J. Chem. SOC., Faraday Trans. I , 1986,82, 3053-3067 13C Nuclear Magnetic Resonance Investigations of Carbon Monoxide in Decationated Zeolites of Type Y Andreas Michael, Wolfgang Meiler, Dieter Michel* and Harry Pfeifer Sektion Physik, Karl-Marx- Universitat Leipzig, DDR- 701 0 Leipzig, German Democratic Republic Dieter Hoppach Sektion Chemie, Karl-Marx- Universitat Leipzig, DDR-7010 Leipzig, German Democratic Republic Jean Delmau Laboratoire de Spectroscopie, U.E.R. de Physique, Universitk Claude Bernard Lyon 1, 69621 Villeurbanne, France The 13C n.m.r. of carbon monoxide adsorbed on decationated zeolites of type Y is characterized by extremely large resonance shifts to lower magnetic field. The strong dependence of this resonance shift on temperature and on coverage can be explained by means of an exchange model described in the 1iterature.lS This treatment yields the resonance shift of carbon monoxide molecules in the adsorption complex (350 ppm relative to gaseous CQ) and the number of sites in zeolites effective for the complex formation (0.07 active sites per cavity). Investigations of the 13C nuclear magnetic relaxation of CO and CO, adsorbed on zeolite NaY and its decationated forms suggest that the large resonance shifts are due to an interaction with Lewis-type centres in the form of extra-lattice aluminium ions.Thus, investigations of the CQ adsorption by means of the I3C n.m.r. method are suitable for a characterization of Lewis-acid sites in zeolites. In a previous paper3 it was shown that the 13C n.m.r. lines of CO adsorbed on decationated zeolites show very large resonance shifts to lower magnetic fields with respect to the gaseous state.For zeolites NaA, NaX and NaY a similar effect could not be detected. Since the n.m.r. shift has a sensitive dependence on the conditions of pretreatment of the decationated zeolites which produce Lewis-acid sites it was suggested that the shifts are due to interactions of carbon monoxide molecules with this kind of acid sites. Although this hypothesis is supported by the results of quantum-chemical ab initio calculations for an idealized A13+/C0 ~ornplex,~ there remains some doubt concerning the influence of paramagnetic impurities. It is the aim of this paper to present more arguments for the interpretation of the n.m.r. shifts observed and for a detailed description of their dependence on coverage and on temperature.In order to understand better thz role of paramagnetic impurities in the zeolites used, the 13C nuclear magnetic relaxation of CO and CO, adsorbed on NaY and its decationated form has been studied. On this basis an interpretation of the observed resonance shifts has been made possible. Experimental 13C n.m.r. spectra were run at 22.63 MHz in the temperature range 120-450 K. The 13C nuclear magnetic relaxation times and T, were measured at 25.1 MHz and at 22.63 MHz using the inversion-recovery and Hahn’s spin-echo method, respectively. To 3053 101-23054 N.M.R. of CO in Zeolites Fig. 1. 13C n.m.r. spectra of carbon monoxide adsorbed on zeolite 30DeNaY as a function of coverage, N, (in molecules per cavity) at 273 K (a) and for different temperatures (K) with a coverage of 0.4 molecules per cavity (b).increase the signal-to-noise ratio, especially for the n.m.r. relaxation time measurements, the adsorbates CO and CO, were enriched with 13C nuclei to values between 50 and 90%. The use of these substances also allowed measurements of 13C n.m.r. spectra at very low coverages. The NaX and NaY zeolites studied in the present paper were purchased from VEBA . Michael et al. 3055 Chemiekombinat Bitterfeld-Wolfen, GDR, and are characterized by Si/Al ratios of 1.35 and 2.6, respectively. The decationated forms of zeolite Y, abbreviated as DeNaY, were prepared from NH,NaY zeolites with different degrees of ammonium exchange through thermal decomposition under self-steaming conditions (deep-bed activation3).For instance, 30DeNaY denotes such a zeolite where 30% of the sodium ions were exchanged by NH:. Results of 13C N.M.R. Shift Measurements Basic Results The 13C n.m.r. shifts of CO and CO, molecules adsorbed on NaX, NaY and NaA zeolites are not very different from the values in the gaseous state3 (- 1 to - 4 ppm for CO and -0.5 to -2 ppm for CO,, where the negative sign indicates shifts for the adsorbed molecules to higher fields). The same result holds for these adsorbate-adsorbent systems if Na+ is exchanged by other alkali cations or T1+. The resonance shifts depend only weakly on the temperature. Since the values for the shifts are of the same order of magnitude as the corrections due to the specimen susceptibility (ca.0.7 ppm to higher fields) and to van der Waals interactions, they are without further interest in this work. Strong 13C n.m.r. shifts occur when CO molecules are adsorbed on decationated DeNaY zeolites. The values are largest for deep-bed treated specimens, but also for other treatments of decationated forms of zeolite Y they are much greater than for zeolites containing alkali cations. No such effects could be observed for CO,.,? The 13C n.m.r. shifts for CO adsorbed on DeNaY zeolites increase monotonically with decreasing coverage N and with increasing temperature between ca. 200 and 280 K. These variations, which are completely reversible, and the influence of the treatment and of other modifications of the zeolite specimen are important for the interpretation of the results.Dependence on Coverage To derive values for the concentration of the active sites interacting with CO molecules in the decationated zeolites and for the n.m.r. shift of a molecule in such a complex, the dependence of the observed n.m.r. shift on the number of adsorbed molecules has to be analysed. With decreasing coverage (which we were able to measure downward to a value of N x 0.1 molecule per cavity) the observed 13C n.m.r. shift increases strongly, while for large coverages ( N > 3) it approaches the value for CO in the gaseous state (fig. 1). This behaviour is typical for an exchange between physisorbed (P) and chemisorbed (C) species being characterized by different resonance shifts dp and d,, respectively.The task is to extract the difference 6, - B p and the number of complexed molecules by treating the complex formation as a chemical reacti0n.l. 2? Two models will be considered. In Model 1 the decomposition of the complex is only determined by its equilibrium constant Kg. If we denote by P a physisorbed molecule, by A an active site (total number NA) and by C a surface complex (total number Nc), the equilibrium is described by (1) P + A e C K* from which the equation3056 N.M.R. of CO in Zeolites i 0.5 Y 0 200 - 100 - A 1 I I 1 2 3 N Fig. 2. Dependence of the n.m.r. shift on coverage for the system C0/30DeNaY at 303 (0) and 273 K (A) (b) and plot of the experimental quantities y us. x [according to eqn (9)] (a). Table 1. The resonance shift, 6,, number, NA, of active sites per cavity in zeolite 30DeNaY and the equilibrium constant, Kg, calculated from the experimentally determined quantities Am (= 6,/6,) and yoa using Model 1 Kg __ T / K Am Yo 6, (PPm) N A _ _ ~ - 300 0.96k0.02 0.08k0.01 310+20 0.07&0.01 400f300 273 0.85f0.02 0.10&0.01 350+30 0.07f0.01 78f25 a yo = (x = 0) from the plot of the experimental quantities y us.x [cf. eqn (9)]. can be derived. Here N - N , is the number of physisorbed molecules which are not bound in complexes and NA - N , is the number of unoccupied active sites. For a fast exchange and N 2 N , the observed n.m.r. shift is given by iVc N-N, N 6 = --6,+N6p. If we refer the n.m.r. shifts to the physisorbed state (6, = 0) we find Combining eqn (2)-(4) we obtain the formula (3)A .Michael et al. 3057 In Model 2 it is assumed that a decomposition of the complex occurs only when an unoccupied site (denoted as F) for physisorption exists. If we denote the total number of physisorption sites by NPh and the number of unoccupied sites by N, = NPh + N, - N we may derive from the equilibrium the equation KL P + A e C + F l-NPh+KLNA KL-l A- - +--(l -A). KL NA KL NA (7) As can be seen from eqn ( 5 ) and (7), a decrease of the coverage (expressed by the shift value of N) does lead to a maximum value for the n.m.r. shift S(N -+ 0) = 6, which, however, must not be equal to 8,. By means of eqn ( 5 ) and (7), we find for S,/& = Am Am = KgNA (Model 1) 1 +KgNA Am = KL(NA/NPh) (Model 2) + KL(NA/NPh) and hence the final equations I X 1 y = --NA- (Model 1) Am A& X 1 Am KL-1 A& y = - - - KL N A P (Model 2). A plot of the experimental quantity y = N(S/6m)2 (1 -S/Sm)-l us.x = N(6/Sm)(1 -S/S,)-l allows the determination of Am and NA or NA[KL/(KL- l)]. Thus, in general, independently of the model used an unambiguous determination of both the n.m.r. shift 6, for the complex and the number NA of active sites is only possible if K , & 1 holds. Both models were applied to analyse the experimental results for the n.m.r. shifts of the system C0/30DeNaY shown in fig. 2. The following conclusions can be drawn.6 (1) In the temperature range 273-300 K, the condition KL $ 1 holds and thus Model 1 is valid. This conclusion results from the relation [cf. eqn (S)] : KL=----- Am NPh 1-Am NA taking the values for Am from table 1 and considering that the number of active sites is much less than the number of sites for physisorption.(2) According to the values derived for the equilibrium constant Kg and for NA (cf. table l), the CO molecules undergo a strong interaction with a small number of active sites, viz. 0.07 centres per cavity. The relatively large uncertainty for the value of Kg is due to the rather large error (up to 50%) for the term Am/(l -Am). (3) For T < 273 K the condition of fast exchange is no longer fulfilled. Temperature Dependence For T < 180 K the 13C n.m.r. shifts are only small (-4 to -6 ppm) and to higher magnetic field with respect to CO in the gaseous state. The values are typical for physisorption and may be explained in terms of van der Waals interactions [cf.also ref. (3)]. With increasing temperature the resonance lines are shifted to lower magnetic field (positive values of S) and broadened [fig. 1 (b) and 31. Similar to the dependence of the3058 N.M.R. of CO in Zeolites t t I L B 3 4 5 6 Fig. 3. Dependence of (a) the 13C n.m.r. shift and (b) the linewidth, A V , , ~ , of the system CO/30DeNaY on temperature. The symbols and A denote experimental values and theoretical data calculated by the formulae of Swift and Connick, respectively. The coverage is 0.5 molecules per cavity. resonance shift on the coverage, this behaviour may be explained by an exchange process, viz. a fast exchange at room temperature and above and a slow exchange at low temperatures, where the influence of the active centres cannot be detected.Since we know from the analysis in the preceding section that the overwhelming majority of molecules is physisorbed (shift 6,) and that only a very small fraction N , 4 N p is involved in complexes with large resonance shifts 6,, the formula of Swift and Connick7 can be applied to analyse the experimental n.m.r. shift 6 and half linewidth Av: Here z denotes the mean residence time of a molecule in region C and < its transverse relaxation time. Avp is the half linewidth in the physisorbed state. For fast exchange occurring at higher temperatures the conditions z 4 T, and TAU, 4 1 are valid, leading to the simple equations: 1 N , 1 6 = ( N , / N ) S , and Av = Av,+---. n N T ,A . Michael et al. At lower temperatures, i.e. for slow exchange (z % TJ, and the linewidth 1 N , 1 AV = A v ~ + - - - 7 r N z is only determined by Av,.For the intermediate 3059 the resonance shift 6 goes to zero (14) region the formulae predict a monotonic increase of-the n.m.r. shift 6 with increasing temperature and a maximum for the linewidth Av as a function of z/&. To describe the temperature dependence of the 13C n.m.r. shift 6 and the linewidth Av for the system C0/30DeNaY400DB with a coverage of N = 0.5 molecules per cavity (fig. 3) we use the values N,/N = 1/6.25 and 6, = 350 ppm, derived from the analysis of the coverage dependence (table 1). The linewidths Avp for the physisorbed molecules are taken from the resonance lines at low temperatures. For simplification we assume that Avp is constant within the temperature interval considered.From the fit of the linewidths and the resonance shifts us. temperature we obtain an Arrhenius plot for the lifetime z and for the transverse relaxation time with activation energies of 18 and 10.5 kJ mol-1 and pre-exponential factors 1.5 ns and 10 ms, respectively. The agreement between the experimental and theoretical curves (fig. 3) is relatively good at low and high temperatures. Deviations up to 25% occur at medium temperatures. The decrease of the n.m.r. shifts at T > 300 K, which is still more pronounced for samples with a higher temperature of pretreatment, cannot be explained in terms of this exchange model. Adsorption Centres for CO According to the interpretation of the 13C n.m.r. shifts, at room temperature a fast exchange between physically adsorbed CO molecules and those which are strongly adsorbed on a very small number ( N , < 0.07 per large cavity for 30DeNaY 400DB) of adsorption sites occurs.It has been suggested2p3 that these latter are Lewis-acid sites (preferentially extra-lattice A13+ ions) which are formed during thermal treatment of decationated zeolites. In accordance with this suggestion we have found6 a correlation between the number of these sites and the magnitude of the 13C n.m.r. shifts. The shifts increase with rising the treatment temperature for zeolites DeNaY and are appreciably smaller for stabilized decationated zeolites with a strongly reduced aluminium content .8 For those stabilized decationated zeolites where extra-lattice aluminium species were completely rern~ved,~ no significant 13C n.m.r.shifts could be observed for the adsorbed CO (6 = +0.5 ppm relative to gaseous CO for N = 1.8). A similarly small shift was also found for CO adsorbed on a special aluminium-free zeolite Y prepared by means of Beyer’s method.1° Furthermore, we have studied zeolites where Na+ and H+ ions at cationic sites were replaced by A13+ ions (formula for the unit cell: A~,,Na,H,A~,,Si,,,O,,,, x + y = 23, ratio Si/Al = 2.42 for the skeleton). Here similar resonance shifts as for CO adsorbed on heat-treated zeolites DeNaY were measured. Appreciable resonance shifts were also observed for H mordenites and zeolite HNaZSM-5 having A13+ ions at extra-lattice sites. In spite of these measurements, which strongly recommend us to identify the adsorption centres responsible for the large resonance shifts of adsorbed CO, with A13+ ions on extra-lattice sites there are two facts which deserve further consideration.(i) The number of extra-lattice aluminium ions determined by 27Al n.m.r. measurements11 (1.1 ions per large cavity for 80DeNaY 400DB) is at least by a factor of about 5-10 larger than N,. This discrepancy may be explained by the well known experimental fact that the number of strong Lewis-acid sites acting in catalytic reactions is much smaller than3060 N.M.R. of CO in Zeolites 10’: 1 lo-’ rA 1 L- 1 o-2 1 o - ~ . l n 1 m 1 m 1 m l m 1 m 1 3 4 5 6 7 8 9 1 0 3 KIT Fig. 4. 13C nuclear magnetic relaxation times, ( I = 1,2), of (a) carbon monoxide and (b) carbon dioxide adsorbed on zeolites NaX (0, T,; 0, T,) and NaY (m, T,; n, T,) as a function of temperature.The coverage is 2 molecules per cavity.A . Michuel et al. 306 1 1 Id m 1 h- 1 0-: 1 . ;@’ , / / / /’O / ;6 / ,o’ 3 4 5 6 7 8 9 103 KIT Fig. 5. 13C nuclear magnetic relaxation times, (I = I , 2), of carbon monoxide (8, T,, N = 6 ; 0, T,, N = 1.2; 0, G, N = 1.2) and of carbon dioxide (0, T,, N = 2; 0, T,, N = 0.7; 0, q, N = 2) adsorbed on zeolite SODeNaY as a function of temperature. the total number of extra-lattice aluminium ions. (ii) Adsorption shifts of ca. 300 ppm relative to gaseous CO are very large compared with known coordination shifts in diamagnetic systems. Hence, it seems reasonable to consider also a possible influence of a small number of paramagnetic extra-lattice sites (such as Fe3+), which could be formed simultaneously with the A13+ sites owing to the presence of paramagnetic impurities in the zeolites used.13C Nuclear Magnetic Relaxation According to the preceding analysis, the strong 13C n.m.r. shifts of CO molecules adsorbed on decationated zeolites are caused by interactions with a small number of active sites ( N , < 0.07 per cavity). The remaining question is whether Lewis-acid centres or paramagnetic impurities at extra-lattice sites are responsible for this behaviour. In the following investigation 13C nuclear magnetic relaxation time measurements will be used to solve this question. and T, of CO and CO, molecules adsorbed on zeolites NaX and NaY are shown as a function of temperature In fig. 4 the longitudinal and transverse relaxation times3062 N.M.R. of CO in Zeolites Table 2.The minimum of the 13C nuclear magnetic relaxation time, T,, and the ratio of TIT, for different interaction mechanisms of CO interaction r/nm (Wrdmin Tmin/S (TI Qmin - ~~ ~ . _ _ _ _ _ _ _ ~~~ ~ 013c.. .13co 0. I 5 (0.2) 0.616 1.68 (9.4) 1.60 27~13+. . .13co 0.206” 26.96 0.278 4.42 23Na+. . . 13C0 0.261 19.0 3.7 4.42 Fe3+. . . l3COC 0.2 1 4.1 x 1.83 CSAd (A0 = 406 ppm)e 1 0.640 3.67 a Ref. (4). Ref. (1 5). Fe paramagnetic centre. CSA = chemical shift anisotropy. Value for free CO at 4.5 K [ref. (16)]. between 400 and 11 5 K. For lower temperatures (T < 210 K), the longitudinal relaxation time shows a minimum, while the transverse relaxation time increases monotonically with rising temperature.At higher temperatures (T > 170 K for CO and T > 300 K for CO,) both relaxation times q and T, decrease. For CO molecules adsorbed on decationated zeolites the 13C nuclear magnetic relaxation times q and T, increase with decreasing temperature in the total interval of measurements. For adsorbed CO,, maxima occur at intermediate temperatures ( T z 180 K for and T z 150 K for T,) (fig. 5). If we compare the temperatures for which the longitudinal relaxation time exhibits its minimum, we find that the mobility of the adsorbed CO, increases in the order NaX + NaY + DeNaY. An analogous statement can be drawn for the adsorbed CO molecules. Obviously, the mobility in decationated zeolites is so high that the minimum of the longitudinal relaxation time is shifted to lower temperatures, outside the interval of measurements. A decisive question for an analysis of the 13C nuclear magnetic relaxation times is which of the possible interaction mechanisms dominates.In table 2 we summarize theoretical results for the contributions q! of various interaction mechanisms to the total longitudinal relaxation rate 7;-l and for the respective ratios T,,/qi at the minima The discussion includes the magnetic dipolar interaction among the 13C nuclei, between the 13C and 27Al or 23Na nuclei or paramagnetic impurities (Fe3+) and the interaction due to an anisotropy of the chemical shift (CSA). In spite of the small amount of paramagnetic impurities (ca. 550 ppm Fe3+ ions for the zeolites X and Y used) this contribution dominates. The influence of chemical shift anisotropy and of the magnetic dipolar interaction between different nuclei is negligible.For instance, the CSA contribution could only be significant if the anisotropy were larger than 1700 ppm, i.e. ca. one order of magnitude higher than for CO-iron(r1r) porphyrin complexes,12 where the chemical shift anisotropy has been found to be dominant for the 13C nuclear magnetic relaxation of carbon monoxide. For I3C nuclear magnetic relaxation processes due to interactions with paramagnetic impurities, the measured relaxation times q (1 = 1 longitudinal, I = 2 transverse) are proportional to the relative amount of CO molecules, N,m/N, near the paramagnetic of Ti. sites: Here NI and N denote the number of paramagnetic ions and the total number of CO molecules per cavity, respectively, and m is the number of CO molecules in the first coordination sphere.The quantities q(ion) denote the relaxation times of a single 13C nucleus near a paramagnetic ion. It is interesting to note that the values of Tmin are about the same for CO and CO, adsorbed on zeolites NaX and NaY (cmin z 40 ms)A , Michael et al. 3063 Table 3. Results of adsorption isotherm measurementsa CO,/NaX 8.2 298b 2.0 0.3 0.1 CO,/NaY 8.1 29gb 2.0 3.3 1.5 CO/NaX 7.1 1 98b 1.7 0.9 0.9 CO/NaY 8.4 273" 2.0 53.3 31.0 CO/SODeNaY 1.7 20Od 1.2 4.5 1.4 CO/SODeNaY 1.7 273d 1.2 59.4 13.5 a The symbols have the following meaning: free sample volume, V,; temperature, T ; total number, N , of molecules per cavity of the given zeolite; adsorbate pressure, p , in the sample and amount of desorbed molecules, N,,,/N.Ref. (17). Ref. (18). Ref. (19). if the same coverage is chosen. Since the number of impurities is comparable for both zeolites, we can conclude that the time constants q(ion)min are the same for CO and CO, and hence that the distances between the 13C nuclei and the paramagnetic ion should be the same. For decationated zeolites the relaxation time qmin of CO, is much greater than for zeolites NaX and Nay. According to fig. 5 a value of Tmin = 800 ms has been found near the minimum at 125 K. Further characteristic properties to be seen in fig. 5 are the large value for the ratio of q/T, near the minimum of q and the decrease of both relaxation times and T, with increasing temperature.The latter behaviour can be explained either by an increasing desorption of the molecules at higher temperatures or the onset of an exchange process. In the case of desorption, both relaxation times should decrease according to eqn (1 5 ) with decreasing N . To check this influence quantitatively, the number of desorbed molecules was evaluated by the aid of adsorption isotherms (table 3).6 The reduction factors are only small for CO, in zeolites NaX and Nay, which is in qualitative agreement with the relatively small decrease of the relaxation times and T, for these zeolites at higher temperatures [cf. fig. 4(b)]. decreases from 46 ms at 200 K to 14 ms at 273 K, which corresponds to a factor of 3.3. The fraction of carbon monoxide desorbed, however, is only 0.3.The discrepancy becomes even more apparent for CO adsorbed on decationated zeolites, where thr; decrease of the relaxation times by more than one order of magnitude between 200 and 273 K must be compared with the desorption which amounts here to ca. 15%. The large values for the ratio of q/T, and the decrease of both relaxation times and T, at higher temperatures for CO and CO, adsorbed on decationated zeolites can be explained by molecular exchange between physically adsorbed molecules bound to a small number of extra-lattice aluminium (subscript ea) and iron ions (subscript ei) which are created during the stabilization of decationated zeolites under self-steaming conditions. The temperature dependence of the linewidth Av which is proportional to G1 has been already explained on the basis of the Swift-Connick formula [c$ eqn (12)] and hence for the present case we have to write For CO adsorbed on NaY the relaxation time3064 N.M.R.of CO in Zeolites T-’/s-’ Fig. 6. Schematic representation of the influence of different sites on the 13C nuclear magnetic relaxation times, 7 (l = 1,2) as a function of temperature. Details are described in the text. For the longitudinal relaxation time it follows [by placing Amj = 0 and exchanging 2 with 1 in eqn (16)]:13 1 where p. = - * 4 NP The decrease of the transverse relaxation time T, with increasing temperature can be explained if in this temperature interval q is controlled by the residence times z,, and zei of the molecules at A13+ and Fe3+ ions, respectively. This implies that the relaxation times qj and/or the inverse values of chemical shifts for molecules at the respective extra-lattice sites (Amj)-’ must be smaller than the residence times zj.This condition corresponds to the case of slow exchange: at lower temperatures the contribution G& dominates. Its value is given by magnetic dipolar interactions with paramagnetic impurities of the zeolite skeleton. Because of the slow exchange, paramagnetic and aluminium extra-lattice ions are ineffective. The contribution of the different mechanisms to the resulting relaxation times is shown schematically in fig. 6. Owing to the characteristic properties of the relaxation times qei and Tea, the resulting longitudinal and transverse & relaxation times are predominantly determined by interactions with extra-lattice paramagnetic (Fe3+) and extra-lattice diamagnetic (A13+) ions, respectively.As was already mentioned, q1 is proportional to the n.m.r. linewidth Av, so that this interpretation is in agreement with that given above for the observed temperature dependence of Av. It is of special importance that in contrast to the quite different temperature dependences for the resonance shifts of CO and CO, the behaviour of their relaxation times is qualitatively the same as can be seen by comparing fig. 4 6 . Obviously, the typical features of the nuclear magnetic relaxation times of CO, adsorbed on decationated zeolites can only be understood if a specific interaction of CO, molecules with extra-lattice sites also occurs which, however, is not reflected in the 13C n.rn.r.shifts. This behaviour allows the following conclusions.A . Michael et al. 3065 Table 4. Comparison of the changes of the electronic charges, Ap, for CO and CO,, linearly attaching an A13+ ion with respect to the isolated moleculesa and theoretical values of the 13C n.m.r. shift complex APcle Apole APAlle 8 (PPmIb ~- ~- ~ 1 3 + . . . oc - 0.847 0.631 0.216 73.1 ~ 1 3 + . . . co 0.006 - 0.346 0.340 - 80.8 ~ 1 3 + . . . oco -0.265 0.488, - 0.41 6 0.192 - a Electronic charges of isolated molecules: C, 5.844; 0,8.156 for CO. C , 5.192; 0,8.404 for CO,. Positive 6 are to lower field. Table 5. Reaction coordinate, R, stabilization energy, and theoretical resonance shift, 6,b calculated by the CND0/2 methodc complex Rjnm AE/kJ mol-l 6 (ppm) o=c=o H+ o=c=o HS o=c=o 0.301 1 0.3323 0.3742 0.1110 0.1049 0.3776 0.3820 0.1177 0.1071 - 127.1 - 65.9 - 17.4 - 1 184.0 - 906.5 - 18.2 -21.3 - 698.9 - 1016.8 - 35.4 - 25.2 + 0.4 - 74.3 - 34.1 - 0.6 - 1.2 - 2.9 -7.1 - a Stabilization energy AE = Ecomplex - ECO,COp.Theoretical resonance shift 6 = ZC0,C02-~complex; u = + ( 2 ~ ~ ~ - 0 ~ , ) . The interatomic distances (0.1 191 and 0.1230 nm for CO and CO,, respectively) were also optimized by CND0/2. - - Conclusions Influence of Extra-lattice Paramagnetic Sites (i) Extra-lattice paramagnetic sites (e.g. Fe3+ impurities) act as relaxation agents and exert approximately the same influence upon the longitudinal nuclear magnetic relaxation times & of CO and CO, in decationated zeolites. According to Johnston and Grant,14 in this case the longitudinal electron spin relaxation time zlS must be larger than the residence time zei of the molecule at the paramagnetic ion.Since the thermal correlation time z, of qei must be equal to or smaller than zei, the effective correlation time, which is given by (q$ + q1)-l,13 cannot be determined by the electron spin relaxation time. In agreement with this conclusion,3066 exhibits a dependence on temperature which is characteristic for thermally activated molecular motions. (ii) If we assume that in contrast to (i) the interaction with extra-lattice paramagnetic sites would lead to an appreciable n.m.r. shift, then its value should be about the same for CO and CO, molecules. This follows from the fact that the local magnetic fields due to the paramagnetic sites must be approximately the same at the carbon nuclei of CO and CO, molecules in accordance with the similar behaviour of their nuclear magnetic relaxation times.Hence this interpretation is clearly ruled out by the experimental result that for CO, only very small values for the n.m.r. shifts could be observed. N.M.R. of CO in Zeolites Interpretation of the N.M.R. Shifts The preceding discussion suggests explaining the large n.m.r. shifts observed for CO molecules adsorbed on decationated zeolites in terms of a change of the electronic charge at the carbon atom due to an interaction with extra-lattice aluminium ions.* To understand why in the case of CO, molecules only very small values for the 13C n.m.r. shifts could be observed, quantum-chemical calculations for complexes of CO and CO, with A13+,4 Na+15 and H+ ions6 were undertaken.Ab initio calculations of CO and CO, complexes with A13+ ions (table 4) revealed that the change of the electronic charge at the carbon atom is large if the CO molecule is attached to the A13+ ion via its oxygen atom, while for CO, only a smaller change could be found. These statements could be confirmed by means of ab initio calculations for the systems Na+/CO, Li+/C0l5 and by semi-empirical CNDO calculations for CO and CO, on different ions.6 The general trend is that in the case of CO, relatively small influences on the electronic density at the carbon atoms appear. Additional calculations of the 13C n.m.r. shifts revealed that for the systems Na+/CO, and H+/CO, nearly no shifts result, in contrast to pronounced shifts for similar complexes of CO depending on the special arrangement (table 5).As described el~ewhere,~ ab initio calculations yield for A13+/C0 complexes strong 13C n.m.r. shifts to lower magnetic field only if the carbon monoxide molecule is attached to the ion uia its oxygen atom. For the opposite structure, where the carbon atom is attached to the aluminium ion, the 13C n.m.r. shift is to higher magnetic field. This behaviour may explain why at higher temperatures the experimental resonance shift decreases again: owing to the increased thermal energy of the molecules the occupation numbers for both structures become approximately equal and the 13C n.m.r. shifts are averaged to zero. References 1 V. Yu. Borovkov, A. V. Zaiko, V. B. Kazansky and W. K. Hall, J. Catal., 1982, 75, 219. 2 H. Pfeifer, W. Meiler and D. Deininger, NMR of Organic Compounds Adsorbed on Porous Solids, Ann. Reports on NMR Spectroscopy (Academic Press, London, 1983), vol. 15, p. 291. 3 A. Michael, W. Meiler, D. Michel and H. Pfeifer, Chem. Phys. Lett., 1981, 84, 30. 4 Th. Weller, W. Meiler, A. Michael, H-J. Kohler, H. Lischka and R. Holler, Chem. Phys., 1982,72, 155. 5 Th. Bernstein, D. Michel, H. Pfeifer and P. Fink, J. Colloid Interface Sci., 1981, 84, 310. 6 A. Michael, Dissertation A (Karl-Marx-Universitat, Leipzig, 1984). 7 T. J. Swift and R. E. Connick, J. Chem. Phys., 1962, 37, 307. 8 V. BosaEek, D. Brechlerova and M. KEivBnek, Adsorption of Hydrocarbons in Microporous Adsorbents 9 V. BosaCek, V. Patzelova, Z. Tvarbzkova, D. Freude, U. Lohse, W. Schirmer, H. Stach and 10 H. Beyer and I. Belenykaja in Catalysis by Zeolites, ed. B. Imelik et al. (Elsevier, Amsterdam, 1980), 11 D. Freude, T. Frohlich, H. Pfeifer and G. Scheler, Zeolites, 1983, 3, 171. 12 T. Perkins, J. D. Satterlee and J. H. Richards, J. Am. Chem. Soc., 1983, 105, 1350. 13 H. Pfeifer, NMR Basic Principles and Progress (Springer-Verlag, Berlin, 1972), vol. 7, p. 53. IZ (Academy of Science of the GDR, Eberswalde, 1982), vol. 2, p. 26. H. Tharnm, J. Catal., 1980, 61, 435. p. 203.A . Michael et al. 3067 14 E. R. Johnston and D. M. Grant, J. Magn. Reson., 1982,47. 282. 15 Th. Weller, W. Meiler, H. Pfeifer, H. Lischka and R. Holler, Chem. Phys. Lett., 1983, 95, 599. 16 J. W. Gleeson and R. W. Vaughan, J. Chem. Phys., 1983,78, 5384. 17 D. W. Breck, Zeolite Molecular Sieves (Wiley-Interscience, New York, 1974). 18 T. A. Egerton and F. S. Stone, J. Chem. SOC., Faraday Trans. 1 , 1970, 66, 2364. 19 A. Michael, D. Michel and H. Pfeifer, Chem. Phys. Lett., 1986, 123, 117. Paper 511452; Received 21st August, 1985