SURFACE INTERACTIONS OF NaY AND DECATIONATED Y ZEOLITES WITH NITRICOXIDE
4163
Surface Interactions of NaY and Decationated Y Zeolites with Nitric Oxide as Determined by Electron Paramagnetic Resonance Spectroscopy by Jack H. Lunsford Department of Chemistry, Texas A & M University, College Station, Texas 77848
(Received M a y 18, 1968)
The epr spectrum of the adsorbed NO molecule was used to determine crystal-field interactions on the surface of NaY and decationated Y zeolites. The crystal-field splitting of the 2pa,* and 2p?r,* levels of N O was found to be 0.21 and 0.60 eV, respectively, for these two zeolites. Comparison of these results with other investigations shows that the extent of interaction is given by the sequence: decationated Y > ZnO > BaY > ZnS 21 MgO > Nay. Spectra of NO on decationated Y zeolites indicated considerable hyperfine coupling with 27Alnuclei at the adsorption sites. The concentration of NO species was found to be a function of the dehydration procedure as well as the extent of exchange for the decationated Y zeolites. A maximum concentration was found following dehydration in the temperature range from 600 to 700'. The importance of these results in establishing a valid model for the catalytic activity of zeolites is considered.
Introduction Recent interest in Y-type zeolites as active catalysts for a variety of organic reactions has prompted this electron paramagnetic resonance study of the interaction between adsorbed nitric oxide and zeolite surfaces. I n previous work the epr spectra of nitric oxide on MgO, ZnO, and ZnS were obtained.'n2 From the g tensor it was possible to deduce the relative interaction of the crystal field with the adsorbed molecule. Similar information using the adsorbed 0 2 - species has also been obtained for 1lgO and Zn03 in addition to N a y and BaY zeolite^.^ Qualitatively, one may conclude from these studies that very large electrostatic fields exist on the surface and that the extent of interaction is given by the sequence: ZnO > BaY > ZnS Rig0 > Nay. Pickert, et aLj5 have shown that there is a definite correlation of the cation charge and size with the catalytic activity for X- and Y-type zeolites containing various monovalent and divalent cations. Calculations show that sizable electrostatic fields exist on the zeolites with divalent cations.6 Benson and Boudart' have determined from an adsorption study that H2 adsorbs on CaY zeolites with a very high heat of adsorption whichL can be related to strong fields at the surface. Ward,8on the other hand, has concluded from his own infrared data and the catalytic results of Benesig that BrZnsted acid sites are responsible for the catalytic activity of decationated Y zeolites. Based upon a study of the effects of various cations on catalytic activity of Y-type zeolites, Richardson'O has proposed that both the electrostatic field and the Br#nsted acidity need to be considered, since the former actually determines the latter. A similar concept was also suggested by Hirschler.11 It was of interest in the present investigation to compare the epr spectrum of KO on very active de-
cationated Y zeolites with the spectrum on a NaY zeolite, which is known to be a relatively inactive catalyst. From the spectra it is possible to obtain not only the crystalline-field interaction but also the number of molecules adsorbed at particular sites and the hyperfine interaction of the adsorbed species with ions at the surface. One of the more important aspects of this study was to determine whether the concentration of active adsorption sites and the catalytic activity had the same functional dependence on the dehydration temperature.
Experimental Section A high-purity sodium form of Y-type zeolite (Lot KO. 11007-73) was supplied by the Linde Co. The sieve was either used in the sodium form, or the sodium ions were exchanged with ammonium ions by one of three different procedures. I n the first of these, 5.0 g of NaY was allowed to equilibrate for 7 days at room temperature in 50 ml of 12.5 M solution. This produced a sieve in which 50% of the sodium ions were exchanged for ammonium ions. To obtain a sample that was more extensively exchanged, 20 g of NaY was repeatedly slurried in hot 1.6 M NH4N03 and then filtered. The extent of exchange in this sieve was 90%. (1) J. H. Lunsford, J. Chem. Phys., 46, 4347 (1967). (2) J. H. Lunsford, J. Phys. Chem., 72, 2141 (1968). (3) J. H. Lunsford and J. P. Jayne, J. Chem. Phys., 44, 1487 (1966). (4) P.H. Kasai, ibid., 43, 3322 (1965). (5) P.E. Pickert, J. A. Rabo, J. A. Dempsey, and V. Schomaker, Proc. I n t . Congr. Catal., Srd, Amsterdam, 1964, 714 (1965). (6) J. A. Rabo, C. L. Angell, P. H. Kasai, and V. Schomaker, Discussions Faraday SOC.,41, 328 (1966). (7) J. E. Benson and M. Boudart, J. Catal., 8, 93 (1967). (8) J. W. Ward, ibid., 9, 225 (1967). (9) H . A. Benesi, ibid., 8, 368 (1967). (10) J. T. Richardson, ibid., 9, 182 (1967). (11) A. E. Hirschler, ibid., 2, 428 (1963). Volume 7 2 , Number 1 2 November 1068
4164 A third sample was prepared by equilibrating 0.23 g of NaY with a 1.8 M solution of NH4N03in DzO for 2 days at room temperature. The hydrogen in the NH4+ ion is known to exchange in a matter of minutes with the deuterium of D20 through the hydrolysis reaction.12 All three of the above samples were washed several times with distilled HzO or D20. After drying in air at loo", the sieves were broken into chips which had a maximum dimension of approximately 2 mm. The sieves were degassed under vacuum according to one of the following temperature programs: (a) 15 min at 100" intervals with 15 min to reach the preset temperature or (b) 105 min at 100" intervals with 15 min to reach the preset temperature, For identification the form of the sieve will be given first, followed by the per cent exchange and the dehydration temperature. After the sieve had been degassed in a 14-ml quartz or Vycor sample tube with a side arm 4 mm in diameter, the NO was admitted onto the sample at room temperature and at pressures ranging from 0.2 to 20 torr. The sample tube was sealed by means of a stopcock at room temperature, and the chips were tapped into the side arm. The side arm was then cooled in liquid nitrogen. I n order to obtain reproducible results, it was necessary to ensure that an adequate amount of NO was adsorbed on the portion of the sieve that was to be in the sensitive region of the microwave cavity. The samples were cooled in liquid nitrogen for at least 30 min before the epr spectrum was recorded. One sample of "4-Y-50 was also studied at 2°K. The 14K0,which had a minimum purity of 98.5% as received from a commercial source, was further purified by the freeze-pump technique. Liquid nitrogen was used to freeze the nitric oxide, The ISNO,enriched to 99.3% lSN,was purified in the same manner. A conventional X-band epr spectrometer with a TE102 mode microwave cavity was employed. The g values were obtained by comparison with the value for 2,2 diphenyl-l-picrylhydrazyl (g = 2.0036) or a phosphorus-doped silicon (g = 1.9987). Absolute measurements on the number of spins per gram of sample were made by using crystals of CuSO4.5H20 or the phosphorus-doped silicon as standardsea The absolute error in the spin concentration is estimated to be f30%, but the relative error between samples in this experiment is considerably less.
Results NO Adsorption on NaY Zeolites. The spectrum shown in Figure l a was observed when 14X0was adsorbed on N a y zeolites that had been degassed at temperatures greater than 400". This paramagnetic species can be easily removed from the surface by degassing the sample for 5 min at room temperature. When nitric oxide was not present on the sample, only a weak line at g = 2.003 was observed in the spectrum. The rather poorly resolved spectrum of Figure l a may The Journal of Physical Chemistry
JACKH. LUNSFORD Q1 = 1.989
100 G A U S S
Figure 1. Epr spectrum of 14N0 on a NaY zeolite (a) and of 14NO on MgO (b).
1
.
\
100 GAUSS
(C)
Figure 2. Epr spectrum of 14N0on a decationated
Y zeolite: (a) recorded a t 77°K; (b) recorded a t 2'K; (c) is a stick-figure representation of six equally spaced epr spectra of 14N0on MgO.
be analyzed by comparing it with the well-characterized spectrum of nitric oxide on magnesium oxide which is shown in Figure lb. It is apparent that the three 14Nhyperfine lines have been greatly broadened. This broadening may be attributed to motional effects of the NO molecules, rapid relaxation of the excited spin state, or magnetic dipole interactions with 23Nanuclei, which have a nuclear spin of s//z and a large magnetic moment. The nitrogen hyperfine splitting for NO on the NaY zeolite (28 f 2 G) is similar to that observed for NO on MgO (33 f 1 G); however, g1 has shifted to a lower value of 1.989 f 0.003. From the broad minimum in Figure la, the value of gil = 1.86 f 0.01 has been determined. NO Adsorption o n Decationated Y Zeolites. The epr (12) A. I. Brodskii and L. V. Sulima, Dokl. Akad. Nauk SSSR, 74, 513 (1950).
SURFACE INTERACTIOXS OF NaY
AND
DECATIONATED Y ZEOLITESWITH NITRICOXIDE
a = I 996
Figure 3. (a) Epr spectrum of l6NO on a decationated Y zeolite; (b) theoretical spectrum that resulted from the summation of the six curves (amplitude reduced) in (c); (c) six equally spaced epr spectra of 16NO on hfgO.
spectrum of nitric oxide adsorbed on decationated zeolites is much more complex, though better resolved than the spectrum of nitric oxide adsorbed on NaY zeolites. When 14N0 was adsorbed on a decationated zeolite, the spectra shown in Figures 2a and 2b were detected at 77 and 2"K, respectively. Again, the spectrum was reduced to that of the zeolite background upon brief evacuation at room temperature. Upon adsorption of W O the spectrum shown in Figure 3a was observed. The experimental spectra of Figures 2 and 3 indicate that the unpaired electron on the NO molecule is interacting with other nuclei as well as with its own nitrogen. The obvious other interaction is between the unpaired electron and 27Alwhich has a nuclear spin of 5/2. (This is the only naturally occurring isotope of Al.) Such an interaction would split each of the nitrogen hyperfine lines into six lines. I n the resulting spectrum there may be an overlap of some of these lines. Using the hyperfine splitting for NO on MgO and the relative amplitude of the maxima in Figure lb, a stick figure has been constructed (Figure 2c) to show how the curves in Figures 2a and b could be generated. This same approach was carried out in more detail in Figure 3c, where six spectra of '"0 on MgO have been displaced equally with respect to one another and then summed to give the resulting curve in Figure 3b. The only adjustable parameter in this procedure was the extent of displacement which is a measure of the 27Al hyperfine interaction. The good agreement between the theoretical and the experimental curves indicates that the KO is indeed interacting with aluminum. The experimental hyperfine coupling of 27Alis 14 2 G.
4165
Consideration was given to the possibility that these experimental spectra resulted from the interaction of KO with a proton ( I = l / 2 ) at the surface. Such an interaction would lead to only six hyperfine lines for 14N0 and four hyperfine lines for 16N0. This is not consistent with the experimental spectra. Furthermore, the deuterated sieve gave exactly the same spectra as the normal NH,Y sieve. It should be pointed out that the deuterium nucleus has a spin of 1 and a much smaller magnetic moment than hydrogen. The origin for the symmetrical displacement of the six spectra in Figures 2 and 3 is the point from which g L should be determined. By using the inflections in Figures 2a, 2b, and 3a, a value of gL = 1.996 A 0.002 was calculated. It is apparent that the high-field minimum, which is related to gll, shifted far downfield when compared with the spectrum of NO on MgO. For NO on decationated Y zeolites the value of g 1 I was 1.95 f: 0.01, While the estimated error of *0.01 in gi 1 is probably valid for an absolute determination, the experimental evidence is convincing that the relative positions of the high-field minimum can be more accurately defined. For example, the minimum for NO on the decationated Y zeolites is distinctly at a lower field than for NO on ZnO, even though the g values are reported as 1.95 and 1.94,2respectively. E f e c t of Dehydration and Extent of Exchange. The shape and magnetic parameters of the NO spectra on both NaY and decationated zeolites were unaffected by the extent and temperature of dehydration (or dehydroxylation), but the number of spins per gram of sample was quite strongly influenced by the dehydration procedure. I n addition, the extent of NH4+ exchange became an important factor in this study. From previous work it is clear that the nitric oxide molecule which contributes to the epr spectra is being adsorbed at a site that produces a strong, low-symmetry From the number of paracrystal-field magnetic species one can determine at least a lower limit for the number of such sites on the surface. Both the NaY and decationated Y zeolites developed an appreciable number of these sites only after the solids had been degassed at 400-500". Upon degassing at higher temperatures the spin concentration went through a maximum. The temperature range for this maximum was from 600 to 700", depending on the length of time that the sample was held at each temperature. These results are summarized in Figure 4. Each of the points for a particular sample was recorded at the same NO coverage, which was sufficiently high so that the number of spins was only a function of the number of special sites but not the total amount of NO adsorbed or frozen on the solid. One should note that the maximum number of spins per gram of zeolite was a function of the heating period as well as the degree of exchange for the decationated zeolites. The maximum spin concentration of 1.1 X 1019 spins/g was observed Volume 7,9, ivumber 12 November 1968
JACK H. LUNSFORD
4166 I
L.UI
tions.'s2 I n summary, the 2pn* orbital, which contains the unpaired electron, is degenerate in the absence of any low-symmetry crystal-field interaction. If the symmetry about the molecular axis is partially removed, as upon adsorption, one of the 2pn* orbitals will have a higher energy than the other. This splitting of energy levels is exhibited in the g tensor of the epr spectrum according to the equations
and
400
500 Dehydration
600
700
800
900
Temperature, "C
Figure 4. Spin concentration as a function of dehydration temperature for several zeolites: A, XH4-Y-90, degassed 105 min at each temperature; 0 , "4-Y-50, degassed 105 min a t each temperature; 0, "4-Y-50, degassed 15 min at each temperature; b , Nay, degassed 15 min a t each temperature.
for "4-Y-90-600, which had been degassed for 105 min at each 100" temperature interval, A cursory X-ray study was made on the NH4-Y-50 sieves to determine if the decrease in spin concentration between 700 and 800" could be attributed to a substantial loss in crystallinity. A comparison was made between the X-ray spectra of a sample dried in air at 120", a sample degassed to 700" followed by exposure to 0.4 torr of S O at room temperature, and a sample degassed to 800". These spectra showed that heating the zeolite to 700" and/or exposure to NO caused a slight loss in crystallinity but that the sample degassed a t 800" showed no further loss in crystallinity. In contrast, McAtee13 has recently shown by an oscillating X-ray diffractometer study that a 111 d line in XaY zeolites undergoes between 700 and 800" a sharp loss in intensity which is much more drastic than the loss in intensity of spins as shown in Figure 4. The effect of adding water back to a sample that had been dehydrated at 800" was also explored. The H2O or D2O was added back in differing amounts onto zeolites maintained at a range of temperatures. The samples were then degassed again, usually at 400". It was found that the spin concentration did not vary much from that observed following dehydration at 800".
Discussion Nature of the Active Site. The theoretical considerations that enable one to quantitatively determine the crystal-field interactions with the adsorbed KO molecule have been discussed in detail in previous publicaThe Journal of Physical Chemistry
where X is the spin-orbit coupling constant, g, is the free-electron g value, and A is the extent of splitting of the energy levels by the crystal field. For the NO molecule, X = 0.0150 eV. Using the experimentally determined values for g1 and gll, it was calculated that A = 0.21 eV for NO on NaY zeolites and A = 0.60 eV for KO on decationated Y zeolites. These values may be compared with A = 0.30 eV for NO on MgO' and A = 0.50 eV for NO on ZnO.* Clearly, the crystalfield interaction on decationated zeolites is the largest yet measured. The results confirm the 02- data4which indicate that NaY zeolites produce the smallest surface interaction yet determined. While the value of A indicates the strength of the interaction at the active site, it reveals very little about the exact configuration of the active site. On the other hand, the hyperfine interaction with 27Alin the decationated zeolites establishes that the NO molecules are either weakly bonded to certain surface aluminum ions or close enough to the aluminum ions for strong magnetic dipole interactions to produce the splitting. If the latter case were truz, the NO molecule would have to be approximately 1.2 A from the A1 n u ~ l e u s . ' ~It appears more likely that the NO is actually bonded to the aluminum, perhaps through a weak charge-transfer complex. The hyperfine interaction between NO and the NaY zeolite was much less, as indicated by the epr spectrum. This suggests that no suitable aluminum ions are available on the surface to act as adsorption centers. It is interesting to compare the results presented in Figure 4 for NH4-Y-50 with the data of Dollish and Halllj for perylene radical ions formed on a decationated zeolite that had been 45y0 exchanged with ammonium ions. They reported that the effect of the evacuation temperature on the ability of the sieve to form cation radicals is quite similar to that shown in Figure 4 for (13) J. L. McAtee, J . Catal., 9, 289 (1967). (14) M . Bersohn and J. C. Baird, "An Introduction to Electron Paramagnetic Resonance," W. A. Benjamin, Inc., New York, N. Y . , 1966, p 63. (15) F. R. Dollish and W. K. Hall, J . Phys. Chen., 71, 1005 (1967).
SURFACE INTERACTIONS OF NaY
AND
DECATIONATED Y ZEOLITES WITH NITRICOXIDE
the formation of the NO species. Furthermore, the 45% exchanged sieve was able to generate a maximum of 9.5 X 10'' cation radicals/g, whereas the "4-Y-50 sieve in this study was able to generate a comparable 5 X 10l8NO radicals/g. Uytterhoeven, Christner, and
4167
ionated surface. This results in defect sites such as those which are active in the formation of radical cations and the NO species. These sites then act inductively on local hydroxyl groups to form strong acids. Such an interaction is depicted by H I
Hall16 suggest that the dehydroxylated form of the zeolite contains cation-anion defect pairs. The cation defect consists of an accessible trigonal aluminum which would be strongly electrophilic. It is probable that the nitric oxide molecule is located at such a site where the trigonal aluminum introduces the hyperfine interaction. Relationship to Catalytic Activity. Benesig has recently shown by a study of the cracking of toluene over decationated zeolites that the conversion increased with dehydration temperature up to a maximum at about 600". Readdition of water to a zeolite after dehydration at 700" restored the activity to 70% of the maximum activity. Venuto and coworker^'^ have shown that a similar dependence on dehydration temperature exists for the ethylene-benzene alkylation reaction. Hughes and Whitela and later Wards used the infrared spectra of adsorbed pyridine to determine the concentration of Brgnsted and Lewis acid sites as a function of dehydration temperature. Both studies showed that the number of Brgnsted acid sites began around 500" to decrease monotonically with increasing pretreatment temperature. The number of Lewis acid sites began to increase at about the same temperature, but there is a discrepancy between the two studies as to whether a maximum was reached at around 600". Hughes and White18observed such a maximum but indicated that it may have been the result of exposure of the sample to water after the 600" heat treatment. Ward8 states that the catalytic data of Benesi correspond very well with the Brgnsted acid population and concludes that the Bronsted acidity is the seat of activity. However, the validity of this correlation is open to question, since the Br@nstedacid concentration, as determined by the pyridinium ion, is clearly decreasing while the catalytic activity is incyeasing in the temperature range between 500 and 600". It is only after the catalytic activity has reached its maximum that the two have the same functional dependence on the dehydration temperature. It appears that a model which includes both a strong electrostatic field and the hydroxyl proton could adequately explain the experimental data. According to this model, which is very similar to that proposed by Richardson'O for the cationic form of zeolites, the active surface is formed by a pavtial dehydration of the decat-
From 400 to 600" the number of catalytic centers is limited by the surface concentration of defect sites, but at temperatures greater than 600" the limiting factor is the number of hydroxyl groups adjacent to the defect sites. Water added back can replace these hydroxyl groups. Of course, at the higher dehydration temperatures the concentration of defect sites also begins to decrease. These relationships are shown in Figure 5
\
,/-7
/ /'
\
\
1
I
i
I I
I
I
I
200
300
400
500
I
600
I.
700
Dehydration Temperature,
I
800
900
O C
Figure 5. Comparison of the relative 3643-cm-l OH ion concentration: the catalytic activity for toluene cracking,g and the NO spin concentration for "4-Y-90 as a function of dehydration temperature: - -, 3643-cm-l OH- concentration; - - -, catalytic activity; , NO concentration.
-
(16) J. B. Uytterhoeven, L. G. Christner, and W. K. Hall, J . Phys. Chem. 69, 2117 (1965). (17) P. B. Venuto, L. A. Hamilton, P. S. Landis, and J. J. Wise, J. Catal., 5, 81 (1966). (18) T. R. Hughes and H. M. White, J . Phys. Chem., 71, 2192 (1967). Volume 72, Number 1.9 November 1968
ALLENR. OVERMAN
4168 where the 3643-cm-’ OH- ion concentration,8 the NO concentration, and the catalytic activitygare all plotted as a function of dehydration temperature. The BrZnsted acidity, as determined by the pyridinium ion concentration, could have been used in place of the OH ion concentration, but the latter is a more fundamental parameter in this model. In a sense it may be argued that Brqhsted acidity is still responsible for the catalytic activity; yet, it is only the very strong acid sites that are active in catalysis. In particular, according to this model, it is only those acid sites that result from the interaction of a defect site with an adjacent hydroxyl group. While the model may be used to explain the activity for reactions that are typically acid catalyzed, the defect site by itself may be active in other reactions
such as dehydrogenation. As demonstrated in this work, these sites are characterized by strong crystalfield interactions with adsorbed molecules. The interactions which split the ?r antibonding orbitals also split the T bonding orbitals. I n a molecule such as ethylene in which only one of the T bonding orbitals is filled these large perturbations could significantly alter the reactivity.
Acknowledgment. The author wishes to acknowledge the contribution of Mr. John Jayne in determining the epr spectrum at 2°K and Dr. Edward RiIeyers in obtaining the X-ray data. This work was partially supported by NASA Grant 239-62. The epr spectrometer was made available through the National Science Foundation, Grant GP-3767.
Convective Diffusion across a Porous Diaphragm
by Allen R. Overman Agronomy Department, University of Illinois, Urbana, Illinois 61801
(Received M a y 13, 1968)
An equation for solute transport across a porous diaphragm has been obtained from the one-dimensional continuity equation for convective diffusion. The equation was tested at 25” by using aqueous KC1 solution and a diaphragm of 5-p nominal pore size. Experiments were conducted at Peclet numbers of 0, -1.38, -2.71, -5.31, and -7.44. Agreement between theory and experiment was better than 3% except at -7.44, where the discrepancy was 13%. Good agreement between the predicted and observed transient time constants was also obtained.
I n convective diffusion one is concerned with the transport of matter by the combined processes of convection and molecular diffusion.’ This phenomenon is of considerable importance in porous media where both processes occur s i m u l t a n e ~ u s l y . ~In~ ~this paper a flow equation is developed for a porous diaphragm, and an aqueous KC1 solution is used to test the range of Peclet number over which the equation is applicable.
Theory
c- -- CL
co - c1
dC D-d2C - r-dX2
dX
=
0
where C is the solute concentration at position X, D is the diffusion coefficient for the system, and P is the average streaming velocity within the porous diaphragm. The solution to eq l subject to the boundary conditions C(X = 0) = Co and C(X = I) = Czis The Journal of Physical Chemistrg
(2)
where 0 = 1V / D is the Peclet number. This solution has been obtained previously4 and in an equivalent form by Mackie and me are^.^ The net rate of transport of solute, QN, at steady state can be written as QN
The approximate one-dimensional continuity equation for steady-state convective diffusion is given by
-1 e’ - 1
e’X/‘
-1-
=
dC
-DA- dX
+ QC
(3)
where A is the pore area of the diaphragm and Q is the flow rate of solvent. If eq 2 is used, in eq 3 there follows, after rearrangement (1) V. Levich, “Physicochemical Hydrodynamics,” Prentice-Hall, Inc., Englewood Cliffs, N. J., 1962. (2) J. Span and M. Ribaric, J . Chem. Phys., 41, 2347 (1964). (3) J. S. Mackie and P. Meares, Discussions Faraday Soc., 21, 111 (1956). (4) A. R. Overman and R. J. Miller, J . Phys. Chem., 72, 155 (1968).