Professional Documents
Culture Documents
J Micromeso 2016 08 018
J Micromeso 2016 08 018
a r t i c l e i n f o a b s t r a c t
Article history: Original and thermally degraded FAU type zeolites were characterized by nitrogen adsorption at 77 K and
Received 30 September 2015 compared with isotherms for amorphous silica from literature. The isotherms were evaluated by the
Received in revised form method of Remy and Poncelet which is an additive composition of the Dubinin-Radushkevich and the
28 December 2015
BET equation. Nonlinear fits typically generate two optimal values. One is attributed to very low, the
Accepted 19 August 2016
other to very high values of the C-constants of the BET part. The conversion of the DR-isotherm to a set of
Available online 21 August 2016
Langmuir isotherms with different affinities shows that the low value is correct. For comparison all data
were compared with those obtained by the t-plot method. The t-plot generally yields a less good fit with
Keywords:
Nitrogen adsorption
the experimental data, lower micropore volumes and deviating external surfaces, especially in the case of
BET equation thermally degraded zeolites with an enhanced degree of mesopores. The entire isotherms from p/
C-term po ¼ 0.5 to 0.98 are covered by an additive combination of the Dubinin-Radushkevich equation and the
t-plot general Aranovich equation. That equation, however, fails for the determination of the external surface.
Micropore © 2016 Elsevier Inc. All rights reserved.
External surface
1. Introduction used pressure range [9]. Furthermore, Galarneau et al. have also
shown that this type of extrapolation has principally to be ques-
A lot of classical procedures exist to evaluate the data of N2 tioned [2], and in addition, the thickness t calculated according to
adsorption isotherms at 77K. They range from methods claming to de Boer is only correct in case of ideally flat surfaces. In reality, the
be universal [1] or with a sophisticated view to the adsorbent surface curvature has an effect which cannot be neglected [2]. The
surface on the molecular level [2]. The aim of this contribution is incorporation of that effect into the t-plot method requires data
the evaluation of a method of data analysis for materials with mi- which are not available in the most cases of routine characteriza-
cropores and an external surface, especially for zeolites. tions of porous materials. Therefore a simple mathematical tool for
Although adsorption data analysis based on molecular simula- the reliable determination of the micropore volume and the
tion of adsorbate-adsorbent interaction, such as NLDFT [3], has external surface is requested.
found a growing scientific and commercial interest, the application For the practical use the method should especially focus on the
of classical methods prevails. The calculation of micropore volumes course of the isotherm describing the transition between micro and
and external surfaces of zeolites is mostly done by the t-plot mesopores. Remy and Poncelet defined that region as an additive
method originally developed by de Boer [4]. Sing has evidenced the superposition of the Dubinin-Radushkevich (DR) equation for
strong influence of the chemical composition of the surface on the micropore adsorption [10] and a BET term [11] for the behaviour on
evolution of the statistical thickness t [5]. It was recommended to the external surface (DR-BET method [12]). The authors have
choose a non-porous material of the same chemical composition compared their results with the external surface obtained by the t-
[6] or characterized by the same C-value of the BET-plot [7]. But plot (2 types of reference systems) and by the usual BET method.
such a look to a reference material is effortful. Mostly the universal Great discrepancies were found. The question remains if the
t-plot accounting for the average surface property of a lot of method of Remy and Poncelet [12] is sound enough to replace
different materials is used [8]. commonly used methods. Based on the DR term in combination
In addition to the question which type of t-plot is appropriate, with BET one has to address the weakness of both equations alone.
the extrapolation procedure generally depends on the arbitrarily The clearness of the BET theory is based on ideal prerequisites
which are the homogeneous solid surface, neglected “horizontal”
interactions in the same adsorbed layer, and identical forces be-
* Corresponding author. tween each layers from the first up to infinite [13,14]. But in reality
E-mail address: buttersack@inc.uni-leipzig.de (C. Buttersack). these conditions are only partly fulfilled. As a consequence it was
http://dx.doi.org/10.1016/j.micromeso.2016.08.018
1387-1811/© 2016 Elsevier Inc. All rights reserved.
64 C. Buttersack et al. / Microporous and Mesoporous Materials 236 (2016) 63e70
Table 1
Characteristic parameters of industrial zeolites and amorphous silica (LiChrospher 1000, data from literature [35]). Total volume Vtot at x ¼ 0.95, Stot by the I-point method [1], t-
plot (0.4 > t > 1 nm) by Equation (1) with the correlation coefficient r. Si/Al and particle size taken from the producer.
t-plot
4. Results and discussion industrial FAU type zeolites. The values of Vmicro and Sext for the CBV
zeolites were similar to those reported by Remy and Poncelet [12].
4.1. Analysis of the DR-BET method In total 165 samples were evaluated with these methods. In all
cases, the high C-values were situated above C > 104 (Fig. S1).
Fig. 1 shows the nitrogen isotherm measured with a X-zeolite. Therefore, as shown in Fig. 2, a change of C in the range of 104e108
The non-linear fitting procedure due to the DR method was has practically no influence on the regression coefficient and the
generally restricted to partial pressures lower than 0.45. This is also optimal values of Vmicro. In that case the BET term is a rectangular
advised by the reason that capillary condensation between the function and only a constant value of Sext is superposed to the DR
powdered samples can occur [33]. In most cases the lowest partial term. The region of 2 < C < 20 is instable and no optimal values
pressure used for the evaluation was not lower than 0.005. This is a were found. The instability is also due to the applied algorithm
consequence of a limited time available for attaining the adsorption (Levenberg and Marquardt [26]) for the non-linear regression. In
equilibrium within the experimental procedure. The fitting due to contrast, some sensitivity is found for the low C-values (Fig. 2)
DR-BET performed with the boundary conditions which means that the BET term is a slow increasing function su-
perposed to the DR-term. Analogous results were found for the
0 < Vext < 0.01 parameters Sext and Eo (Fig. S2). For all 165 samples the regression
0 < k < 0.1 coefficient r2 of the second optimum is only marginally greater than
and that of the first optimum. The mean ratio is r2/
0<C r1 ¼ 1.00032 ± 0.00092.
The high C-values correspond to the mathematical property of
results in very great optimal values of C > 10000 and leads to the the BET Equation (5). For very large C-values, C 1 approaches C,
blue curve. However, when the additional condition C < 10 was and the BET-Equation (5) is reduced to
used, the optimal value for C was small (C ¼ 6.4) leading to the
green curve. These 2 optima were found for all samples. The ob- V2 ¼ Vext =ð1 xÞ (11)
tained characteristic parameters were shown in Table 2 for 11
At this point no dependence on C is possible. Exactly that limit
value of the BET equation, where C approaches infinity, was called
“point of inversion” and proposed as a reference point for the
calculation of the total specific surface area [34].
Fig. 3 shows the data of Fig. 1 within a half-logarithmic plot. The
measured data clearly obey the DR term within a large pressure
range with values of p/po down to 106, as expected independent
on the C-term. The underlying experimental data was obtained
with extremely long times of equilibration for each measuring
point within the low pressure region. All other zeolite samples
were measured with lower effort, and only the data above x ¼ 0.005
were used for the regression analysis.
In order to get information for the external surface of the zeo-
lites, a set of data from amorphous silica [35] is presented here. It
stems from the literature and was originally measured for the
purpose of having a reliable reference for the so-called as-plot [36].
Both regression types of the DR-BET-isotherm can be applied for
the pressure range from x ¼ 106 to 0.4 but as shown in Fig. 4, the
fitting with low C-values is significantly better than that with high
values. The C-values of 12000 and 12.6 resp. are in the same order
Fig. 1. Nitrogen adsorption (77K) on a 13X zeolite. Fitting according to the DR-BET
than those found for the zeolites. Although the material was
isotherm (8) (0.0005 < x < 0.4) and the DR-Aranovich Equation (10) claimed to have a total amorphous character, the application of the
(0.0005 < x < 0.98). DR-BET method yields a fraction of micropores (Table 2) which
66 C. Buttersack et al. / Microporous and Mesoporous Materials 236 (2016) 63e70
Table 2
Characteristic parameters of industrial zeolites obtained by the DR-BET Equation (8) for 0.004 < x < 0.4 with the correlation coefficients r1 and r2. In addition the DR-BET results
of an amorphous silica (LiChrospher 1000, data from literature [35]) for 106 < x < 0.5 are listed.
HSZ390 0.333 62.9 2.19 399986 0.99953 0.353 47.1 2.26 2.72 0.99978
DAY 0.316 16.9 2.37 471985 0.99909 0.321 20.7 2.40 0.97 0.99961
Y100 0.338 43.8 2.22 319684 0.99974 0.352 32.9 2.27 2.58 0.99988
Y756 0.316 29.5 2.26 1720670 0.99978 0.325 24.1 2.30 2.30 0.99988
Y530 0.314 38.8 2.25 196720 0.99900 0.327 30.9 2.29 1.94 0.99916
Y411 0.314 34.7 2.39 711570 0.99934 0.323 25.3 2.48 5.45 0.99985
Y307 0.302 26.0 2.47 713633 0.99943 0.308 18.9 2.55 6.28 0.99989
CBV780 0.285 182.0 1.95 1611450 0.99904 0.345 122.1 2.20 6.96 0.99988
CBV760 0.307 179.0 2.02 1286929 0.99911 0.357 102.0 2.23 6.99 0.99987
CBV500 0.308 25.5 2.51 572890 0.99981 0.316 20.6 2.55 2.38 0.99993
13X 0.320 15.6 3.07 394415 0.99990 0.325 8.5 3.10 6.41 0.99995
LiChrospher 0.0075 11.5 1.26 11774 0.99980 0.0081 12.6 1.81 6.35 0.99995
and the silica are similar with respect to the chemical nature. The
zeolite surface should be characterized by an imperfect crystal
structure and contain silanol groups such as the amorphous
counterpart. However, the principal difference of the zeolite surface
is established by the presence of the pore entries. The respective
surface fraction is the region where micropores and external sur-
face are overlapping. The larger surface area of the amorphous
material is therefore reasonable but the result has to be supported
by the comparison with further materials.
To evaluate the correctness of the C-value one can also compare
the DR-term with the BET-term of the same experimental data. The
C-value expresses the affinity of the monolayer formation as a
Langmuir process. It is shown here that it is mathematically
possible to express the DR-equation as a sum of Langmuir
isotherms
Fig. 3. Nitrogen adsorption (77K) on the 13X zeolite. Fitting according to the DR-BET
isotherm (8) (low optimum) (106 < x < 0.4).
Vmicro X
n
ki x
V ¼ Vmicro expf kðlnð1=xÞÞ2 g ¼ (12)
n i¼1 1 þ ki x
even so amounts to 23% of the total volume and is significantly
greater than that obtained by the t-plot (Table 1). For comparison, where ki can be found by nonlinear regression. By that way the DR
the usual BET-plot using the data between x ¼ 0.005 and 0.4 is also isotherm calculated for a typical dealuminated Y zeolite was con-
shown in Fig. 4. The C-value is 250 and similar to those reported in verted to a series of expansion with 10 or 12 terms (Fig. S3,
the literature [34,37], but due to the model used here the value of Table S1). As shown in Fig. 5 the first molecules were adsorbed with
12.6 is correct. The external surface of the dealuminated zeolites ki ¼ 109 while the last ones are adsorbed with a constant of about
25. The values of ki can directly be compared with the C-values.
C. Buttersack et al. / Microporous and Mesoporous Materials 236 (2016) 63e70 67
300
-1
C = 3.6 ) / m g
2
250
200
150
50
0
0 50 100 150 200 250 300
S ext (lower optimum) / m2 g-1
Table 3
Characteristic parameters of industrial zeolites obtained by the t-plot and by the DR-Aranovich Equation (10) and the simplified Equation (14) with m ¼ 0.36. In addition the
DR-BET results of an amorphous silica (LiChrospher 1000) for 106 < x < 0.5 are listed.
Type DR-Ara (1) x ¼ 0.0005 …. 0.99 DR-Ara (2) m ¼ 0.36, x ¼ 0.0005 …. 0.40
HSZ390 0.360 819.5 2.50 0.050 0.025 0.99787 0.353 88.3 2.26 1.290 0.99986
DAY 0.320 42.9 2.42 0.916 0.316 0.99922 0.321 54.0 2.40 0.391 0.99960
Y100 0.355 90.5 2.23 0.369 0.434 0.99944 0.352 63.3 2.27 1.179 0.99989
Y756 0.327 177.0 2.27 0.122 0.246 0.99974 0.325 63.8 2.30 0.694 0.99991
Y530 0.328 62.3 2.27 0.631 0.460 0.99917 0.327 71.9 2.29 0.773 0.99921
Y411 0.328 376.4 2.39 0.066 0.000 0.99943 0.323 46.5 2.47 2.064 0.99994
Y307 0.311 43.3 2.49 0.617 0.395 0.99945 0.309 33.9 2.54 2.377 0.99997
CBV780 0.342 173.4 2.24 5.123 0.326 0.99988 0.347 172.2 2.19 3.893 0.99995
CBV760 0.353 145.3 2.27 5.239 0.373 0.99863 0.358 143.4 2.22 3.978 0.99993
CBV500 0.316 197.3 2.56 0.191 0.000 0.99993 0.316 44.9 2.55 0.872 0.99992
13X 0.325 11.6 3.11 6.089 0.307 0.99984 0.325 12.7 3.10 3.705 0.99996
LiChrospher 4.76 105 28.2 1.77 1.41 0.292 0.99986
progressive smoothing of the liquid-covered surface during the a constant or simply m ¼ 1. The application of the DR-Aranovich
multilayer filling process. isotherm is therefore regarded as an interesting empirical tool,
The overall effect of the surface compression can be comprised but according to the present knowledge it has no relevance for the
in a correction function f (x) which connects the Aranovich function characterization of porous systemsin practice.
with the BET term according to The full adsorption isotherm of the silica up to x ¼ 0.97 again can
be described by the combination of the DR-isotherm with the
Aranovich equation (Equation (10)). According to the low value of
1 f ðxÞ m ¼ 0.29 the C-value of 28 is greater than that obtained with DR-
¼ (13)
ð1 xÞm ð1 xÞ BET (Table 3, Fig. S6).
The correcting function f (x) principally must have an impact on the
fitting procedure similar to the findings presented in Fig. 2, also in 4.4. Comparison of the DR-BET-method with the t-plot
case of a limitation to small values of x. To illustrate that effect the
exponential term is expressed by a Taylor series of expansion Fig. 8 shows the striking disadvantage of the t-plot compared
1 ðm 1Þm 2 ðm 1Þmðm þ 1Þ 3
f ðxÞ ¼ ¼ 1 þ ðm 1Þx þ x þ x þ/
ð1 xÞm1 2! 3! (14)
; m>0
After truncation of the series (14) behind the second term Equation with the DR-BET isotherm. The t-plot is the linear extrapolation of
(10) can now be modified to the linear part of the plot of V versus t. The linear part of t (see
Equation (1)) was found to exist between 0.4 and 1 nm for most
Sext sv Cxð1 þ ðm 1Þx Þ
V ¼ Vmicro expf kðlnð1=xÞÞ2 g þ (15)
ð1 xÞð1 þ ðC 1Þx Þ
As shown in Table 3, the values of m obtained by fitting with
Equation (10) over the full pressure range of 0.005 < x < 0.98 were
found to spread from 0 to 0.5 for 11 different FAU type zeolites. It
seems to be improbable to assign those differring m-values to
multilayer compression alone. Especially at high partial pressure
capillary condensation in the void between aggregated zeolite
particles may influence the isotherm [33] and thus allow an influ-
ence by the sample preparation. Nevertheless, the validity of the
DR-Aranovich isotherm (Equation (10)) is remarkable. The value of
m has no essential influence on the evaluation of the micropore
volume, but a strong influence on the external surface and the C-
term exists. To evaluate that effect, Equation (15) is applied in the
same lower pressure region used for the DR-BET-method
(0.005 < x < 0.4) and a constant value of m ¼ 0.36, being the
average of those belonging to all zeolite samples with m > 0.1, was
inserted as well. Compared with the DR-BET method, the regres-
sion coefficients are similar, but the external surface is doubled,
while the C-terms are halved (Table 3). To compare the surface data Fig. 8. Comparison of the DR-BET isotherm (low optimum, 0.0004 < x < 0.60) with the
of similar zeolites with each other, it is therefore essential that m is t-plot (0.4 < t < 10 nm) based on the experimental data for zeolite Y756.
C. Buttersack et al. / Microporous and Mesoporous Materials 236 (2016) 63e70 69
Acknowledgement
450 dx.doi.org/10.1016/j.micromeso.2016.08.018.
2
400
350 References
300
[1] A.K. Ladavos, A.P. Katsoulidis, A. Iosifidis, K.S. Triantafyllidis, T.J. Pinnavaia,
250 P.J. Pomonis, Micropor. Mesopor. Mater. 151 (2012) 126e133.
200 [2] A. Galarneau, F. Villemont, J. Rodriguez, F. Fajula, B. Coasne, Langmuir 30
(2014), 13266e13224.
150 [3] J. Landers, G.Y. Gor, A. Neimark, Coll. Surf. A 437 (2013) 3e32.
100 [4] J.H. de Boer, B.G. Linsen, T.J. Osinga, J. Catal. 4 (1965) 643e648.
[5] S.J. Gregg, K.S.W. Sing, Adsorption, Surface Area and Porosity, Academic Press,
50
1982, p. 98.
0 [6] G.D. Parfitt, K.S.W. Sing, D. Urwin, J. Coll. Interf. Sci. 53 (1975) 187e193.
0 50 100 150 200 250 300 [7] A. Lecloux, J.P. Pirard, J. Coll. Interf. Sci. 70 (1979) 265e281.
2 -1 [8] J.H. de Boer, B.C. Lippens, B.G. Linsen, J.C. Broekhoff, A. van den Heuvel,
external surface S ext (DR-BET low optimum) / m g T.J. Osinga, J. Coll. Interf. Sci. 21 (1966) 405e414.
[9] F. Rouquerol, J. Rouquerol, K.S.W. Sing, P. Llewellyn, G. Maurin, Adsorption by
Fig. 10. Comparison of the external surface obtained by the DR-BET method (low Powders and Porous Solids, Academic Press, 2014, p. 306.
optimum) with that of the t-plot. [10] N.D. Hutson, R.T. Yang, Adsorption 3 (1997) 189e195.
[11] S. Brunauer, P.H. Emmett, E. Teller, J. Am. Chem. Soc. 60 (1938) 309e319.
[12] M.J. Remy, G. Poncelet, J. Phys. Chem. 99 (1995) 773e779.
[13] ref. [5] p. 49.
zeolite samples. That region corresponds to a pressure range of [14] C.S. Dutcher, X. Ge, A.S. Wexler, S.L. Lang, J. Phys. Chem. C 115 (2011)
0.15 < x < 0.75. The extrapolated micropore volume lays outside 16474e16487.
that region. In contrast, the DR-BET plot of most zeolite samples is [15] D.D. Do, Adsorption Analysis: Equilibria and Kinetics, Imperial College Press,
1998 (Chapter 4).
based on the region of 0.0005 < x < 0.5 covering the transition
[16] M.J. Remy, D. Stanica, G. Poncelet, E.J.P. Feijen, P.J. Grobet, J.A. Martens,
between micropore filling and external adsorption. The calculated P.A. Jacobs, J. Phys. Chem. 100 (1996) 12440e12447.
micropore volume is located inside that region. In Fig. 8 an example [17] F. Collignon, M. Mariani, S. Motreno, M. Remy, G. Poncelet, J. Catal. 166 (1997)
53e66.
is given for both procedures.
[18] M.J. Remy, A.C. Vieira-Coelho, G. Poncelet, Micropor. Mater 7 (1996) 287e297.
Fig. 9 shows the comparison of the micropore volume evaluated [19] E.F. Sousa-Aguiar, A. Liebsch, B.C. Chaves, A.F. Costa, Micropor. Mesopor.
with DR-BET and the t-plot of all 165 zeolite samples. The t-plot Mater 25 (1998) 185e192.
values are about 8% lower. Thermally degraded zeolites with low [20] J. Horniakova, M. Kralik, A. Kaszonyi, D. Mravec, Micropor. Mesopor. Mater 46
(2001) 287e298.
micropore volume show the greatest discrepancy. The external [21] G. Larsen, H.S. Silva, R. Venturini di Silva, J. Non-cryst. Solids 271 (2000) 1e11.
surfaces are extremely different. Here, a split was observed. Some [22] A. Aerts, A. van Isacker, W. van Huybrechts, et al., J. Appl. Catal. A 257 (2004)
results of the t-plot were greater, most of them smaller than those 7e12.
[23] S.P.B. Kremer, C.E.A. Kirschhock, M. Tielen, F. Collignon, P.I.J. Grobet,
obtained by DR-BET (Fig. 10). P.A. Jacobs, J.A. Martens, Adv. Funct. Mater 12 (2002) 286e292.
[24] Z.M. Wang, Z. Liu, N. Yamashita, H. Kanoh, K. Oi, Langmuir 18 (2002)
1957e1962.
5. Conclusion [25] A. Marcilla, A. Go mez-Siurana, F.J. Valde s, Ind. Eng. Chem. Res. 48 (2009)
10820e10826.
[26] P. Gill, W. Murray, M. Wright, Practical Optimization, Academic Press, 1981.
Based on the nitrogen adsorption data of 165 FAU-type zeolites [27] M.M. Dubinin, E.D. Zaverina, L.V. Radushkevich, Zh. Fiz. Khim 21 (1947)
with different Si/Al and different strong exposure to thermal stress 1351e1362.
[28] M. Polanyi, Trans. Faraday Soc. 28 (1932) 316e333.
the DR-BET method is superior to the t-plot. The method yields 4 [29] G.L. Aranovich, Langmuir 8 (1992) 736e739.
characteristic parameters, which are Vmicro, Eo, Sext, and C. The curve [30] G.L. Aranovich. Zh. Fiz. Khim. 63(1989) 1527e1534.
fitting principally yields two optima, that with the low C-value [31] G.L. Aranovich, M.D. Donohue, J. Coll. Interf. Sci. 173 (1995) 515e520.
[32] P. Schneider, Appl. Catal. A 129 (1995) 157e165.
being the correct one. By the conversion of the DR-term to a sum of
[33] E.L. Fuller, Micropor. Mesopor. Mater 98 (2007) 102e106.
Langmuir equations it is possible to compare the affinity of the [34] J. Pomonis, et al., Micropor. Mesopor. Mater 69 (2004) 97e107.
adsorption in the micropores with that on the external surface. [35] M. Jaroniec, M. Kruk, J.P. Olivier, Langmuir 15 (1999) 5410e5413.
Some interdependence of the optimal external surfaces and the [36] M. Jaroniec, K. Kaneko, Langmuir 13 (1997) 6589e6596.
[37] D.A. Payne, K.S.W. Sing, D.H. Turk, J. Coll. Interf. Sci. 43 (1973) 287e293.
optimal C-values exists. For the comparison of similar samples the [38] B.P. Bering, M.M. Dubinin, V.V. Serpinsky, J. Coll. Interf. Sci. 21 (1966)
optimization with constant C-values is recommended. 378e393.
70 C. Buttersack et al. / Microporous and Mesoporous Materials 236 (2016) 63e70
[39] B.P. Bering, M.M. Dubinin, V.V. Serpinsky, J. Coll. Interf. Sci. 38 (1972) [42] A. Seri-Levy, D. Avnir, Langmuir 9 (1998) 2523e2529.
185e194. [43] P. Vajda, A. Felinger, J. Chomatogr. A 1324 (2014) 121e124.
[40] J. Lorens, M. Pera-Titus, J. Coll. Interf. Sci. 331 (2009) 302e311. [44] J.J. Fripiat, L. Gatineau, H. van Damme, Langmuir 2 (1986) 562e567.
[41] R. Roque-Malherbe, Micropor. Mesopor. Mater 41 (2000) 227e240. [45] G.F. Cerofolini, L. Meda, Surf. Sci. 416 (1988) 403e422.