A. V. DEO,T. T. CHUANG~AND I. G. DALLALANA
234 this produces only a slight change in lattice constant but a large increaise in cohesive energy. Calculations using the tight binlding approximation indicate that the presence of ti p oirbital in these sites significantly increases the overlap of the d wave functions of the metal atoms. We suggest that nitrogen and, to a lesser extent, hydrogen adsorbed in these sites on (1OO)W or Mo stabilizes the surface in an analogous fashion. The geometry of adjacent adsorbate atoms on the (100) plane is simillar to that of these atoms in the bulk compounds, and it is expected that the interactions should also be simila,r.
Sum,mary There are close similarities between nitrogen and
hydrogen as adsorbates on (100)Mo and between (100)Mo and (1OO)W as substrates for these gases; this implies that similar bonding mechanisms are operative. Yet the differences between the four systems, different LEED structures of hydrogen and different binding energies on the substrates, show definite effects of the “chemical” nature of the substrates. Precise experimental comparison of simple adsorption systems promises to provide an important means of elucidating the nature of solid surfaces and their interactions with adsorbates. However, the evidence to date indicates that, while ideas such as those discussed here may be possibilities, chemisorption i s complex enough that additional correlation between systems will be necessary to establish the character of adsorption bonds.
Infrared Studiies of Adsorption and Surface Reactions of Some Secondary Alcohols, C, to C,, on ?-Alumina and ?-Alumina Doped with Sodium Hydroxide by A. V. Deo, T. T. Chuang, and I. G. Dalla Lana* Department of Chemical and Petroleum Engineering, University of Alberta, Edmonton, Alberta, Canada (Received M a y 18, 1970) Publication costs assisted by the National Research Council of Canada
Infrared studies of the adsorption of secondary alcohols, Zpropanol, 2-butanol, and 3-pentanol, on y-alumina and y-alumina doped with sodium hydroxide catalysts, in the temperature range from room temperature to 300” are described. On y-alumina four types of surface species were observed: (i) alcohol hydrogen-bonded to surface hydroxyl groups; (ii) an alkoxide-type structure chemisorbed on Lewis acid sites, Ala+; (iii) a partial double bond type structure which on desorption gives an olefin; and (iv) a carboxylate species which is stable even at high temperatures. The physically adsorbed species are mainly dehydrated, forming a mixture of olefinic isomers. On sodium hydroxide doped y-alumina, in addition to the species i, ii, and iv, a ketonelike species was detected which on desorption gave a ketone. Two types of carboxylate species may exist on Isodium hydroxide doped catalyst. A mechanism for the formation of carboxylate species, identical for both of the catalysts studied, is proposed. The three alcohols follow similar mechanisms for dehydration and dehydrogenation, with 3-pentanol reacting at a lower temperature than encountered with 2-butanol and %propanol.
The Journal of Phyaka,l Chemistry, Vol. 76, No. 3, 1971
IR STUDIES OF .ADSO:RPTION OF SOME SECONDARY ALCOHOLS technique indicated the presence of three types of surface complexes : physically adsorbed alcohol, an alkoxide type species, and a carboxylate type species. Our previous r e s u h l on sodium hydroxide doped yalumina showed the presence of a conjugated hydrocarbon species in addition to the above three surface species. While studying the adsorption of isopropyl alcohol on alumina, Yakerson, et al.,6 suggested that hydrogenbonded alcohol yields propylene while alcohol bonded to a lattice aluminum atom forms acetone, with the formation of a complex with acetate character as a side reaction. Knozinger, et al.," studied the kinetics of dehydration of primary and secondary alcohols and reported that the stable alkoxide is an intermediate in the formation of ether while the olefin is formed through a secondary elinnin,ation-type reaction intermediate. Tamaru and ~oworkers,~ studying the decomposition of 2-propanol over alumina, suggested that the alkoxide is an intermediate for the function of both ether and olefin. Pines and Manassen* reported that the olefin formation also goes through a secondary elimination reaction intermediate. Our present paper discusses the results of adsorption of secondary alcohols from C3 to C5 on y-alumina and sodium hydroxide doped y-alumina catalyst, systems from room temperature to 300".
Experimental Details The procedures for preparing y-alumina (Cabot Corp., Boston, Mass.) and 2% sodium hydroxide doped y-alumina catalyst wafers, their pretreatment, ,and the recording of their infrared spectra on a Perkin-Elmer Model 621 infrared spectrophotometer were similar to those reported earlier.' All of the alcohols, ketones, and alkenes used in this work were of spectroscopic grade and were purified by freezing and thaTwing under vacuum. Differential spectra were recorded by placing an identical cell, but without the catalyst wafer, in the reference beam to eliminate any gas phase interference. To evaluate the extent of reaction, tlhe gas phase was analyzed by infrared and gas chromatographic or mass spectral analysis. After recording the base line spectrum, i.e., the spectrum obtained for the catalyst wafer in an evacuated cell, the reactant vapor was introduced to a pressure of 10 Torr. The reactant was exposed to the wafer, and the cell was then either pumped off or heated to various temperature levels. Depending on the sequence of experiments being performed, various combinations of hesting and/or degassing cycles were performed. The ir spectra were measured a t room temperature during the experimental cycles. Figures 1 and 2 show selected representative spectra and the significant changes in them which were noted during this study. To improve differentiating between spectra,
235
I
j
3800
I
I
I
3600
/
.
(
3400" 3000
,
,
I
2800
I
I
I
2600'1800
FREQUENCY
I
/
I
1600
I
1
1400
)
1
1200
I
IO
(CM-')
Figure 1. 2-Propanol on 7-alumina: 1, base line; 2, ?-alumina and 2-propanol at 10 Torr and room temperature; 3, 7-alumina and Zpropanol at 300" for 1 hr; 4, y-alumina and water vapor a t 5 Torr and room temperature; 5, degassing of 3 at room temperature for 1 hr; 0, degassing of 5 at 300' for 1 hr.
the successive spectral curves are shown displaced from one another on an arbitrary transmission coordinate scale. The curves drawn in the region from 1000 to 1800 cm-l are shown with the base line spectrum subtracted from the recorded adsorption spectrum.
Results and Discussion The base line spectra of y-alumina and sodium hydroxide doped y-alumina were discussed previously. * The base line of y-alumina was interpreted using the model proposed by Peri. 9,10 The level of sodium hydroxide doping which was used results in the disappearance of the high-frequency 3785-cm-l hydroxyl bond suggesting a reaction to form =A1-0-Na. If this interpretation is correct it is inconsistent with Peri's view that the higher acidity is associated with the lower frequency hydroxyl groups. I . ,%'-Propanol on Pure y-Alumina. Figure 1 shows the spectra of 2-propanol adsorbed on pure y-alumina (4) R. 0 . Kagel, J. Phya. Chem., 71, 844 (1967). (5) Y.Yakerson, L. Lafer, and A. Rubinshtein, Dokl. Akad. Nauk SSSR, 174, (l),111 (1967). (6) H. Knozinger, H.Buhl, and E. Ress, J . CataE., 12, 121 (1968). (7) Y.Soba, T. Onishi, and K. Tamaru, Trans. Faraday Soc., 65, 2215 (1969). (8) H. Pines and J. Manaasen, Advan. Catal., 16,49 (1966). (9) J. B. Peri and R. B. Hannan, J . Phgs. Chem., 64, 1526 (1960). (10) J. B. Peri, ibid., 69, 20 (1965). The Journal of Phusieal Chemistrg, Vol. 76, No. 9, 1971
A. V. DEO,T. T. CHUANO,AND I. G. DALLA LANA
236
on adsorption merge into a single broad band at 1170 em-'. These observations suggest that 2-propanol is physically adsorbed through the hydrogen bonding between the -C-OH of the alcohol and the surface hydroxyl groups (as shown in structure I). I n addition to the hydrogen-bonded species, the close similarity between the spectrum of the adsorbed species and that of aluminum isopropoxide suggests the presence of an alkoxide214type of structure, 11. Pumping at room temperature and up to 80" removed most of the physically adsorbed hydrogen-bonded species, leaving the alkoxide type structure on the surface. The formation of such an alkoxide species on the electronabstracting Lewis acid site, A13+,is in agreement with other work. l12s4
H
I
H&-C-CHs
H
I
9, :H
/AtI
100
3600
I
3000
1
'
* I
2800
FF!EQUENCY
h00
1600
1400
1200
11
1
(CM-')
Figure,2. 2-Propanol on NaOH doped 7-alumina: 1, base line; 2, doped alumina and 2-propanol at 10 Torr and room temperature; 3, doped alumina and 2-propanol at 300" for 1 hr; 4, doped alumina and acetone at 10 Torr and room temperature; 5, degassing of 3 at room temperature for 1 hr; 6, degassing of 6 at 300" for 1 hr; 7, degassing of 4 at 300' for 1 hr.
from room temperature to 300". Up to 80" (curve 2), the surface hydroxyl groups are displaced considerably to the lower frequency with the formation of a large broad hydrog en-bonded hydroxyl region. On adsorption, the three bandls observed in the gas phase at 2981, 2972, and 2888 cm-l have shifted to the lower frequencies at 2960, 2930, and 2870 cm-l. The C-H bending region shows almost no change in frequencies except that the intensity of the band at 1465 cm-l has increased twofold over that in the gas phase in contrast to the band at 1373 om-l. These changes are similar to those observed during the transition from gaa to liquid phaser;.6 The three distinct bands at 1245, 1145, and 1065 em-1 in the 0-H bending and C-0 stretching regions, (observed in the gas phase spectrum, The Jozirnal of Phgsysical Chemistry, Vol. 76, No. 3, 1971
&-cH, I 0
1
O/H I
.I-_i_ll+i
I
H,C-
+H
n
On heating above 100' and up to 300" (curve 3), all of the band structure attributable to these two species disappears, giving propylene in the gas phase and leaving adsorbed water on the alumina surface. This is indicated by the strong band at 1635 em-', the 0-H bending vibration. For comparison, curve 4 shows this band when water vapor is adsorbed on y-alumina. This adsorbed water can be removed by pumping, even at room temperature. The dehydration reaction* can pr\oceed as shown in reaction 1.
-1
A1 / \
/O\
Above 130°, two new bands appear at 1575 and 1465 cm-1, and they increase in intensity with increasing temperature. The observed bands could be assigned to the symmetrical and asymmetrical stretching vibrations of the COO- groups of an acetate.2 The formation of a carboxylate from the alkoxide species as explained by Greenler2 and Kagel4 has been subject to criticism by Fink;ll Le., it seems unlikely that three hydrogen atoms will split from the alkoxide and an adjacent hydroxyl group. However, the adsorption of acetone on y-alumina at 220" produced the same (11) P.Fink, Rev. Roum. Chim., 14, 811 (1969).
IRSTUDIESOF ADSORPTION OF SOME SECONDARY ALCOHOLS infrared characteristics that were observed for the proposed carboxylate species IV obtained with 2-propanol. Although acetone was not detected as an intermediate product from 2-propano1, the adsorbed acetone may have been present in the form of species I11 preceding the formation of the carboxylate structure. Additional evidence12 supporting this explanation arises from the studies with the sodium hydroxide doped yalumina. The doping eliminates or modifies active surface hydroxyl groups. Under these conditions, large amounts of the adsorbed acetone-like species I11 can be found by adsorbing 2-propanol at 130". Upon further heating above 220", large amounts of the carboxyla,te species were detected. Reaction 2 seems plausible for the formation of carboxylate. CH3 I
/?\
/O\
237
propionate' structure. Methane accompanied the dehydration products and propionic acid was found upon hydrolysis at 220". These observations are consistent with the explanation for the acetate structure formation shown in reaction 1. It would thus be anticipated that a secondary alcohol may form some carboxylate and an alkane fragment upon heating in the presence of y-alumina. III. 3-Pentanol on y-Alumina. The spectrum of 3-pentanol adsorbed on y-alumina at room temperature is similar to those for 2-propanol and 2-butanol, showing the presence of both hydrogen-bonded and alkoxide species. On heating a8t80", a stable new band was observed at 1685 em-' accompanied by the formation of a trace amount of mixture of cis- and trans-2-pentene. This 1685-cm-' band occurs at a frequency too high to be attributable to the 0-H bending band of water formed on dehydration. An alternative explanation attributes this 1685-cm-l band to some partial double bond character similar to but not necessarily identical with that associated with an adsorbed olefin. This species may be represented by structure V, which could be an intermediate in the dehydration reaction.
t PH \
If1
Iv
The presence of methane in the gas product was confirmed by high-resolution mass spectral analysis. I n addition, the acetate structure IV could be partially hydrolyzed by water above 220" to form acetic acid. These observations further support the proposed reaction steps (2). The formation of ether observed elsewherel6particularly at low temperatures, was not observed in this study. It is possible that the ether is present in undetectable amounts or that it rapidly forms the alkoxide species.la Moreover, the formation of ether from secondary alcohols it thermodynamically less favorable than the formation of' 01efins.l~ I I . 2-Butanol on y-Alumina. The spectra obtained from the adsorption of 2-butanol on y-alumina up to 300" suggested that the surface reactions are similar to those observed with 2-propanol. The products from dehydration were delected in the gas phase to be a mixture of 1-butene and trans- and cis-2-butene. Consistent with thermodynamics, their distribution appeared in the order trans-2- > cis-2- > 1-butene. Little of the carboxylate species was noted even at a temperature of 300". Since the carboxylate species is formed from the alkoxide species, the reduced amount of alkoxide initially presept could account for this. Two infrared bands detected at 1561 and 1479 cm-l and analysis of the gas phase confirmed the presence of a
0I
/*\
P\V
A similar band at 1675 em-l wae observed on adsorbing either cis-2-pentene or trans-2-pentene on y-alumina, stable even on prolonged pumping at room temperature. Because of its position near to this frequency and because it was stable and reproducibly obtained only from 3-pentanol adsorption, the 1685-em-' band was not attributed to an adsorbed olefin, However, the existence of the intermediate V does require further substantiation. Above 80°, the dehydration proceeds at a faster rate, making it more difficult to observe the 1685-em-' band. The 1640-em-' band due to 0-H bending of adsorbed water increases with temperature and above 220", 1-pentene is formed. At 130°, the carboxylate species is formed along with the alkoxide species, shown by an increase in the intensity of the broad band at 1150 em-1. The carboxylate structure (in this case, propionate) remains stable even when pumping at 300" and would be expected to produce ethane, analogous to the methane formed in reaction 2. I V . $-Propanol on NaOH Doped y-Alumina. Fig(12) T.Chuang and I. G. Dalla Lana, unpublished results. (13) E. Heiba and P. Landis, J . Catal., 3, 471 (1964). (14) H. Knoainger and R. Kohne, ibid., 3, 559 (1964).
The Journal of Physical Chemktry, Vol. 76,No. 9,1971
238 ure 2 shows the spectra obtained from adsorption of 2-propanol on sodium hydroxide doped y-alumina from room temperature to 300". At temperatures up to 130" (curve 211,the hydrogen-bonded species I is formed to a much lerrser degree than was observed on the yalumina. The elimination of many surface hydroxyl groups by re>Iction to form =A1-O-Na would account for this reduced physical adsorption. The presence of the allcoxidel species I1 was shown by the deep band a t 1165 cm- l. Above 130") dehydrogenation occurs to form acetone along with mino, dehydration indicated by a trace of propylene. The bands at 1705, 1640, 1440, 1360, 1300, and 3227 em-' (curve 3) originate with the acetone adsorbed on R13+ sites. The band at 1640 cm-l results from the adsorbed acetone as well as from some adsorbed water formed during dehydration. In addition to this band structure, four more bands are observed a t 1572, 1456, 1330, and 1170 em-' (curve 3). The pair of bandEi al, 1572 and 1456 em-' are attributed to the carboxylate species, similar to those found on y-alumina. The remaining two bands at 1330 and 1170 em-', which also always appear as a pair and occur in the cttrboxylate frequency range, were tentatively assigned to a second type of carboxylate structure. On pumping from room temperature (curve 5) to 300" (curve 6), the banid structure due to acetone completely disappears, leaving only the four carboxylate bands. Additional atucliesl2 with only the ketone and carboxylite adsorbed species present on the doped alumina indicated, upon ireahing, a decrease in the intensity of ketone bands :dong with an increase in intensity of both pairs of carboxylate bonds. Further, mass spectral analysis of the dega,ssedphase showed only methane to be formed, in Rubstmtial agreement with the proposed mechanism shown in reaction 2. On this basis, the assignment of th~eadditional band pair at 1330 and 1170 cm-I to a second carboxylate structure seemed logical. The presenre of these two pairs of bands may be the result of differences in charge distributions and in symmetry considerations for oxygen atoms and their energy of bonding to the surface.lG By doping y-alumina with sodium hydroxide, it was observed that the high-frequenoy hydroxyl groups disappeared. This, according to Peri's model, should change the charge distribution of oxygen atoms which are responsible for the carboxylate formation. Ace tone was adsorbed on unused sodium hydroxide doped r-alumina catalyst, and the system was then heated to 130" (curve 4) to compare adsorbed acetone and its behavior with the intermediate 111which led to the formation of acetone. This spectrum is very similar to that observed on heating the catalyst in the presence of 2-propanol (curve 3). Upon degassing the catalyst expoised to pure acetone a t 300", only the carbo xylate structure remained (curve 7). This conThe Journal of Physical Chemistry, Vol. 76, No. 2, 1971
A. V. DEO,T. T. CHUANG, AND I. G.DALLA LANA firms the formation of an adsorbed acetone-like species, 111, like that obtained from 2-propanol and which on desorption gives acetone. V . ,%Butanol on NaOH Doped y-Alumina. The spectra of 2-butanol adsorbed on sodium hydroxide doped y-alumina from room temperature to 300" were also studied but are not shown because of their similarity to Figure 2. At temperatures up to SO", the alkoxide species is formed with very little hydrogenbonded species appearing. Near 130", the dehydrogenation reaction starts to form species 11 as shown by the increased band intensity at 1730 cm-l attributed to the C=O stretching frequency. After heating at 300", the carboxylate structure spectrum became dominant. Heating of adsorbed 2-butanone on a newly prepared catalyst wafer gave spectra similar to those from 2-butanol. Both 2-butanol and 2-butanone produced carboxylates with bands identical with those observed with l-propanol' on the doped y-alumina. VI. 3-PenCanol and NaOH Doped y-Alumina. The various spectra of 3-pentanol adsorbed on sodium hydroxide doped y-alumina from room temperature to 300" were also compared. At room tempertl.ture, the main species is again the alkoxide type. The dehydrogenation reaction occurs a t a still lower temperature, 80", and continues up to 300" with the formation of the stable carboxylate species. Adsorption of 3-pentanone formed by heating shows spectral characteristics comparable to those expected from reaction 2.
Conclusions 1. y-Alumina. Secondary alcohols are primarily dehydrated by hydrogen bonding to surface hydroxyl groups on y-alumina a t temperatures ranging from 80" for 3-pentanol to higher temperatures for 2-butanol and 2-propanol, respectively. At lower temperatures, an alkoxide surface species also appeared and reacted a t higher temperatures to form a surface carboxylate structure, stable to 300". Contrary to other reportsl7#la the alkoxide did not lead to the formation of detectable amounts of ether. With 3-pentanol, a hydrogen-bonded surface intermediate suspected of possessing a partial double bond character was postulated to form during the dehydration but before formation and desorption of the olefin occurs. A mixture of olefins was obtained with some catalytic selectivity exhibited. Up to 80", only 2pentenes were observed and above SO", significant 1pentene also appeared. I I . NaOH Doped y-Alumina. Doping of y-alumina with sodium hydroxide reduced the number of high-frequency surface hydroxyl groups, thus leading to a reduction in the hydrogen-bonded surface species formed from secondary alcohols. Correspondingly increased amounts of alkoxide surface species were detected for the three (15) L. Bertsch and H. Habgood, J. Phys. Clzem., 67, 1621 (1963).
ZWANZIGTHEORY OF DIELECTRIC FRICTION
239
alcohols studied. At high temperatures, minor dehydration pas accompanied by increasing dehydrogenation to ketones. h - 1 adsorbed ketone-like intermediate was observed analogous to the adsorbed secondary alcohol. With further heating to 300" the ketone-like intermediate formed a stable carboxylate surface structure accompanied by release of methane from the secondary alcohols, 2-propanol and 2-butanol. The doping of y-etlumina with sodium hydroxide alters the catalytic sctivity from dehydration to dehydro-
genation. Since two types of sites are involved, elimination of the more a,ctive surface hydroxyl groups by doping enables the less active aluminum ion sites to assume the primary catalytic role. It appears unlikely that a product degassed from the surface carboxylate species will appear in other than trace amounts. Acknowledgment. Financial support for A. V. D. and T. T. C. from the National Research Council of Canada and the University of Alberta is gratefully acknowledged.
An Examination of the Zwanzig Theory of Dielectric Friction. I1 by Roberto Fernindez-Prini and Gordon Atkinson* Department of Chemistry, University of Maryland, College Park, Maryland
$0740
(Received September 26, 1970)
Publication costs assisted by Ofice of Saline Water, Department of the Interior
The recent revision of the Zwaneig theory for the dielectric friction effect on ion mobility has been examined for representative monovalent ions in a number of pure solvents and solvent mixtures. The theory is shown to predict quantitatively the dependence of mobility on ion radius in dipolar aprotic solvents. The correct trends are predicted in protic solvents but the absolute magnitudes of the mobilities show systematic devia. tions from the predictions. Some possible explanations for these deviations are advanced.
introduction One of the most unappealing features of the classical electrolyte theory is the primitive view taken of ionsolvent interactioiis The Debye-Fuoss-Onsager solvent is a classical continuum completely characterizable by macroscopic paratmeters such as dielectric constant, viscosity, and density. Generally it is also assumed that the solvent parasmetersare those of the pure solvent unaffected by the presence of the ions. Since most research workers in the field had little faith that this was an adequate model, the search for more sophisticated views of ion-solvent interaction has gone on simultaneously with the search for a more complete treatment of ion-ion interactions. The ionic conductance a t infinite dilution, Xio, has been a favorite measure of ion-solvent interaction for many years. However, the basis for the discussion was commonly taken as Eltokes' law
Since it became clear that Ti was rarely- equal to the crystallographic radius, a literature developed on the discussion of solvation in terms of the discrepancy, T i (Stokes) > T i (crystal). The discussion was often couched in terms of solvation numbers. The discussion, of course, was somewhat muted in the cases where T i (Stokes) < Ti (crystal). However, it is well known that most of this type of calculation is of dubious value since the frictional force act~ingon a moving ion depends not only on the viscosity of the medium but on its dielectric pr0perties.l Boyd2 and Zwanzig3 calculated the dielectric frictional force assuming the relative velocity between the ion and any volume element in the medium is constant. This frictional force arises from the energy dissipated by the ion in orienting the solvent dipoles it encounters as it moves through the medium. Combining this with Stokes' law gives a Walden product expression of the form (Xioq)-l
I\io =
~tF'/(6xN~iq)
(1)
(all symbols are defjined in the Appendix) which leads to the often-used Walden rule (Xioq)
=
-.
252
6xN1.i
(2)
=
Ari
+
B/Ti3
(3)
where A and B characterize the viscous and dielectric (1) ( a ) M. Born, 2.Phgsik., 1 , 221 (1920); (b) R. M. Fuoss, Proc. Nat. Acad. Sei., 45, 807 (1959). (2) R. H. Boyd, J. Chem. Phys., 35, 1281 (1961); 39, 2376 (1963). (3) R. Zwanzig, ibid., 38, 1603 (1963).
The Journal of Physical Chemistry, Vol. 76, N o . 2, 1971
