1616
L. H. LITTLE
Vol. 63
INFRARED SPECTRA OF HYDROCARBONS ADSORBED ON SILICA SUPPORTED METAL OXIDES BY L. H. LITTLE’ Department of Colloid Science, University of Cambridge, England Received March 3,1969
Infrared studies have been made of surface species formed by the adsorption of ethylene on the oxides of nickel, copper and palladium supported in optically transparent porous Vycor silica glass. Quantitative adsor tion measurements showed that polymerization of ethylene occurred on the metal oxides. Concurrent measurements of t i e spectra of the gas phase and the adsorbed phase have been made to study the equilibrium between these phases. trans-Butene-2 was formed in the gas phase for ethylene admitted to nickel oxide samples. The surface species on these samples were composed of methyl, methylene and unsaturated groups. However, ,the surface s ecies formed by ethylene on copper oxide contained only methylene groups, while that on oxidized palladium containel methyl and methylene groups. I n the latter s stems the gas phases remained ethylene. The integrated absorption intensities of the surface species were measured anxcompared to hydrocarbons in normal liquid environments. This comparison provided knowledge about the perturbations of molecular vibrations due to adsorption of molecules on to surfaces. For the following hydrocarbon-supported metal oxide systems only physically adsorbed molecules were detected and no bands characteristic of new chemisorbed species were found: acetylene on silica supported nickel, copper, palladium and silver oxides; ethylene on eilica supported silver oxide and ethane on silica supported nickel oxide. Quantitative measurements showed physical adsorption occurred largely on the silica support.
The application of infrared spectroscopy to the study of molecules adsorbed on solid surfaces has been reviewed recently by Eischens2 and by Sheppard.a The value of this technique for investigation of adsorption problems lies in the direct evidence that it may provide for the structure of adsorbed surface species. This knowledge is of the greatest importance in the understanding of mechanisms involved in heterogeneous catalysis. Difficulties in the use of this technique arise largely from scattering and absorption of energy from the infrared beam by the sample. In the case of supported catalysts this occurs at both the adsorbent and its support. The sensitivity may be enhanced by decreasing the catalyst particle size which reduces scattering and also increases the surface area and hence the amount of adsorbed material in the sample. Finely divided metal catalysts supported on high area silica powder have been used extensively by Eischens.2 I n the present work Vycor porous silica glass was used to support the finely divided oxides in the manner described prev i o u ~ l y . ~This material has the advantage that it scatters less infrared radiation than does silica powder, although the available thicknesses of porous glass restrict the usable region of the spectrum to frequencies above 1500 cm.-l (>2000 cm.-l for samples 1 mm. thick) because of the strength of silica absorption bands. The present study extends the results for the chemisorption of ethylene on supported nickel oxide that have been reported earlier.4 Additional infrared spectra are reported for ethylene chemisorbed on copper and palladium oxides and for the gas phase in equilibrium with the adsorbents. No chemisorption was detected on supported silver oxide, and in this case, as in the cases of ethane and acetylene on the other oxide-glass systems, only physical adsorption was detected. These results are reported more briefly. (1) Division of Applied Chemistry, National Research Council, Ottawa, Canada. (2) R. P. Eischens, 2.EEektrochem., 60, 782 (1956). R. P. Eischens and W. A. Pliskin, “Advances in Catalysis and Related Subjecta,” VoI. X, Academic Press Inc., New York. N. Y.,1958, p. 1. (3) N. Sheppard, “Molecular Spectrosoopy,” Hydrocarbon Research Group Conference on Molecular Spectroscopy, London, February 1958 ( t o be published by Pergamon Press, London). (4) L. H. Little, N. Sheppard and D. J. C. Yates, to be published.
Experimental The samples were tubular specimens of porous Vycor glass, 1 mm. wall thickness, 20 mm. diameter and 5 cm. long. These were cleaned by heating to 400’ in oxygen, evacuated, cooled and weighed quickly before water was adsorbed from the atmosphere. The sample then was immersed in a nitrate solution suitable to give 2% by weight of metal oxide. The metal nitrate was decomDosed to oxide bv heating under vacuum and the sample Gas then cooled, Eeweighed to find the amount of metal oxide and then transferred to a glass infrared-adsorption cell having a furnace wound on one portion. The sampIe was free to slide in this cell between the furnace section and a section fitted with sodium chloride windows for measuring the spectrum. After heating and evacuating at 300°, the sample was cooled and the background spectrum recorded. Then a dose of gas was admitted and the amount of adsorption measured. The spectrum was again recorded and the absorption of the surface species found. The infrared cell permitted the measurement of the spectrum of the gas phase in contart with the sample, and throughout this work simultaneous study was made of the spectra of the adsorbed species and the gas with which it was in equilibrium. Unfortunately it was necessary to detach the cell from the adsorption system in order to measure the spectra. Therefore intermediate pressures during slow adsorption processes had to be obtained from intensities and op.tica1 densities of the gas phase via previously determined calibration ciirves. At the end of the series of infrared measurement8sthe cell was replaced on the adsorption system and the final equilibrium gas pressure was found. Finally the amount of adsorption on the metal oxide was obtained from the difference between the total adsorption and the physical adsorption on the glass without oxide present. The infrared spectra were recorded on a Perkin-Elmer Model 21 Spectrophotometer fitted with either a sodium chloride or a calcium fluoride prism. The following spectral parameters were measured for absorption bands of surface species: frequency Y (cm.+); band half width Av1/2.(cm.-1); optical density D = log (10/1);extinction coefficient e = l / C l log ( Z o / I ) (mole-‘ 1. cm. -1); integrated intensity T = T log (lo/I)dv(cm.-l); molar integrated intensity M = l / C l Tlog(I0/1) dr(mo1e-11. cm.-2). Here l o and I are the incident and transmitted radiation intensities at a particular frequency; C is the concentration of adsorbed material in terms of moles of adsorbate per liter of samp!e, where the sample volume is defined by the external dimensions of the porous Vycor support; E is the total path length traversed in the sample in cm. The extinction coefficients were not corrected for the effect of finite spectrometer resolvine Dower5 because of the overlamina - - - nature of the bands. The integrated intensities are probably within 10% of the true values. The two greatest sources of error in the quanI &
( 5 ) D. A. Ramsay, J . Am. Chem. SOC.,74,72 (1952).
Oct., 1959
1617
SPECTRA OF HYDROCARBONS ADSORBED ON SILICA SUPPORTED METALOXIDES
titative results were (1) the measurement of the amount of material adsorbed on the surface of the metal oxide and (2) the construction of the background apectrum relative to which the spectrum of the adsorbed molecules was meat+ ured. Estimated errors in integrated intensities are given in the tables of reaults. Throughout this work the band intensities of the adsorbed species have been measured to investigate the effect of perturbation of molecular vibrations by the solid surfaces. The intensities of a number of halogenated hydrocarbons were measured (Table VI) and previously published Gas and liquid intensities have been converted into the units employed here (Table VII) for comparison with those of the surface species. At the completion of the infrared investigations the metal oxides were reduced with hydrogen to metals on which hydrogen, carbon monoxide and oxygen adsorption measurements were made to determine the surface areas of the dispersed materials (Table VIII).
Results Ethylene Adsorption on Nickel Oxide in (i) Porous Glass. (a) Adsorption on Non-stoichiometric Nickel Oxide.-The nickel oxide samples prepared as described above were black and were presumed to be the non-stoichiometric oxide.6 Under certain conditions this could be partially reduced to the greenish stoichiometric oxide, which will be referred to below. Oxygen admitted to the latter sample returned it to the original black colour. When ethylene was admitted a t 20' to a nonstoichiometric nickel oxide sample, absorption bands appeared at 3020,2970,2940 and 2885 cm. -I. In Fig. l a the bands are shown 80 minutes after admitting ethylene. Arising from the porous Vycor support an intense absorption band was observed between 3800 and 3400 cm.-l; this was due to the V O H vibration of the hydroxyl groups on the surface of the glass. The bands at 2760 and 2540 cm.-l belong to the skeletal silica vibrations.' Quantitative results for the spectrum in Fig. l a are given in Table I. TABLE I C2H,ON NICKEL OXIDEIN POROUS GLASS NaC1 prism, 16 ern.-' slit width, % nickel oxide 2.7, sample thickness 0.28 cm.
Gas pressure (cm.) transC2H4 butene4
1.90
0.48
4.85
- Phyaioal = Chemisorbed 0.55
4.30
ethylene. Physical adsorption of ethylene on the glass surface during the adsorption process led to the perturbation of the VOH band of the surface hydroxyl groups on the glass, as seen by the difference between the two dashed curves in Fig. l a . Composition of the Gas Phase,-As described above the intensity of the surface species increased with increasing adsorption. Concurrently the gas phase was found to change from ethylene to transbutene-2. These aspects of the adsorption process are shown in Table 11. The ethylene component of the gas spectrum has been shown dotted in Fig. l b and the intense maximum superimposed a t 2960 cm.-l was due to trans-butene-2. TABLE I1 CzHION N I C ~ EOXIDE L IN POROUS GLASS Time after ndmitting
CzHd
Pressure (om.) transCaH4 butene-2
v c
3020 6
2970 12
2940 12
2885 7
T
M
146
1800
The pressure in the cell decreased continuously over a period of five days when the ethylene was allowed to stand at 20' in contact with the sample. The intensity of the bands increased as adsorption proceeded. At the same time the relative intensity of the bands changed, probably indicating that more than one adsorbed species was present on the surface. By comparing the spectra with those obtained by Sheppard and Yates7 it may be seen that no bands were attributable to physically adsorbed
M
146 187 238
1800 2040 2 100
(b) 2nd Series 1.15 1.34 0.85 2.44 0.31 .44 .65 3.36 .35 4.27 68 .05 5.47 .59 0 5.89 .43
72 113 168 200 233 270
2900 2400 2600 2450 2200 2400
40 min. 70 min. 105 min. 235 min. 475 min. 48 hr.
0
(6) S. J. Teiohner and J. A. Morrison, Trans. Faraday Soc., Si, 961 (1955). (7) N. Sheppard and D. J. C. Yates, Proc. Rov. SOC.,(London), A ~ S 69 , (1956).
0
Total intensity
T
0.48 0.64 0.68
5.20 3.6 2.6 1.1 0.2
CZHIadsorbed (cc./g. glass) Chemi-
1st Series 0.55 4.30 0.40 4.80 0 5.99
80 min. 1.90 130 min. 1.55 14 hr. 0.20
Absorption bands Total intensity
,-,i! 2600
-----
(4
C2H4 adsorbed (cc./g. glass) Total
3000
34M)
Fig. 1.-(a) spectrum of the ethylene surface species formed on non-stoichiometric nickel oxide in porous silica , background spectrum of sample without glass; adsorbate. (b) gas phase (ethylene and trans-butene-2) u1 equilibrium with (a); . . . . . ., spectrum of ethylene component of gas phase.
I
Physical
sorbed
To confirm the presence of trans-butene-2 in the gas phase, the gas had been compressed and the more intense spectrum so obtained permitted a comparison of all absorption bands between 4000 and 650 cm.-' to be made with known standard spectra. No cis-butene-2 mas found, but the maximum concentration of this isomer that could have escaped detection was approximately 20%, as shown by the absence of the characteristic C-H out of plane deformation band at about 680 When trans-butene-2 in the gas phase was re-
L. H. LITTLE
1618
I-
C2WI
ON NICKEL OXIDE
IN THIN POROUS G L A S S
Fig. 2.-Ethylene admitted to nickel oxide in thin porous silica glass: (a) background spectrum of sample of porous silica glass containing nickel oxide and without adsorbate; (b) combined spectrum of sample surface s ecies and gas phase; (c) gas hase spectrum; (d\ obtained%y subtracting (c) from (b); obtained after evacuating gas phase.
6)
moved from the cell, more was desorbed from the sample showing that this gas was involved in a physical adsorption equilibrium. Indications are that this was most probably physical adsorption on the hydroxyl groups of the silica glass. Stepwise out-gassing of the trans-butene-2 from the adsorption cell resulted in the disappearance of the absorption band a t 3020 cm.-' as well as some reduction of the 2970 and 2940 cm.-' bands. The former band was thus ascribed to physically adsorbed trans- butene-2. Ethylene Adsorption on Nickel Oxide in Thin Glass Samples.-In Fig. 2 the spectrum is shown of the species formed by ethylene on a sample of non-stoichiometric nickel oxide in 0.1 mm. thick porous silica glass. Bands appeared at 2960, 2920 and 2860 cm.-'. Weak absorption occurred above 3000 cm.-l which disappeared on evacuation. This probably was due to a species weakly adsorbed on the sample. The molar integrated intensity of the surface species on this sample was slightly smaller than that shown in Table I. The adsorption process ceased after 18 days even though a considerable amount of ethylene still remained in the system. Hydrogen added to the system after the completion of the adsorption led t o the hydrogenation of the remaining ethylene. The spectrum of the surface species was unaffected by the hydrogen. Samples of 0.1 mm. thickness were transparent down to but not below 1475 crn.-'. Thus it was not possible to detect C-H deformation bands of the adsorbed hydrocarbon. With the present samples no absorption bands appeared between 1500 and 1700 cm.-l in the region characteristic of the absorption of unsymmetrically substituted carboncarbon double bonds. Figure 2 shows the V O H absorption of the surface hydroxyl groups of the glass. The sharp band a t 3735 cm.-' corresponds to absorption by free hydroxyl groups on the glass surface and the broad band a t lower frequency to hydrogen bonded hydroxyl groups. It may be seen that due to physical adsorption of ethylene on the hydroxyl groups, the V O H vibration was perturbed (spectrum a to b) with
Vol. 63
an increase of intensity of the lower frequency hydrogen bonded component. Reproducibility of Catalyst for Adsorption.It was possible to remove the species from the surface of the nickel oxide by heating a t 350" for 12 hours with oxygen. A fresh dose of ethylene was then admitted and, apart from a small decrease in rate, the adsorption process occurred as before, with trans-butene-2 formed in the gas phase and the spectrum of the surface species was the same as that in Fig. la. Table IIa shows the rate of formation of the surface species on a fresh nickel oxide sample and Table IIb the rate after cleaning and admitting more ethylene. The sample then was reduced completely by hydrogen a t 320" t o nickel metal. Ethylene standing for two weeks over this sample did not produce absorption bands due to surface species on the nickel. After reoxidizing by heating with oxygen at 300" for 12 hours it was possible to repeat the process with ethylene. However, there was a great decrease in the rate of formation of the surface species after reduction and reoxidation, possibly due to loss of surface area of the nickel oxide. (b) Adsorption of Ethylene on Stoichiometric Nickel Oxide.-Non-stoichiometric samples were normally prepared by decomposing nickel nitrate in the pores of the silica glass. It was possible partially to reduce these to give light green samples which were presumably the stoichiometric oxide.6 A premixed dose of hydrogen and ethylene (Table 111)was admitted to the sample at 20" and allowed to stand to investigate (1) whether the adsorption was the same on this sample as on the previous ones and (2) whether ethylene would be hydrogenated over nickel oxide. PREMIXED CzH,
+ Ha
TABLE I11 ON
STOICHIOMETRIC NICKELOXIDEIN
POROUS GLASS
CaFz prism 8 ern.-' slit width, % nickel oxide 2.3, sample thicknesa 0.22 cm., original gas mixture HZ9.41 cm. and C2H44.71 em. ~~
Time after admisslon
Pressure (cm.) CaHi adsorbed Absorption bands trans(cc.(g. glass) e Total Physi- Chemi- (2970 ButeneCzHd 2 CzHs cal sorbed om.-])
20min. 4.2 0 80min. 3 . 5 0.29 130min. 3 . 0 .34 5hr. 2.1 .35 20.5hr. 0 . 5 .35 0 Accuracy f 20%.
..
0 0.85 0 .. 0 -70 0.47 40 30 4900 0 .65 0.94 27 50 4100 0.40 .55 1.56 25 88 4300 1.3 .15 2.70 21 129 3700 b Only physical adsorption on glass.
The early stages of the adsorption proceeded in a similar manner to that on the non-stoichiometric samples and trans-butene-2 was produced. Later, however, the reaction CzH4 Hz CzHs commenced and the rate of production of ethane was soon faster than that of tran,s-butene-2. Eventually all the ethylene was consumed by hydrogenation and by adsorption. The final gas phase contained trans-butene-2, ethane and excess hydrogen and possibly butane. The spectrum of the adsorbed species on this sample showed a strong band at 3020 cm.-l, which persisted even after evacuation at room temperamm. This ture for 3 hours at a vacuum of was unlike the 3020 cm.-l band in the non-stoichio-
+
-
SPECTRA OF HYDROCARBONS ADSORBEDON SILICASUPPORTED METALOXIDES
Oct., 1959
metric samples which was easily removable by pumping and which was demonstrated to be due to physically adsorbed trans-butene-2. Throughout the course of the adsorption measurements the sample remained green. When air was admitted a t 20' the sample rapidly became black indicating non-stoichiometric nickel oxide. The absorption intensity of the surface species was unaltered by this process. (ii) Ethylene Adsorption on Copper Oxide in Porous Glass.-A surface species having absorption bands a t 2920 and 2860 cm.-' was produced when ethylene was allowed t o stand over a sample containing copper oxide prepared in the same manner as the nickel oxide samples. The intensity of these bands increased with time as more ethylene was adsorbed. I n Fig. 3 the spectrum of the adsorbed species has been measured and replotted on an optical density scale. The quantitative results are shown in Table IV. TABLEIV C Z HO~N COPPEROXIDEI N POROUS GLASS NaCl prism 13 cm.-l dit width, o/o copper oxide 2.1, sample thickness 0.20 cm. Fig.3
Time
CERl pressure (om.)
a b c d e
45min. 19 hr. 43 hr. 43days 67days
6.1 5.8 5.7 4.4 3.9
Vol. adsorbed (cc./g. glass) ChemiPhysical sorbed
1.30' 1.25 1.20 1.00 0.90
0.37 .67 .99 2.73 3.49
Concn. (mole/l.)
0.026 ,047 .069 .19 .24
Absorption bands 2920
om-1
(Auilz =
60 om.-')
2860
cm.-1
(AUIIZ =
45
cm-1)
Total intensity
M D E D e T a 0.050 10 0.03 5 5 1200 b .11 12 .05 5 11 1200 c .16 11 .08 6 18 1300 d .74 19 .39 10 72 1900 e .99 20 .54 11 100 20oob a Physically adsorbed CzHd on glass (concn. 0.091 rnole/l.). Absorption bands: 3090 cm.+ 3000 cm.-l Total intensity € = 3.5 € = 3.5 T M 11 600(rt50%) Accuracy A20%.
In addition to the bands at 2920 and 2860 cm.-', weak bands appeared a t 3090 and 2990 cm.-' due to ethylene physically adsorbed on the glass.' The gas phase remained ethylene during the adsorption process. Electron diffraction and X-ray studies on the porous glass containing copper oxide failed to reveal the oxidation state of the copper. It was presumed that the sample contained cuprous oxide as this material is usually more active in catalysis than cupric oxide. For example ethylene is readily adsorbed by cuprous oxide.8 (iii) Ethylene Adsorption on Oxidized Palladium in Porous Glass.-Bands appeared a t 2960, 2940 and 2870 cm.-l and the molar integrated intensity was about 1500 mole-' 1. cm.-2 when ethylene was admitted to a porous glass supported sample of palladium, oxidized by heating in ( 8 ) W. E. Garner, F. S. Stone and P. F. Tiley, PTOC.Roy. SOC. ( / , o n d o n ) . M i l , 472 (1952).
t o 4 0IN POROUS GLASS
/ \A
1019
If'
Fig. 3.-Speotmm of ethylene surface species on copper oxide in porous silica glass at the following times after admission: (a) 45 minutes. (b) 19 hours; (c) 43 hours; (d) 4 3 da a; (e) 67 days; comparison spectrum of long chain hy&ocarbon -CH2CH2-. .
if)
.
air. The intensity of the bands increased with time over a period of several days as adsorption proceeded. (iv) Ethylene Adsorption on Silver Oxide in Porous Glass.-The amount of ethylene adsorbed on the silver oxide sample (prepared by decomposing silver nitrate impregnated in porous glass) was equal to that adsorbed on a clean glass sample a t the same pressure. Absorption bands appeared at 3090 and 2990 cm. -l corresponding to physically adsorbed ethylene on the glass of the sample. The intensity of the bands was similar to that shown a t the foot of Table IV for ethylene on the copper oxide sample. No other bands appeared in the spectrum even after standing for two weeks. (v) Ethane Adsorption on Nickel Oxide in Porous Glass.-Only physical adsorption occurred when ethane was admitted a t 20" to a non-stoichiometric sample of nickel oxide in porous glass. The amount of adsorption on the sample was equal to that on a clean glass sample at the same pressure. No absorption bands persisted when the gas phase was evacuated or when the sample was heated to 150' with ethane. The intensity of the physically adsorbed ethane is given in Table V. TABLE V PHYSICAL ADSORPTION O N NICKELOXIDE-POROUS SILICA GLASS NaCl prism 15 cm.-* slit width, sample thickness 0.28 cm. Gas
Pressure (om.)
1'01. adsorbed (co./g. glass)
Concn. (mole/l.)
CzHa CaH6
3.68 4.80
1.16 0.51
0.079 0.035
Absorption bands Y
CsHe
Avilp
3240 48 GH:s 2950 a Accuracy f 2 5 % .
..
Total intensity
D
e
T
Ma
1.20 0.50
50 52
77 58
3600 6000
(vi) Acetylene Adsorption on Porous Glass Containing Oxides of Nickel, Copper, Palladium or Silver.-Acetylene was admitted to porous glass samples containing the above metal oxides and in each case a well resolved intense band appeared at 3240 cm.-I due to acetylene physically adsorbed
1620
L. H. IATTLE
on the glass of the sample.' No bands were found at lower frequencies. The VOH vibration of the surface hydroxyl groups of the glass was perturbed by this adsorption and the low frequency hydrogen bonded component of the hydroxyl absorption band was increased. The gas phase remained acetylene even after prolonged standing. Quantitative results of the band of physically adsorbed acetylene on a non-stoichiometric nickel oxide sample are given in Table V. When this sample was evacuated at lou6 mm. and 20' for 45 minutes the spectrum showed a considerable amount of acetylene remaining. Intensity measurements of the 3240 cm.-l band of acetylene on the samples containing other metal oxides were in good agreement with that given in Table V. No other absorption bands appeared when hydrogen was added to the oxide systems containing acetylene. The acetylene in the gas phase was also unaffected by adding hydrogen. Hydrogen at 20" reduced the very transparent silver oxide samples to silver which was opaque to visible light. Discussion Ethylene Adsorption on Metal Oxides in Porous Silica Glass ; Assignment of Vibrational Bands.-The bands at 2970 and 2885 cm.-' in the spectrum of the ethylene surface species on nickel oxide samples (see Results ia) have been assigned to C-H stretching vibrations of methyl groups and the band a t 2940 cm.-l to the asymmetric methylene C-H stretching vibration by comparison to normal hydrocarbon spectra. These assignments also agree within narrow frequency limits with the spectra of halogen substituted hydrocarbons (Table VI). The methylene vibrational band a t 2850 cm.-l did not appear in the spectrum of the ethylene-nickel oxide system and this band is indeed weak or poorly resolved in the spectra of the halogenated hydrocarbons such as 1,2-dichloroethane, 1,2-dibromoethane and the n-propyl halides. Absorption bands above 3000 cm.-' are characteristic of C-H stretching modes of normal unsaturated hydrocarbon^.^ However in certain circumstances the presence of polar substituents in a saturated hydrocarbon may raise the frequency of a C-H stretching mode above the normal limits. In such cases frequencies above 3000 cm.-' may not indicate unsaturation. For example methyl iodide (Table VI) has an absorption band a t 3053 cm.-'. The absorption band a t 3020 cm.-l in the spectrum of the ethylene on non-stoichiometric nickel oxide (see Results ia) has been shown to belong to physically adsorbed trans-butene-2. In view of this and the correspondence of bands below 3000 cm.-l with those of normal hydrocarbons it seems likely that the 3020 cm.-l band on the stoichiometric nickel oxide (Results ib) was due to an unsaturated species. The frequencies of the ethylene species on oxidized palladium suggest the presence of methyl and methylene groups (Results iii). I n contrast to the spectrum of the surface species on the palladium and nickel oxide samples the spec(9) R. N. Jones and C. Sandorfy in "Chemical Applications of Spectroscopy." Vol. IX, Intersoienoe Publishers, New York, N. Y., 1956.
Vol. 63
TABLE VI INTENSITIBB OF G H STRETCHINQ MODEL-HALOOBNATED HYDROCARBONS
CaFz prism 8 cm. -1 slit width Compound
Methyl iodide
Ethyl bromide
Total intensity M
t
Y
3053 2951 2814
l Z } 1.3
2980 2927 2870
16.8 12.8 7.8
554
1
1660
l12-Dichloroethane 2960 2880 2846
13.1 2.2} 1.9
%-Propylchloride
71
1
25 Shoulder
1
%-Propyl iodide
2973 2945 2885 2855 2970 2938 2880 2849
24 13
814
4530
trum of the ethylene surface species on copper oxide (Results ii) corresponds to the structure -CH2CHz- . . . The frequencies and contour of the bands on the copper oxide sample may be compared to the spectrum of Apiezon grease, shown in Fig. 3f, which contains a large proportion of long unbranched chains -CH2-CH2-. . . The Adsorption Process and Nature of the Surface Species.-The ratio of the number of ethylene molecules adsorbed to the total number of nickel atoms in the catalyst in the later stages of the adsorption process (Results ia and b) exceeded 1: 1, so that even assuming maximum surface area of the nickel oxide it would not be possible to accommodate the ethylene in a single adsorbed layer. However, it would be possible to accommodate the ethylene as a polymeric species extending outward from the surface of the nickel oxide. The formation of one type of dimer in the nickel oxide system was shown conclusively by the presence of trans-butene-2. Measurement of the surface area of the nickel, after reduction from nickel oxide, confirmed the occurrence of polymerization. Thus from Table I1 the maximum amount of ethylene adsorbed on the nickel oxide was 6 cc. per gram of glass, while from Table VI11 the hydrogen adsorption on the sample, after reduction to nickel, was 0.37 cc. per gram of glass. The hydrogen molecule adsorbs as atoms on two surface metal atoms'O and assuming that two metal sites were covered by the hydrocarbon chain it was probable that the surface species w7as comprised of fifteen ethylene molecules. The amounts of hydrogen and carbon monoxide adsorbed on the copper oxide sample, and also on the sample after reduction to copper, were very small (Table VIII). It has been shown that the adsorption of hydrogen on clean copper surfaces is very small16 and therefore it is probable that the (10) B. AT. W. Trapnell, "Catalysis," Vol. 3, Reinhold Publ. Corp., New York, N. Y.,1965, p. 1.
Oct., 1959
SPECTRA OF HYDROCARBONS ADSORBED ON SILICASUPPORTED METALOXIDES
1621
greatly perturbed. This has been done by comparing the absorption intensities with those of molecules of similar basic structure, but in normal liquid on gas environments. Total intensity Gas Ref. M Before making this comparison some idea of the Acetylene 11 3000 intensity variations of the C-H stretchingvibrations Ethylene 12 1670 of liquid hydrocarbons may be obtained by comEthane 13 7370 paring the intensity of the C-H stretching mode of Liquid chloroform in dilute solution in carbon tetrachloride CHS-CH, 14 7680 (174 mole-' 1. cm.-2)16 with that found for liquid 14 7160 chloroform (407 mole-1 1. cm.+). This twofold CHrCHr intensity increase illustrates the effect of changing surface area of the copper in the present sample was the polar nature of the solvent environment. Francis14has measured the intensityof thestretchgreater than indicated in Table VIII. I n view of this, an estimate of 0.3 to 0.4 cc. (per gram of glass) ing vibration of tertiary C-H groups in aliphatic of hydrogen to completely cover the copper sur- hydrocarbons and a value of 1210 mole-' 1. cm.-2 face was made by comparison to the amount of hy- has been calculated from his results. It may be drogen adsorbed 011 a nickel sample with a similar seen that the intensity of a tertiary C-H group is concentration of metal. Since the amount of ethyl- considerably decreased by the presence of adjacent ene adsorbed on the sample was much greater than polar substituents. Similar conclusions may be this figure, polymerization of ethylene must have drawn from Tables VI and VI1 where the intensioccurred on the sample. If the polymer chain was ties of halogen substituted molecules are one-fifth to assumed to cover two surface metal atoms in the one-tenth as large as those of unsubstituted moleoxide, then the chain length of the -CH2-CH2-- . . . cules with similar basic structure. It is probable surface species for maximum coverage (Table IV) from these considerations that great variations of was about ten ethylene molecules long. Similarly C-H intensities may be caused by varying the enadsorption measurements showed that ethylene was vironment of the C-H groups. The intensities of the ethylene surface polymers also polymerized on oxidized palladium samples. (Tables I, 11, 111 and IV) were intermediate between those of the halogen substituted hydrocarTABLE VI11 bons (Table VI) and those of the unsubstituted ADSORPTIONON METALOXIDES AND METALSIN POROUS molecules shown in Table VII. In view of this it GLASS was concluded that the adsorption of the polymer Surface area of total sample (B.E.T. method using argon at,7Q°K.) on to the metal oxide surface introduced no greater Nickel oxide Copper oxide Surface area sample sample perturbation of C-H stretching modes than that (m.Vg.) 105 179 Chemisorption studies (17O) produced by substitution of halogen atoms into Nickel oxide sample Copper oxide sample similar molecules. Vol. chemisorbed Vol. chemisorbed GaS (cc./g. glasa) (cc./g. glms) In the examples where acetylene and ethane were co 0.13 0.09 admitted to porous glass samples containing metal Ha 0.04 0.02 oxides, the intensities of these molecules, physically adsorbed on the samples (Table V), were in good Samples reduced to metals agreement with the intensities of the corresponding Ht 0.37 0.03 molecules in the gaseous state (Table VII). Any co .. .07 chemisorption of acetylene and ethane on to the 02 0.89 .69 metal oxides, and also of ethylene on to silver oxide, Absorption Intensity of Surface Species.-Polymust have occurred to less than monolayer covermerization of ethylene has been shown to occur ages. A monolayer of material with intensity comon the oxides of nickel, copper and palladium with parable with those reported in Tables I-IV would the formation of surface species, which are for the have been detected readily by the spectrum. greater part saturated and composed of methyl Acknowledgments.-The author wishes to thank and methylene groups. This assignment of struc- Dr. N. Sheppard for suggesting and supervising ture was made on the basis of the band frequencies this research and Dr. D. J. C. Yates for advice. and contours as described above. From these ob- Awards of a Hackett Postgraduate Studentship and servations it is apparent that in the adsorption a Consolidated Zinc Corporation Bursary are grateprocess the ethylene molecules have been perturbed fully acknowledged. and reacted to give new chemical species. DISCUSSION It is of interest to see whether the C-H stretching modes of the new polymeric surface species, A. C. ZETTLEMOYER(Lehigh University).-Is it possible through adsorption on t o the surface, have been you have water present in the nickel oxide sample? TABLE VI1 INTENSITIES OF C-H STRETCHING MODES-GASES LIQUIDHYDROCARBONS
AND
(11) D. F. Eggers, I. C. Hisatsune and L. Van Alten, THISJOURNAL. 59, 1124 (1955). (12) R. C. Golike, I. M. Mills, W. B. Person and B. Crawford, J . Chsm. Phyr., 26, 1266 (1956). (13) I. M. Nyquist, I. M. Mills, W. B. Person and B. Crawford, ibid., 26, 552 (1057). (14) 8. A. Francis, ibid.. 18, 861 (1950). (15) G. L. Kington and J. M. Holmes, Trans. Faraday Soc., 49, 417 (1953).
1,. H. LITTLE.-odY adsorbed water which is unaffected by evacuation for several hours a t 300' could be present on the sample. The spectrum of the porous silica support shows an intense h droxyl absorption band even after prolonged evacuation. &th samples of 0.1 mm. thickness it is possible to resolve (16) N. S. Bayliss, A. Chem., 8, 26 (1055).
R. H. Cole and L. H. Little, Avatralian J.
1622
AAGE SOLBAKKEN
AND
this band into a high frequency component due to free hydroxyl groups on the surface and a broad low frequency component due to hydrogen bonded surface hydroxyl groups. Strongly adsorbed water which remains after evacuation
LLOYDH. REYERSON
Vol. 63
would be hydrogen bonded and i t would not be possible from the infrared spectrum to distinguish betwcen this water and the surface hydroxyl groups which are involved in hydrogen bonding.
SORPTION AND MAGNETIC SUSCEPTIBILITY STUDIES ON NITRIC OXIDE-SILICA GEL SYSTEMS AT A NUMBER OF TEMPERATURES BY AAGESOLBAKKEN' AND LLOYD H. REYERSON Contribution from the School of Chemistry, University of Minnesota, Minneapolis, Minnesota Received March 8, 1969
Sorptions of nitric oxide by silica gel were determined a t 181, 193, 273 and 293°K. A movable magnet made it possible to measure the magnetic susceptibility of the sorbed nitric oyide at each point on the sorption isotherm. The results showed that the nitric oxide was sorbed in the same average magnetic state at 273 and 293'K. as it existed in the gas phase. HOWever, at 181 and 193°K.) the nitric oxide, up to nearly monolayer coverage, was sorbed entirely in a state m t h a magnetic moment equal to the excited I r a / , state. Additional sorbed nitric oxide had the same susceptibility as the gas at the same tem erature. This seems to be positive evidence that the first molecules sorbed a t these lower temperatures were all present on t i e surface in an excited state.
I n an earlier study in this Laboratory,2 the magnetic susceptibility of NO2 sorbed on silica gel was determined at two temperatures over a considerable sorption range. The results showed the NO2 was physically adsorbed but that it existed on the surface of the silica gel almost entirely as the nitrogen tetroxide dimer. During July, 1958, in a private communication to one of us, Professor J. H. deBoer reported that, in studies on the sorption of N O on silica gel at the Laboratories of the States Mines in Holland,* the sorbed system turned a greenish color ' and that this color intensified as sorptions at 0 were carried out a t lower and lower temperatures. A lighter color was observed when N O was sorbed on alumina gel. This suggested that the sorption surfaces might have catalytically changed the sorbed nitric oxide. If so, then by following the magnetic susceptibility of the strongly parama.gnetic nitric oxide, as it was sorbed by the silica gel, it should be possible to determine whether or not any change had taken place. The following experiments record a number of very interesting results obtained by such a study at temperatures ranging from 181 to 293°K. Experimental Procedure The general schematic layout of the eguipment used.in these experiments is shown in Fig. 1 and it was much like that reported in the previous workzexcept that a new, betterdesigned and better-cooled electromagnet had been built and installed in the travelling magnet frame. The field strength of this magnet was about 9000 oersteds when operating at full current strength and the pole pieces, as constructed, gave a very inhomogeneous field. A spherical glass bucket, containiing 0.1732 g. of very pure silica gel, was suspended on a quartz fiber from the quartz spiral spring. A glass jacket surrounded the tube containing the spiral spring as well as the mercury manometer. This jacket was kept at a constant temperature of 30" by circulating water from a thermostat maintained at this temperature. The diagram shows how the sample was maintained at a constant temperature by moving large quantities of air through a thermostat, then up a large insulated tube to a jacket surrounding the tube containing the sample, and then down through an outer jacket and out. Calibrated thermometers (1) Graduate Norwegian Fellow at the University of Minnesota from the Norwegian Institute of Teohnology, Trondheim, Norway. (2) L. H. Reyerson and John Wertz. THISJOURNAL, 63,234 (1949). (3) door C. Bokhoven and P. Zwietering, Chem. Weekblad, 62, 83 (1956).
determined the thermostat temperatures, and thermocouple junctions placed in the thermostat and in the jacket surrounding the sample made it possible to maintam the sample a t a constant temperature. Using solid COS in acetone, together with a bimetallic control to an immersion heater, permitted measurements to be made at 181 and 193"K., while the use of ice in water made possible the determinations at 273°K. A t 293"K., the air was blown through water thermostated to the desired temperature. A sample of the same silica gel that was used in the previous study2 was carefully dried and weighed into the sample bulb. It was thoroughly outgassed before each run and its magnetic susceptibility showed it to be free of paramagnetic impurities. The NO, that was transferred to the large flask, had been prepared for spectroscopic studies and proved to be free of other oxides of nitrogen by all the tests that were made on it. By surrounding the tip of the large flask with liquid nitrogen, the gaseous NO could be fyozen and thus handled easily during its transfer and use UI the sorption process. The sorption system was thoroughly outgassed while the flask of NO was shut off from it. The isotherm at a given temperature was obtained by measuring the extension of the calibrated quartz spring as the NO reached equilibrium with the silica gel, following the addition of successive increments of gas from the flask. Once sorption equilibrium was established for each increment or decrement of gas, the magnetic field was turned on and the precision lift raised or lowered the magnet until it exerted its maximum force on the sample with the sorbed NO. This permitted the determination of the magnetic susceptibility of the sorbed NO. The magnetic field had been calibrated by using 173.2 mg. of CuO in a Pyrex bulb weighing 62.5 mg. The calibration was carried out in an atmosphere of argon at 50 mm. and the results are given in Table I.
TABLE I CALIBRATION OF MAGNETIC FIELD Temp.. OK.
Force on CuO, dyne
(Ha H / b S ) "
2.0338 5.486 X loe 181 193 2.1711 5.466 X lo8 273 2.6171 5.378 X lo8 293 2.6732 5.340 X 10" a Calculated using magnetic susceptibility of Pyrex = and interpolated values of magnetic SUB-0.337 X ceptibilities of CuO from reference 5.
Results and Discussion The equilibrium sorption values for NO sorbed by 0.1732 g. of silica gel are given for the four temper& (4) J. M. Honig, Thesis. Univ. of Minn., 1952. (5) N. Perakis, A. Serres and T. Karantassis, J . phys. radium, 17, 134 (1956).
