Influence of the Partial Hydrophobization of Fumed Silica by

10816

Langmuir 2003, 19, 10816-10828

Influence of the Partial Hydrophobization of Fumed Silica by Hexamethyldisilazane on Interactions with Water V. M. Gun’ko,*,† V. V. Turov,† V. M. Bogatyrev,† B. Charmas,‡ J. Skubiszewska-Zie¸ ba,‡ R. Leboda,‡ S. V. Pakhovchishin,† V. I. Zarko,† L. V. Petrus,† O. V. Stebelska,† and M. D. Tsapko§ Institute of Surface Chemistry, 17 General Naumov Street, 03164 Kiev, Ukraine, Faculty of Chemistry, Maria Curie-Sklodowska University, 20031 Lublin, Poland, and Kiev Taras Shevchenko University, 03030 Kiev, Ukraine Received March 26, 2003. In Final Form: July 23, 2003 Fumed silicas unmodified (SBET ) 378 m2/g) and modified (379-285 m2/g) by hexamethyldisilazane reacting with silanols (concentration of grafted trimethylsilyl groups CTMS ) 0.09-0.79 mmol/g) were studied by means of the NMR, IR, differential thermogravimetry, atomic force microscopy, microcalorimetry, adsorption, and theoretical methods. Variation in the surface composition and changes in the structural characteristics of primary and secondary particles lead to the nonlinear dependence of the Gibbs free energy of interfacial water, the heat of immersion of silicas in water, and the chemical shift δH(T) of water adsorbed from air on CTMS, despite a nearly linear decrease in water adsorption (at p/p0 ≈ 0.8 for 24-72 h) with CTMS. An increase in CTMS causes alterations in the structure of the hydrogen bond network in the interfacial water and other properties of water strongly (700-300 mg/g silica) and weakly (1400-500 mg/g) bound to the silica surfaces.

Introduction Chemical modification of silica by alkylsilanes or other organosilicon compounds (OSCs) is widely used to change the surface properties1-10 important in iteraction with polar and nonpolar substances.1,3,10,11 This modification (hydrophobization) results in the diminution of the surface * Corresponding author. E-mail: [email protected]. Fax: +38044-4243567. † Institute of Surface Chemistry. ‡ Maria Curie-Sklodowska University. § Kiev Taras Shevchenko University. (1) Vrancken, K. C.; De Coster, L.; Van Der Voort, P.; Grobet, P. J.; Vansant, E. F. J. Colloid Interface Sci. 1995, 170, 71. (b) Vansant, E. F.; Van Der Voort, P.; Vrancken, K. C. Characterization and Chemical Modification of the Silica Surface; Elsevier: Amsterdam, 1995. (2) Sindorf, D. W.; Maclel, G. E. J. Phys. Chem. 1982, 86, 5208. (3) Tertykh, V. A.; Belyakova, L. A. Chemical Reaction Involving Silica Surface; Naukova Dumka: Kiev, 1991. (4) Gun’ko, V. M.; Voronin, E. F.; Pakhlov, E. M.; Chuiko, A. A. Langmuir 1993, 9, 716. (5) Tripp, C. P.; Kazmaier, P.; Hair, M. L. Langmuir 1996, 12, 6407. (b) Tripp, C. P.; Kazmaier, P.; Hair, M. L. Langmuir 1996, 12, 6404. (6) Ericksonn, J. C.; Ljunggren, S.; Claesson, P. M. J. Chem. Soc., Faraday Trans. 2 1989, 85, 163. (7) Jesinowski, T.; Krysztalkewicz, A. Colloids Surf., A 2002, 207, 49. (8) Tsutsumi, K.; Takahashi, H. Colloid Polym. Sci. 1985, 263, 506. (b) Fuji, M.; Fujimori, H.; Takei, T.; Watanabe, T.; Chikazawa, M. J. Phys. Chem. B 1998, 102, 10498. Fuji, M.; Machida, K.; Takei, T.; Watanabe, T.; Chikazawa, M. Langmuir 2000, 16, 3281. (9) Blitz, J. P.; Gun’ko, V. M. In Encyclopedia of Surface and Colloid Science; Hubbard, A. T., Ed.; Marcel Dekker: New York, 2002; pp 29392950. (b) Gun’ko, V. M.; Vedamuthu, M. S.; Henderson, G. L.; Blitz, J. P. J. Colloid Interface Sci. 2000, 228, 157. (c) Gun’ko, V. M.; Sheeran, D. J.; Augustine, S. M.; Blitz, J. P. J. Colloid Interface Sci. 2002, 249, 123. (d) Gun’ko, V. M.; Zarko, V. I.; Sheeran, D. J.; Blitz, J. P.; Leboda, R.; Janusz, W.; Chibowski, S. J. Colloid Interface Sci. 2002, 252, 109. (e) Blitz, J. P., Little, Ch. B., Eds.; Fundamental and Applied Aspects of Chemically Modified Surfaces; Royal Society of Chemistry: Cambridge, 1999. (10) Unger, K. K. Porous Silica: Its Properties and Use in Column Liquid Chromatography; Elsevier: Amsterdam, 1995. (11) Jaroniec, M.; Madey, R. Physical adsorption on heterogeneous solids; Elsevier: Amsterdam, 1988. (b) Jaroniec, C. P.; Gilpin, R. K.; Jaroniec, M. J. Phys. Chem. B 1997, 101, 6861. (c) Kruk, M.; Jaroniec, M.; Gilpin, R. K.; Zhou, Y. W. Langmuir 1997, 13, 545.

affinity to water or other polar compounds, but for nonpolar organics, the opposite effect may be observed. However, the nitrogen (which is nonpolar) adsorption energy and adsorption potentials of the silica gel and fumed silica surfaces modified by different OSCs decrease with increasing amounts of grafted functionalities or growing of their CH chains.9,11 Hydrophobic properties of silylated glass particles appear (in the floating rate and the contact angle) at a surface coverage (Θ) between 0.5 and 0.6.8b The heat of immersion of silylated Aerosil A-380 in water significantly reduces from approximately 50 J/g at low Θ values to 10 J/g at high Θ values.8a Nevertheless, the infrared (IR) and 1H NMR spectroscopy data show that significant amounts of adsorbed water are observed on similar materials.12,13 It has been shown12 that water adsorbed on a surface of mesoporous glasses (modified by hexamethyldisilazane, HMDS) from air is in pores in the form of small clusters with an average number of molecules close to 2. These clusters are linked to residual silanols locating between trimethylsilyl (TMS) groups. Silylation of a surface of both porous (e.g., silica gel) and nonporous (e.g., fumed silica) silicas by different OSCs has been investigated intensively by means of IR and NMR spectroscopies and other methods showing changes in the reactivity of residual silanols and in the properties of the modified surface as a whole depending on the content of the grafted functionalities.1-12 Initially, the reaction of HMDS with silica results in a random distribution of wellseparated TMS groups.2 One can assume that a portion of the silanols remaining after silica modification by OSC in contact zones between adjacent primary particles or in narrow pores are difficult to access for OSC molecules; however, these surface patches can be easily accessible for water molecules with the size significantly smaller than that of OSC. Therefore, the 1H NMR spectroscopy (12) Takei, T.; Yamazaki, A.; Watanabe, T.; Chikazawa, M. J. Colloid Interface Sci. 1997, 188, 409. (b) Takei, T.; Chikazawa, M. J. Colloid Interface Sci. 1998, 208, 570. (13) Turov, V. V.; Mironiuk, I. F. Colloids Surf., A 1998, 134, 257.

10.1021/la0301238 CCC: $25.00 © 2003 American Chemical Society Published on Web 11/22/2003

Partial Hydrophobization of Fumed Silica

Langmuir, Vol. 19, No. 26, 2003 10817

Table 1. Structural Characteristics of Initial and Modified Fumed Silicas sample

CC, wt %

Ctms, mmol/g

A-380 SM1 SM2 SM3 SM4 SM5 SM6

0 0.33 0.52 0.44 0.83 1.52 2.84

0 0.09 0.14 0.12 0.23 0.42 0.79

DCH at 2964 cm-1

SBET, m2/g

S K, m2/g

SDS, m2/g

Vp, cm3/g

VBJHa, cm3/g

VDS, cm3/g

xDS, nm

DAJ

0.018 0.055 0.046 0.122 0.241 0.688

378 379 362 372 345 330 285

316 313 327 312 299 310 300

35 34 17 20 7 1 0

0.778 0.776 0.850 0.775 0.726 0.771 0.678

0.970 0.972 1.395 0.969 0.900 0.987 0.794

0.143 0.136 0.116 0.129 0.113 0.093 0.063

2.03 1.95 2.14 2.06 2.14 2.29 2.58

2.589 2.593 2.586 2.588 2.574 2.547 2.490

investigations show that a similar surface can disturb a relatively thick interfacial water layer (unfrozen at T < 273 K), and its thickness can exceed that of a similar layer at a hydrophilic surface.13,14 One of the possible reasons of the observed effects is polarization of the heterogeneous surface with different charging of the hydrophobic and hydrophilic patches13 and changes in the structure of the water layers near totally hydrophobic and hydrophilic surfaces.6 Besides, the nonuniform electrostatic field of the oxide surface can reorient dipoles of water molecules, which results in the enhancement of their rotational mobility and disordering of the hydrogen bond network in the interfacial water,6 especially near hydrophilic/hydrophobic surface patches. Therefore, the boundary water remains unfrozen at T < 273 K. In the case of totally hydrophobic surfaces, the number of hydrogen bonds between water molecules in the first interfacial layers becomes greater.6 Consequently, one can expect the essential dependence of the structure of the interfacial water on amounts of grafted hydrophobic TMS groups. The use of the 1H NMR spectroscopy in combination with freezing out of bulk water and level-to-level freezing out of the interfacial water allows one to determine the thickness of the water layers disturbed differently by the surface and frozen at different temperatures, changes in the Gibbs free energy of these layers, radial dependences of interaction forces between the surface and the aqueous medium, and amounts of strongly and weakly bound waters.13-18 (14) Leboda, R.; Turov, V. V.; Gun’ko, V. M.; Skubiszewska-Zie¸ ba, J. J. Colloid Interface Sci. 2001, 237, 120. (15) Turov, V. V. In Chemistry of Silica Surface; Chuiko, A. A., Ed.; ISC: Kiev, 2001; Vol. 1, pp 510-607. (16) Turov, V. V.; Leboda, R. Adv. Colloid Interface Sci. 1999, 79, 173. (17) Gun’ko, V. M.; Voronin, E. F.; Pakhlov, E. M.; Zarko, V. I.; Turov, V. V.; Guzenko, N. V.; Leboda, R.; Chibowski, E. Colloids Surf., A 2000, 166, 187. (18) Gun’ko, V. M.; Turov, V. V. Langmuir 1999, 15, 6405. (b) Gun’ko, V. M.; Zarko, V. I.; Turov, V. V.; Leboda, R.; Chibowski, E. Langmuir 1999, 15, 5694. (c) Turov, V. V.; Gun’ko, V. M.; Zarko, V. I.; Bogatyrov, V. M.; Dudnik, V. V.; Chuiko, A. A. Langmuir 1996, 12, 3503. (d) Gun’ko, V. M.; Turov, V. V.; Zarko, V. I.; Voronin, E. F.; Tischenko, V. A.; Dudnik, V. V.; Pakhlov, E. M.; Chuiko, A. A. Langmuir 1997, 13, 1529. (e) Gun’ko, V. M.; Leboda, R.; Skubiszewska-Zie¸ ba, J.; Turov, V. V.; Kowalczyk, P. Langmuir 2001, 17, 3148. (f) Gun’ko, V. M.; Turov, V. V.; Zarko, V. I.; Dudnik, V. V.; Tischenko, V. A.; Voronin, E. F.; Kazakova, O. A.; Silchenko, S. S.; Chuiko, A. A. J. Colloid Interface Sci. 1997, 192, 166. (g) Gun’ko, V. M.; Zarko, V. I.; Turov, V. V.; Leboda, R.; Chibowski, E.; Holysz, L.; Pakhlov, E. M.; Voronin, E. F.; Dudnik, V. V.; Gornikov, Y. I. J. Colloid Interface Sci. 1998, 198, 141. (h) Gun’ko, V. M.; Zarko, V. I.; Turov, V. V.; Leboda, R.; Chibowski, E.; Gun’ko, V. V. J. Colloid Interface Sci. 1998, 205, 106. (i) Gun’ko, V. M.; Zarko, V. I.; Turov, V. V.; Leboda, R.; Chibowski, E.; Pakhlov, E. M.; Goncharuk, E. V.; Marciniak, M.; Voronin, E. F.; Chuiko, A. A. J. Colloid Interface Sci. 1999, 220, 302. (j) Leboda, R.; Turov, V. V.; Charmas, B.; SkubiszewskaZie¸ ba, J.; Gun’ko, V. M. J. Colloid Interface Sci. 2000, 223, 112. (k) Gun’ko, V. M.; Leboda, R.; Turov, V. V.; Villie´ras, F.; SkubiszewskaZie¸ ba, J.; Chodorowski, S.; Marciniak, M. J. Colloid Interface Sci. 2001, 238, 340. (l) Turov, V. V.; Gun’ko, V. M.; Leboda, R.; Bandosz, T. J.; Skubiszewska-Zie¸ ba, J.; Palijczuk, D.; Tomaszewski, W.; Zie¸ tek, S. J. Colloid Interface Sci. 2002, 253, 23. (m) Leboda, R.; Turov, V. V.; Tomaszewski, W.; Gun’ko, V. M.; Skubiszewska-Zie¸ ba, J. Carbon 2002, 40, 389. (n) Gun’ko, V. M.; Leboda, R.; Turov, V. V.; Charmas, B.; Skubiszewska-Zie¸ ba, J. Appl. Surf., Sci. 2002, 191, 286.

Despite published results of the investigations of the surface properties of modified silicas and the corresponding interfacial water layers,1-19 many questions related to the relationships between the structural characteristics of modified oxides and the properties of bound water remain unclear. Therefore, the aim of this work was to study the relationships between the characteristics of the fumed silica surface partially modified by HMDS at different TMS loadings (CTMS) and the properties of the interfacial water layers in different media (air, liquid and frozen water, and weakly polar chloroform) at different temperatures by means of adsorption, desorption, spectroscopic, and theoretical methods. Materials and Techniques Materials. Controlled hydrophobization of fumed silica (A380, Pilot plant at the Institute of Surface Chemistry, Kalush, Ukraine) was carried out by HMDS (“Siliconpolymer”, Zaporozhye, Ukraine) reacting with silanols

2(tSiOH) + [(CH3)3Si]2NH f 2[tSiOSi(CH3)3] + NH3 Before reaction, HMDS was purified by distillation, and fumed silica was heated at 723 K for 2 h to remove adsorbed compounds. HMDS chemisorption occurred in a reactor (Figure 1) with IRspectroscopy monitoring of the synthesis. Before HMDS chemisorption, the silica powder and plate were heated in the reactor in air at 673-693 K for 1.5 h. Then, the sample was degassed for 10-15 min and cooled to ambient temperature. HMDS vapor (dosed in dependence on a desirable level of surface modification) interacted with the silica for 15-18 h (such a long time of adsorption was caused by the necessity to diminish the difference in the HMDS reaction with silica powder and pressed silica samples) at ambient temperature, and then the sample was heated and degassed at 473 K for 2 h. The carbon contents (CC) in modified samples SM1-SM6 (Table 1) were determined by chemical analysis. The CC values were used to estimate the amounts of grafted TMS groups (CTMS). IR Spectroscopy. IR-spectroscopy monitoring was realized at all modification stages (silica drying, HMDS adsorption, reaction, degassing, and water adsorption from air) by means of a Specord M-80 (Karl Zeiss, Jena) spectrophotometer. The concentration of TMS functionalities grafted on the silica surface was also estimated from the intensity of an IR band of CH3 groups at νCH ) 2964 cm-1 (Figure 2). This νCH band overlaps with a broad νOH band of adsorbed water and disturbed (residual) t SiOH groups that leads to diminution of the exactness of definition of the corresponding baseline. Comparison of the IR spectra of modified samples before (in a vacuum) and after (in air) hydration reveals (Figure 2) that the water adsorption results in the reduction of the intensity of the νCH band. Because both the IR (19) Lagerge, S.; Ken, E.; Partyka, S.; Lindheimer, M. J. Colloid Interface Sci. 2000, 227, 412. (b) Lagerge, S.; Rousset, P.; Zoungrana, T.; Douillard, J. M.; Partyka, S. Colloids Surf., A 1993, 80, 261. (c) Douillard, J. M. J. Colloid Interface Sci. 1997, 188, 511; 1996, 182, 308. (d) Douillard, J. M.; Zoungrana, T.; Partyka, S. Petrol. Sci. Eng. 1995, 14, 51. (e) Malandrini, H.; Sarraf, R.; Faucompre, B.; Partyka, S.; Douillard, J. M. Langmuir 1997, 13, 1337. (f) Partyka, S.; Rouquerol, F.; Rouquerol, J. J. Colloid Interface Sci. 1979, 21, 681. (g) Douillard, J. M. In Interfacial Dynamics; Kallay, N., Ed.; Surfactant Science Series; Marcel Dekker: New York, 1999; Vol. 88, pp 313-350.

10818

Langmuir, Vol. 19, No. 26, 2003

Gun’ko et al.

Figure 1. Sketch of the reactor used for the modification of fumed silica by HMDS under IR spectroscopy monitoring.

Figure 3. (a) TG and (b) DTG graphs for initial and modified silicas.

Figure 2. IR spectra of modified fumed silicas (a, c) hydrated in air, and (b) with no contact with air in the reactor; (d) DCH and CTMS versus carbon content CC.

Figure 4. Relationships between amounts of water adsorbed on initial and modified silicas for 24, 48, and 72 h and TMS loading.

absorption intensity and the baseline slope depend on the concentration of adsorbed water, a special procedure used to define the intensity of a given band for the unmodified sample (I0) was also applied to determine the optical density of the chosen band for modified samples as follows:

MOM, Budapest) DTG apparatus. Oxide samples (pressed at 180 MPa and milled) were placed (without additional treatment) in a corundum crucible at room temperature and heated to 1280 K in air at a controlled heating rate (10 K/min) (Figure 3). To control the adsorptive ability of modified silicas with respect to water, samples SM1-SM6 (120-250 mg, heated at 453 K for 4 h, initial silica was heated at 673 K for 2.5 h) were hydrated at a relative water vapor pressure p/p0 ≈ 0.8 for 24, 48, and 72 h before TG measurements (Figure 4). Microcalorimetry. A study of unmodified and modified samples was carried out by means of a DAC 1.1A (EPSE, Chernogolovka, Russia) differential automatic calorimeter. Before measurements of the heat of immersion (∆Him), samples (25-50 mg) were degassed at 393 K and 0.01 Pa for 2 h. Used amounts of samples were 50 mg/3 mL of distilled water exposed

DCH ) log(I0/I)

(1)

The DCH values (Table 1) were determined with consideration for the difference in the sample weights recalculated to 20 mg using experiments for several samples of 13-35 mg. Differential Thermogravimetry. Differential thermogravimetry (DTG) analysis and thermogravimetry (TG) were carried out by means of a Q-1500D (Paulik, Paulik & Erdey,


Langmuir, Vol. 19, No. 26, 2003 10819

Table 2. Characteristics of Interfacial Water Layer in Aqueous Suspensions of Initial and Modified Fumed Silica sample

-∆Gs, kJ/mol

-∆Gw, kJ/mol

Cuws, mg/g

Cuww, mg/g

Vuw, cm3/g

γv,a J/g

γS,b mJ/m2

A-380 SM1 SM2 SM3 SM4 SM5 SM6

2.3 3.0 3.3 2.6 2.5 3.0 2.7

0.4 0.4 0.4 0.4 0.35 0.7 0.4

700 600 500 450 400 400 300

1000 1400 1000 750 700 500 900

1.7 2.0 1.5 1.2 1.1 0.9 1.2

32.7 32.8 33.9 42.4 39.1 59.6 29.2

150 170 140 135 125 160 120

a γ denotes the total changes in the Gibbs free energy/g of v unfrozen water at T f 273 K. b γS is the total change in the Gibbs free energy per unit of surface area of oxide.

for several hours. The average errors of the ∆Him measurements were (3%. Similar investigations of different silicas and other materials were described in detail in the literature.12,19 1H NMR Spectroscopy. For recording the 1H NMR spectra of water bound to the silica surface in the gas or liquid media, a high-resolution WP-100 SY (Bruker) NMR spectrometer with a bandwidth of 50 kHz was used. Relative mean errors were (10% for the signal intensity and (1 K for the temperature. The amounts of interfacial unfrozen water (Cuw) in the suspensions of silica samples frozen at 200 < T < 273 K were estimated by comparison of an integral intensity (Iuw) of a 1H NMR signal of unfrozen water with that (Ic) of water adsorbed on silica powder from the gas phase using a calibrated function Ic ) f(Cc), assuming Cuw ) CcIuw/f(Cc). The signals of silanols and the water molecules from ice were not detected because of features of the measurement technique and the short time (∼10-6 s) of the cross-relaxation of protons in solids. Changes in the Gibbs free energy on water adsorption (γS in mJ/m2 or J/g) were calculated (with relative mean error (15%) using the known (tabulated) dependence of changes in the Gibbs free energy of ice [∆Gi(T)] on temperature.15-18 One can assume that water is frozen (T < 273 K) at the interfaces when G ) Gi; that is, ∆G ) G - G0 is equal to ∆Gi ) Gi(T) - Gi|T)273K, and it corresponds to a decrease in the Gibbs free energy of the interfacial water due to its interactions with the solid surfaces (G0 refers to the Gibbs free energy of the bulk undisturbed water). The 1H NMR technique applied with freezing-out of the bulk water was described in detail elsewhere.15-18 Using this approach, one can calculate the amounts of strongly (Cuws) and weakly (Cuww) bound unfrozen waters (because they are frozen at different temperatures), a maximal decrease in the Gibbs free energy of strongly (∆Gs) and weakly (∆Gw) bound waters, and the Gibbs free energy of water adsorption (γS) corresponding to the total changes in the Gibbs free energy of the interfacial water disturbed by the adsorbent at T f 273 K per m2 of the oxide surfaces

γS ) K

∫

Cuwmax

0

∆G dCuw

Figure 5. (a) Temperature dependence of amounts of unfrozen interfacial water (Cuw) in the aqueous suspensions of initial and modified silicas and (b) changes in the Gibbs free energy versus Cuw.

(2)

where K is a constant dependent on the units of γS and ∆G and Cuw is the total amount of unfrozen water at T f 273 K (Table 2, Figure 5). The γS can also be recalculated per gram of unfrozen water (γv). ∆G in eq 2 is equal numerically to the differential work of adhesion ∆G ) -Wa. Besides, γS is equal to the total work of adhesion and defines the integral change in the Gibbs free energy of the system, consisting of the adsorbent and a disperse medium, stipulated by the presence of the phase boundary. The magnitude of the adhesion force can be calculated from the ratio F ) ∆G/X, where X is the thickness of a bound water layer. Graphs of the radial dependences of the adhesion forces for initial and modified silicas are shown in Figure 6. The characteristics of the water layers bound to a surface of explored adsorbents are shown in Table 2 for aqueous 5 wt % suspensions. Adsorption. The specific surface area SBET calculated according to the standard BET method20,21 and the pore volume Vp (estimated at p/p0 ≈ 0.98, where p and p0 denote the equilibrium and saturation pressures of nitrogen, respectively), converting the adsorbed amount a0.98 (in cm3 STP of gaseous nitrogen/g of

Figure 6. Radial dependences of the adhesion force for initial and modified fumed silicas in the aqueous suspension. the adsorbent) to the liquid adsorbate Vp ≈ 0.001 546 8a0.98; Table 1), were calculated from the nitrogen adsorption-desorption isotherms (Figure 7a) recorded at 77.4 K by a Micromeritics ASAP 2405N adsorption analyzer. The pore volume at the pore radius Rp between 0.85 and 150 nm (VBJHa) was determined by the Barrett-Joyner-Halenda method applying the Micromeritics software to the nitrogen adsorption data. The specific surface area of mesopores (SK) was calculated using the Kiselev equation.22 The specific surface area (SDS), pore volume (VDS), and (20) Adamson, A. W.; Gast, A. P. Physical Chemistry of Surface; Wiley: New York, 1997. (21) Gregg, S. J.; Sing, K. S. W. Adsorption, Surface Area and Porosity; Academic Press: London, 1982.

10820


Gun’ko et al. and agglomerates), respectively; w ) 1 for slitlike pores and 2 for cylindrical pores; rk(p) is determined by the modified Kelvin equation

rk(p) )

σs wγνm cos θ + t(p, Rp) + 2 RgT ln(p0/p)

(4)

and t(p, Rp) can be computed using the modified BET equation

cz × (1 - z) 1 + (nb/2 - n/2)zn-1 - (nb + 1)zn + (nb/2 + n/2)zn+1

t(p, Rp) ) tm

1 + (c - 1)z + (cb/2 - c/2)zn - (cb/2 + c/2)zn+1

Figure 7. (a) Nitrogen adsorption-desorption isotherms for initial and modified fumed silicas. (b) Relative changes ∆Z/Z at Z ) SBET, SK, SDS, Vp, and VDS versus loading of TMS groups.

tm ) am/SBET; b ) exp(∆/RgT); ∆ is the excess of the evaporation heat as a result of the interference of the layering on the opposite particle surface or wall of pores (∆ ≈ 2.2 kJ/mol);25 t(p, Rp) is the statistical thickness of the adsorbed layer; am is the BET monolayer capacity; c ) cs exp[(Qp - Qs)/RgT]; cs is the BET coefficient for adsorption on the flat surface; Qs and Qp are the adsorption heats on the flat surface and in pores, respectively; z ) p/p0; n is the number (noninteger) of statistical monolayers of adsorbate molecules, and its maximal value for a given Rp (or pore half-width x) is equal to (Rp - σs/2)/tm; and σs is the collision diameter of surface atoms.25 Desorption data were utilized to compute the f(Rp) distributions with eq 3 and a modified regularization procedure26 under the nonnegativity condition [f(Rp) g at any Rp] at a fixed regularization parameter R ) 0.01 using a model of cylindrical pores at Rp > 0.7 nm and slitlike pores (gaps between adjacent primary particles) at 0.2 nm < Rp < 0.7 nm. This hybrid model of pores gives the PSDs close to that calculated for a model of gaps between spherical particles.23 The f(Rp) functions were renormalized in comparison with that calculated according to25,28,29

f(Rp) )

Figure 8. Size distributions of the gaps between primary particles in secondary structures (aggregates and agglomerates) for initial and modified fumed silicas. average radius (xDS) of the narrow pores were estimated using the Dubinin-Stoeckli (DS) equation.23 Calculation of the fractal dimension (Table 1, DAJ) was fulfilled using the Frenkel-HalseyHill equation24 applied to the nitrogen adsorption data at p/p0 e 0.85. This boundary pressure is close to the onset of the hysteresis loops of the isotherms for the studied adsorbents (Figure 7a). Pore size distributions [PSDs, f(Rp); Figure 8] were calculated using the overall isotherm equation25

a)

∫

rk(p)

rmin

f (Rp) dRp +

∫

rmax

rk(p)

w t(p, Rp) f (Rp) dRp (3) Rp - σs/2

where rmin and rmax are the minimal and maximal half-widths (or pore radii or gaps between primary particles in aggregates (22) Kiselev, A. V.; Dreving, V. P.; Runov, A. D. Dokl. Akad. Nauk Ukr. SSR 1945, 46, 310. (23) Dubinin, M. M.; Stoeckli, F. J. Colloid Interface Sci. 1980, 75, 34. (24) Avnir, D.; Jaroniec, M. Langmuir 1989, 5, 1431.

(5)

Rp fi(R)Vp

∫R

p fi(Rp)

(6)

dRp

to produce the PSDs close to being dependent on the pore volume increment on Rp. Notice that the PSDs calculated using the density functional theory (DFT) and the modified CONTIN procedure with eqs 3-6 (based on the combination of the modified Kelvin equation and the statistical adsorbed film thickness) are closely related over the 0.2-100-nm range for such adsorbents as fumed silica, silica gel, and active carbons.9,30 Additionally, this approach gives the PSDs for silicas close to those computed using nonlocal DFT (NLDFT).31 Notice that certain parameters for silica surface potentials of interaction with nitrogen and the corresponding equations were the same in NLDFT31 and on computation using eqs 3-6. (25) Nguyen, C.; Do, D. D. Langmuir 1999, 15, 3608. (26) Provencher, S. W. Comput. Phys. Commun. 1982, 27, 213; 229. (27) Gun’ko, V. M.; Zarko, V. I.; Voronin, E. F.; Turov, V. V.; Mironyuk, I. F.; Gerashchenko, I. I.; Goncharuk, E. V.; Pakhlov, E. M.; Guzenko, N. V.; Leboda, R.; Skubiszewska-Zie¸ ba, J.; Janusz, W.; Chibowski, S.; Levchuk, Y. N.; Klyueva, A. V. Langmuir 2002, 18, 581. (b) Gun’ko, V. M.; Voronin, E. F.; Mironyuk, I. F.; Leboda, R.; Skubiszewska-Zie¸ ba, J.; Pakhlov, E. M.; Guzenko, N. V.; Chuiko, A. A. Colloids Surf., A, in press. (c) Gun’ko, V. M.; Leboda, R.; Zarko, V. I.; Skubiszewska-Zie¸ ba, J.; Grzegorczyk, W.; Pakhlov, E. M.; Voronin, E. F.; Seledets, O.; Chibowski, E. Colloids Surf., A 2003, 218, 103. (d) Gun’ko, V. M.; Seledets, O.; Skubiszewska-Zie¸ ba, J.; Leboda, R.; Pasieczna, S. Colloids Surf., A, in press. (28) Nguyen, C.; Do, D. D. Langmuir 1999, 15, 3608. (b) Nguyen, C.; Do, D. D. Langmuir 2000, 16, 7218. (c) Do, D. D.; Nguyen, C.; Do, H. D. Colloids Surf., A 2001, 187-188, 51. (29) Gun’ko, V. M.; Do, D. D. Colloids Surf., A 2001, 193, 71. (30) Kowalczyk, P.; Gun’ko, V. M.; Terzyk, A. P.; Gauden, P. A.; Rong, H.; Ryu, Z.; Do, D. D. Appl. Surf. Sci. 2003, 206, 67. (b) Murphy, M. C.; Patel, S.; Phillips, G. J.; Davies, J. G.; Lloyd, A. W.; Gun’ko, V. M.; Mikhalovsky, S. V. In Studies in Surface Science and Catalysis; Rodriguez-Reinoso, F., McEnaney, B., Rouquerol, J., Unger, K., Eds.; Characterisation of Porous Solids VI; Elsevier Science: Amsterdam, 2002; Vol. 144, pp 515-520. (31) Ravikovitch, P. I.; Haller, G. L.; Neimark, A. V. Adv. Colloid Interface Sci. 1998, 76-77, 203.


Langmuir, Vol. 19, No. 26, 2003 10821

To compute the nitrogen adsorption energy distributions (AED), one can use the regularization procedure applied to the equation

Θ(T, p) )

∫

xmax

0

θl(T, p, x) f (x) dx

(7)

with the Fowler-Guggenheim equation as the kernel (local isotherm θl) describing the localized monolayer adsorption with lateral interaction as follows:

θl(p, E) )

Kp exp(zwΘ/kT) 1 + Kp exp(zwΘ/kT)

(8)

In this case, K ) K0(T) exp(E/kT) is the Langmuir constant for adsorption on monoenergetic sites, and the preexponential factor K0(T) is expressed in terms of the partition functions for an isolated gas and a surface phase. The nitrogen pressure is denoted by p, z ) 4 is the number of nearest neighbors of an adsorbate molecule, w is the interaction energy between a pair of nearest neighbors, k is the Boltzmann constant, and zw/k ) 380 K.11,32,33 A maximal p/p0 value on the f(E) computation corresponds to coverage Θ ) a/am ≈ 0.99, where a is the adsorption and am is the BET monolayer capacity. Cumulative pore volume distributions fCV(V) were calculated versus the pore size and the adsorption energy of nitrogen filling the corresponding pores

fCV(x) )

∫

x

xmin

f (Z) dZ

(9)

where Z denotes the pore size or the adsorption energy [f(E) is normalized by Vp as well as f(Rp)]. One can assume that a higher adsorption energy corresponds to adsorption in narrower pores.20,21 This assumption allows one to define a connection between f(E) and f(Rp) and to calculate the cumulative PSD as a function of E. A similar procedure was also applied to ∆G(Cuw) and f(Rp), assuming that water filling narrower pores is frozen at lower temperatures and the specific density of water unfrozen in pores corresponds to that of free fluid water at 273 K. These assumptions allow one to calculate dependences of ∆G on Rp or Vp for pores filled by water unfrozen at different temperatures. Quantum Chemical Calculations. Silica clusters with four or eight SiO4/2 tetrahedrons and four or eight tSiOH groups for the unmodified “surface” (Figure 9) were calculated by using the GAMESS34 (Version 6.2) and Gaussian 9435 program packages. The magnitudes of the NMR shielding σiso (or chemical shift δH,iso referenced to tetramethylsilane calculated with the same basis set) were computed with the gauge-independent atomic orbital (GIAO)36 methods using the Hartree-Fock theory and DFT (DFT was used with the combined B3LYP exchange and correlation functional) methods35 with the basis set 6-31G(d,p) (6-31G(d,p) was applied to optimize the geometry and B3LYP/6-31G(d,p) was used to calculate σiso for clusters with this geometry). The GIAO method is described in detail elsewhere36,37 provides gauge(32) Hill, T. L. Introduction to Statistical Thermodynamics; AddisonWesley: Reading, MA, 1960. (33) Choma, J.; Jaroniec, M. Langmuir 1997, 13, 1026. (34) Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S. T.; Gordon, M. S.; Jensen, J. J.; Koseki, S.; Matsunaga, N.; Nguyen, K. A.; Su, S.; Windus, T. L.; Dupuis, M.; Montgomery, J. A. J. Comput. Chem. 1993, 14, 1347. (35) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.; Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Petersson, G. A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski, V. G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski, J.; Stefanov, B. B.; Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala, P. Y.; Chen, W.; Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. P.; Head-Gordon, M.; Gonzalez, C.; Pople, J. A. Gaussian 94, Revision E.1; Gaussian, Inc.: Pittsburgh, PA, 1995. (36) Wolinski, K.; Hilton, J. F.; Pulay, P. J. Am. Chem. Soc. 1990, 112, 8251. (37) Hinton, J. F.; Wolinski, K. In Theoretical Treatments of Hydrogen Bonding; Hadzi, D., Ed.; John Wiley & Sons: New York, 1997; pp 7593. (b) Sierka, M.; Eichler, U.; Datka, J.; Sauer, J. J. Phys. Chem. B 1992, 102, 6397.

Figure 9. Chemical shift δH,iso for adsorbed water molecules and silanols calculated by the GIAO method with BLYP3/631G(d,p)//6-31G(d,p).

Figure 10. Relationship between the free energy of solvation ∆Gs [calculated using the SM5.42R method with 6-31G(d) and B3LYP/6-31G(d) or IEFPCM with B3LYP/6-31G(d,p)] and substitution degree for tSiOH to tSiCH3. invariant results for the magnetic shielding tensor. The free energy of solvation ∆Gs (Figure 10) and structural and electronic parameters of the silica clusters with varied numbers of tSiOH and tSiCH3 groups were calculated using a solvation model SM5.42R/HF/6-31G(d) and SM5.42R/B3LYP/6-31G(d) by means of the GAMESOL (Version 3.1, modified by the addition of the DFT method) program package.38 Additionally, the free energy of solvation was calculated using the IEFPCM method39 with the basis set B3LYP/6-31G(d,p) with the geometry calculated by the HF/6-31G(d) method. It should be noted that similar cluster models are frequently used on ab initio calculations of interaction (38) Xidos, J. D.; Li, J.; Zhu, T.; Hawkins, G. D.; Thompson, J. D.; Chuang, Y.-Y.; Fast, P. L.; Liotard, D. A.; Rinaldi, D.; Cramer, C. J.; Truhlar, D. G.. GAMESOL, Version 3.1; University of Minnesota: Minneapolis, 2002, based on the General Atomic and Molecular Electronic Structure System (GAMESS) as described in ref 33. (b) Cramer, C. J.; Truhlar, D. G. Chem. Rev. 1999, 99, 2161. (39) Cances, M. T.; Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 107, 3032. (b) Cossi, M.; Barone, V.; Mennucci, B.; Tomasi, J. Chem. Phys. Lett. 1998, 286, 253.

10822


Gun’ko et al.

between the adsorbate (one or a few molecules) with a solid surface18,40 as well as for consideration of the solvation effects,18,38 which give quite appropriate results for many energetic and structural parameters.

Results and Discussion An increase in the concentration of TMS groups (Table 1, CTMS, CC, and DCH, and Figure 2) leads to changes in the interaction with water (Figures 3-6). These changes are clearly seen in the TG and DTG curves at T lower than 473 K (Figure 3). The first maximum observed on the DTG curves (Figure 3) corresponds to the desorption of water at T < 520 K, but the second one at 650-720 K (marked only for SM4-SM6) is linked to the destruction of TMS groups. An increase in CTMS for SM5 and SM6 leads to diminution of the first maximum because surface modification reduces the amounts of silanols, which are adsorption centers for water. For SM1-SM4, diminution of the amounts of water adsorbed in air is not observed (Figure 3). However, in the case of water adsorption on additionally heated samples at a fixed water vapor pressure p/p0 ) 0.8 for 24-72 h, amounts of adsorbed water depend stronger (nonlinearly for SM1-SM3) on CTMS (Figure 4). These changes can be connected not only with increased hydrophobicity but also with changes in the spatial structure of the silica surface. An increase in CTMS leads to reduction of the specific surface area (SBET). The contribution of micropores (SDS) depends on a CTMS stronger than that for mesopores (SK) as well as a decrease in VDS that exceeds that for Vp (Figure 7b). Notice that small relative changes in SK and Vp versus CTMS are closely related in contrast to greater alterations of SBET, SDS, and VDS (Figure 7b). These results are connected to features of the morphology of nonporous primary and “porous” secondary particles of fumed silica.41 The contact zone (narrow gaps) between adjacent primary particles in aggregates [with the bulk density Fa,b close to 30% of the true density (F0) of silica] becomes less accessible for nitrogen molecules (or other adsorbates, e.g., water) after deposition of TMS groups, which are significantly larger than substituted H atoms in tSiOH groups. Relative changes in the pore volume Vp are lower than that for SBET (Figure 7b) because Vp corresponds only to a minor portion of the empty volume (Vem) of fumed silica powder (because the bulk density of the powder is Fb ) 0.03-0.05 g/cm3 and Vem ) 1/Fb - 1/F0), and chemical modification leads not only to a reduction of the accessible surface area but also to a rearrangement of secondary particles (aggregates and agglomerates).17 This is also seen (40) Pacchioni, G. Solid State Sci. 2002, 2, 161. (b) Sahai, N.; Tossell, J. A. Geochim. Cosmochim. Acta 2001, 65, 2043. (c) Tossell, J. A.; Sahai, N. Geochim. Cosmochim. Acta 2000, 64, 4097. (d) de Dios, A. C. Prog. Nucl. Magn. Spectrosc. 1996, 29, 229. (e) Civalleri, B.; Garrone, E.; Ugliengo, P. Chem. Phys. Lett. 1999, 299, 443. (f) Civalleri, B.; Garrone, E.; Ugliengo, P. Chem. Phys. Lett. 1998, 294, 103. (g) Roggero, I.; Civalleri, B.; Ugliengo, P. Chem. Phys. Lett. 2001, 341, 625. (h) Ferrari, A. M.; Garrone, E.; Spoto, G.; Ugliengo, P.; Zecchina, A. Surf. Sci. 1995, 323, 151. (i) Zhou, M.; Zang, L.; Lu, H.; Shao, L.; Chen, M. J. Mol. Struct. 2002, 605, 249. (j) Soscun, H.; Castellano, O.; Hernandez, J. J. Mol. Struct. (THEOCHEM) 2000, 531, 315. (k) Murashov, V. J. Mol. Struct. 2003, 650, 141. (41) Basic Characteristics of Aerosil, Technical Bulletin Pigments, N11; Degussa AG: Hanau, 1997. (b) The Surface Properties of Silicas, Legrand, A. P., Ed.; Wiley: New York, 1998. (c) Iler, R. K. The Chemistry of Silica, Wiley: Chichester, 1979. (d) Gun’ko, V. M.; Mironyuk, I. F.; Zarko, V. I.; Turov, V. V.; Voronin, E. F.; Pakhlov, E. M.; Goncharuk, E. V.; Leboda, R.; Skubiszewska-Zie¸ ba, J.; Janusz, W.; Chibowski, S.; Levchuk, Y. N.; Klyueva, A. V. J. Colloid Interface Sci. 2001, 242, 90. (e) Proceedings of International Conference on Silica Science and Technology “Silica 98”, Mulhouse, France, Sept 1-4, 1998. (f) Proceedings of International Conference on Silica Science and Technology “Silica 2001”, Mulhouse, France, Sept 3-6, 2001.

Figure 11. Cumulative pore volume distributions versus (a) nitrogen adsorption energy and (b) pore size and (c) nitrogen AEDs for the initial silica and SM6.

in the f(Rp) graphs (Figure 8). The contribution of “micropores” (i.e., narrow gaps between adjacent primary particles) at Rp < 1 nm decreases (see also ∆SDS/SDS and ∆VDS/VDS versus CTMS), but the contribution of narrow “mesopores” at Rp < 10 nm (as well as larger pores) increases for some samples. Only a minor portion of the large gaps between primary particles in secondary ones can be entirely filled by nitrogen even at p/p0 f 1 (Figures 7a, 8, and 11) because Vp , Vem. However, the filling of large pores gives a marked contribution to total adsorption because the volume VBJHa determined for pores at Rp over the 0.85-150-nm range


Figure 12. Comparison of the heat of immersion of silicas in liquid water at room temperature and changes in the Gibbs free energy of interfacial water (in J/g of oxide) at T f 273 K.

is significantly larger than Vp. This difference is the largest for SM2, as VBJHa - Vp ) 0.545 cm3/g (Table 1). SM2 is also characterized by maximal alterations in f(Rp), and a shoulder (absent for other samples) is observed at Rp ≈ 55 nm (Figure 8). Silylation of silica leads to a reduction of the nitrogen adsorption potential because a high-energy AED peak at 11-12 kJ/mol decreases (Figure 11). The contribution of the first monolayer (providing the second, at 7-8 kJ/mol, and third, at 11-12 kJ/mol, AED peaks due to adsorption on different sites placed in broad and narrow pores) to the adsorbed amounts of nitrogen is less 10%. Notice that, according to calculations of adsorption complexes by using B3LYP/6-31G(d,p), the difference in the interaction energy of a nitrogen molecule with silanol and siloxane bonds is equal to approximately 3 kJ/mol, which can affect the AED for the first monolayer. A cumulative volume distribution versus the adsorption energy fCV(E) (Figure 11a) reveals a significant contribution of adsorbed nitrogen with low-energetic binding to the surface (because of the low-energetic nonspecific dispersive interaction on the secondary volume filling of mesopores) to its total adsorbed amounts. This shows a rather “mesoporous” character (Figure 8) of the secondary particles of fumed silica. Therefore, the differences in the AEDs or PSDs for the initial silica and maximum modified sample SM6 are relatively small; however, a high-energy AED peak is significantly lower for SM6. A tail of the high-energy AED peak is connected with nitrogen adsorption in narrow gaps between adjacent primary particles [first f(Rp) peak at Rp < 1 nm). Thus, on the basis of analysis of the results of nitrogen adsorption, one can assume that water interaction with SM6 should be weaker than that for the pristine silica surface. However, the difference in the binding of water molecules to the initial and modified silicas may be greater than that for nitrogen molecules because of the interaction of water with surface silanols by the hydrogen bonds and with tSiOSi(CH3)3 by weaker dispersive forces. Results obtained on water desorption (Figure 3) are confirmed by the microcalorimetric and 1H NMR data (Figure 12), showing nonlinear changes in the heat of immersion and the Gibbs free energy of the interfacial water with increasing CTMS and with their significant diminution observed only for SM6. One can assume that interaction of partially hydrophobic silicas with water in the aqueous suspensions (in contrast to interaction with water adsorbed from air and desorbed upon TG measurements or upon interaction with nonpolar nitrogen) can lead to the significant rearrange-

Langmuir, Vol. 19, No. 26, 2003 10823

Figure 13. Particle size distributions in a sonicated aqueous suspension of initial fumed silica at a concentration of 3 wt % with respect to the intensity of the scattered light, particle volume, and particle number measured by the photon correlation spectroscopy method.

ment of the secondary particles of fumed silica.41,42 These particles are instable in the aqueous suspension and can be decomposed upon different treatments (mixing, stirring, ball milling, or sonication).41,42 For instance, a major portion of the particles observed in the sonicated aqueous suspension of initial fumed silica (3 wt %) correspond to individual primary particles (Figure 13, particle number curve). Notice that the hydrodynamic diameter of the particles estimated from the photon correlation spectroscopy measurements (Figure 13) is larger than the geometrical diameter of these particles only by 10-15%.42 Hydrophobic patches on primary particles formed upon chemical modification of their surfaces by HMDS can cause rearrangement of secondary particles (aggregates and agglomerates) in the suspension to provide maximal shielding of patches of hydrophobic TMS groups from fluid water that can also lead to an increase in the size of the secondary particles.9,42 Similar effects were observed for blends of hydrophilic and hydrophobic fumed silicas.13,43 Therefore, the interaction energy between unmodified silica patches (which have been inaccessible, between adjacent primary particles, for HMDS molecules upon surface modification but are accessible for water molecules because of their smaller size and the mentioned rearrangement of secondary particles in the suspension) and water can be great and gives a significant contribution to ∆Him or γS for the modified samples (Figure 12 and Table 2). Notice that the shape of the ∆Him(CTMS) graphs differ from that described previously.8a The magnitudes of the Gibbs free energy of the interfacial water γS (calculated from the data shown in Table 2) per a mass unit of adsorbent and the values of the heat of immersion ∆Him measured for the same samples by the microcalorimetry method show consistent changes with CTMS for practically all samples except SM6 because of a greater contribution of changes in enthalpy as opposed to entropy on water interaction with these solids. A small ∆Him value for SM6 is caused by its poor wetting on calorimetric measurements, but in the NMR experiments, it has been suspended using alcohol/water followed by decanting. Notice that the maximal γS values were typically observed for complex (42) Gun’ko, V. M.; Zarko, V. I.; Leboda, R.; Chibowski, E. Adv. Colloid Interface Sci. 2001, 91, 1. (43) Mironyuk, I. F.; Gun’ko, V. M.; Zarko, V. I. Dopov. Nats. Akad. Nauk. Ukr. 1999, No. 3, 149.

10824


adsorbents with a low amount (0.5-3 wt %) of the second oxide or pyrocarbon grafted on the oxide matrix.18 Consequently, a maximum of the ∆Him and γS values for SM1 corresponds to a general regularity observed for the interfacial water at the surfaces of heterogeneous adsorbents, and one can assume that these materials possess a maximal surface nonuniformity at low amounts of grafted (second) material. To elucidate changes in the properties of the interfacial water versus CTMS, we have calculated the ∆Gs values for the silica cluster with varied amounts of OH and CH3 groups (nOH + mCH3 ) 4 at nOH and mCH3 varied from 4 to 0) using different methods (Figure 10). There is a difference in -∆Gs (Figure 10) and ∆Him or γS (Table 2 and Figure 12) that may be caused by too great a ratio between the “surface” and the “volume” of the used cluster in comparison with primary particles because their average diameter for A-380 is approximately 7.2 nm, and the first water monolayer (which gives the lion’s share to ∆Him)19 corresponds to approximately 12 wt % silica. Consequently, ∆Him (or γS) can be lower by 1 order of magnitude than -∆Gs calculated for the small cluster having only “surface” atoms. With consideration for this difference, the -∆Gs value is in agreement with the ∆Him and γS values (recalculated to kJ/mol of the interfacial water). Calculations of ∆Gs with [SM5.42R/B3LYP/6-31G(d) and IEFPCM/B3LYP/6-31G(d,p)] or without [SM5.42R/HF/6-31G(d)] consideration for the electron correlation effects show that ∆Gs is a linear function of the substitution degree (calculated as mCH3/4; Figure 10). Consideration for the electron correlation effects leads to a slightly steeper lowering of -∆Gs with an increasing concentration of CH3 groups. These results are also in agreement with the experimental data.8a Thus, for the same texture of an adsorbent, one can anticipate a linear decrease in the amounts of disturbed adsorbed (bound) water (unfrozen at T < 273 K) with an increasing CTMS value. Deviation from a similar linear dependence (e.g., nonlinear dependence of amounts of adsorbed water on CTMS < 0.2 mmol/g observed in Figure 4) can be caused by the textural nonuniformity of modified silicas, which can be characterized by, for example, the fractal dimension DAJ (as well as other structural characteristics) dependent nonlinearly on CTMS (Table 1, Figure 7b). Both the long-range [corresponding to the small slope of F(X) curves at X > 1 nm] and the short-range (greater slope at X < 1 nm) components of the surface forces appear in the aqueous suspensions (5 wt %) of all samples, which is clearly seen from the obtained results (Figures 5, 6, and 14 and Table 2). The average thickness of a strongly bound water layer is 0.8-1.2 nm (corresponding to the shortrange forces), and the total thickness of the disturbed water layer can exceed 3 nm (Figures 6 and 14). The amounts of strongly bound water diminish with CTMS, while for weakly bound water, such a correlation is not revealed (Table 2) because of rearrangement of secondary particles and the nonuniformity of partially modified surfaces. The difference in the dependence of F on X (Figure 6) and ∆G on Rp (Figure 14) is caused by a difference in the models of interfacial water layers used on the calculations of F(X) and ∆G(Rp). F(X) corresponds to the pure radial dependence for the spherical water layers around individual primary particles, but ∆G(Rp) corresponds to the structure of gaps and their volume between primary particles packed in aggregates and agglomerates, and the interfacial water fills these gaps by nonspherical layers. A significant portion of the interfacial water unfrozen at T f 273 K (Table 2, Cuw ) Cuws + Cuww, Vuw, γS, and ∆Him are maximal for SM1) fills the “pore” volume

Gun’ko et al.

Figure 14. Relationships between changes in the Gibbs free energy of interfacial water layers and the (a) pore volume and (b) pore radius.

in aggregates (which is close to the empty inner volume in the aggregates plus the volume of the adsorbate monolayer on the outer surface of these aggregates); however, Vp is smaller than the volume of unfrozen water Vuw (but VBJHa is close to Vuw). This difference is caused by a stronger interaction of the silica surface with polar water than nonpolar nitrogen molecules. Consequently, a portion of the disturbed water partially fills the gaps in the agglomerates because the total empty volume in the agglomerates is greater than 10 cm3/g because of low their bulk density of 0.06-0.07 g/cm3, which is close to that of fumed silica powder (0.03-0.05 g/cm3) than to that of aggregates (Fa,b ) 0.6-0.7 g/cm3).41 Hence, the unfrozen interfacial water (with different thicknesses) is nonuniformly distributed in narrow mesopores in aggregates due to confinement of their size and at the outer surfaces of these aggregates, where the opposite “pore walls” are too distant. Despite diminution of the concentration of silanols (main surface centers for water adsorption)41 for modified silicas by several times (e.g., if a maximal concentration of tSiOH groups on a fumed silica surface is approximately 2.5 µmol/m2,1,3,41 i.e., 9.4 mmol/g, for SM6, the residual amount of tSiOH groups is approximately 16% of their initial content, and the IR band νOH at 3750 cm-1 is not yet observed; Figure 2c), changes in the Gibbs free energy of the interfacial water layer in the aqueous medium are less 20%. This result is in agreement with previous data obtained for aqueous suspensions of modified fumed silicas.17,18 A great nonuniformity of partially modified surfaces causes disorder of several monolayers of interfacial water (Figures 6 and 14) and enhances the rotational mobility of water molecules that leads to reduction of the freezing temperature (Tfr) of bound water.


Langmuir, Vol. 19, No. 26, 2003 10825

Figure 15. The 1H NMR spectra of water adsorbed on surfaces of modified silicas (a) SM4, (b) SM5, and (c) SM6 at different temperatures (1H signals of CHCl3 and tetramethylsilane are shown).

Notice that an increase in the electrical conductivity for an alternating current (increase in bound charges, e.g., dipoles of molecules) and its decrease for a constant current (mobile charges, e.g., protons and other ions) were observed for the suspensions of a blend of hydrophilic and hydrophobic (silylated) fumed silicas.42 Consequently, changes in polarization of interfacial water (enhancement of bound charges) can affect its free energy because of the influence of the electrostatic field of the surface on the rotational mobility of the water molecules (i.e., bound charges dipoles of molecules) and the hydrogen bond network as a whole. In the case of contact of a hydrated silica surface with air (water vapor), the structure of adsorbed water is determined not only by the adsorbent surface (i.e., phase boundary of silica/water) but also by the phase boundary of water/air. The appearance of these phase boundaries leads to a reduction of the free energy of the interfacial water accompanied by a lowering of its freezing temperature.15-18,44 The information on a structure of the adsorption complexes at a surface of oxide adsorbents can be obtained from temperature dependences of the chemical shift of protons (δH) of interfacial water molecules. It is necessary to take into account that δH is defined by the strength of the hydrogen bonds between the water molecule and the active surface sites and depends on the amounts of hydrogen bonds per water molecule.18 The chemical shift of protons in water molecules that are not participating in hydrogen binding is 1.5-1.7 ppm.45 If water molecules form icelike polyassociates in which the volume of each molecule forms four hydrogen bonds (two of them are formed by hydrogen atoms and the two remaining are formed with the participation of the lone electron pair of the oxygen atom), δH ) 7 ppm.46 Besides, (44) Mironyuk, I. F.; Gun’ko, V. M.; Turov, V. V.; Zarko, V. I.; Leboda, R.; Skubiszewska-Zie¸ ba, J. Colloids Surf., A 2001, 180, 87. (45) Pople, J. A.; Schneider, W. G.; Bernstein, H. J. High-Resolution Nuclear Magnetic Resonance; McGraw-Hill: New York, 1959.

it is necessary to take into consideration that the binding of H2O‚‚‚H-OSit influences the δH of the water molecule weaker than that for H‚‚‚OX. The formation of the later hydrogen bonds enhance the δH value by 2.7 ppm. The 1H NMR spectra of the hydrated powders of SM4-SM6 (in chloroform) are shown at different temperatures (Figure 15). For SM1-SM3 (which are not shown in Figure 15), the δH(T) graphs are akin to that for SM4. Because chloroform is a poor proton donor, its molecules are not capable of competing with water molecules upon interaction with residual silanols on modified samples to form the complexes tSiO(H)‚‚‚H-X. Therefore, one can assume that chloroform weakly affects the structure of the water clusters adsorbed on the hydrophilic surfaces. On the other hand, chloroform hampers the proton exchanging between the water molecules interacting with different active surface sites and allows us to specify the δH values in adsorbed water.15,16,44 In the case of partially hydrophobic surfaces, chloroform can promote the formation of larger water droplets weakly interacting with hydrophobic groups but strongly interacting with the remaining hydrophilic patches.18 This circumstance causes enhancement of contribution of the 1H NMR signal intensity corresponding to fluid water at δH ≈ 5 ppm (Figure 15) with increasing CTMS; i.e., the size of the water droplets located between aggregates in agglomerates (observed in the atomic force microscopy, AFM, image of the initial fumed silica shown in Figure 16) increases with CTMS. For samples SM1-SM4, adsorbed water gives the signal with a minimal width at T ) 240 K. The signal intensity decreases at low temperatures because of layer-by-layer freezing out of the interfacial water. An increase of the signal width with lowering temperature is caused by reduction of the molecular motility of adsorbed water,47 (46) Kinney, D. R.; Chaung, I.-S.; Maciel, G. E. J. Am. Chem. Soc. 1993, 115, 6786.

10826


Figure 16. AFM image (2 × 3 µm, height 500 nm, obtained by means of NanoScope III apparatus, Digital Instruments, U.S.A.) of the initial silica demonstrating aggregates (100200 nm) of primary particles and agglomerates (>1 µm) of aggregates.

and the high-temperature effect is due to the acceleration of the proton exchanging between water molecules and surface hydroxyls.48 The chemical shift δH of adsorbed water is observed over the 3-4.5 ppm range (Figure 15), and δH decreases with elevating temperature [δH(T) graphs for water adsorbed on SM1-SM5 are shown in Figure 17 at different hydrations at CH2O between 50 and 300 mg/g of oxide]. Additionally, the signals of H atoms from chloroform CHCl3 are observed at δH ) 7.26 ppm, CH3 groups of tetramethylsilane (used as a standard) correspond to δH ) 0 ppm, and the low-intensity signal of water dissolved in chloroform is observed at δH ) 1.7 ppm. The main signal of adsorbed water splits with increasing CTMS (Figure 15), and a narrow line at 5 ppm corresponding to fluid water becomes intensive. Practically only this signal is observed for water adsorbed on SM6 (Figure 15c). This can also correspond to an increase in the average number of hydrogen bonds per molecule in the first interfacial water layers near the hydrophobic surfaces.6 Additionally, the intensity of an IR band at 3660 cm-1 (corresponding to silanols in sites difficult to access even for water molecules)41 increases with CTMS (Figure 2c). Most probably for modified fumed silica, adsorbed water forms relatively large nanoscaled droplets in loosely packed agglomerates between aggregates (see the AFM image of the initial fumed silica in Figure 16) of primary particles as a result of capillary condensation with weak contacts with remained silanols. There are certain regularities in the δH(T) graphs at T > Tfr of adsorbed water, as δH decreases with elevating temperature and the diminution of the water concentration CH2O in samples (Figure 17). A minimal value δH ) 1.8 ppm is observed for SM4 at CH2O ) 52 mg/g and T ) 330 K (Figure 17d). A maximal value δH ≈ 4.8 ppm is observed at a high content of adsorbed water and a temperature close to 270 K, and at lower temperatures, the δH value decreases because of the freezing of water. (47) Mank, V. V.; Lebovka, N. I. Nuclear Magnetic Resonance Spectroscopy in Heterogeneous Systems; Naukova Dumka: Kiev, 1988. (48) Turov, V. V.; Zarko, V. I.; Chuiko, A. A. Ukr. Khim. Zh. 1995, 69, 1262.

Gun’ko et al.

A linear portion of the δH(T) plots can be found for all the samples at any CH2O. Prolongation of these lines to lower temperatures gives their crossing at 115 K at a point corresponding to δH ) 7 ppm characteristic for molecules possessing four hydrogen bonds, that is, in the icelike structure. It is possible that this temperature corresponds to an ordered structure of adsorbed water akin to that of free ice, which, however, is not observed in our experiments because of features of the used measurement technique. A small δH value at T close to 320 K testifies to changes in the structure of water clusters adsorbed on the surface of modified silica in comparison with that observed at lower temperatures in consequence of the diminution of the average number of hydrogen bonds per molecule with elevating temperature and because of the enlargement of the hydrogen bonds with T, which results in the reduction of δH.18,27,41 A similar picture was observed earlier for the initial silica.18,44 So, small δH values for water molecules taking part in the hydrogen bonds with the surface hydroxyls can be also explained by their participation in complexes tSiO-H‚‚‚OH2 as an electron donor; that is, a marked portion of protons of adsorbed water molecules (forming small clusters with only several molecules) do not participate in the hydrogen bonds. Additionally, water interaction with siloxane bonds (tSi)2O‚‚‚H-OH results in lower δH values than that for tSiO(H)‚‚‚H-OH (Figure 9). Therefore, small water clusters located under “umbrellas” of TMS groups near siloxane bonds can be characterized by relatively small δH values. Calculations of two types of the adsorbed water clusters around a TMS group (Figure 9c) and “under” this group (Figure 9d) give lower potential energy for the second structure (∆Etotal ) -61 kJ/mol), which indicates a greater probability of the formation of similar complexes. Localization of a water droplet near a siloxane bond results in the diminution of δH (Figure 9d). Notice that the δH value for H from silanols participating in the hydrogen bonds tSiO-H‚‚‚OH2 is greater (>7 ppm) than that for water molecules HO-H‚‚‚OH2 (4-5 ppm). Similar δH values of about 7 ppm were also observed in the NMR spectra of silica.49 Thus, the “downtake” of the water droplets to the siloxane bonds under “umbrellas” of TMS groups (part d akin to part c of Figure 9) may lead to the diminution of the δH values in comparison with that for the initial silica because of a decrease in the number of water molecules in these clusters in confined spaces between and under TMS groups, diminution of the polarization influence of modified surface on the electronic structure of adsorbed molecules, and low ability of the siloxane bonds to form the hydrogen bonds.18,41 A similar effect of the δH diminution is observed in the experimental 1H NMR spectra of modified silicas (Figures 15 and 17). It should be noted that the δH dependence on a charge value of H atoms (qH) is nonlinear (Figure 18) because δH depends also on the surroundings of the H atoms,18 and δH(qH) for the water molecules (δH only for H in the hydrogen bonds) differs significantly from that for the tSiOH groups (free and bonded to water molecules; Figure 18). One can assume that water adsorbed in large pores of hydrophobic materials or in large gaps between primary particles in secondary ones tend to form larger droplets in larger pores to reduce the surface area of contacts between water and hydrophobic surfaces. Therefore, partially or totally hydrophobic and porous materials with large pores possess a greater ice-forming ability in comparison with microporous hy(49) Doremieux-Morin, C.; Heeribout, L.; Dumousseaux, C.; Fraissard, J.; Hommel, H.; Legrand, A. P. J. Am. Chem. Soc. 1996, 118, 13040.


Langmuir, Vol. 19, No. 26, 2003 10827

Figure 17. Temperature dependences of δH for (a) SM1, (b) SM2, (c) SM3, (d) SM4, and (e) SM5 at different amounts of adsorbed water and (f) different modified samples.

drophobic materials. This circumstance can be important for adsorption processes occurring (in the atmosphere) on industrial nanoscaled aerosol particles with the participation of water. Conclusions Fumed silicas partially modified by HMDS and containing different amounts of TMS groups (from 0.09 to 0.79 mmol/g) are characterized by the nonlinear dependences of changes in the structural, adsorptive, and other

parameters such as the specific surface area, the pore volume (total and micro- and mesopous), the PSDs, the concentrations of weakly and strongly bound water, the Gibbs free energy of these waters, the heat of immersion in liquid water, and the chemical shift δH(T) of water adsorbed from air. Upon water adsorption on modified silicas from air, the size of the water droplets (located between aggregates of primary particles in agglomerates) increases with amounts of grafted TMS groups, and for SM6 maximum hydrophobic, the main 1H NMR signal of adsorbed water becomes close to that for fluid water.

10828


Gun’ko et al.

water and other adsorbates) occurs with increasing CTMS. The formation of small water clusters between grafted TMS groups causes the appearance of the 1H NMR signal with the low chemical shift of 1H (in adsorbed water molecules) because of the formation of the (tSi)2O‚‚‚HOH complexes (characterized by smaller δH in comparison with that for HO-H‚‚‚OH2 or tSiO(H)‚‚‚H-OH) and as a result of a smaller number of hydrogen bonds per water molecule in the small water clusters in comparison with that in larger water droplets on maximum hydrophobic silica.

Figure 18. Relationship between the chemical shift δH,iso (calculated by the GIAO method) and charge of the H atoms [BLYP3/6-31G(d,p)//6-31G(d,p)].

Acknowledgment. This research was supported by STCU (Grant 1946). R.L. is grateful to the Foundation for Polish Science for financial support. V.M.G. is grateful to Dr. T. L. Petrenko for the use of the Gaussian 94 program package.

Enhancement of the intensity of the IR band νOH at 3660 cm-1 (corresponding to silanols difficult of access for

LA0301238

Influence of the Partial Hydrophobization of Fumed Silica by

Recommend Documents