J. Phys. Chem. 1991, 95, 6309-6316
6309
'H NMR Spectroscopy of Water Adsorbed on Synthetic Chrysotlle Asbestos: Mlcrotubes with Acidic and Basic Surfaces Sumio Ozeki,* Department of Chemistry, Faculty of Science, Chiba University, 1-33 Yayoi-cho. Chiba 260, Japan
Yuichi Masuda,' Hirotoshi Sano,
-
Department of Chemistry, Faculty of Science, Tokyo Metropolitan Uniuersity, 2- 1 1 Fukazawa, Setagaya, Tokyo 158, Japan
Hiroko ski, Analysis Center of Chiba University, 1-33 Yayoi-cho, Chiba 260, Japan
and Kenta Ooi Government Industrial Research Institute, Shikoku, 2-3-3 Hananomiya-cho, Takamatsu 761, Japan (Received: December 13, 1990)
Adsorbed water on a synthetic chrysotile asbestos was studied by IH NMR spectroscopy at -80 to +30 "C. Two kinds of the longitudinal relaxation rate ( R I )were observed. The fast (R,?and slow (Rl') relaxation rates are ascribed to water adsorbed on an inner S O 2 and an outer Mg(OH)2surface, respectively, of a tubular chrysotile crystal with a cylindrical mesopore with a 7-nm diameter. At low coverage 0, preferential adsorption of water onto the inner surfaces seems to occur. Water with RIf mode has three states, the first, second, and third, and the higher layer on the inner surfaces. At 0 > 3, water molecules with Rlrare adsorbed in the mesopores by the capillary condensation to form a liquidlike phase that may be supercooled to -60 O C . On the outer surfaces, at least the first layer of adsorbed water are solidlike and water in fourth and higher layers behaves like liquid down to -30 O C . The above ideas are also discussed by the half-height width and the integral intensity of the NMR absorption peaks.
Introduction
Water adsorbed on metal oxides and porous materials has been investigated by various methods,'-15 such as dielectric loss and permittivity, 'H NMR and IR spectroscopies,etc. Adsorbed water is immobilized in a monolayer like ice and in some cases behaves as liquid on flat surfaces and in wide pores. Molecules in a potential field, as in pores, may have peculiar properties. Recently, it was reported that when water was adsorbed on porous jarosite and alunite, which have slitlike micropores of ca. I-nm width, the ( I ) McIntosh, R. L. Dielecrrlc behavior of physically adsorbed gases; Marcel Dckker: New York. 1966; Chapter 5. (2) MaCaffeetety, E.;Ravdic, V. H.; Zettlemoyer, A. C. J. Collold Inrerface Scf. 1970, 34, 452; Dfscuss. Farahy Soc. 1971, 52, 239. (3) Dransfeld, K. J. Chem. Phys. 1%2,36, 1574. Baldwin, M. 0.;Morrow, J. C. Ibfd. 1962, 36, 1591. (4) (a) Morimoto, T.; Iwaki, T. J. Chem. Soc., Faraday Trans. I 1987, 83, 943. (b) Iwaki, T.; Morimoto, T. Ibid. 1987, 83, 957. (c) Iwaki, T.; Morimoto, T. Longmuir 1987, 3, 282. (5) Kaneko, K.; Serizawa, M.; Ishikawa, T.; Inouye, K. EuN. Chem. Soc. Jpn. 1975, 48, 1764. (6) Zimmerman, J. R.; Lasater, J. A. J . Phys. Chem. 1958, 62, 1157. (7) Hall, P. 0.;Williams, R. T.; Slade, R. C. T. J . Chem. Soc., Faraday Trans. 1 1985,81, 847. (8) Brey, W. S.,Jr.; Lawwn, K. D. J . Phys. Chcm. 1964,68, 1474. (9) (a) Ruing, H. A.; Thomson, J. K.; Krebr, J. J. J . Phys. Chem. 1964, 68, 1621. (b) Foster, K. R.; Resing, H. A. Ibld. 1976,80, 1390. (IO) Li icas, M.; Straley, C.; Catanzo, P. M.; Giese, R. F., Jr. J . Colloid Inretface El. 1985, 107, 221. (1 I ) Pearson, R. T.; Dtrbyshiete, W. J. Colloid I n r e r f i Scf.1974,46,232. (12) Clark, J. W.; Hall, P. 0.;Pidduck, A. J.; Wright, 0. J. J . Chem. Soc., Faraday Trans. I 1985,81, 2067. (13) Kurata, M.; Kaneko, K.; Inouye, K. J . Phys. Chem. 1984,88,2119, (14) Kaneko, K.; Fujiwara, Y.; Nishikawa, K. J . Colloid Inrerjiace Sci. 1989, 127,298. Suzuki, T.; Kaneko, K. Carbon 1990,28, 103. (I5) (a) Ozeki, S. J. Chem. Soc., Chem. Commun. 1968,1039; Longmufr 1989,5,181. (b) Ozeki, S.; Masuda, Y.; Sano, H. J. Phys. Chem. 1989,93, 7226.
0022-3654/91/2095-6309S02.50/0
motional mode of adsorbed water changed dramatically with the micropore filling.Is Such changes of motional mode of adsorbed water will depend strongly on an array of adsorption site pore shape, surface fractal, etc. Chrysotile asbestos (MgbSi4010(OH)8)has a unique tubular structure16 (Figure l), whose inner diameter is about 7 nm. The crystal walls are constituted by repeating layers of basic Mg(OH)2 and acidic Si02(Figure 2). Thus the inner and outer surfaces of chrysotile crystals are composed of Si02and Mg(OH)* layers, respectively. The cylindrical mesopores have a substantial homogeneous size and thus are useful to test an adsorption theory on capillary condensation.17 The two different, completely separated surfaces are also useful for examining the accessibility of molecules onto the adsorption sites, -MgOH and -SiOH; e.g., NO was preferentially adsorbed on the outer surfaces in low NO pressure and on the inner ones in high pressure.'* It is well-known that mineral asbestos has carcinogenic activity, e.g., lung cancer.I9J" Therefore, serious attention has been devoted - ~ ' gases (SO2,NO, to the interaction of proteins in ~ a t e r ~ l and (16) E.g.: Yada, K. Acta Crysrallogr. 1971, A27,659. Yada, K.; Iishi, K. Am. Mlneral. 1977,62,958. Tanji, T.; Yada, K.; Akatruka, Y. Clays Clay Mineral. 1984,32,429. Breck, D. W. Zcollre Molecular Sieves; Wiley: New York, 1974; p 37. (17) Scholten, J. J. F.; Beers. A. M.; Kiel, A. M. J . Coral. 1975, 36, 23. De Wit, L. A,; Scholten, J. J. F. Ibld. 1975, 36, 30, 36. (1 8) Uchiyama, H.; Kaneko, K.; Ozeki, S.J . Chem. Soc., Faraday Trans. I 1989,85, 3833. (19) Selikoff, 1. J.; Lee, D. H. Asbesros and Diseases; Academic Press: New York, 1978. (20) Wagner, W. L.; Rom, W. N.; Merchant, J. A. Healrh Issues Relared ro Metal and Nonmeralllc Mining, Butterworths: Boston, 1977. (21) Jones, 8 . H.; Edward, J. H.; Wagner, J. C. Er. J . Ind. Med. 1972, 29, 287. (22) Morgan, A. Environ. Res. 1974, 7, 330. (23) Light, W. 0.;Wei, E. T. Envlron. Res. 1977, 13, 135. (24) Valerio, F.; Balducci, D.; Scarabclli, L. Enulron. Res. 1986,4I, 432.
Q 1991 American Chemical Society
6310 The Journal of Physical Chemistry, Vol. 95, No. 16, 1991
Ozeki et al.
10
20
30
50
40
60
2 m u Kk) I *
Figure 3. X-ray diffraction pattern of a prepared chrysotile asbestos. I
100 nm Figure 1. Transmission electron microscopic image of a prepared chrysotile asbestos. Outer surface
-
P 0 E S
1
1
I
I
I A /#
I
I
P 'P.
Figure 4. Adsorption isotherm of N2for a prepared chrysotile asbestos at 77 K: 0, adsorption; 0 , desorption.
Si
0
Inner surface
Figure 2. Schematic structure of chrysotile asbestos.
NH,, etc.)'-28 with mineral asbestos. However, surface properties such as molecular adsorptivity would be affected by some impurities in a mineral asbestos. Additionally, surface Mg2+ions are apt to be leached, leading to very defective outer surfaces of mineral chrysotile^.^^*^^*^^^^ When a chrysotile is synthesized under well-defined conditions, the synthetic chrysotiles could give more reliable surface properties and adsorption mechanisms.17.18,31,32 (25) Murphy, W.J.; Ross,R. A. Cluys Cluy Miner. 1977.25.78; J. Phys. Chem. 1977,81,135; Can. J . Chem. 1978.56.1847. Ross, R. A.; Todd, J. E.Surfuce Technol. 1983,18,1. Ross,R. A.; Vishwanathan, V. Ind. J. Chem. 1983,22A, 138. (26) Bonneau, L.;Suquet, H.; Malard, C.; Pezerat, H. Emiron. Res. 1986, 41, 432. (27) Young, G. J.; Healey, F. H.J. Phys. Chem. 1954,58,881. Healey, F. H.; Young, G. J. Ibid. 1954, 58, 885. (28) Fripiat, J. J.; Faille, M. D. Cluys. Cluy Miner. 1966, 15, 305. (29) Choi, I.; Smith, R. W. J. Colloid Interface Sci. 1972, 40, 253. (30) Pam, P. F.; Cangh, L. V.; Fripiat, J. J. Bull. Soc. Chim. 1976, 1021. (31) Ozeki, S.; Takano, I.; Shimizu, M.;Kaneko, K. J. Colloid Interfuce Sci. 1989, 32, 523. (32) Ozeki, S.; Oowaki, Y.; Kaneko, K. Colloids Surf. 1989, 37, 329.
The behavior of water in the vicinity of surfaces of chrysotile, *~~J3-~~ as well as surface properties in aqueous ~ y ~ t e m ~ , ~is~very important for understanding the activity of chrysotile in vivo. However, such investigation is very limited, and only a few reports on heat of wetting and adsorption isotherms are a~ailable?~SThe chemical and geometrical features of the surface may influence the structure and dynamic properties of water in the vicinity of walls of chrysotile pores. Therefore, NMR signals of water adsorbed could give some information on the pores and surfacesi5 as well as the dynamic properties of adsorbed water and the accessibility of surfaces for water. In this paper, dynamic prop erties of water adsorbed on a synthetic chrysotile asbestos, obtained by 'HNMR spectroscopy, are related to the tubular structure: the two kinds of surfaces and cylindrical mesopores. Experimental Section
Chrysotile asbestos was synthesized according to the method of Noll et al.37 Aerosil 380 (Nippon Aerosil Co., Ltd.) dissolved in aqueous NaOH solution was mixed with aqueous MgCI, solution. The gel mixture (pH 13), which contained stoichiometric amounts of Si and Mg, was autoclaved in a gold-coated nickel cell at 300 OC and 65 atm for 24 h. The product was washed with abundant aqueous saturated Mg(OH)2solution, rinsed with water, and then freeze-dried. The prepared material was characterized by X-ray diffraction (Rigaku Denki Geigerflex 2038) and transmission electron microscopic observation (Hitachi H-800 transmission electron microscope). N2 and H 2 0 adsorptions on the chrysotile, dried at 110 "C and Torr for 15 h, were measured by a gravimetric method at -196 and +30 O C , respectively, as previously reported.15 'H NMR was measured by a JEOL FX-200 Fourier transform spectrometer under the magnetic field, 4.7 T (200 MHz for the (33) Martinez, E.; Zucker, G. L. J. Phys. Chem. 1960,60,924. (34) Seshan, K. Environ. Heulth Perspect. 1983, 53, 143. (35) Pundsack, F. L. J. Phys. Chem. 1955, 58, 143. (36) Zettlemoyer, A. C.; Young, G. J.; Chessick, J. J.; Healey, F. H. J. Phys. Chem. 1953,57,649. (37) Noll, W.; Kircher, H.; Sybertz, W . Kolloid-2. 1958, 157, 1.
NMR Spectra of Adsorbed Water on a Microtube 5
I
I
I
I
I
The Journal of Physical Chemistry, Vol. 95, No.16, 1991 6311
I
4
*a & I
n '
E
TU 3 0
L
z
P
2
1
1
5
3
7
Radius I n m
Figure 5. Pore size distribution for a prepared chrysotile asbestos calculated by the Cranston-Incley method.
proton resonance). The spin-lattice relaxation time ( T , ) of the 'H nucleus of water was determined by the saturation-recovery method, as previously described.Isb The temperature of the samples (-80 to +30 "C) was controlled within 0.5 O C with a variable-temperature control unit (JEOL PVTS2 type). About 200 mg of the chrysotile sample was placed into a 5-mm-diameter NMR cell, which was dried by the procedure similar to the H 2 0 adsorption, and then was sealed off after the sample had adsorbed definite amounts of water at various relative pressures and 30 OC.
Results The XRD pattern of the prepared sample had the characteristic peaks of chrysotile, as seen in Figure 3. The TEM image shows that an average size of a tubular particle has a length ( l ) of 600 nm and inner (di) and outer (do) diameters of 7 and 30 nm, respectively (Figure 1). Figure 4 shows the adsorption isotherm of N2 for the chrysotile, which reveals a hysteresis above ca. 0.7 in relative pressure. The specific surface area (a,) of the prepared chrysotile, 92 m2/g, was calculated by the BET method. The a, value is comparable with a geometrically estimated value, 70 m2/g, which was obtained as a sum of inner and outer surface areas, using the inner (3.5 nm) and outer radii (1 5 nm) of a tube and assuming the density of chrysotile crystal 2.5 g/cm3. This means that almost all inner surfaces are opened to N2 (H20)molecules. Pore size distribution (Figure 5) was calculated by means of the Cranston-Incley method for a cylindrical pore,3s using the reference curve in ref 39 for thickness of surface film of N2.39aMaximum porosity appears at 7-nm diameter and agrees approximately with the inner diameter estimated from the TEM image. The 3-nm pore arises from interstices between cylindrical particle^.^^*^^ The chrysotile sample has no micropores of less than ca. 1.2 nm in diameter, since the t of N2 adsorption exhibited no break points on the curve. Figure 6 illustrates the adsorption isotherm of water vapor for the chrysotile at 30 O C . The isotherm also has a hysteresis loop. The monolayer capacity of water, u,/(mg/g) = 22.2 f 0.9 (an average value of four runs), was obtained from the BET plot. The specific surface area from u,, awris 78.7 m2/g, using the molecular area of water 0.106 nm2. The surface coverage of water, 8, is (38) Cranston, R. W.; lnkley, F. A. Ado. Carol. 1957, 9, 143. (39) Gregg, S.J.; Sing. K. S.Adsorpllon Area and Poroslry; Academic Prm: New York, 1982; (a) p 93, (b) pp 94-97; (c) p 4. (40) Lippens, B. C.; De. Boer, J. H.J . Caral. 1965, 125, 356.
'0
aa
05 PIP,
a75
1
Figure 6. Adsorption isotherm of water vapor for a prepared chrysotile asbestos at 303.15 K: 0,adsorption; 0 , desorption.
defined by u/u,, where v is an amount of adsorbed water at humidity p / p o . Here p and po are a vapor pressure and the saturation vapor pressure of water at 30 OC, respectively. and inteFigure 7 shows variations of half-height width grated intensity (fa = peak height X u112) of N M d signals of adsorbed water with temperature T. I, is approximately independent of temperature for 8 < 2 and decreases abruptly below -40 OC for 8 > 3, especially at 8 = 5.7 (Figure 7A). u I / z for all 8, except for 8 = 5.7, is almost constant above -40 O C (Figure 7B). At -80 OC uIi2for all B increases steeply, and the values approach those (2.7 kHz at 30 OC to 4.3 kHz at -30 OC for 8 = 0) of the skeletal OH groups of chrysotile. The results suggest that water for 8 > 3 may be immobilized or frozen below -40 O C . The longitudinal relaxation rate, R I (=TI-l, where TI is the longitudinal relaxation time), was obtained from amplitude measurements of free decay signals of magnetization. An analysis of the data was made according to the expression
WO- M(OI/MO =f,exp(-tRI') + (1 -PIexp(-tRI?
(1)
where Moand M ( t ) are the thermal equilibrium magnetization and the magnetization at a time t , f is the fraction of hydrogens with the slow relaxation rate, and RI8and R I fare the slow and fast relaxation rate, respectively. Figure 8 shows examples of curve fitting for free decay curves using eq 1. The agreements between the experimental and calculated ones were fairly good in all cases. The parameters, RIM, Rlr,a n d j , which were obtained from eq 1, are plotted in Figures 9 and 10. The temperature dependence of R I is very small, except for RIf for 8 = 5.7. The R I fvalues (60 s-I) for 8 = 0.75, which remain unchanged with temperature, are comparable with those for 8 = 5.7 and less than -50 OC: water in the monolayer is immobilized like ice formed on the surfaces. T h e y values (0.65-0.75), except for 8 = 5.7, remain roughly unchanged with temperatureabove -60 OC (Figure lo), indicating that hydrogens of more than 65% of adsorbed water molecules have a motional mode of the slow relaxation, irrespective of temperature and 8. fa,uIl2,R,, a n d p are replotted against 0 in Figures 11-13. I,, roughly speaking, increases linearly with increasing 0 above -30 OC; subtle break points on the curves appear in the region of 0 = 1.4. At 8 < 3, f, for -60 and -80 OC changes in ways similar to those for other temperatures but becomes almost constant at
Ozeki et al.
6312 The Journal of Physical Chemistry, Vol. 95, No. 16, 1991 0
,
.
h
O
0.4
A
ai
0
Time Is
Figure 8. Examples of a curve fitting for a free decay curve of magnetization of hydrogen atoms in adsorbed water on a prepared chrysotile asbestos by using eq 1: circle, experiment; dotted line, calculation. Conditions and used parameters (8, temperature/OC, R1'/s-', R,'/s-',f): a, (0.75, 30, 4.428, 60.92,0.7284); b, (5.65, 30, 3.907, 16.58,0.7463). I
'
I
I .0Q -40 0 40 Temperature I "C
B
\
I-
I
4
I
4
'' -bQ
I
1
- 40
I
I
I
1
0
40
Temperature I @C
Figure 9. RI' and RIfof water adsorbed on a preparedchrysotile asbestos as a function of temperature. B: 0,0.75; 0, 1.44, 8, 2.55; 0, 3.3; 0, 5.7.
I
I
I
I
I
I
I
I
-40 0 40 Temperature I 'C F i m 7. Variations of integrated intensity I , (A) and half-height width v I p(B)of NMR signals of adsorbed water on a prepared chrysotile with temperature T. B: 8, 0; 0, 0.75; 0, 1.44; 8, 2.55; 0, 3.3; 0, 5.7. -00
B > 3, suggesting that water molecules in the first, second, and probably third layers in a multilayer remain as mobile as those for B < 3 even below -60 "C. Y ~ for / ~more than -30 "C increases abruptly below B 1, At -60 "C, v I l 2is markedly different from
-
- 80
-
40 0 Tempcratur I 'C
40
Figure 10. f of water adsorbed on a prepared chrysotile asbestos as a function of temperature. B: 0, 0.75; 0, 1.44; 8, 2.55; 0, 3.3; a, 5.7.
The Journal of Physical Chemistry, Vol. 95, No. 16, 1991 6313
NMR Spectra of Adsorbed Water on a Microtube I
I
I
I
0 '
I
I
t A
-
70
I
, I
I
I
I
I
I
I
-
50
30
-
a- 1 0 1
5l1 -
C . '
4
c-
- -_-_- - - --- --
A-
......... ...
..........
@.e;,
,.f/
2
0
*-a:
.............
6
4
& Figure 12. Variation of RI of water adsorbed on a prepared chrysotile with surface coverage 8. Temperature/°C: 0 , 3 0 ; 0.0; 0,-30; 8,do,
0,-80.
I
0
I
I
2
I
I
6
4
& Figure 13. Variation of) of water adsorbed on a prepared chrysotile with surface coverage 8. Temperature/°C: 0, 30; 0 , O ; 0 -30; 8, do,0, -80.
-
I
I
I
2
1
4
1
I
6
b Y ~ (B) / ~ of water adsorbed on a prepared chrysotile with surface coverage 8. Temperature/°C: 0 , 30; 0 , O ; 0 , -30 8, -60: @, -80.
Figure 11. Variations of I. (A) and
that at room temperature only for B > 3, where adsorbed water is frozen. At -80 OC,water seems to be immobilized in the whole B ranp R, depends more strongly on B than does R,',whose B dependence is significantly small (Figure 12). The RIfvalusabove -30 O C decrease gradually with B through a vague plateau in the
region of B = 2.5; Le., the state of adsorbed water chan ts across 8 2.5. This change appears more clearly on the R,-8 curves for -60 and -80 O C , which deviate upward from the curves for the other temperatures around B = 2.5. Figure 13 shows that f decreases with increasing B at low coverages and increases with B of more than 1.4. The extrapolated f value (0 = 0) Po= 0.83 corresponds approximately to a ratio of the outer surface area to the total surface area, f = d,,/(d, t di) = 300/(300 t 70) = 0.81. Therefore, the variation off with 0 seems to reveal an adsorption process of water onto the inner and outer surfaces.
f
Discussion Two phasecl of Adsorbed Water. Various factors, such as crca relaxation, may cause a non-single-exponential decay curve of a magnetization, as shown in Figure 8. We will, however, interpret it on the basis of a peculiar structure of chrysotile: the two different relaxations may arise from the inner and outer surfaces of chrysotile microtubes.
6314 The Journal of Physical Chemistry, Vol. 95, No. 16, 1991 In general, the exchange among water molecules, as long as they are adsorbed on the same surface, is very rapid even at different sites as well as in different layers. Therefore, only an averaged RI value is usually obtained, unlike the case of the chrysotile. On the other hand, a motion of water on the outer surfaces of the tubes may be distinguished in TI from that of water on the inner surfaces, because the average distance which a water molecule travels during a time scale of TI (10-Ll s), 10L104 nm or less$l is comparable with the average length of the chrysotile particle, lo3 nm. f is in proportion to number of 'Hatom having a relaxation rate Rls,because (i) I, increases approximately in proportion to an amount of water adsorbed, (ii) the line width Y ~ is/ almost ~ constant above -60 OC, (iii) spectra seem to be a single absorption peak, Le., the line widths of the two components (R18and R,') are not so different. Therefore, the agreement between thef, value at 0 = 0 (0.83) and thefvalue (0.81) that is the geometrically expected fraction of the outer surface is understandable if water with R1' is on the outer surfaces of the chrysotile. We thus suppose that the slow and rapid relaxations may arise from water adsorbed on the outer and inner surfaces of chrysotile, respectively. Then, as mentioned above, since the exchange between water molecules or protons in the various states could be rapid, both RIsand RIf,respectively, could be a value averaged, if any, among different states of water. When water u/(mg/g) is adsorbed on the chrysotile, the observed R16and Rlf obey the following equation:
where ui is the amount of adsorbed water in the i state having RIi." For example, assuming a three-state a,0, and y states, eq 2 is converted to R1
Ria
for o < U,
R1 = (uu/u)(Rla- RIP) + RIP Rl
(l/U)(U,(RI"
for
UP
(3a)
< u < U,
+ up
(3b)
- RIT) + ~p(Rl' - RIY)} + RIY for u,
+ u8 < u (3c)
When the observed R I is plotted against I/u, the plot could be composed of the three straight-line portions for the three-state model. In practice, for the chrysotile system u should be replaced by u(l -f)for RIfand uf for RI1. Figure 14 demonstrates that the plots for R l f a r e apparently composed of two linear sections, although those for R: are unclear: there are two (or three) states of water with each relaxation mode. This also seems to support the idea that the rapid and slow relaxation modes arise from water adsorbed on the outer and inner surfaces. Water on the Inner Surfaces (in Mesopores). Water in a monolayer is markedly different from that in a multilayer, judging from the break points around B = 1 in the Ia- 8 curves and the steep drops of uIl2and R I f for B < 1. Water for 0 = 0.75 is immobilized even at rmm temperature and water for 0 < 3 is unfrozen even at -60 OC (Figure 7 and 9). At B = 3.3 as well as 5.7 water seems to be frozen, like a liquid, at -30 to -40 OC. The temperature dependence of RIfis very small, except for B = 5.7 (Figure 9). Such a small temperature dependence or a broad maximum of R I may occasionally occur when some motional modes are coupled, when a motion has large anisotropy, and when a correlation time distributes ~ i d e l y . ~ *The ' ~ shape of the (41) Strictly speaking, a satisfactory condition should be I >> (A9)Il2, where I is the length of a chrysotile particle and is the minimum duplaament within a TIfor water. The displacement was rtimated by wing Einstein's relation (air) = 2Dt, where t is the time and D is the diffusion coefficient, which is assumed to be D < 1od cm*/s for adsorbed water on a silica gel (Morariu, V. V.; Mills, R. Z . Phys. (Munich) 1972, 79, 1) and D = 2.3 X 10-' cm2/s for pure water at 25 OC. (42) Derbyshire, W. In Wotrr, Franks, F.,Ed.;Plenum Press: New York, 1982; Vol. 7, p 339. (43) Almagor, E.; Belfort, G. J . Colloid Interface Sei. 1978, 66, 146.
Ozeki et al. I
I
I
I
0.05 Ql l/Vfs /mg-'g Figure 14. R l s(bottom) and Rlf (top) of water adsorbed on a prepared respectively. Temchrysotile as a function of (uf)-l and (D( 1 -f)\-I, perature/C 0,30; 0 , 0; 0 , -30; 0 , -60; 0, -80.
maximum around 0 "C, if so, is considerably flat, e.g., in comparison with the RI-T curve expected for a motion with an activation energy of 4 kcal/mol (hydrogen bond), assuming that T = ool = (200 MHz X 27r)-I at 0 OC: R I ( T OC)/R,(O "C) changes between 0.6 and 1 (estimated) and between 0.8 and 1 for B = 1.4 and 3.3 in the range -30 to +30 O C . One of characteristics of Rlf is that the values are significantly large, more than 17 s-I, in comparison with other systems; e.g., the R I values of water in mesopores of silica gel are 7-8 S-I for 10-nm-diameterpores and 12-13 s-l for 3-nm-diameter pores at 0 OC.I1 However, the R,'vaIues of 17-29 s-I for 3.3 < B < 5.7 in the range -30 to +30 "C are comparable with the R1 value, ca. 23 s-I (25 "C), of pure liquid water, which is estimated from the BPP theory:
and the condition Wo7 = 1, if Rlf is around an R1 maximum. Here M 2 is the second moment which is calculated from the BPP equation, using lil = 0.28 s-' and 7 = 2.5 p for pure liquid water at 25 OC, wo is the resonance frequency (2n X 200 MHz), and T is the correlation time. Therefore, water in a mesopore of chrysotile would be liquidlike above 8 = 3.3. Figure 14 shows that the plots are composed of two linear sections, indicating that water with an RIfmode has apparently two states; e.&, RIf 29 and 3 s-I for 0 OC. The second state appears in the region B = 2.6-2.9, which is corrected by considering that the pore diameter diminishes as adsorption progresses, as discussed later. Unfortunately, the number of the data points in Figure 14 is not enough to conclude anything, especially 8 < 2, but we suppose that the first layer on the inner surfaces should be distinguished from others, referring to the other data here. Thus, it is assumed that water on the inner surface is in a threstate: first, second and third, and higher layers (or condensed phase). Then, the parameters estimated using eq 3 are listed in Table I.
-
~~
(44) Levin,
~~~~
~
~
Y. K. J . Chem. Phys. 1974,60, 2890.
(45) Connor, T. M. Trans. Forodoy Soc. 1964, 60, 574.
NMR Spectra of Adsorbed Water on a Microtube
The Journal of Physical Chemistry, Vol. 95, No. 16, 1991 6315
TABLE I: Relaxation Rates (It1')and Adsorption Capncities (v,) of Water in Different Adsorption States on Synthetic Chrysotile Asbestosa relaxation rate/& (adsorption capacity)(mg/g)) first layef temp, O C ,%phaseb ?phaseb (a-phaseb) second and third layersC 0 to +30 -30 -60 -80
0 to +30 -30 -60 -80
25 f 3 (16 i 2) 28 (19) 30 (20) 38 (13) 4.7 f 0.6 (39 f 4) 5.2 (47) 5.3 (36) 5.0 (40)
R/
0.02 12
1.9
60.5 (6.7) 56 (6.7) 58 (6.7) 59 (6.7)
27 26 25 30
f 2) ** 535 (11.7 (13.3 f 2) (11 f 3) 5 (6 4)
RI'
3.3 3.6
R,' and of were estimated by using eq 3 and Figure 14. ba-,B-, and y-phases correspond to those in eq 3. CRlfvalues at 0 = 0.75 are assumed to be q u a l to those of the first layer or a-phase, because eq 1 shows the fact that Rlf for the first layer is constant if the surface is homogeneous. Here the adsorption capacity of the first layer is assumed to be om(] -f)= 22.2(1 0.70) = 6.7 mg/g.
-
Judging from Rlf, water in first and second and third layers on the inner surfaces is in almost the same state from 30 to -30 O C , while water in the higher layers changes markedly its state between 0 and -30 O C (Figure 14). The RIfvalue, 0.02 f 1.9 s-I, obtained by extrapolation of Rlf for 0 and 30 "C to infinite u, is comparable with the R , , 0.25-0.59 s-I, of pure liquid water at the same temperature region. On the other hand, the extrapolated RIfvalue for -30 O C , ca. 12 s-l, agrees roughly with RI of supercooled water in microemulsion around -40 O C (Figure 15).
The temperature dependence of RIffor 8 = 5.7 is similar to that of RI for silica4 and charcoalh under saturation vapor (Figure 15). The R,'maximum appears around -70 O C , which agrees remarkably with that in the charcoal system." The increase and decrease in RIf(R,) around the maximum, respectively, seem to correspond to behaviors of a supercooled water" and an ice,48as seen in Figure 15. Therefore, we suppose that liquidlike water adsorbed on the inner surfaces behave like superoooled water below -30 OC and is converted to ice around -70 O C . Water Adsorbed on the Outer Surfaces. No motional changes of water having an RI1mode with 8 between the first and sixth layer are clearly indicated by any data here. However, a vague break point around 0 -2.6 in the RIs-(uy)-' plots (Figure 14) for 0 and 30 O C seems to appear and the extrapolated R1'values to (of)-' = 0 are 3.4 s-I, which is in the same order as a supercooled water (Table I). Also, when log R I sfor 8 = 3.3 and 5.7 is plotted against T I (Figure 15), a maximum appears around -50 O C as in R,'. These suggest that layers higher than the fourth in a multilayer exhibits liquidlike properties, i.e., a transition of supercooled water to ice which is not significant because water adsorbed on the outer surfaces seems to be immobilized even at room temperature: RIsvalues are in the range 3.7-5.3 s-l. The extremely small dependences of R I son 8, especially 8 C 3, are not easily understandable. Water molecules in a first layer on the smooth, flat outer Mg(OH)2surfaces are strongly adsorbed through hydrogen bond~,Z~-Mg(0H)~--OH~, and thus their rotation and/or two. (or three-) dimensional translation would lead to small RI', which is comparable to R I for an ice. Then, we assume that only the first layer in a multilayer is responsible for the measured R I svalues (in which the relaxation of water in a solidlike system is often determined by a motion with the most rapid relaxation through spin diffusion) and presumably that water molecules in the second and third layers are also forced to be in the same state as those in the first layer or more immobilized state, as in a growing c I ~ ~ s t e r . ~Otherwise ~fl as the thickness of adsorbed water layer increases, water molecules in the multilayer would (46) W o e " , D.E.J. Chcm. Phys. 1963, 39, 2783. (47) Hindman, J. C.; Svirmickas, A.; Wood,M.J . Chrm. Phys. 1973,59, 15 17. Angell, C. A. In Water; Franks, F., Ed.;Plenum Press: New York, 1982; Vol. 7, p 1 . (48) Glasel, J. A. In Wutrr; Franks, F., Ed.;Plenum Press: New York, 1972: Vol. I . D 215. (49) Meakhfi, R. J.; Sack, R. A. Ausr. J . Sci. Res. 1951, Ad, 213. Sack, R. A. Ibid. 1952, AS, 135.
Figure 15. log R , of adsorbed water, supercooled water, and ice as a function of TI.RIf for water adsorbed on the chrysotile: 0,8 = 3.3; 0 . 0 = 5.7. RI1for water adsorbed on the chrysotile: 0, B = 3.3; 0 , 8 = 5.7. Water adsorbed on a charcoal" and a silica gel47under saturation vapor: A, charcoal; A, silica gel; 0 , supercooled water;47 . , ice fb.u
be liable to be, on the average, more mobile, and thus RI1could depend appreciably on 8 and temperature, as seen for 6 > 3. This assumption seems to be consistent with the fact that water having RI6relaxation mode remains unfrozen down to -80 O C at least at 0 < 3, as observed often for a monolayer.'J5 The assumption presupposes a proton channeling (a rapid exchange of protons), as in an ice, between adsorbed water molecules which array regularly on the flat, homogeneous Mg(OH)2 surface. However, the measured uI12 values might be somewhat small for the solidlike model. Adoorption Pmesses of Water. As discussed above, ally values are comparable with thefvalue. But it seems thatf changes appreciably with 8 around 1.4. The extrapolated valuef, = 0.83 would reflect the surface fraction of the chrysotile. The expected
6316
J . Phys. Chem. 1991, 95, 6316-6322
value 77.7 m2/gS0 obtained by using theyo value, agrees with the experimental a, (78.7 m2/g), which is smaller than aN (92 m2/g). Therefore, we infer that the change i n p reveals an adsorption process or accessibility of the surfaces for water molecules: at 6 < 1.4 preferential adsorption of water onto the inner surfaces with increasing 6. This is consistent with the fact that the isosteric heat of adsorption of water on chrysotile decreased with increasing 6 through a maximum around 6 = 0.6.5’ It is likely that water adsorbs more strongly on silica surfaces by hydroxylation. The f minima near 6 = 1.4 might relate to the fact that the completion of water adsorption onto surface OH on a silica may occur around 8 = 1.5.’ The freezing of liquidlike water for 6 = 5.7 was markedly observed not in R I Sbut only R,f. This suggests that water in multilayers differs from that in capillary condensed phase. The thickness of water film for 6 = 5.7 is ca. 1.7 nm = 5.7 X 0.30 nm, a,
(50) If the difference between o, (92 m2/g) and (I, (78.7 mZ/g) of the chrysotile coma from the inner SiOl surfaces, the expected a,value 77.7 m2/g is estimated from 92f, + 92(1 -yo)X (27.4/150.5), by using o, = 150.5 and ow = 27.4 m2/g for a nonporous silica (Aerosil 130; Nippon Aerosil Co., ref 51). (51) Ozeki. S.;Wakai, C. Unpublished data.
where the thickness of water monolayer (tw) is assumed to be 0.30 nm. Although the film thickness (1.7 nm) seems to be too thin to form a meniscus of liquid water in the cylindrical mesopore (7-nm diameter), the thickness could be substantially underestimated. Because the film thickness is defined on the basis of monolayer capacity, while the diameter of the mesopore would diminish as water adsorption progresses. Additionally, when water is adsorbed by capillary condensation, the concept of 6 will be inapplicable. In fact, the pore volume of a chrysotile particle, ~ 1 ( 4 / 2 ) ~38.51 nm3, agrees approximately with the volume of water adsorbed at 6 = 5.7 on the inner surface, 2rl(di/2)(tw6) 37.61 nm3;i.e., the mesopores of chrysotile are completely filled with water near 6 = 5.7 @/p0 0.83), probably through the capillary condensation. At 6 = 3.3 about 60% of the total pore volume of the tubes would be filled with water. The liquid phase around 6 = 3.3 arises from condensed water in all pores of less than 5.3-nm diameter, determined by the Kelvin equation. Considering that the chrysotile sample would have considerably homogeneous pores, the capillary condensation would also occur in the interstices of about 3-4-nm diameter (Figure 4) which can be made by cylindrical crystals ordering in parallel.
-
-
-
Registry No. Water, 7732-18-5.
Electron Transfer from a Surfactant-llke Zinc Porphyrin to a Covalently Attached Ru02 Cluster Ute Rescb and Marye Anne Fox* Department of Chemistry, University of Texas at Austin, Austin, Texas 78712 (Received: December 19, 1990) The photophysical properties of a surfactant-like zinc porphyrin and its diacid derivative, covalently attached, through a bipyridine unit, to a Ru02 cluster, are described. Intramolecular electron transfer from the excited zinc porphyrin to the bound Ru02particles has been studied by fluorescenceand flash photolysis techniques in homogeneous solution and in both anionic and cationic water-in-oil micrcemulsions. Coordination of the zinc porphyrin to Ru02 decreases both the porphyrin fluorescence intensity and triplet yield by 95-97%. The lifetime of the zinc porphyrin triplet state is unaffected by binding to Ru02,whereas the singlet state lifetime is greatly shortened, suggesting electron transfer from the porphyrin excited singlet state to the Ru02cluster. For the correspondingdiacid porphyrin-Ru02 system, a diminution of the fluorescence intensity by 91% is observed.
Introduction Within the last few years, several attempts to mimic natural photosynthesis by covalently linked donoracceptor systems have been described.’-’ An efficient photosynthetic system must contain a chromophore that strongly absorbs visible light, an electron-transfer pathway that separates the electron-hole pair, and catalytic sites for the transformation of the charge carriers into reduced and oxidized products. In such a system, the effi( 1 ) (a) Tabushi, I.; Koga, N.; Yanagita, M. Tetrahedron Lrrr. 1979,20, 257-260. (b) Kong, J. L. Y.; Loach, P. A. Nature 1980,286,254-256. (c) Migita, M.:Okada,T.; Mataga, N. Chem. Phys. Lrtr. 1981,84,263-266. (d) Kong, J. L. Y.; Spears.P. A.; Loach, P. A. Photochem.Phorobiol. 1982,35, 545-553. ( e ) Bcrgkamp, M. A.; Dalton, J.; Netzel, T. L. J . Am. Chem. Soc. 1982. 104, 253-259. (f) Nishitani, S.;Kurata, N.; Sakata, Y.; Misumi, S. J . Am. Chem. Soc. lW, 105,7771-7772. (g) Mataga, N.; Karen, A.; Okada, T.; Nishitani, S.J . Phys. Chem. 1984,88, 5138-5141. (h) Wasielewski, M. R.; Niemczyk, M. P.; Svtc, W. A.; Pcwitt, E. B. J . Am. Chem. Soc. 1985, 107, 1080-1082. (i) Leland, B. A.; Joran, A. D.; Felkcr, P. M.; Hopfield, J. J.; Zewail, A. H.; Dervan, P. B. J . Phys. Chem. 1985,89, 5571-5573. (j) Scaler, J. L.; Johnson, M.R.; Lin, T.-Y.; Creager, S.E. J . Am. Chem. Soc. 1988, 110, 3659-3661. (2) Hamman, A.; Porter. 0.;Wilowska, A. J. Chem. Soc., Faraday Trons. 2 1984, 80, 191-204. (b) Blondel, 0.; De Keukeleire, D.; Harriman, A.; Milgrom, L. R. Chem. Phys. k i t . 1985,118, 77-82. (c) Kanda, Y.;Sato, H.; Okada, T.; Mataga, N . Chem. Phys. Len. 1986, 129,306309. (d) Kaji, N.; Aono, S.;Okura, 1. J . Mol. Coral. 1986, 36, 201-203. (3) (a) Elliott, c. M.; Freitag, R. A,; Blaney, D. D. J . Am. Chem. Soc. 1985, 107, 4647-4655. (b) Daniebn, E.;Elliott, C. M.; Merkert, J. M.; Meyer, T. J. J . Am. Chem. Soc. 1987, 109, 2519-2520.
0022-3654/91/2095-63 16$02.50/0
ciency of energy conversion depends on the quantum yield of formation of the initial charge separation and on the ratio of the rates of product formation from this initially formed radical-ion pair to that of back electron transfer. Numerous attempts to slow back electron transfer have been In some applications, back electron transfer is prevented by the use of sacrificial electron donors or acceptors to trap the separated charge centers. An alternate known strategy for increasing the yield of photoproducts in such systems involves light-induced electron transfer across a phase boundary, e.g. in membranes, micelles, or microemulsion^.^^^ In these systems, the electrostatic field a t the interface can assist in the separation of the primary charge carriers (4) (a) Fendler, J. H. Acc. Chem. Res. 1976, 9, 153-161. (b) Brugger, P.-A.; Infelta, P. P.; Braun, A. M.; Graetzel, M. J. Am. Chem. Soc. 1981, 103.320-326. (c) Graetzel, M., Ed. Energy Resources rhrough Phorochemisrry ond Carolysls;Academic: New York, 1983. (d) Mandler, D.;Dcgani, Y.;Willner, 1. J. J . Phys. Chem. 1984,88,4366-4370. (e) Steinmueller, F.; Rau, H. J . Phorochem. 1985,28,297-308. (f) P i h i , M.P.; Lerebours, B.; Brochette, P.; Chevalier, Y. J. Phorochem. 1985,28, 273-283. (e) Seta, P.; Bienvenue, E.; Moore, A. L.; Mathis, P.; Bemasson, R. V.; Liddel, P.; Pedsiki, P. J.; Joy, A.; Moore,T. A.; Gut,D. Narun 1985,316,653-655. (h) Metsuo, T. J . Phorochem. 1985,29,41-54. (i) Nagamura, T.;Takeyama, N.; Tanaka, K.; Matsuo, T. J . Phys. Chem. 1986, 90,2247-2251. (j) Colaneri, M.J.; Kcvan, L.; Schmehl, R. J . Phys. Chem. 1989. 93, 397-401. (5) (a) Willner, I.; Ford, W. E.; Otvos, J. W.; Calvin, M. Narun 1979, 280, 823-824. (b) Turro, N.J.; Graetzel, M.; Braun, A. M. A n e w . Chem. 1980, 92, 712-734. (c) Willner, 1.; Goren, 2.;Mandler, D.; Maidan, R.; Degani, Y . J . Photochem. 1985, 28, 215-228.
0 1991 American Chemical Society