Desorption-diffusion of ammonia in molecular sieves: IV. Decationated ZSM5 zeolite Lucio Forni, Francesco P. Vatti, and Emanuele Ortoleva Dipartimento di Chimica Fisica ed Elettrochimica, Universit(~ di Milano, Milano, Italy The analysis of t.p.d.a, spectra from ZSM5 is reported, assuming various chemical or physical steps as rate-determing. After saturation, quite long isothermal desorption times (up to 50 h) were needed, before starting the temperature ramp, to obtain reproducible data. Intracrystalline diffusion was found to be the probable rate-determining step with apparent activation energy of ca. 13 kcal/mol. A comparison of the present results with the ones obtained with HNaY zeolites provided information on the most probable hindering phenomena characterizing the overall diffusive process. Keywords: Temperature-programmed desorption of ammonia from ZSM5
INTRODUCTION Temperature-programmed desorption of ammonia (t.p.d.a.) has become one of the most popular techniques for measuring surface acidity of solids, as confirmed by Rabo and Gajda 1 in a recent excellent review paper. In spite of the complications arising from the intrusion of diffusion effects, 1 most of these studies are still carried out by following the procedure suggested 25 years ago by Cvetanovi~ and Amenomiya, 2"3 even in some very recent literature. 4'5 However, that procedure was proposed for largepored materials, such as amorphous aluminosilicates, where diffusion effects are negligible. It is not suited for those solids, such as zeolites, in which an extended network of low-diameter pores is present. Indeed, in such a case, at least some of the precisely defined assumptions, on which the method is based, namely, energetically homogeneous surface sites, first-order desorption rate with respect to the adsorbed species, linear temperature increase, and absence of intraparticle diffusion effects, are far from being met. Furthermore, in t.p.d.a, experiments, as reported so far, 6--1° it has been usual to start the programmed temperature increase when only a quite short isothermal desorption time, t~, has elapsed, after the presaturation of the solid with the gaseous base. In a previous paper, ll we proposed a procedure for t.p.d.a, data processing, based on a different approach, i.e., by taking into account different chemical or physical phenomena as rate-controlling steps of the overall process. This procedure has been Address reprint requests to Dr. Forni at the Dipartimento di Chimica Fisica ed Elettrochimica, Universitd di Milano, via C. Golgi, 19 20133 Milano, Italy. Received 20 May 1991; accepted 15 July 1991
© 1992 Butterworth-Heinemann
tested by us essentially on Y zeolite samples, at different ion-exchange degrees. 12'la In the present work, the method has been extended to a ZSM5 zeolite, i.e., to an acid solid characterized by a much higher SIO2/A1203 ratio. This results in a lower concentration of surface acid sites, although of higher strength. Hence, the experimental problems connected with the low signal/ noise ratio of the detector and with the disturbance due to impurities in the carrier gas have become much more stringent than with the previous Y zeolites.
EXPERIMENTAL Zeolite The zeolite employed was prepared hydrothermally, following a well-known Mobil patent, 14 and then transformed into the acidic form by ion exchange with refluxing aq. CH3 COONH4, followed by drying and calcining at 823 K in air. The value of the SIO2/A1203 ratio (mol/mol), as determined by gravimetric analysis, was 33.6. An example of SEM micrographs of this zeolite is shown in Figure I. The XRD analysis (Debye method, CuK0c radiation, Nifiltered) showed the typical MFI structure, 15'16 with a very high degree of crystallinity. The BET surface area was '299 m2/g, and t h e p o r e volume for the Rp < 100 .~, pores was 0.14 cmS'/g; both referred to dry zeolite, i.e., to the sample calcined overnight at ca. 1 Pa and 573 K.
Apparatus and procedure The t.p.d, apparatus has been described in detail elsewhere. Z2 T.p.d. experiments were carried out
ZEOLITES, 1992, Vol 12, January 101
T.p.d. of ammonia from ZSM5: L. Forni et al. 12,
4
tO',~,.
"" ~ A~
'-- L~' " ~ .: ;.-,~'
"
"
#
. ,
,=
O3 I 0
,
>
iO
E .
,- , ~
.,
- r ~, a~..==
.
¢,..
-J,.
~.-
/.~h~/,_,'.-~'~
/
.
".-
\
:'-
_
=r~i
.,,
=
8 0
30
60
tid (h)
lota~
Figure 2 Effect of carrier gas impurities o n t h e o v e r a l l peak area ( m V x s). To = 423 K. (r-I) in presence and ( ~ ) in a b s e n c e o f purifying trap
the presence or in the absence of the trap. The results for the lowest value of To (423 K) are given in Figure 2. The effect of the trap is more evident for longer tid. Since the collection of reproducible experimental data at the lowest To required li~l values in excess of 30 h, it was decided to carry out all the experiments in the presence of the purifying trap.
RESULTS AND DISCUSSION
INN
Figure 1 ployed
Typical SEM micrographs of the Z S M 5 zeolite em-
with a He flow rate of 30 Ncm:~/min and a temperature increasing rate of 13 = 10 K/min. Between the t.p.d, runs, the sample was always kept at 823 K under He flow of 5 Ncm3/min to avoid contamination. After presaturation with ammonia (SIAD, transistor grade, >_ 99.995 vol%, employed as supplied) at the desired initial temperature, To, a sufficient isothermal desorption time, lid, was allowed to elapse before starting the t.p.d, run, to remove the excess base from the bed of particles.
The experimental t.p.d, spectra recorded with the present zeolite always showed single peaks, very similar to those given by Y zeolites. 12'r3 A typical example is shown in Figure 3. After saturation with ammonia at a .given To, the sample must be left at that temperature m the flowing carrier gas until the detector signal returns to the base line. This is not easy to verify, because the amount of ammonia leaving the sample rapidly becomes very small. However, if the sample is left in the flowing carrier gas for increasing values of lid, the area subtended by
•
• ~o
o e
~2E-4
o o
O 0) N
e
= o
Carrier gas purification The carrier gas (SIAD ultrapure grade, -> 99.9999 vol%) was further purified carefully by means of a glass trap filled with Linde 3A zeolite beads, cooled in liquid N2, and frequently regenerated. The lower the temperature of the sample, the more important is the purification because of the affinity of the zeolite for gas impurities, especially moisture. The efficiency of the purification system was verified by carrying out some t.p.d, experiments with increasing tid either in
102
ZEOLITES, 1992, Vol 12, January
03
o o
z iE-4 ~
°
°o o
o o
°% =o=
= o
O E + O ~" 42O Figure 3
'
1
620 T,K
°°=¢~e¢~=-.-
I
820
E x a m p l e o f e x p e r i m e n t a l t.p.d, peak. To = 423 K
T.p.d. of ammonia from ZSM5: L. Forni et al. ,
0.28
,
.
.
,
,
,
,
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Table 1 Comparison between N® for ZSM5 and Y zeolites
.
[]
N® ( m m o l NH3/g zeolite)
,+.=
To
,r-4
r.
~0.14
° x
O
o
,~
o
(K)
NaY a
HNaY30 b
NHaY55 b
ZSM6
423 448 473 498 523
0.101 0.050 0.030 0.020
0.195 0.106 0.051 0.034 0.021
0.145 0,091 0.055 0.037
0.188 0.135 0.089 0.062 -
aData from Ref. 13 bData from Ref. 11. Figures refer to % ion-exchange of the original Na ÷ with H ÷ ions 0.00
0
.........
' .........
I000
' .........
2000 minutes
3000
Figure 4 Change of the a m o u n t of a m m o n i a remaining adsorbed with increasing t~d. To (K): 423 (E3); 448 (C)), 473 (A), 498 (+)
the following t.p.d, spectrum can be a much more reliable probe to verify if the stationary condition has really been attained. T h e latter condition is fulfilled when a constant value of the area is measured for increasing t/d. Our data showed that the lower the To, the longer the t/d must be in order to obtain a constant value of the area. For instance, at To = 423 K, 2 d are needed to reach the steady state. For such high values of t/d, the error in peak area due to the adsorption of impurities from the carrier gas begins to become significant. Furthermore, it should be remembered that under the usual working conditions of the zeolite as catalyst, all the very weak sites remain free, while the strongest ones are permanently occupied. As a consequence, only the medium-strength sites are involved in the catalytic reaction. These are the reasons why To -- 423 K has been chosen as the lowest value for the presaturation temperature. The upper limit (To -- 498 K) corresponds to the lowest reproducible peak area that we could measure with our detector system. T h e amount N= (mmol/g zeolite) of the ammonia remaining adsorbed at different temperatures was determined by carrying out a series of t.p.d, runs, with different starting temperatures To, 25 K apart, after the proper t/d had passed. The experimental results are shown in Figure 4.
corresponding ones obtained ]2'13 for variously protonated Y zeolites is shown in Table I. One may observe that the overall amount of lower-strength acid sites (lower To data) is higher for the Y zeolite samples than for the ZSM5 zeolite, although the former samples are only partially protonated. However, with increasing the strength of the centers (higher and higher To data), their concentration decreases more rapidly for the Y zeolites, so that it quickly becomes lower than for the ZSM5 sample. This confirms not only that high-silica MFI zeolites possess a lower amount of acidic sites, due to lower A1 concentration, but also that in this zeolite the sites are of considerably higher strength.
Mechanistic analysis and equations employed The analysis for determining the most probable rate-determining step (r.d.s.) of the t.p.d, process has been carried out ]l either on the experimental t.p.d. spectra (e.g., Figure 3) obtained for different values of the initial temperature To or on the difference curves, obtained from the previous ones by subtracting point-by-point the values of the (To, + AT) curve from those of the To. one. Three hypotheses have been made about the r.d.s.: desorption without significant readsorption of ammonia, desorption with
-
1.50
Dependence of A T = on To In our previous papers lt-la concerning Y zeolites, the following equation: N= = exp (A1 + A2/To)
8 Z
c
-2.50
(1)
was found to best correlate the experimental data, after comparison of a set of five different simple interpolating functions. The same Equation (1) was also found to fit satisfactorily the present experimental data. Such an equation can be readily linearized. Its parameters, optimized by the leastsquares method (Figure 5), were A1 = - 9 . 5 5 +_ 0.36 and A2 = (3.36 + 0.17) x 10s . A comparison of the values of N= with the
-3.50 1.90
'
'
'
' . . . . 2.10 103/T (K -1)
2.30
Figure 5 Dependence of N® on To. Data interpolated by Equation (1). Curves mark the borders of 95% confidence limit area
ZEOLITES, 1992, Vol 12, January 103
T.p.d. of a m m o n i a from Z S M 5 : L. Forni et al. Table 2
8E-5
To (K) 423 448 473
0 N ,
[]
[]
~4E-5
OE+O 430
To + AT(K)
Ea (kcal/mol)
A (s-1)
448 473 498
11.3 12.4 12.4
1902 5255 2927
sion coefficient, N(Ti) can be calculated by the following equations:
630
530 T,K
Figure 6 Typical example of fit between experimental difference curve and the corresponding one calculated through Equation (2) and the optimized parameters of Table 2. To = 448
K, To + A T = 473 K
free readsorption, and intracrystalline diffusion of the desorbed base. The kinetic equations corresponding to such hypotheses are the following: (i) Process controlled by desorption without significant readsorption: -d(Aq)/dt = A Aq exp (-Ea/RT),
(2)
where Aq is the amount of ammonia held by energetically homogeneous sites present on the zeolite surface at temperature T; A, the preexponential factor; and Ea, the apparent activation energy for the desorption process. (ii) Process controlled by desorption with free readsorption of the desorbed base: Aq0 ln(Aq/Aqo) + Aq0 - Aq = [(Aqo - Aq)/Aq] (dAq/dT)exp(AHd/RT) j'ff-0exp (-AHa/RT)dT,
(3)
where Aq0 is the amount of ammonia adsorbed at temperature T = To; AHa, the enthalpy change connected with the desorption process; and R, the gas constant. (iii) Process controlled by intracrystalline diffusion of the desorbed base: -dN/dt = MAt~At,
(4)
where N is the amount of ammonia still absorbed on the sample and MAt is the amount of ammonia leaving the zeolite during the time interval At. For small values of At, corresponding to temperatureincreasing steps of 1 K, we have: o0
MAt = [N(ti) - N=(T~,)] (1 - 6/~ 2) X {(1/n 2) exp[-(D flR 2) (7"*) n27~2Ati]},
n=l
(5)
in which N(Ti) is the amount of ammonia still adsorbed at T = Ti, and N=(T*), the amount still adsorbed at T = 7"* = (Ti + 0.5) K. Rc is the zeolite crystal radius, and De, the apparent effective diffu-
104
Optimized parameters of Equation (2)
ZEOLITES, 1992, Vol 12, January
N(T3 = N=(Ti)
i= 1
(6)
N(Ti) = N=(Ti-I) + Mat,_,
i > ].
(7)
Desorption controlling with no readsorption Equation (2) was employed for drawing the calculated difference curves to be compared with the experimental ones obtained as previously mentioned. To obtain difference curves representative of energetically homogeneous acid sites, the temperature interval AT between the initial value To of the successive t.p.d, peaks was put as low as 25 K. T h e values of Ea and A have been calculated by the usual nonlinear regression techniques. A typical example of the fit between calculated and experimental curves is shown in Figure 6, while the optimized values of Ea and A are reported in Table 2.
Desorption controlling with free readsorption Equation (3) was employed to analyze each difference curve in several points. T h e integral was calculated numerically, through the Simpson's rule. Similar results were obtained for all the difference curves and a typical example of the results is shown in Table 3.
Intracrystalline diffusion controlling When intracrystalline diffusion is the r.d.s. Equations (4) - (7) can be employed directly on the primary experimental spectra (Figure 3 curves). De, appearing in Equation (5), changes with temperature. This dependence can be evaluated by the following equation, proposed elsewheretT: De/R 2 = Aa exp ( - E d R T ) ,
(8)
where A~ and Ea are the preexponential factor and the apparent activation energy, respectively, for the diffusion process. O f course, this kinetic parameter, in addition to the physical restriction due to the Table 3 Typical results obtained when assuming desorption controlling with free readsorption [Equation (3)] Point (T, K) considered 448 473 515 a 523 548 573 To=423K; To+AT=448K a Peak maximum
Alia (kcal/mol)
109.6 37.2 22.4 20.6 14.9 10.7
T.p.d. o f a m m o n i a f r o m Z S M 5 : L. Forni et al.
Table 4
Optimized parameters of Equation (8)
To (K)
2E-4
Ea (kcal/mol)
A. (S -1)
13.2 13.5 13.0 13.6
8444 8646 8426 8805
423 448 473 498
q)
"6 bl
~8E-5 narrowness of the zeolitic pores, takes into account the slowing-down phenomena connected with the interaction of the diffusing base molecules with the acid sites and/or cationic sites present on the walls of the pores. A typical example of the fit between experimental data and the optimized curve, calculated by means of Equations (4) - (8), is shown in Figure 7. The results are collected in Table 4.
DISCUSSION
Table 5 Comparison of the v a l u e s o f DelR~ and apparent activation energy for the intracrystalline diffusion of desorbed NH3 f r o m HZSM5 and variously decationated Y zeolites
o~/~ (s -1)
HZSM5 HNaY55 a NHaY30 a NaY
Ref.
This work 12 12 13
0E+0 430
530
630
730
T,K Figure 7 Typical example of fit between experimental t.p.d. spectrum and the corresponding curve, calculated t h r o u g h Equations (4) - (8) and the optimized parameters of Table 4. T o = 448K
The results of the analysis can be discussed by comparing what is expected from the calculation, based on the various r.d.s, assumed, and the actual results. For desorption controlling, without readsorption, Equation (2) must lead to increasing values of Ea with the starting temperature To of each difference curve. Indeed, by increasing To, increasingly stronger sites may still retain the base, the weaker ones becoming progressively free. For desorption controlling with free readsorption, Equation (B), employed for the whole set of points forming a given difference curve, must lead to the same value of AHa. Indeed, as previously pointed out, these curves are representative of energetically homogeneous sites, provided the AT between the starting temperature To of the two t.p.d, peaks, leading to the actual difference curve, is sufficiently small. For intracrystalline diffusion controlling, since diffusion should not depend directly on the strength of acid sites, Equation (8) must lead to a common value of E, and of A~ for the whole set of t.p.d, curves. The data shown in Table 3 point to the exclusion of desorption with free readsorption as r.d.s. In fact, AHd decreases monotonically with increasing the temperature of the given point along the same difference curve. No random distribution of data
Zeolite
o
423 K 9,7 6.6 3.3 6.5
x x x x
10 -4 10 -10 10 -1° 10 -11
E.
498 K 1.1 8.4 8.1 2,9
x x x x
10 -2 10 -8 10 -e 10 -8
(kcal/mol) 13.4 27,1 30.8 34.0
+_ 0.3 +_ 2.8 ___4.0 + 7.0
a Figures refer to % ion exchange of the original Na + with H + ions
around a common average value was ever observed for any of our difference curves. Hence, the overall desorption-diffusion process cannot be represented by this model. As for the first model, it can be noted that the data in Table 2 seem to reach a plateau, more than showing steadily increasing values of Ed with increasing the starting temperature, To, of the difference curve considered, as expected. As a consequence, this model is not able to represent the process correctly. Finally, as can be seen in Table 4, the values of Ea and Aa are practically constant, as expected, for each of the t.p.d, spectra, with an average value of Ea = lB.4 + 0.B kcal]mol and o f l n A = 9.0 + 0.1. Therefore, only the model considering intracrystalline diffusion as the r.d.s, seems to interpret correctly the present data. If so, the optimized values of Ea and A~ can be employed for calculating the corresponding values of De, provided the value of Rc is known. This c a l c u l a t i o n was p e r f o r m e d in o u r p r e v i o u s papers, 12'Ia where the values of Rc were evaluated by the analysis of SEM micrographs. Unfortunately, in the present case, this evaluation is not possible, since Figure I shows that our ZSM5 sample cannot be considered as made of separated crystals. Indeed, the figure shows only aggregates of microcrystals, the sizes of which are not easily measurable. Therefore, the only parameter we can safely calculate is Dt/R~. For the examined 423--498 K temperature range, a value ranging from 9.7 x 10 -4 to 1.1 x 10 -2 s -1 is obtained from the present data. These values are compared in Table 5 with those obtained in our previous papers 12'la with differently decationated Y zeolites. The latter also showed intracrystalline diffusion as the likely r.d.s. It appears immediately that diffusion in ZSM5 is several orders of magnitude faster than that in any of the Y zeolite samples. Moreover, the apparent activation energy of the process results are much lower (13.4 vs. ca. 30 kcal/mol). These results confirm that intracrystalline diffusion of N H s seems to be affected only to a minor
ZEOLITES, 1992, Vol 12, January
105
T.p.d. of ammonia from ZSM5: L. Forni et al.
extent by the size of the zeolitic pores or by the strength of protonated sites. The main restriction seems to be connected with the presence of Na + sites. This apparently surprising result can be rationalized as follows: It is well known that in the intracrystalline void space of Al-rich zeolites, very large electrostatic field gradients can be measured. In the 0c-cages of Y zeolite, the latter can reach up to 1 V/A at a distance of 2.5 ~ from the center of site-2 bivalent cations, is This, in turn, provides large areas of negative electrostatic fields on the oxygen rings. As a consequence, NaY zeolite can polarize and even ionize a variety of molecules, forming stable ionic clusters occluded in the zeolite crystal. As an example, the ionization of NO + NO2 radicals has been observed, with formation of NO + and NO2 ionic complex. 19 The energy required for a so strongly endothermic reaction (ca. 140 kcal/mol) is provided by the zeolite. An important practical result of this so-called solvent effect I is the enhancement of the concentration of adsorbates within the zeolite crystal. 2° The effect is centered on the polarity of the O - N a group and on the polarizability of the AI-O bond, so that both high A1 and high cation concentration contribute to increase it. On the contrary, Al-poor zeolites display minimal polarity-based interaction, due to the relatively smaller polarizabihy of the silica-rich zeolitic framework, l So, the principle restriction for a diffusing basic molecule within the pores of these Al-poor zeolites is connected with the acid-base interaction with the protonic site. Our results are perfectly in line with these findings. In partially decationated Y zeolites, at least two different types of sites do exist. One of them corresponds to protonated centers, and the other, to Na + centers. The interaction between Na + and NHa seems undeniable, since even for the unexchanged NaY zeolite the amount of NHa adsorbed on these sites is far from being negligible. On the contrary, it is comparable to the amount adsorbed on the variously ion-exchanged samples (Table 1). Furthermore, the energy of the acid-base interaction between ammonia and the protonated sites must be much stronger than that of the interaction between the gaseous base and the cationic Na + centers. However, the presence of the ionic clusters forming in correspondence of the Na + sites seems to constitute the most important obstacle to the diffusion of NH3, likely because of the strong restriction of the pore size. To sum up, the intracrystalline diffusion of the desorbed ammonia has to overcome (i) the geometric restriction due to the narrowness of the pores; (ii) the hindrance due to acid-base interaction with the protonic sites; and (iii) the hindrance connected to the presence of the ionic clusters forming by interaction of the diffusing molecules with the strong electrostatic field generated by the alkali cation sites and due to the strong polarizability of the AI-O linkage. These three effects play a different but comparable role in each of the zeolitic samples examined. However, the first effect should be more
106
ZEOLITES, 1992, Vol 12, January
important for the narrower-pored ZSM5 than for the Y zeolite samples; the second one should grow in importance with increasing the Si/AI ratio and with increasing the degree of ion exchange of a given zeolite sample. Finally, the third one should be important only in the highly polarizable framework Al-rich (Si/A1 _< 2.5) zeolites and should grow in importance with increasing the concentration of Na + sites, i.e., the sites around which the cumbersome ionic clusters can form.
CONCLUDING REMARKS The results of the present work confirm the following points: (i) After presaturation of the zeolite sample with ammonia, the lower the starting temperature, To, of the t.p.d.a, ramp, the longer must be the isothermal desorption time, t/d, which has to elapse before starting the ramp. Values of tia of the order of 1 h or less cannot give reproducible data. (ii) The analysis of t.p.d.a, spectra from zeolites should be carried out by taking into account several chemical or physical steps as possibly ratedetermining, irrespective of the concentration or strength of the acid sites. (iii) Incrystalline diffusion of the desorbed ammonia should be considered as one of the most probable controlling steps for all the acid solids in which a framework structure of low-diameter channels is present, irrespective of the Si/A1 ratio of the zeolite and of the degree of ion exchange of the original Na + ions with H +.
REFERENCES 1 Rabo, J.A. and Gajda, G.J. Catal. Rev. ScL Eng. 1990, 31,385 2 Amenomiya, Y. and CvetanoviE, R.J.J. Phys. Chem. 1963, 67, 144 3 Cvetanovi6, R.J. and Amenomiya, Y. Adv. Catal. (Eds. D.D. Eley, P.W. Selwood and P.B. Weisz) Academic Press, New York, 1967, Vol, 17, p. 103 4 Sawa, M., Niwa, M. and Murakami, Y. Zeolites, 1990, 10, 307 5 Hunger, B., Hoffmann, J., Heitzsch, O. and Hunger, M. J. Therm. Anal 1990, 36, 1379 6 Itoh, H., Hattori, T. and Murakami, Y. Chem. Lett. 1981, 1147 7 Ghosh, A.K., Keats, N.G. and Curthoys, G. J. Catal. 1984, 96, 288 8 Post, J.G. and van Hooff, J.H.C. Zeofites 1984, 4, 9 9 Hidalgo, C.V., Itoh, H., Hattori, T., Niwa, M. and Murakami, Y. J. CataL 1984, 85, 362 10 Nelson, H.C., Lussier, R.J. and Still, M.E. AppL CataL 1983, 7, 113 11 Forni, L. and Magni, E. J. Catal. 1988, 112, 437 12 Forni, L., Magni, E., Ortoleva, E., Monaci, R. and Solinas, V. J. Catal. 1988, 112, 444 13 Forni, L., Peratello, S., Ortoleva, E., Solinas, V., Monaci, R. and Ferino, I. Gazz. Chim. Ital. 1990, 120, 535 14 Argauer, R.J. and LandoR, G.R. US Pat. 3 702 886 (1972) 15 von Ballmoos, R. Structure Commission of International Zeolite Association, Butterworths, London, 1984 16 Meier, W.M. and Olson, D.H. Atlas of Zeolite Structure Types, 2nd ed., Butterworth, London, 1987 17 Satterfield, C.N. Mass Transfer in Heterogeneous Catalysis, MIT Press, Cambridge, MA, 1970 18 Rabo, J.A., Angell, C.L., Kasai, P.H. and Shomaker, V. Discuss. Faraday Soc. 1966, 41,328 19 Kasai, P.H. and Bishop, R.J. ACS Monogr. 1976, 171,350 20 Poutsma, M.L. ACS Monogr. 1976, 171,437