J. Phys. Chem. 1991,95, 9999-10004 TABLE 111: Interatomic Distances of Framework Atoms from Water
Moleeules
atoms All-01
All-011
dist, A 2.056
AII-OVI AI 1-0111
1.940 3.974 4.003
AII-OV
4.440
A12-01 AI2-0111
3.898 4.276
A13-OVl AI 3-01 I I
4.02 1 4.367
atoms PI41 P1-011 P1-01' PI-OVI PI-0111
dist, A 3.840 3.907 4.153
P2-01 P2-0111 P2-011 P2-OVI
3.570 3.658 4.163
P3-OVI P3-0111
3.810 4.259
4.369
4.452
3.98 I
"Calculated from atomic positions given in ref 1 I . distinguish by N M R is problematical. Since the hydrogen positions are not known with certainty, we can only infer that P2 and P3 are sufficiently differently affected by the water environment to yield separate resonances. Phosphorus cross-polarization measurements at high water contents show that the resonance peak a t -23.9 ppm is enhan~ed.~JO This was interpreted as evidence that the AIV1was in the 6-MR. In contrast, Maistriau et aI.l3 chose this peak as indicating phosphorus in the 4-MR. However, it is now known that water indeed coordinates to aluminum in the 4-MR, and by the arguments presented under Results the 4-MR phosphorus peak is at -34.1 ppm. Most likely, the resonance at -23.9 ppm is that due to P2 since it is the one most influenced by water molecules. One other point requires discussion. Both we4sSand Duncan et a1.6 have found that the AI/P ratio in H1 and AIP04-8 is greater than 1. If this aluminum is in the framework, it would create AI-0-AI linkages or AI-OH H0-AI type sites. Alternatively, the excess aluminum could be within the cavity and would be expected to be 6-coordinate. Evidence for these types of aluminum have not been found in the N M R spectra. However, it has also been shown4 that H1 hydrolyzes in water much more readily,than VPI-5 and that more phosphate than aluminum is solubilized. This hydrolysis would be expected to create the hydroxyl type sites within the aluminum phosphate framework and on dehydration to create Lewis acid sites. Thus, we feel that such sites already exist in HI and are responsible for the ready conversion of H I to AIPOp-8.4*6~'8*'9
9999
In addition, the missing framework phosphate is thought to be responsible for the different ratio of peaks in the spectrum (-0% water) from the expected 1:2:3 ratio. In contrast to H1, VPI-5 has an aluminum/phosphorus ratio very close to 1 so that the number of open framework sites is quite small. Thus, VPI-5 hydrolyzes to a much smaller extent in water than H1 and is much more thermally stable but on extensive exposure to water eventually loses enough phosphate that conversion to AIP04-8 can take place. Annen et aLZ0have shown that VPI-5 does not convert to AlP04-8 on heating in vacuo or in the absence of water. However, we find that stirring VPI-5 in water for up to 24 h converts about 50% to AIP04-8. Breaking the A I U P bond by dehydration as indicated by our proposed mechanism is considered to be akin to the hydrolysis process and probably is keyed by strain within the 18-ring framework as is phosphate hydrolysis. Dehydration is accompanied by a decrease in the value of the a axis and an increase in the c axis, which may well alleviate some of the strain. According to our picture, phosphate must be removed to form A1P04-8 from VPI-5. In the absence of water there is no mechanism for forming H3P04 and P20, would have to volatilize for the transformation to take place. This volatilization is an inherently very energetic process and unlikely to occur given the separation of phosphate groups in the framework. Thus, it is not surprising that VPI-5 is thermally stable in the absence of moisture.z0 One final caution is necessary. We notice that in many recent publication^'^^'^^^^^^^^^^ the authors have synthesized what we refer to aqH1. Since the behaviors of H1 and VPI-5 are different, it would be well for authors to give a detailed description of their preparations accompanied by analytical data. This would avoid much confusion in dealing with 18-ring systems.
Acknowledgment. We thank the Regents of Texas A & M University for financial support of this work under the Materials Science and Engineering Initiative. Registry No. H20,7732-18-5. (18) Prasad, S.;Balakrishnan, I. Inorg. Chem. 1991, 30, 4830. (19) Vogt, E. T. C.; Richardson, J. W . ,Jr. J. Solid Srare Chem. 1990,87, 469. (20) Annen, M. J.; Young, D.; Davis, M. E.; Cavin, 0. C.; Hubbard, C. R.J . Phys. Chem. 1991, 95, 1380. (21) Liu, X.;Klinowski, J. J . Phys. Chem. 1991, submitted for publication.
Monolayer Structures of Niobic Acids Supported on SiOp and Their Catalytic Activities for Esterification of Acetic Acid with Ethanol Masayuki Shirai, Kiyotaka Asakura, and Yasuhiro Iwasawa* Department of Chemistry, Faculty of Science, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113, Japan (Received: April 10, 1991)
New submonolayer niobic acid catalysts were prepared by the reaction of Nb(OCzH5)5with surface OH groups of SOz, followed by H20treatment. The surface structures were characterized by means of EXAFS, XRD, XRF, and R - I R . The niobic acids on S O z up to 8.0 wt 9% Nb loadings were found to grow in a monolayer mode. These catalysts showed activities 20 times as large as that of a niobic acid bulk catalyst for the esterification of acetic acid with ethanol. The activity was referred to Lewis acid sites in the monolayer structure, which was stable even after evacuation at high temperatures such as 813 K.
Introduction Niobium has recently attracted much attention as a catalyst material.'+' Nb acid bulk behaves as a solid acid which catalyzes (1) For example, articles in:
KO,E. I. Cafal. Today 1990, 8 .
0022-365419 112095-9999$02.50/0
isomerization of butene, hydration of ethene, esterification of acetic acid with ethanol, and so on.1*12 N b acid bulk has the advantage (2) Nishimura, M.; Asakura, K.; Iwasawa. Y. J . Chem. Soc., Chem. Commun. 1986. 1660.
0 1991 American Chemical Society
loo00 The Journal of Physical Chemistry, Vol. 95, No. 24, 1991
Shirai et al.
SCHEME I: Preparation Steps for-NTobic Acid Layer on SiOl H H H
0 0 0
+
Nb(OC2H5)5
in hexane
473 K
293 K
evac
7iZ
’
evac
373 K
A
W N i o b i c - a c i d layer
of not being deactivated by water, unlike many other solid acids. It has been reported that Bronsted acid sites on the N b acid bulk play an important role in the above catalyses.’*12 It is well-known that Ni/Nb205 and Pt/Nb205 show SMSI phenomena.13-15 Asakura et al. made a Nb205/Si02sample by the reaction between Nb(OC2H5), and silanol groups, followed by calcination, and they demonstrated that niobium oxide had a one atomic layer structure on Si02surface by EXAFS spectroscopy.M They also reported that Pt particles dispersed on the one atomic layer Nb20S(Pt/Nb205/Si02) did not show a SMSI phenomenon. Rather, Pt/Nb20S/Si02 pretreated at 773 K showed a higher activity than that pretreated at 473 K for the hydrogenation of ethene.,” However, KO and his co-workers reported that the Ni/Nb205/Si02 catalyst showed a weak SMSI phenomenon as compared to Ni/Nb205.1b18 Wachs and Hardcastle prepared Nb205/Si02,Nb2O5/AI2O3, and Nb20J/Ti02 samples by an impregnation method using methanol solution of Nb(OC2H5)s.19 They claimed by Raman spectroscopy that niobia formed monolayer structures on A1203 and Ti02, while Nb205 particles were made on SiOz. Irrespective of the discrepancy in chemistry of niobium species on SO2,one atomic layers on the high surface area oxides provide a potential way to produce a new class of surface catalyst with high efficiency. Moreover, the one atomic overlayers interact directly with supports to generate a unique interface catalysis. In the present paper, we report the structure and growth mode of niobic acid layers on a S i 0 2 surface prepared according to different preparation procedures from those for the niobium oxide monolayers and their catalysis: the active sites for the esterification reaction of acetic acid with ethanol are also discussed.
Experimental Section Catalyst Preparation. Si02-attached niobic acid monolayers were prepared as shown in Scheme I. The reaction of Nb(0C2H5)s with the O H groups of Si02 was carried out under high-purity (99.9999%) Ar atmosphere or under vacuum. Nb(3) (4) (5) (6) (7) 8, 57. (8)
Nishimura, M.; Asakura, K.; Iwasawa, Y. Chem. Lerr. 1987, 573. Asakura, K.; Iwasawa, Y. Chem. Lerr. 1988, 859. Asakura, K.; Iwasawa, Y. Chem. Lerr. 1988,6339. Asakura, K.; Iwasawa, Y. J . Phys. Chem. 1991, 95, 171 1. Shirai, M.; Ichikuni, N.; Asakura, K.; Iwasawa, Y.Catal. Today 1990.
Ichikuni. N.; Asakura, K.; Iwasawa, Y. J . Chem. SOC.,Chem. Commun. 1991, 112. (9) Wada, Y . ; Morikawa, A. Bull. Chem. SOC.Jpn. 1987, 60, 3509. (10) lizuka, T.; Ogasawara, K.; Tanabe, K. Bull. Chem. Soc. Jpn. 1983, 56, 2927. (11) Ogasawara, K.;lizuka, T.; Tanabe, K. Chem. Leu. 1984, 645. (12) Chen, 2.;lizuka, T.; Tanabe, K. Chem. Lerr. 1984, 1085. (13) Tauster, S.J.; Fung, S.C. J . Catal. 1978, 55, 29. (14) Tauster, S. J.; Fung, S. C.; Garden R. L. J . Am. Chem. SOC.1978, 100. 170. (1 5) Baker, R. T. K., Tauster, S.J., Dumesic, J. A., Eds. Srrong MetalSupporr fnreracrions;ACS Symposium Series 298; American Chemical Society: Washington, DC, 1986. (16) Blackmond, D. G.;KO.E. 1. J . Carol. 1985, 94, 343. (17) KO,E. 1.; Bafrali, R.;Nuhfer, N . T.; Wagner, N . J. J . Catal. 1985, 95, 260. (18) Burke, P. A.; KO,E. 1. J . Catal. 1989, 116, 230. (19) Wachs, E. I.; Hardcastle, F. D. Proc.-Inr. Congr. Caral., 91h 1988, 3, 1449.
0
-
0
373 473 573 673 773 073
Evac. Temp./ K
Figure 1. Initial rates of the esterificationreaction on a niobic acid bulk and on the 2 wt % A-Nb205/Si02catalyst (0)as a function catalyst (0) of the pre-evacuation temperature.
(OCZHS)S(Soekawa Chemicals) was reacted in a dry hexane with silica (Aerosil200, surface area, 220 m2 g-I), which was annealed for 2 h at 473 K in situ before use as support, to control the concentration of silanol groups to be 5 O H nm-2. After this attachment reaction, the sample was washed by a dry hexane three times to remove unreacted Nb(OC2Hs)s in a flow of Ar. Following decantation of the residual hexane in a flow of Ar, the attaching reaction was completed by heating at 473 K under vacuum for 2 h. Then, the sample was treated with a 1 mol/L ammonium solution, followed by filtration and evacuation at 473 K for 2 h. The obtained samples were further exposed to H20 vapor at 373 K and evacuated at given temperatures in situ before use as catalyst. Impregnated Nb205/Si02catalyst were also obtained according to traditional impregnation method using a methanol solution of NbCIs (Wako Chemical). We denote the niobic acid layers attached on S i 0 2 as ANb205/Si02, the impregnation catalysts as I-Nb205/Si02, and the bulk niobic acid as B-Nb205.nH20. Esterification Reaction between Ethanol and Acetic Acid. The catalytic reactions were carried out in a closed circulating system (dead volume, 194 cm’). The initial pressures of both ethanol and acetic acid were 1.6 kPa. Products were analyzed by gas chromatograph (Shimazu GC-9A) using a DOS column (Gasukuro Kogyo Inc.). The catalytic activities of the samples were estimated from the initial rates for ethyl acetate formation. Pyridine Adsorption. To examine the acidic properties, we measured IR spectra of adsorbed pyridine. IR spectra were recorded with JASCO FT-IR 7000. The sample wafers were pre-evacuated at given temperatures in an IR cell. Then, they were exposed to pyridine vapor at 293 K, followed by evacuation at 423 K to remove physisorbed pyridine. XRD and XRF Spectroscopy. X-ray powder diffraction (XRD) patterns of the catalysts were measured on a diffractometer (Rigaku Geigerflex RAD-IA). The amounts of N b on silica were determined by X-ray fluorescence (XRF) measurements by using Seiko Sea 2001. EXAFS Spectroscopy. N b K-edge EXAFS spectra were measured at BL-IOB of Photon Factory in the National Laboratory for High Energy Physics (KEK-PF). The storage ring was operated at 2.5 GeV with 150-200 mA. The spectra were taken at 80 or 293 K. EXAFS oscillations were extracted for the EXAFS raw data by using a cubic spline method and normalized with the edge height. The k3-weighted EXAFS spectra were Fourier transformed to R space over the range from 40 to 130 nm-’. The inversely Fourier filtering data were analyzed by using a curvefitting technique on the basis of the single-scattering plane-wave theory.20 Empirical phase shifts and amplitude functions were used. The model compounds were NbSi2, LaNb04, and Nb2OS for Nb-Si, Nb-0, and Nb-Nb bonds, respectively. (20) Teo, B. K. EXAFS, Basic Principles and Data Analysis; SpringVerlag: Berlin, 1986.
The Journal of Physical Chemistry, Vol. 95, No. 24, 1991 1OOO1
Niobic Acids Supported on Si02
t - o0
5
Lewis Acid 1450 cm-'
10
Amount of Nbl w t %
0
Figure 2. Initial rates of the esterification reaction on A-Nb2OS/SiO2 as a function of the amount of N b loading.
Evac. Temp. I K
Figure 4. Intensities of the Lewis acid band and the Bronsted acid bands as a function of the pre-evacuation temperature.
10 20 30 40 50 28 I degrees
60
Figure 3. XRD patterns for the samples annealed at 773 K in air: (a) SiO,; (b) 15 wt % A-Nb,0,/Si02; (c) 8 wt % I-Nb205/Si02; (d) TNbZO,. Results Surface Transformationof Nb Species on SO,. The preparation steps for the Si02-attached niobic acid monolayers are shown in Scheme 1. After the N b species were attached, the intensity of the O H groups of the Si02surface remarkably decreased. In contrast, C-H stretching vibrations for C2H5 moieties appeared. The C-H stretching vibration peaks disappeared after treatment with aqueous NH3 solution, accompanied with the appearance of a new peak at 1452 cm-' which is attributed to the N-H bending mode of NH4+. This peak disappeared by evacuation at 473 K, suggesting the decomposition of NH4+ species. XRF analysis indicated that all Nb(OC2H5)5reacted with the OH groups of SO2. However, we could not get the sample with more than 8.0 wt % of N b loading by the preparation including a washing of the sample with hexane. The niobic acid monolayer catalysts with 10 and 15 wt % of N b were prepared similarly to scheme I but without washing with hexane. Esterification Reaction between Acetic Acid and Ethanol. The product in the esterification of acetic acid on B-Nb20S.nH20and on A-Nb205/Si02with various N b loadings was only ethyl acetate. No byproducts such as ethene, diethyl ether, and acetaldehyde were detected. Figure 1 shows the dependency of the initial rates at 423 K on the pre-evacuation temperature of B-Nb205.nH20. The catalytic activity decreased sharply with an increase in the pre-evacuation temperature to 473 K, followed by a relatively gentle decrease in the activity in the temperature range 473-773 K, and eventually at 873 K the activity of B - N b z O p H 2 0 becomes nearly zero. Typical results for A-Nb2OS/S1O2with 2 wt % N b are also shown in Figure 1. In contrast to the B-Nb205-nH20catalyst,
4
6
8
10
12
k / 10 nm' Figure 5. EXAFS oscillations for A-NbzOS/SiOz:(a) 0.4 wt %; (b) 4.6 wt % (c) 7.0 wt 8;(d) 10 wt W.
no deactivation by preheating of the catalyst to 873 K was observed, showing a highly thermal stability. The catalytic activity of A-Nb205/Si02linearly increased as the amount of N b attached on Si02increased to 8 wt % ' of N b loading as shown in Figure 2. At N b loadings higher than 8 wt % the activity gradually decreased with an increase in N b loading, showing a break in the two lines in Figure 2. XRD Measurement. Figure 3 shows the XRD patterns for the samples treated at 773 K in air. There was no diffraction peak corresponding to Nb2OSparticle for I5 wt % A-Nb20S/Si02 calcined at 773 K. In the case of 8 wt % I-Nb2OS/SiO2,however, diffraction peaks were observed, suggesting the presence of TNb20Sparticles on Si02as shown in Figure 3c,d. Pyridine Adsorption. Pyridine adsorbed on A-Nb20S/Si02with various N b loadings exhibited four peaks a t 1613, 1548, 1491, and 1450 cm-I. The peak at 1450 cm-I is assigned to pyridine coordinated to N b sites with Lewis acid character, and the peak of 1548 cm-' is assigned to pyridinium ion formed from pyridine adsorbed on Bronsted acid sites2' The intensities of the bands at 1548 and 1450 cm-' are plotted as a function of pre-evacuation temperature for A-Nb20s/Si02 (4.6 wt %) in Figure 4. The peak intensity at 1548 cm-' decreased with pre-evacuation temperature. On the other hand, the intensity of the peak at 1450 cm-' remained constant or slightly increased against the pretreating temperature. Almost the same results were observed with other A-Nb205/Si02 catalysts. EXAFS Measurement. EXAFS oscillations of the ANb20s/Si02 catalysts with various N b loadings are shown in (21) Parry, E.P.J . Caial, 1963, 2, 371.
Shirai et al.
10002 The Journal of Physical Chemistry, Vol. 95, No. 24, 1991 TABLE 1: Curve-Fitting Analysis for the EXAFS Data of A-NbzOdSiOz bond distance," nm loading, N M Nb-O Nb-Nb Nb-Nb wt %
0.40
0.191 coord no.b 2.1
2.0
0.336 0.5
0.392 0.6
0.327 0.9
0.194 0.210 1.7
0.335 0.7
0.392 0.6
0.328 1.0
0.190 0.209 1.6
0.336 0.4
0.390 0.5
0.328 0.9
coord no.b 2.5 4.6
coord no.b 2.1 7.0
0.193 0.209 1.7
0.335 0.4
0.392 0.4
0.328 0.9
coord no.b 2.2
0.193 0.209 1.3
0.334 0.8
0.395 0.5
0.330 0.7
0.194 0.210 coord no.b 2.1 1.4
0.333 0.7
0.394 0.6
0.329 0.7
coord no.* 2.0 10
I5
Nb-Si
0.208 1.7
"f0.003.bf0.4.
Figure 5. There were characteristic shoulder peaks in the proximity of 55 cm-I, which have been demonstrated to arise from the Nb-Si bond." The curve-fitting analyses for the EXAFS data over k = 40-130 nm-' are listed in Table I.
Discussion Nature of Active Sites of the Silica-AttachedNiobic Acid Layer Catalysts. The initial rate per N b atom of the niobic acid layer on SiO, pretreated 473 K for the esterification at 423 K was 20 times higher than that of the niobic acid bulk catalyst in Figure 1. The initial rates per surface Nb2O5 area of the A-Nb205/Si02 catalysts pretreated at 423 and 573 K were also higher by factors of 3 and 6 than those of the bulk catalysts pretreated at 423 and 573 K, respectively. This superiority of the layer catalyst to the bulk catalyst may be referred to new active sites formed on the SiOz surface. Furthermore, the activity of A-Nbz05/Si02remained almost unchanged against an increase of the pretreatment temperature, but the activity of B-Nbz05.nH20was very sensitive to the pretreatment temperature, the activity markedly decreasing with the temperature. Iizuka et al. reported that the active site of the bulk catalyst is a Bronsted acid site from pyridine adsorption experiments.I0 On the other hand, the amount of Lewis acid sites on ANb205/Si02did not decrease with an increase in the pre-evacuation temperature, while Bronsted acid sites decreased rapidly and disappeared at 773 K (Figure 4). These results suggest that the active sites of A-Nb205/Si02are Lewis acid sites, although there may be a small contribution of Bronsted acid sites because the activity slightly decreased in the pre-evacuation temperature range 473-773 K as shown in Figure 1. The surface area of B-Nb205*nHz0decreases by pre-evacuation at high temperatures. The catalytic deactivation of the niobic acid bulk catalyst has been ascribed to the decrease in the number of Bronsted acid sites.1° On the contrary, Lewis acid sites on A-Nb205/Si02 were thermally stable up to 873 K so that the A-Nb2O5/SiO2catalysts are applicable as a good catalyst at a wide range of reaction temperatures. Layer Growth of Nb Species on Silica Surface. The catalytic activity of A-Nb2OS/SiO2pre-evacuated at 573 K for the synthesis of ethyl acetate from acetic acid and ethanol at 423 K increased linearly with N b loading in the range 0-8.0 wt % as shown in Figure 2. The results suggest a uniform distribution on niobic acid species on Si02. The I-Nb2O,/SiOz catalyst with 8.0 wt % N b prepared by an impregnation method had one-third the activity of the corresponding A-NbzO5/SiO2 catalyst. The XRD results in Figure 3 revealed that the N b species in I-Nb20S/Si02were aggregated to form three-dimensional T-Nb20Sparticles. On the contrary, no XRD pattern was observed with the A-Nb2OS/SiO2catalyst as shown in Figure 3. The drastic difference between the two catalysts in the growth mode of N b species on SiOz should be
0
5
10
15
Amount of N b l wt% Figure 6. Number of Nb-Si bonds contained in 1 g of the A-Nb205/ S O 2catalyst as a function of the Nb loading.
ascribed to the formation of chemical bonds between the N b species and the Si02surface in the A-Nb205/SiOz catalyst. The EXAFS oscillations for all A-Nb205/SiOzcatalysts with 0.4-8.0 wt % N b loadings showed the shoulder peak at 55 nm-I due to Nb-Si bond (Figure 5).ks The curve-fitting analysis for the EXAFS data in Table I confirmed the presence of the Nb-Si bond at 0.328 nm (i0.003 nm) in A-Nbz05/Si02. The number of Nb-Si bonds contained in 1 g of the catalyst was plotted against the amount of N b loaded in the A-Nb205/SiOzcatalyst in Figure 6, which shows the linear relation between the number of Nb-Si bonds and the N b quantity (58.0 wt %) in the catalyst. There was a break at 8.0 wt % of N b loading, followed by a line with a different slope. This is consistent with the observation for the esterification activities as a function of N b loading in Figure 2. These results suggest the monolayer growth mode of niobic acid deposited on the SO2surface up to 8.0 wt %. No XRD peak was observed also for 15 wt % A-Nb20S/SiOz in Figure 3. The structure of this sample is not clear, but 8 wt % niobic acid would form a monolayer on Si02and the others may be dispersed as small particles on the monolayer. From the results of the esterification activity and the EXAFS analysis, the N b species of more than 8.0 wt % seem not to be supported on Si02in a monolayer form. The molecular area of niobium ethoxide employed for the preparation of the catalyst is estimated to be ca. 0.40 nmz. Thus, the full monolayer coverage for the surface N b sub-ethoxides should be ca. 8.0 wt %, taking into account the surface area of Si02 (220 m2 g-l). In the attachment reaction of Nb(OC2H5),with the surface O H groups, once niobium ethoxide covered the Si02 surface, no more niobium ethoxide could react with the OH groups. The bulky ethoxyl groups prevent Nb(OC2H,), from further reacting with SO2. In fact, Nb(OC2H5), beyond the amount corresponding to 8.0 wt % as Nb/SiOz was only physisorbed (not forming chemical bonds with the surface) and readily removed by washing with hexane as described under Catalyst Preparation. We should have impregnated Nb(OC,H,), in hexane without washing with hexane to obtain samples having the N b species of more than 8.0 wt %. In preparation of the catalyst, the number of surface hydroxyl groups on silica was controlled to be ca. 5 OH nnr2 by annealing at 473 K for 2 h. The number of surface O H groups is enough to make a full monolayer catalyst. The size of the bulky Nb(0C,H,), is almost twice as large as that of [Nb02,,]. Consequently, there remained areas uncovered with niobic acid layers on Si02 surface as illustrated in Scheme I. Active Structure of the SO2-Attached Niobic Acid Layers. According to the EXAFS analysis, the Nb-Nb bond was already observed with the A-Nb205/Si02sample of a low coverage of 0.4 wt %. This means that islands of niobic acid were already formed on the SiOzsurface at low coverages. Bradley reported that metal alkoxide exists in its oligomer form bridged by an alkoxide group.22 Niobium alkoxide may also form a cluster composed of several Nb(OCZHS),molecules in hexane. After the adsorption of Nb(OC2H,), on S O 2 ,it tends to make ensembles on the surface even at low coverages. (22) Bradley, D. C. Prog. Inorg. Chem. 1960, 2, 303.
The Journal of Physical Chemistry, Vol. 95, NO. 24, 1991 10003
Niobic Acids Supported on Si02 0.392 nm K
SCHEME 11: Reaction Scheme for the Esterification on A-Nb2OJSiOI1
0.335 nm
*
r Noh" I
Nb,O,Nb A
Figure 7. Proposed structure for the silica-attached niobic acid layer catalyst. a
.
CzHsOH(g) + b,(y
u/
CH3COOCzHs+HzO
\
CHCOOHa)
NbA ,Nb simply represents the local structure around N b in Figure 7. 0
4Mx)
3500
3033
2503
Wave Number I c i f ’
Figure 8. IR spectrum of C2H50Hadsorbed on the 7 wt % A-Nb205/ Si02 catalyst at 383 K.
(a)
1800
From the results of the EXAFS curve-fitting analysis, we can propose a model structure for A-Nb205/Si02with 0.4-8.0 wt % N b loadings in Figure 7. There are two kinds of Nb-Nb bonds: one is observed at a distance of 0.392 nm and the other at 0.335 nm as shown in Table 1. Judging from a known crystalline Nb2O5 structure, the shorter Nb-Nb bond is assigned to the bond bridged with two oxygen atoms (like edge-shared) and the longer bond is ascribed to a one-oxygen-bridged Nb-Nb bond as shown in Figure 7. The full monolayer of niobium oxide (23.7 wt %) on Si02has been synthesized by a two-stage attaching technique,4g5where two Nb-Nb distances were also observed. The Nb-Nb bond lengths in these samples were 0.342 and 0.378 nmeC6 The activity of the niobium oxide monolayer catalyst for the esterification at 423 K was one-third of the esterification activity of A-Nb205/Si02. The Nb-Nb bond distance (0.335 nm) bridged with two oxygen atoms in A-Nb205/Si02is shorter than 0.342 nm for the niobium oxide monolayer. The Nb-Nb separation (0.392 nm) bridged with one oxygen atom in A-Nb205/Si02is longer than 0.379 nm for the niobium oxide monolayer. There is a bigger difference between the niobic acid layer catalyst and the niobium oxide monolayer catalyst in Nb-0-Nb separation than in
separation. The longer distance of Nb-Nb observed with ANb205/Si02may give rise to a larger vacant space on the Nb5+ ion with a Lewis acidic nature to which reactant molecules are readily accessible. Furthermore, the active phase of the ANb205/Si02catalyst composed of small two-dimensional islands is dispersed on the Si02surface as already mentioned, resulting in the formation of many periphery sites of the niobic acid islands, which also makes the coordinatevly unsaturated Nb5+ ions with a Lewis acid character. Esterification Reaction between Acetic Acid and Ethanol on A-Nb205/Si02. Figure 8 shows an IR spectrum of C 2 H 5 0 H adsorbed on 7 wt % A-Nb205/Si02.A broad peak of 3428 cm-I has been assigned to the Nb-OH group.‘ The peaks at 2983,2940, 2897, and 2885 cm-l are assignable to the C-H stretching vibrations of the ethoxide (C2H50)group.23 The amount of adsorbed C 2 H 5 0 Ha t saturation on A-Nb205/Si02at 423 K was found to be 0.5 C 5 H 5 0 H molecule/Nb atom. The Nb-0-Nb (23) Nishimura, M.; Asakura, K.; Iwasawa, Y. froc.-Int. Congr. Coral., 9th 1988, 4, 1842.
17m
1600
1500
1400
1300
Wave Number/ cm’
Figure 9. IR spectra of CH3COOH adsorbed on the 7 wt % ANb205/Si02catalyst at 423 K: (a) 7 wt % A-Nb205/Si02;(b) CHjCOOH adsorbed on (a); (c) after exposure of (b) to C2H50Hfor 15 min at 423 K; (d) after exposure of (b) to C2H50Hfor 40 min at 423 K.
bond in the Si02-attached N b dimers was not broken upon the dissociative adsorption of C2H50H according to the EXAFS analy~is.8,~~ Ethanol may adsorb on the twooxygen-bridged site as shown in Scheme 11. Figure 9 shows IR spectra after CH3COOH adsorption (b) and the subsequent reaction with ethanol (c, d) on the A-Nb205/Si02 catalyst with 7 wt % Nb loading. There are two types of adsorbed CH3COOH. The peaks at 1400-1600 cm-I have been assigned to bidentate acetate, and the peak at 1700-1800 cm-’ has been assigned to molecularly adsorbed acetic acid.25 When ethanol was introduced to the acetic acid preadsorbed system at 423 K, the peak at 1740 cm-’ almost disappeared, whereas the acetate peaks decreased slightly (Figure 9c,d). Upon the introduction of C2H50Hat 423 K, ethyl acetate was produced in addition to a partial exchange of adsorbed CH3COOH with ethanol. When acetic acid was introduced to the surface preadsorbed with ethanol (hydroxyl and ethoxyl groups) at 423 K, the ethoxyl and hydroxyl peaks disappeared (the hydroxyl peak did not completely disappear), forming ethyl acetate and giving a spectrum similar to spectrum b of Figure 9. These results suggest that molecularly adsorbed CH3COOH and dissociatively adsorbed ethanol reacted to form ethyl acetate as shown in Scheme 11, whereas the dissociatively adsorbed acetate was stable under the reaction conditions. The sites for the dissociative adsorption of CH3COOH to form acetate are not clear, but CHICOOH may dissociate on the peripheral Nb sites of the niobic acid islands with high coordination unsaturation. The ratio of molecularly adsorbed acetic acid to bidentate acetate on the A-Nb205/Si02catalyst was estimated to be roughly 70%:30%, judging from the amount of ethyl acetate produced by the reaction of saturated acetic acid (CH3COOH/Nb = 1) with an excess of ethanol.
Conclusion 1. Silica-attached niobic acid layer catalysts (A-Nb205/Si02) were prepared by the reaction between Nb(OC2H5)5and surface (24) Ichikuni, N.; Asakura, K.; Iwasawa, Y. Unpublished results. (25) Schiavello, M.; Auguliaro, V.; Couccia, S.; Palmisano. L.; Sclafani, A. Photochemistry on Solid Surfaces; Studies in Surface and Catalysis 47; Elsevier: Amsterdam, 1988; p 149.
J . Phys. Chem. 1991, 95, 10004-10009
10004
OH groups of Si02, followed by chemical treatments with aqueous ammonia and water vapors and by treatments at given temperatures. 2. On Si02surface niobic acids grew in a two-dimensional mode, forming islands in the range of N b loadings from 0 to 8.0 wt %. 3. The local structure around the N b atom in the one atomic layer showed direct bonds of Nb-Si, Nb-Nb, and Nb-Nb at 0.328, 0.335, and 0.392 nm, respectively. 4. The catalytic activity of the A-Nb205/Si02catalyst for the
esterification from ethanol and acetic acid was 20 times higher than that of a niobic acid bulk catalyst. The activity per surface Nb of the A-Nb205/Si02catalyst at 573 K was also 6 times higher than that of the bulk catalyst. 5 . The esterification proceeded on the Lewis acid sites which were stable even at 873 K, suggesting the application of this surface layer catalyst in wide reaction conditions. Registry No. Nb(OC,H,),, 3236-82-6; SO2, 7631-86-9; Nb205, 13 13-96-8; C2HSOH,64-17-5; CH,COOH, 64- 19-7; pyridine, 1 10-86- I .
Ultraviolet-Laser-Induced Ablation of Poly(ethylene terephthalate) Peter E. Dyer,+ Geoffrey A. Oldemhaw,*?*and Jagjit Sidbut*o Department of Applied Physics and School of Chemistry, University of Hull, Hull, HU6 7RX England (Received: April I O , 1991; In Final Form: July 17, 1991)
The ablation of poly(ethy1ene terephthalate) (PET) by 308-nm XeCl laser irradiation and by 193-nm ArF laser irradiation has been studied. In this process etching of the polymer occurs for laser fluences above a threshold value FT.Stress waves exhibited by thin films of PET during XeCl laser irradiation at fluences below FThave a form characteristic of thermoelastic stress and demonstrate that relaxation of the absorbed energy to heat is rapid on the time scale of the laser pulse. In films irradiated with the XeCl laser at fluences above FT, the initial part of the thermoelastic stress wave is succeeded by a compressive stress due to ablation, which starts during the laser pulse. The major gaseous products of both XeCl laser ablation and ArF laser ablation are CO, COz,CH,, C2H2,CzH4, C4H2, C4H4,C6H6, and CH,CHO, which are also found in the cozlaser ablation of PET. Yields of gaseous products of the XeCl laser ablation are low for fluences below 2FT, but in the case of ArF laser irradiation greater fragmentation of the ablated PET occurs and yields of gaseous products in the fluence range FTto 2FT are appreciable. The mechanism of ablation is discussed and it is concluded that XeCl laser ablation of PET occurs by rapid relaxation of the initial electronic excitation to heat, resulting in thermolysis of the polymer. Thermolysis is probably also an important factor in ArF laser ablation, although a contribution from direct photolysis cannot be excluded. The greater fragmentation of ablation products observed in ArF laser ablation is attributed to secondary photolysis in the ablation plume.
The ablative photodecomposition of an organic polymer by pulsed ultraviolet radiation from an excimer laser was first demonstrated in 1982.'*2 In this process material is forcibly ejected from the polymer surface and etching occurs at the site of irradiation when the laser fluence exceeds a threshold value. The threshold fluence, FT, is characteristic of the polymer and the wavelength of irradiation, and the depth of the etched hole 1 is approximately a linear function of the logarithm of the laser fluence F for a limited range above FT3 1 = k;I In ( F / F T ) (1) where k, is the effective absorption coefficient for the laser radiation. Etching at fluences below threshold, while sometimes detectable, is very small. Photoablation produces etching directly without the need for chemical development and has potentially important applicationsc6 in microlithography and micromachining as well as in surgery. Since the original observation there has been increasing interest in, and work on, this phenomenon, involving studies of several different polymer^.^-^ The time scale of the ablative process has been established by using poly(viny1idene fluoride) (PVDF) film piezoelectric transducers to study the acoustic response of UVlaser-irradiated polymers.'** These experiments show that excimer-laser-induced ablation occurs rapidly, starting during the lifetime (- 20 ns) of a typical laser pulse. One of the interesting mechanistic questions about excimer laser ablation is whether decomposition occurs by direct photolysis or by thermolysis of the polymer following rapid degradation of the
'Department of Applied Physics.
*School of Chemistry. I R a e n t address: Sowerby Research Centre, British Aerospace plc, Filton, Bristol, U.K.
initial electronic excitation to heat.24*9-13 In the case of poly(ethylene terephthalate) (PET) irradiated by the XeCI, KrF, and ArF lasers, it has been shown,14by measurements of the thermal loading of thin films, that at fluences below threshold essentially all the absorbed energy is converted to heat. This implies a substantial temperature rise near the polymer surface owing to the small absorption depth. Above threshold the thermal loading remains approximately constant as the excess energy is carried off by the ablated material. It has also been foundI5 that the threshold fluence for ablation of PET by the XeCl laser is substantially reduced by preheating the polymer with COz laser radiation. (1) Srinivasan, R.; Mayna-Banton, V. Appl. Phys. Lerr. 1982, 41, 576. (2) Srinivasan, R.; Leigh, W. J. J . Am. Chem. Soc. 1982, 104, 6784. (3) Andrew, J. E.; Dyer, P. E.; Forster, D.; Key, P. H. Appl. Phys. Lerr. 1983, 43, 717. (4) Srinivasan, R. Science 1986. 23, 559. (5) Ych. J. T. C . J . Vac. Sei. Technol. 1986, A4, 653. ( 6 ) Dyer, P. E.; Sidhu, J. Opr. Loser Eng. 1985, 6, 67. (7) Dyer, P. E.; Srinivasan, R. Appl. Phys. Leu. 1986, 48, 445. (8) Dyer, P. E. In Phoroacousric and Phororhermal Phenomena; Hess, P., Petzl, J., Eds.; Springer Series in Optical Sciences; Springer-Verlag: Heidelberg, 1988; Vol 58, p 164. (9) Srinivasan, R.; Braren, B.; Dreyfus, R. W.; Hadel, L.; Seeger, D. E. J. Opr. Soc. Am. 1986, 83, 785. (10) Brannon, J. H.; Lankard, J. R.;Baise, A. 1.; Burns, F.; Kaufman, J. J . Appl. Phys. 1985, 58, 2036. (11) Dijkkamp, D.; Gozdz, A. S.;Venkatesan, T.; Wu, X . D. Phys. Reo. k r r . 1987, 58, 2142. (12) Srinivasan, R. Phys. Reu. Letr. 1988, 60, 381. (13) Venkatesan, T.; Gozdz, A. S.;Wu X.D.; Djikkamp, D. Phys. Reu. Lerr. 1988, 60, 382. (14) Dyer, P. E.; Sidhu, J. J. Appl. Phys. 1985, 57, 1420. ( I 5) AI-Dhahir, R.K.; Dyer, P. E.: Sidhu, J.; Foulkes-Williams, C.; 01dershaw, G. A. Appl. Phys. 1989, B49, 435.
0022-3654/91/2095-10004%02.50/0 0 1991 American Chemical Society
