounding inctal atoms and
Onc niight, wonder whether thcic is sufficient nietal- hence the conipctition for elcctrons will vary.
carbon orbital overlap to produce orbitals like the
Iliickel orbitals calculated here. E'irst, it is noted Applications
that evcn as the resonance intcgral which should 1 . Bands for CO on Ni. A Si adsorbent atom in a
be proportional to the overlap, is greatly dccreased, 111 plane face or a 100 plane face has nine or eight
the qualitative pictui,ct reniains unchanged and the surrounding Si atoms, respectively. With this large
energy of jL2 reniains below that of an electron in a number of ligands competing for electrons, the amount
metal d-orbital. .I second and perhaps more con- of electronic charge in the $2 molecular orbitals of an
vincing argument stems from a considcration of the adsorbed CO nioleculc would be expected to be small
o1)scrvcd carbon -oxygen stretching frequencies. If with thc result that the C-0 stretching frequency would
oidy a u-bond is fornicd by the lonc-pair elcct'rons on bc above 2000 c ~ n . -as~ in the unsubstituted carbonyls.
t h o cartmn atom, the carbon-oxygen stretching frc- However, for adsorbcrit atonis located at edges of
qucncy for carbonyls would be expected to be vcry planes, corncrs, and at dislocations, the nurnbcr of
closct to that of free carbon monoxidc. In fact, this surrounding nickel atoms can be around four, five,
frequency is shiftcd to soincwhat lower values. This or six. On thcse sitcs the competition for electrons to
is taken as an indication that> a-bonding does occur go into the $2 molecular orbitals of the inctal-adsorbate
with soiiic charge occupying an orbital like $2 which system is materially reduced with the result that thc
weakens thc carbon--oxygen bond. Cotton and Krai- C-0 stretching band will occur a t lower wave numbers.
hanz('lS,"have shown that as carbon monoxide in hexa- By analogy to the substituted carbonyls, the C-0
cartmiyls is rcplacted with ligands which do not x -
h n d , tho availability of electrons to go into n-orbitals
RfOLECULAR ORBITAL VIEW OF CHEMISORBED CARBON ~ f O K O X I D E 2775
stretching frequency for these sites would be expected graph studies of deBoer and Coenen17 reported by
to be below 2000 cm.--Land perhaps down to 1800 cm.-I. Eischens. The adsorption sites on our silica-sup-
The spectra for CO on Ni usually show a band or bands ported Ni should mostly be of the type to produce the
around 2060 cm.-l and a broad band or bands around high-frequency band. The evaporated metal is con-
1940 cm.-l. According to this molecular orbital densed in the vapor phase and the particles immediately
view, the bands above 2000 cm.-' are regarded as due quenched in oil. This would be expected to result in
to CO adsorbed on regular crystal fares where the a more amorphous material with less well-crystallized
adsorbent Xi atoms have a high coordination number particles. These particles should then have a larger
while the bands around 1900 em.-' represent CO ad- proportion of sites which produce the low-frequency
sorbed on sites where the Ni atonis have a somewhat band. Thus this theory explains some of the differ-
lower coordination number. The low coordination ences we have observed in our samples.
number sites would be expected to t w somewhat more A number of other reported spectra will now be com-
variable and poorly defined by virtue of the extra mented on in the light of the arguments herein, but
freedom granted by the low coordination number the conclusions are rather less clear. Eischens and
when compared to the high coordination number sites. workers2Js have reported spectra for CO chemisorbed
For this reason, it is expected and found that the lorn on silica-supported nickel in some of which the low-
frequency bands are somewhat broader and more vari- frequency band is the more intense while in others the
able than the high frequency bands. The occurrence high frequency band is more intense. Their published
of se\.eral different bands in both regions is a result of reduction procedures are a bit uninformative, but ap-
there being a number of different types of sites that cart parently a batch reduction procedure with tempera-
produce bands in both regions. Also, as expected, tures from 200 to 350' was used for unspecified lengths
there is a continuous variation of sites so that the bands of time. Apparent,ly, their samples were treated
in the two regions which seen1 to represent the largest at lower temperatures for shorter times than ours and
number of sites overlap each other. It should be so would have less crystal growth and sintering oc-
noted that all of this is on the basis of only one adsorbate curring. This would produce more sites that give the
structure, Le., a linear metal-carbon-oxygen system. low-frequency bands. Thus we could explain the
A word of caution is due on the subject of trying to variability in observed bands on the basis of sintering
determine with any degree of accuracy the relative and crystal growth rather than the usual arguments
number of surface sites of different types by comparing about completeness of reduction. Eischenslg and co-
the intensities of the different C-0 bands. As already wwkers have also reported spectra for CO adsorbed
mentioned, the extinction coefficient is expected to on silica-supported Pt, evaporated Pt, and alumina-
increase as the electronic chargv in the IC.s-orbital is supported Pt with the high-frequency band at 2070,
incrvased. Since this effect is'also used to explain the 2050, and 2050 cm.-l, respectively. If these data are
different frequencies, the extinction coefficient for the to be explained on the basis of the degree of crystal-
various bands is expected to incwase as the frequency linity of the Pt, it would be stated that the evaporated
decreases. This has not been exainined experimentally and alumina-supported samples are a little less crystal-
for chemisorbed CO, but if it occurs as predicted, the line so that the ligand Pt atoms are less effective in
number of low coordination number sites could be con- competing for electrons in these samples with the result
siderably less than a casual glance a t the spectra would that the C-0 frequency is lower. Following this line,
suggest. the alumina-supported Pt has a larger low-frequency
2. Supported and Evaporated Metals. The in- hand and a broader high-frequency band than the silica-
frared spectra of Cc? adsorbed on silica-supported6 and supported Pt sample. A lower crystallinity for the
evaporated6 Xi have been obtained in this laboratory. alumina sample is also supportNedby the statement of
These spectra show the bands around 2060 em.-' EiscE.ens2 that the alumina-supported Pt is more dif-
more intense than the band near 1900 em.-' for the ficult to reduce than the silica-supported Pt. The
silica-supported samples while the reverse intensity evaporated Pt sample showed no lorn-frequency band
pattern is observed for the evaporated samples. but the intensity of the high-frequency band was low,
I n producing the silica-supported Ni samples, the so it is difficult to tell if one should have been observed.
sample is heated for over 12 hr. at 400' in a stream of
Hz. This should produce sintered, well-defined crystal- (17) J. H. deBoer and J. W. E. Coenen, as reported in ref. 1, p. 15.
lites which have a maximum of plane faces exposed (18) R. P. Eischens, W. A. Pliskin, and S. A. Francis, J . Chem. Phys.,
2 2 , 1786 (1954).
and a minimum of edge atoms, dislocations. and amor-
(19) R. P. Eischens, -4cccptance Address for American Chemica
phous material. This is supported by electron micro- Society Award in Petroleum Chemistry, San Francisco, Calif., 1958.
Voliolllme 68, S u m b e r IO October. 1964
Thc shift in frequcncy with changing support could Whcn chemisorbed CO is removed by pumping with
also be interpreted in t e r m of elcctron-donating prop- or without heating, the bands are removed in just the
erties of the support. If it is assumed that alumina reverse order in which they appear. This would seem
donates elcctrons, this would shift the bands for the to indicate that in the adsorption process the low
alumina sample to lower frequencics than for the silica energy sitcs are indeed occupied first.
sample. At the present time, there are no data avail- 4. EJect of Olher Gases on Adsorbed CO Bands.
able to assess the extent of such an electronic effect. Thc cffect here is in terms of whether the added gas
Yang and Garlandz0have observcd that for CO ad- adsorbs with addition or rcmoval of electrons from the
sorbed on alumina-supported Rh, sintering decreases metal substrate. If electrons are removed from the
the intensity of a band a t 1925 ern.-’. They stated metal, the occupancy of the &orbitals is reduced with
that this fact is not consistent with the assignment of cmsequent raising of the C-0 frequency. Addition
the 1925-cin.-’ band to a simple bridge structure. of electrons gives the opposite effect. Eischens and
This fact is consistent with the interpretation that eo-workers have shown that the chemisorption of
sintering incrcases the degree of crystallinity and 0 2 , which is cxpectcd to dissociate and form oxide ions
thereby decreascs the number of sites giving the low- on the surface. thcreby removing electrons, shifts the
frequency band. bands for CO chemisorbed on Fe and on Cu to higher
I n a similar manner, the data of O’Seill and Yateszl frcqucncies. This is in accord with the model of
on the effect of using silica, alumina, or titania to bonding proposed. Eischens and c o - w o r k e r ~ ~have
~‘~
support Xi for CO chemisorption studies can be dis- also shown that when Hz is added to CO on Pt the C-0
cussed in tcrnis of the degree of crystallinity of the sup- band is shifted to lower frequencies. From magnctic
ported S i . Thcy used a batch Hz reduction treatment mcasurementsz2 and electrical resistance datalz3it is
for only about 2 hr. generally believed that adsorbed hydrogen contributes
3. Band Position as a Function of Coverage. It elcctrons to thc metal. The observed shift in the C-0
has been observcd that as the surface coverage in- band is in accord with hydrogen contributing electrons
creases, the frequencies of both the high- and low-fre- to the metal.
qucncy bands shift to higher Two possible Yates and Garlandz4 have observed the cffect of
kinds of behavior can be predicted for the frequency as adsorbing mercury vapor on thc bands of C(3 cherni-
a function of coverage. Onc is that a particular band sxbed on Xi. They observcd that adsorption of
will simply shift its position to higher frequencics Hg first caused the disappearance of the low-frequency
in a continuous manner as thc coverage increases. band. Herc this band has bcen ascribed to a site in
The proposed model of the bonding explains this by which the S i atom has a low coordination number.
noting that as the numbcr of adsorbed molecules That means that this site is the most exposed arid gives
incrcases, the competition for the clectrons of the the highest energy of interaction so the Hg might well
surface atoms incrcases so that there is less charge be expected to displace CO from these sites first. They
available to put into each $2-orbital with the conse- further observed that the high-frequency band was
quence that the C-0 frequencies increase. shifted to a lower frequency, the bond strength of the
A second kind of behavior is predicted if the types Ni-C bond increased for the remaining CO after the
of sites are distinct and the adsorbed CO is mobile Hg adsorption, and the extinction coefhcient for the
enough to find the sites which give the highest heat of remaining CQ was greatly increased. The authors were
adsorption. The heat of adsorption would be expected unable to explain these results. Since Hg has its outer
to be grcatest on the low coordination number mctal d-orbitals completely filled, it would be expected to
atom sitcs which give the lower frequency hands. contribute electrons to the Ni substrate upon adsorp-
Eischcnsz has indced observed that the chemisorbed tion. This would put, more electrons into +orbitals
CO which gives the bands below 2000 em.-‘ is the of the remaining CO groups with the observations just
most tightly held. lifter the lowest cncrgy type of given following as a natural course.
site is filled, the next lowest type which will b’ we a 5 . C120 and Cl3O Desorption. Eischens, Francis,
band a t a higher frequency will be filled. Thc result
will be a band with its maximum a t a high frequency (20) A. C. Yang and C. W. Garland, J . Phys. Chsm., 61, 1504
and one or more shouldcrs on the low-frequency side. (1957).
L4ctunlly, both kinds of behavior arc obscrvcd. On (21) C. E. O’Neill and D. J. C. Yates, ibid.. 6 5 , 901 (1961).
(22) P. W. Sclwood, “Adsorption and Collective l’aramagnetisrn,”
nickcl thc second kind of behavior is usually observed Academic Press, New York, N. Y., 1062.
although the scparation of bands and designation of (23) It. Suhrrnann, %. Elcktrochem., 6 0 , 804 (1956).
shoulders is somctimes a matter of personal taste. (24) J. T. Yates and C. W. Garland. , I . P h g s . Chem., 6 5 , 017 (1961).
b/ZOLECULAR ORBITAL l r I E W O F CHEMISORBED C A R B O N bfOSOX11)E 2777
and PliskinZ observed that when a mixture of chemi- Acknowledgnient. Acknoivlcdgmcnt is niadc to the
sorbed C l 2 0 and C130 is slowly desorbed by pumping donors of the Petroleum Research Fund, administered
and heating from a Pt surface, the intensity for the C130 by thc hincrican Chcnlical Socicty, for partial support
band first increascs and then decreases while thc Gl2O of this research.
band first decreases rapidly and then more slowly.
They proposed to explain this result on the basis
of dipole-dipole interaction of adjacent chemisorbed Discussion
GO groups. An alternative explanation seems pos-
F. A. MATSON(1Tniversit,y of 'rexas). The filling of the anti-
siblc. On the basis of thc modcl of bonding proposed bonding orbital should have a signific:tnt effect on the intmxity
here, the extinction coefficicnt for chemisorbed CO of ot)served t)ands. Can such an efTeot tie obsc?rved?
should increase as the coverage decreases. For slow
G. ~ ~ L Y H O L I ~ EII ~expect
. t h a t such an effwt c,orilti be o b -
desorption of Gl2O and C130 a kinetic isotope effect served, and, by analogy to t,he work of Jones (ref. 10 of the ptiper),
is expected with the lightcr isotope coming off the it is expected t o t)e a sizable effect.
surface faster. For the chemisorbcd CO a t almost
I<;. H~:TUIINSON (Stanford I-niversity, California). Woriltf yon
completc coverage, the heat of adsorption is low and the expect that the efTec,t of sintering on the geometry of :t "c~xitral"
isotope cffcct could be fairly large. If the rate a t titoin is sufficiently pronounced to makc it) worthwhile to study
which C130 is being rcmovcd from the surface is less the chaiiges in t.he spectra of carbon rnorioxide iidsorkmi on
than the rate a t which the extinction coefficient is evaporated or s u p p o r k d thin metal films as ii func*tionof time and
being increased due mainly to C l 2 0 desorption, the temperature of sintering'?
intensity of the C130 band will initially increase. As G. F~I,YJWI,I)EB. Yes, and we expect to pursue this lint of at-
CO is rcmoved from the surface, thc heat of adsorption tack in t,he near future.
for the CO increascs so the activation energy for dc- ,J. .'l' Yxrk:s, Jrt. (National Bureau of St:mdtwds, Washirigton,
sorption increascs, with the rcsult that the difference I). C.) (comniunic*ated). ( 1 ) I h . Blyholder's view is that the
in ratc of desorption for C120 and C i 3 0 becomcs less. spectrum of chcrnisort)ed CO on Si and other traiisit ion metals
As C130 is removed from thc surface a t morc nearly may be interpreted in terms of linear CO spec.ies :tdsort)ed on
nielal atonis having various degrees of c~oordinationwith othrr
the rate a t which C l 2 0 desorbs, the intensity of the metal :ttoiiiR in the supported surfa(~c,the lower frcquc?ii(*yt):inds
GI30 band dccrcases even though the extinction being due to CO t)onded to Si atonis having less t h a n thc: niasi-
coefficient may be increasing. The relative roles niuin number of neighbors. Although his t2ieorctic~:rl dtt1uc:-
of dipole interaction and kinetic isotope effect in the tion8 tlnsed on t,he spevtrii of subsf itutcd nirtA carbonyls niay be
observcd phenomcna do not seem to be readily de- qualit:tt.ively cwrrect, to sonir extent, he ignores a body of cxperi-
incrit:il ctat,a which supports t,he oppositr point of view--ri:iniely,
terminable a t prescnt. that) bot,h linear and hridgcd CO specaies :ire fornieti on c'r)
6. Band Positaon on Alloys. The only data Xi surfaces. Thus the work of Yates a n d Garl:md ( J . I'hys.
giving the band position for CO on alloys that havc Cih,evi., 6 5 , G17( 1!)81 ) ) e~t~ablishctl t h a t bofh the low-frrqiiency
been published are those of Eischensz5for a 90% Ni- CO hand writered at 1060 c ~ n . - liind Lhr h:rnti :it 2O:S cni. - - I in-
creasrd in relative intensity as the Si crystallite :~vcr:igesim was
10% Cu alloy. Since Cu has its d-orbitals filled, it
in(-reased by incwasing the percmtsgc o f Si i n ilie sripportrd
would be expected to donate elcctrons to the system saniples (a8 verified by X-ray line hroadenirig st,iidics). At tht:
which would put morc charge in the $z-orbitals with same t,inie the relative intensit,)- of the highest freqiiriic~yIxmd at
a consequent lowcring of the C-0 stretching frcqucncy. 2080 (.in.-' (which in my view is due 10 liricar CO on tlispcrsrtl Si
The data of Eischcns show the CO bands on the alloy sites) dccreaued due t,o the devrease i n the popnlat ion of t1isl)ersrtl
to be very similar to those on pure Ni except that the sites as the percentage of Xi was incrcasc.ti. Chritrary t o this, tlir
IXlyholder model would predict that t h r rrl:itiw intcvisity of t t i c s
alloy bands are indeed shifted to lower frequencies l ~ l ~ O --I ~ line
~ ~ nwould
. derwase :is the nunil)er of frill!. c,ciortiiiirtted
by about 40 em.-'. Si atoms i n the surfwe was iricrrtised. I h . Rlyholdcr'sargument
I n suniniary, thc moic~cular orbital niodel of the is psrtittlly based on H. cwiiparison of s p w t r a of CO o n silica-
metal-carbon- oxygen x-bonding system succcssfully supported Si iind Si particles siispentlctl in oil. ('iiii~i(i~!r;~ti(iiis
qualitatively cxplains . (1) the occiirrcnrc of several of surfac:e rleiinliricss in the ( w e of the oil-suspt~riclc~tlsitniples
would seem t,o vi1 iittc this c:onipitrisori.
carl~on-oxygcri stretching frcqucncic~sin the infrarcd ( 2 ) It is suggested t)y I)r. 131yholtlcr thiit :t I:irgc, kinciti(. isotope
spcctra of C'O adsorbed on metals; ( 2 ) diffcrenccs in effert, esistb in ttw desorption of C I W ' 6 arid ClxOlfi i i i i d ih:tt this
spcctra of C:O adsorbed on cvaporatcd and supported rrii~yexplain the chirige in rrlLttivc i n t m s i t y o f thcl t w i isotopc
inct,als; (x) infrarcd hand positions as a function of brtnds duriiig drsorpt ion from I't as ol)scrvtd l)y I~;iwti(ws, ( h l -
covcragc'; (4) thc effcct of adsorbing other gases in cwlations h e e d on the calassical inwluiiiiw of t i I i i i w r t hrcc4xidy
os(dlator witti end body of infinite ITIiiss ( A . Atiel, I'h!/s. A'w., 45,
addition to pwviously chemisorbed C'O; and (5) 56( l!)B4))show that A c h m g r froin C'?O16t o ( : I 3 0 1 6 should result
band shifts in going from adsorption on pui*c Si to
a Ni-Cu alloy. (25) 1L. P. Eischens, Z. EIcklrochem., 60, 782 (195G).
