F. Rodriguez-Reinosoet al. (Editors), Characterization of Porous Solids I1 0 1991 Elsevier Science Publishers B.V., Amsterdam
685
THE ADSORPTION OF WATER VAPOUR BY MICROPOROUS SOLIDS
P.J.M. CARROTT~, M.B. KENNY, and C.R. THEOCHARIS
R.A.
ROBERTS^,
K.S.W. SING
Dept. of Chemistry, Brunel University, Uxbridge, Middlesex, UB8 3PH, U.K. 1 Faculdade de Cibncias, Universidade de Lisboa, Rua Escola Politecnica, 58, 1200 Lisboa, Portugal 2 Dowty Environmental and Safety Products, Adderbury, Banbury, Oxfordshire, OX17 lHJ, U.K. SUMMARY The adsorption of water vapour has been studied with a range of microporous carbons, zeolites and aluminophosphates in order to elucidate the relative influence of surface chemistry, pore size and pore shape upon the form of the water isotherm. It was possible to separate the adsorbents into three groups on the basis of their affinity and capacity for water vapour. The porous carbons were further examined using the BET and Dubinin-Serpinsky equations. The results show that the adsorption of water vapour at low p/po is largely dependent upon specific adsorbent-adsorbate interactions whilst at higher relative pressures the micropore size and shape control the extent of adsorption. It is proposed that hydrogen-bonded layers of water can be more readily accommodated in the narrow slit shaped pores (-0.5nm) of molecular sieve carbons than in tubular pores of similar width (e.g. Silicalite/ZSM-5). INTRODUCTION The many investigations of the adsorption of water vapour carried out over the last 20 years (refs. 1-5) have brought to
light several unusual features. Thus, large differences have been found in the shape of water isotherms determined with various porous solids such as carbons, oxides and zeolites. It is well known that the low polarisability, resulting from the small size of the water molecule, gives rise to weak non-specific interactions with graphitic carbon and dehydroxylated silica whilst the presence of a permanent dipole enables water molecules to undergo enhanced adsorbent-adsorbate interactions with surfaces where polar or cationic sites are exposed (refs. 7-8). The role played by the porosity of the adsorbent, particularly the microporosity, is much less clear. Conflicting reports have been
686
published regarding the mechanism of water adsorption in terms of the relative influence of the concentration of specific adsorption centres and the pore size distribution (refs 3 - 6 ) . The work reported here was designed to provide a systematic investigation of the adsorption of water by a number of well characterised microporous carbons, zeolites and aluminophosphates. In this way a more complete picture of the adsorption of water by microporous solids could be obtained and thereby allow a basis for the analysis of water isotherms in terms of texture and surface chemistry. EXPERIMENTAL Water vapour adsorption and desorption isotherms were determined gravimetrically at 298 K with the aid of quartz spring balances of the McBain-Bakr type. Prior to measurement each of the microporous carbons was outgassed at 573 K, the other samples torr. The adsorbents at 673 K, for 16 hours to a vacuum of <
used in this study are listed in Table I; they have all been employed in other related studies and their properties are described elsewhere (ref. 9). RESULTS AND DISCUSSION To provide a means of comparison between the different water isotherms the amounts of water vapour adsorbed, expressed as liquid volumes, at p/po = 0.01 and 0.95, are given in Table I as Vw(~.ol) and Vw(o.95) respectively. The apparent pore volume of each adsorbent was assessed from other adsorption measurements
(e-g. nitrogen at 77K) which allows a comparison of the different levels of fractional pore filling by water vapour, 8(o.01) and t3(o.g5), at p/po = 0.01 and 0.95. On the basis of these results the adsorbents in Table I are separated into three groups. The zeolites CaA, NaX and NaY are placed in Group I as each gives rise to a Type I isotherm in the IUPAC classification. These isotherms exhibit a pronounced rectangularity as indicated by the fact that both 8(o.01)and 8(o.95)are close to unity. An example of the Group I isotherms, on CaA, is given in Figure 1. The considerable affinity for water vapour shown by these zeolites results from the high concentration of specific cationic sites arising from their large Al contents The increase in uptake at higher p/po and the (ref. 10).
687 TABLE I
Adsorption 01
ADSORBENT
W d t r r Vapour
at Relative I’rcssures of 0.01 and 0.93
vw ( 0.01)
vw ( 0.95 ) ( ~ r n ~ q - ~ ) (cm39-1)
e(0.01)
e ( 0.95)
GROUP I ZEOLITES CaA NaX NaY
-
0.192 0.258 0.209
0 270 0.356 0.360
0.67 0.70 0.60
0.97 1-01 1.04
0 002 0.001 0.001 0.005 0.008 0.014 0.003 0.004 0.006
0.175 0.810 1.458 0 272 0.922 0.485 0.360 0.195
0.01 0.00 0.00 0.01 0.04 0.01 0.01 0.01 0.03
0.89 0.99 0.62 0.82 1.05 0.97 1.00 0.79 0.62
0.019
0.320 0.361
0.10 0.08
1.33 1.16
0.019
0.01 0.01
0.11 0.20
GROUP TJ. MICXOWROUS CARBONS
Takeda C 5%, Carbosieve Ax21
KCCl JF005 JF518 JF144 JFOlO JF025
charcoal cloth
f
0.338
-
ALUMINOPHOSPHATES ALPO-5 VPI-5
0.021
GROUP I11 ZSH-5 ZEOLITES SILICALITE I HZSM-5
0.001 0 004
-
0.038
characteristically small hysteresis loops are largely associated with intercrystalline effects. The microporous carbons and aluminophosphates are placed in Group I1 as they show a much lower affinity for water vapour (ref. This is mainly 11), as illustrated by the low values of @ ( o .ol). a consequence of the comparatively small number of specific sites present in these materials. The uptakes at p/po = 0.95, however, are generally equivalent to the amount required to give complete pore filling. In fact for the aluminophosphates the values of @ ( o . 9 5 ) are greater than unity as water is able to fit into the narrow six-rings which are unavailable to larger molecules (ref. 12). Each of the isotherms in Group I1 may be regarded as essentially Type V but in most cases an ill-defined Point B is The discernible in the region of low uptake (Figures 1 and 2). magnitude of this low pressure knee and the location of the point of inflexion varies widely between the adsorbents (Table I1 and Figure 2). It is also interesting that the aluminophosphates,
688
-I
0.3
W
c
E
= 0
5
0.2
I
v
>-
0.1
0
0.2
0.6
0.4
0.8
-
1.0
0
PIP'
PIP' Fig. 2.
Water vapour isotherms for the Group II
Fig. 1. Reduced water vapour isotherms for Group I ( 0 0 ) CaA and Group II (Om) JF025; [AA] adsorbents. Takeda 5A and lo*) KCCl adsorbents.
10.)
VPI-5 and (0.1 ALPO-5
which theoretically have neutral frameworks, show more rapid upswings at lower p/po than the carbons, possibly due in part to the presence of -OH groups at defect sites. Furthermore, VPI-5 is Low unusual in that it has a clear step at p/po = 0.02-0.06. pressure hysteresis was observed for each of these adsorbents, a small hysteresis loop was found even with the narrowest molecular sieve carbons. For the microporous carbons, it was noted that the hysteresis loop broadens as the pore width increased indicating that the high-pressure hysteresis is related to the process of capillary condensation. Again VPI-5 is unusual as it has a double hysteresis loop which closes at p/po - 0.07: the loop at lower p/po is associated with the step and cannot be explained by capillary condensation. Finally, the zeolite HZSM-5 and its aluminium-free analogue Silicalite I are placed into Group 111. These adsorbents may be termed truly hydrophobic since their affinity for water vapour remains low over the entire range of p/po as shown by the small values of 8(o.01)and 0(0.95)in Table I. The low uptake of water vapour exhibited by these zeolites is exemplified in Figure 3
689
0
0.2
0.A
0.6
0.8
PIP’ Fig- 3. Reduced water vapour and nitrogen isotherms for the Group 111 adsorbents. Silicalite I).Oi Water, (AA) nitrogen: . HZSM-5
1.0
Fig. 4. A sketch o f the hydrogen-bonded structure of water in a slit-shaped pore.
The circles
represent the positions of the oxygen atoms.
where the fractional pore filling by water and nitrogen are compared. The intracrystalline pores of the ZSM-S/Silicalite structure are tubular and of - 0.55 nm diameter. It is not surprising then that a three-dimensional array of hydrogen-bonded water molecules cannot easily be accommodated in such pores without considerable distortion of the directional hydrogen-bonds (ref. 13). On the other hand, as shown in Figure 4 , a thin slab of water can more easily develop in the slit shaped pores of molecular sieve carbons of similar width (e.g. Takeda 5A) and also the water structure is clearly able to form in the wider tubular pores of ALPO-5 ( - 0.8 nm) and VPI-5 ( - 1.2 nm). However, if favourable cationic or -OH groups are present on the internal surface of the Group I11 adsorbents, as in HZSM-5 (Si/Al = g o ) , then a limited number of water molecules may be adsorbed at low p/po through enhanced adsorbent-adsorbate interactions. A s shown in Figure 3 , the nitrogen and water isotherms are similar in the multilayer region and therefore, the hysteresis loops and much of the water adsorbed by Silicalite I is likely to be associated with secondary mesoporosity and intercrystalline effects.
690
In order to further elucidate the complex pore filling process in microporous carbons, the empirical Dubinin-Serpinsky (DS) equation (refs. 14-15) was used to assess the influence of polar sites on the shape of the isotherm. This equation was developed from the concept of adsorption of water molecules at uniform high energy primary adsorption centres. Molecules adsorbed on these sites act as secondary adsorption centres via a hydrogen-bonding mechanism. Thus, this model does not refer explicitly to the role played by pore size. The DS equation may be written in its modified form (ref. 15) as: P - - -
n
PO
D(no+n)(l-kn)
I
where n is the amount of water adsorbed at p/poI no is the concentration of primary adsorption centres, D is the ratio of the rates of adsorption and desorption, whilst k is a constant dependent on the uptake at p/po = 1. A best fit was achieved when the equation was applied in its quadratic form (ref. 16) but a reasonable fit was still only obtained for the Type I11 part of the isotherm, so that the range of fit was found to be dependent on the pore width of the adsorbent (Figures 5 and 6). Deviations at high pressures coincide with the plateau regions and at low pressures arise from the assumption that the primary sites are of one strength. This brings the validity of no into question, making it only significant from a comparative stand-point. The values of no and D calculated from these fits are given in Table 11. Also given are the BET monolayer capacities, nm, and the values of nm/ABET, where ABET is the BET area given by nitrogen adsorption. Of course, nm only gives an approximate indication of the number of primary centres as more than one water molecule may be associated with a particular site. Examination of Table I1 reveals no close agreement between nm and no and no obvious correlation between these values, or nm/ABET, and the location of the point of inflexion. However, there does appear to be some degree of correlation between the location of the point of inflexion and the values of D, which are in fact related to the pore widths (ref. 17). As can be seen for the charcoal cloths, the inflexion point is more strongly dependent on the percentage burn-off which directly controls the pore width. Therefore, the
691
80
60
40
20
0
8
0
24
16
Quadratic fit of the Dubinin-Serpinrky
equation to JF144 experimental data.
I
/
Wide Pore Width
c p/p'=0.446
0
32
n [mrnol g-'l Fig. 5.
v 1
20
40
60
00
100
n Immol g-'l
~ i6. ~ ~ . ~ f i t of the ~ Dubinin-Serpinsky d ~ equation to KCC1 experimental data.
number of specific sites appears to govern the uptake at p/po < 0.2 whilst the pore width controls the uptake at higher relative pressures. Furthermore, as noted above, the increase in width of the hysteresis loop with pore size also demonstrates the dominant It role played by micropore size distribution at higher p/po. Adsorption Of Water Vapour By Microporous carbons
TABIS 11 ADSOXBENT
Pore Width
"m
BET
4-l)
(-1
"mIABET
( m o l m-2) ~10-4
Inflexion Point (p/pO)
DS "0
(-1
D
¶-I)
KCCL
W
0.66
2.22
0.78
1.03
1.42
Ax21
W
0.31
0.91
0.68
0.47
1.63
JFSl8 (7lrBO)
W
1-90
10.58
0.68
4.09
1.31
Takeda c 5A
0.35
9.02
0.50
0.99
2.28
0.81
7.26
0.46
0.94
2.08
0.32
2.72
0.43
0.31
2.52
JF14C (40tBO)
N N N N
1.41
11.41
0.40
1.16
N
1.46
21.64
0.37
--
2.62
JF025 (15rBO) JF005 (lOtB0)
N
1.32
14.98
0.32
1.50
2.76
JFOlO ( 5 0 t B O )
Carbosieve
Samples with the prefix Jf are charcoal cloths.
BO = Burn-off
--
~
~
692
would seem from the above that the calculated values of no and D have no real theoretical significance but do help to define the mathematical form of the water isotherm over a limited range. Lastly, the fact that VPI-5 has an inflexion point at lower p/po but contains uniformly wider pores than ALPO-5 suggests that a three-dimensional water array can form more easily in wider channels, assuming that -OH groups play a relatively minor role. Hence, within limits, somewhat wider tubular pores and narrower slit shaped pores result in micropore filling at low p/po. ACKNOWLEDGEMENTS
The authors wish to thank Dr. J.J. Freeman, Professor K.K. Unger and Dr. A. Venero for provision of samples, Dr. E.L. Short for assistance with curve fitting routines and the Ministry of Defence and the SERC for financial assistance. REFERENCES 1.
H.F. Stoeckli, F. Kraehenbuehl and D. Morel, Carbon, 21
2.
R.C. Bansal, T.L. Dhani and
(1983) 589.
389.
S.
Parkash, Carbon, 16 (1978)
5.
M.M. Dubinin, Carbon, 18 (1980) 355. S . S . Barton, M.J.B. Evans and B.H. Harrison, J. Colloid Interface Sci., 45 (1973) 542. S . S . Barton and J.E. Koresh, J. Chem. SOC., Faraday Trans. 1,
6.
J.R. Dacey, J.C. Clunie and D.G. Thomas, Trans. Faraday SOC.,
7.
A.M. Youssef, T.M. Ghazy and Th. El-Nabarawy, Carbon, 20
8.
F.S. Baker and K.S.W. Sing, J. Colloid Interface Sci., 55
9.
R.A. Roberts, Ph.D. Thesis, Brunel University, Uxbridge, Middlesex, UK, 1988. D.W. Breck, "Zeolite Molecular Sieves", Wiley, 1973. C.R. Theocharis, M.R. Gelsthorpe and D. Yeates, J. Chem. SOC., Faraday Trans. 1, 85 (1989) 2641. M.E. Davis, C. Montes, P . E . Hathaway, J.P. Arhancet, D.L. Hasha and J.M. Garces, J. Am. Chem. SOC., 111 (1989) 3919. M.B. Kenny and K.S.W. Sing, Chem. Ind. (London), 2 (1990) 39. M.M. Dubinin, E.D. Zaverina and V.V. Serpinsky, J. Chem. SOC., (1955) 1760. M.M. Dubinin and V.V. Serpinsky, Carbon, 19 (1981) 402. S.S. Barton, M.J.B. Evans, J. Holland and J.E. Koresh, Carbon, 22 (1984) 265. M.M. Dubinin, K.M. Nikolaev, G.A. Petukhova and N.S. Polyakov, Izv. Akad. Nauk SSSR (Ser. Khim.), (1984) 743.
3.
4.
79 (1983) 1147. 54 (1958) 250.
10. 11.
12. 13. 14. 15. 16. 17.
(1982) 113.
(1976) 605.