Controlling Diffusion in Sol-Gel Derived Monoliths

Mandakini Kanungo and Maryanne M. Collinson*

Department of Chemistry, Kansas State University, Manhattan, Kansas 66506-3701

Received October 7, 2004. In Final Form: December 13, 2004

Redox probes were trapped within a silica monolith prepared in part with organoalkoxysilanes containinga quaternary ammonium functional group. The diffusion coefficients of the entrapped molecules were mea-sured as the gels were slowly dried using chronoamperometry and cyclic voltammetry with ultramicro-electrodes. Gel-entrapped cobalt(II) tris(bipyridine) (Co(bpy)3

2+) diffuses at rates similar to that measuredin the sols by incorporating a small amount of the positively charged functional group in the matrix. Incomparison, the diffusion coefficient of gel-entrapped ferricyanide (Fe(CN)6

3-) drops an order of magnituderelative to its value in the sol soon after gelation. These results demonstrate the ease at which diffusionin hydrated gels can be easily controlled by simply changing the charge on the walls of the silica host.

IntroductionThe sol-gel process1 provides a versatile means to

prepare inorganic and organic-inorganic host structuresfor catalysis,2,3 chemical analysis,4 ion-exchange appli-cations,5-7 solid-state electrochemical devices,8,9 and pho-tonics.10,11 In this process, a sol is first made through thehydrolysis and condensation of metal alkoxides (i.e.,tetramethoxysilane) and then typically doped with areagent or receptor.12,13 After a certain period of time, thesol gels, thus trapping the reagent in a hydrated matrix.During drying, solvent evaporates, the gel shrinks, thesurface-area-to-volume ratio in the pores increases, andinteractions between the entrapped reagent and the wallsof the host structure become increasingly important.14 Ofutmost importance to many applications is the rotationaland translational freedom at which the gel-entrappedreceptor can react with an analyte species in solution.Equally important is the rate at which a diffusing analytespecies can move through the porous matrix to react withthe entrapped receptor. Understanding how moleculesdiffuse in these materials and developing procedures toalter the rate at which reagents diffuse in a solid host isa necessary first step to improve the performance of sol-gel-based devices.

Several methods have been used to measure the diffu-sion coefficients of various guests entrapped in sol-gelderived materials. Fluorescence correlation spectros-copy,15-17 fluorescence microscopy,18 Raman spectros-copy,19,20 and time-dependent methods21 have been usedto study the diffusion of dyes and other molecules in thin

silicate films and thoroughly dried monoliths. Electro-chemical methods have also been used to evaluate the dif-fusion of redox-active reagents in hydrated monoliths.22-27

Our work, in particular, has utilized gel-immobilized ultra-microelectrodes to measure the diffusion coefficients ofredox molecules of varying size and charge trapped withinhydrated gels prepared under different conditions.25-27

This work has attested to the importance of intermolecularinteractions over constrained pore environments on dif-fusion inpartiallydriedgels. In this study,asimplemethodto influence or “control” the rates at which cationic andanionic probes diffuse in these solids is described. Thismethod involves the manipulation of electrostatic forcesbetween the gel-entrapped receptor and the walls of thematrix.

Experimental Section

Reagents. Tetramethoxysilane (TMOS, 99%) and ferrocenemethanol (FcCH2OH) were purchased from Aldrich. Trimethox-ysilylpropyl-modified poly(ethylene imine), 50% in 2-propanol,(polyamine-Si), and N-trimethoxysilylpropyl-N,N,N-trimethy-lammonium chloride, 50% in methanol (QAPS), were obtainedfrom Gelest Chemicals. Hydrochloric acid (HCl), potassiumferricyanide (Fe(CN)6

3-/4-), potassium chloride, methanol, and2-propanol were purchased from Fisher Scientific. All reagentswere used as received without further purification. Cobalt(II)tris(bipyridine) (Co(bpy)3

2+/3+) and ferrocenylmethyl trimethy-lammonium hexafluorophosphate (FcN+PF6

-) were synthesizedas described previously.26 Water was purified to type I (18 MΩ)using a Labconco four-cartridge system.

* To whom correspondence should be addressed. E-mail:mmc@ksu.edu. Phone: 785-532-1468. Fax: 785-532-6666.

(1) Brinker, J.; Scherer, G. Sol-Gel Science; Academic Press: NewYork, 1989.

(2) Blum, J.; Avnir, D.; Schumann, H. CHEMTECH 1999, 32-38.(3) Schubert, U. New J. Chem. 1994, 18, 1049-1058.(4) Collinson, M. M. Crit. Rev. Anal. Chem. 1999, 29 (4), 289-311.(5) Hsueh, C.; Collinson, M. M. J. Electroanal. Chem. 1997, 420

(1-2), 243-249.(6) Wei, H.; Collinson, M. M. Anal. Chim. Acta 1999, 397 (1-3), 113-

121.(7) Petit-Dominguez, M. D.; Shen, H.; Heineman, W. R.; Seliskar, C.

J. Anal. Chem. 1997, 69, 703-710.(8) Dunn, B.; Farrington, G. C.; Katz, B. Solid State Ionics 1994,

70/71, 3-10.(9) Rolison, D. R.; Dunn, B. J. Mater. Chem. 2001, 11, 963-980.(10) Knobbe, E. T.; Dunn, B.; Fuqua, P. D.; Nishida, F. Appl. Opt.

1990, 28 (18), 2729-2733.(11) Levy, D. Chem. Mater. 1997, 9, 2666-2670.(12) Lev, O.; Tsionsky, M.; Rabinovich, L.; Glezer, V.; Sampath, S.;

Pankratov, I.; Gun, J. Anal. Chem. 1995, 67 (1), 22A-30A.(13) Avnir, D. Acc. Chem. Res. 1995, 28, 328-334.(14) Dunn, B.; Zink, J. I. Chem. Mater. 1997, 9, 2280-2291.

(15) McCain, K. S.; Harris, J. M. Anal. Chem. 2003, 75, 3616-3624.(16) McCain, K. S.; Schluesche, P.; Harris, J. M. Anal. Chem. 2004,

76 (4), 930-938.(17) Mahurin, S. M.; Dai, S.; Barnes, M. D. J. Phys. Chem. B 2003,

107 (48), 13336-13340.(18) McCain, K. S.; Hanley, D. C.; Harris, J. M. Anal. Chem. 2003,

75, 4351-4359.(19) Nikiel, L.; Hopkins, B.; Zerda, T. W. J. Phys. Chem. 1990, 94,

7458-7464.(20) Watson, J.; Zerda, T. W. Appl. Spectrosc. 1991, 45, 1360-1365.(21) Koone, N.; Shao, Y.; Zerda, T. W. J. Phys. Chem. 1995, 99, 16976-

16981.(22) Audebert, P.; Griesmar, P.; Hapiot, P.; Sanchez, C. J. Mater.

Chem. 1992, 2, 1293-1300.(23) Audebert, P.; Griesmar, P.; Sanchez, C. J. Mater. Chem. 1991,

1, 699-700.(24) Audebert, P.; Sallard, S.; Sadki, S. J. Phys. Chem. B 2003, 107,

1321-1325.(25) Collinson, M. M.; Zambrano, P. J.; Wang, H.; Taussig, J. S.

Langmuir 1999, 15 (3), 662-668.(26) Howells, A. R.; Zambrano, P. J.; Collinson, M. M. Anal. Chem.

2000, 72 (21), 5265-5271.(27) Kanungo, M.; Collinson, M. M. Anal. Chem. 2003, 75 (23), 6555-

6559.

827Langmuir 2005, 21, 827-829

10.1021/la047518r CCC: $30.25 © 2005 American Chemical SocietyPublished on Web 01/06/2005

Procedures. The QAPS/TMOS sols were prepared by mixing7.5 g of the TMOS with either 0.75, 1.1, or 1.8 g of QAPS. A7.35-mL portion of the resultant mixture was mixed with 3.8 mLof methanol, 8.2 mL of water, and 1.2 mL of 0.01 M HCl. Theredox species and the supporting electrolyte (KCl) were addedto the sol to give a final concentration of 5 mM and 0.15 M,respectively. For the case of Fe(CN)6

3- (0.75 g of QAPS), 0.6 mLof 0.01 M HCl was added and the water amount was increasedto 8.8 mL.

The polyamine-Si/TMOS sols were prepared by mixing 7.5 gof the TMOS with either 0.80 g of polyamine-Si (for Co(bpy)3

2+)or 0.25 g of polyamine-Si (for Fe(CN)6

3-). A 7.35-mL portion ofthe resultant mixture was added to 3.8 mL of 2-propanol, andthe pH of the sol was adjusted to around 6-7 by adding anappropriate amount of dilute HCl. The pH of the sol was veryimportant for the gelation of the sol. The 5 mM redox species and0.15 M KCl were added to the sol as described above. TMOS-onlysols were prepared as described.26

All the doped sols were vigorously stirred for 30-45 min andthen poured into polystyrene vials. The microelectrode assembly,which consisted of a Pt microelectrode (r ) 13.3 µm) and a Ag/AgCl wire (serves as reference and counter electrode), wasinserted into the sol and the resultant electrochemical cell securedin a darkened Faraday cage. The electrochemical data werecollected at regular intervals as the gels were slowly dried at60-70% relative humidity.26 The sols gelled within 12 h. Therelative change in mass of the gel during the slow drying periodwas ∼15% over a 2-week period for all the gels. “Ambigels”28

formed from 3-week-old gels had approximately the same averagepore diameter (3.8 nm) calculated from the desorption branch ofa N2 adsorption-desorption isotherm using the Barrett, Joyner,and Halenda (BJH) method.

Results and Discussion

Diffusion in porous solids can be significantly morecomplex than that measured in solution because ofpotential interactions between the entrapped guest andthe walls of the porous network and because of confinementeffects. The surfaces of the pore walls are complex andcontain different functional groups including siloxane (Si-O-Si), silanol (Si-OH), and siloxide (SiO-) groups.14,29

In addition, the walls will be negatively charged undermost conditions, as the isoelectric point (pI) of silica isaround 2.30 A few studies have demonstrated the impor-tance of intermolecular interactions on dopant mobilityin both hydrated and fully dried gels, particularly withregard to charged molecules or ones that are able to hy-drogen bond with the matrix.19,29,31-33 One simple methodfor altering the diffusion rates of charged molecules wouldthus be to manipulate the charge on the porous matrixand change the extent of electrostatic interactions.

In this study, positively charged functional groups wereincorporated into the negatively charged silica matrix bycohydrolyzing and condensing tetramethoxysilane (TMOS)with either QAPS or polyamine-Si. The structures of thetwo precursors are shown in Chart 1. Prior to gelation, anultramicroelectrode (r ) 13.3 µm), a reference/counterelectrode, the redox molecule (either Fe(CN)6

3-, Co(bpy)32+,

FcCH2OH, or FcN+), and potassium chloride are added tothe sol. After a certain period of time, the sol turns intoa gel, and then, the gel is slowly dried under a controlledhumidity environment. Throughout the course of theexperiment (typically 2 days to 3 weeks), the gels are suf-

ficiently hydrated that electrochemistry can be done inthe matrix. By utilizing ultramicroelectrodes, the apparentdiffusion coefficient (Dapp) of entrapped redox-active mole-cules can be easily measured as the gel dries by couplingslow scan cyclic voltammetry with chronoamperometry.26

Figure 1 shows the steady-state cyclic voltammograms(CVs) of gel-encapsulated Fe(CN)6

3- in a matrix prepared

solely from TMOS (Figure 1A) and from QAPS (0.75 g)/TMOS (Figure 1B). In the TMOS-only gel, the steady-state limiting current increases slightly as the gel dries,whereas, in the modified gel, it drops quickly. For anultramicroelectrode, the steady-state current depends onthe product DappC, where Dapp is the apparent diffusioncoefficient and C is the concentration.34 As the gel dries,C slowly increases and Dapp will either drop or remain thesame. Since C does not significantly increase during thefirst few days, most of the changes in the voltammetrycan be attributed to Dapp (at least early on). For Fe(CN)6

3-

in a TMOS-only gel, it appears that Dapp stays the same,while, in the QAPS-modified gel, it obviously drops. Justthe opposite is observed for gel-entrapped Co(bpy)3

2+

(Figure 1C (TMOS) and D (QAPS(1.8 g)/TMOS)). In theTMOS-only gel, the limiting current for Co(bpy)3

2+ quicklydecreases (indicative of Dapp significantly decreasing),whereas, in the QAPS-modified gel, it stays relativelyconstant (Dapp likely does not vary too much) after thefirst hour. Nearly identical results were obtained withgels prepared from the cationic polymer, polyamine-Si(i.e., the CVs of Co(bpy)3

2+ and Fe(CN)63- showed the same

behavior as they did in the QAPS gels).

(28) Harreld, J. H.; Dong, W.; Dunn, B. Mater. Res. Bull. 1998, 33(4), 561-567.

(29) Shen, C.; Kostic, N. M. J. Am. Chem. Soc. 1997, 119, 1304-1312.

(30) Iler, R. K. The Chemistry of Silica; John Wiley and Sons: NewYork, 1979.

(31) Sieminska, L.; Zerda, T. W.J. Phys. Chem.1996, 100, 4591-4597.(32) Badjic, J. D.; Kostic, N. M. J. Phys. Chem. B 2000, 104, 11081-

11087.(33) Badjic, J. D.; Kostic, N. M. J. Mater. Chem. 2001, 11, 408-418.

(34) Wightman, R. M.; Wipf, D. O. Voltammetry at Ultramicroelec-trodes; Marcel Dekker: New York, 1989; Vol. 15, pp 267-353.

Chart 1

Figure 1. Cyclic voltammograms (2 mV/s) of gel-encapsulatedFe(CN)6

3- (A and B) and Co(bpy)32+ (C and D). Gels were

prepared from a QAPS/TMOS sol (B and D) or solely from TMOS(A and C). The potentials are with respect to a Ag/AgCl wirereference electrode.

828 Langmuir, Vol. 21, No. 3, 2005 Letters

The apparent diffusion coefficient can be calculatedwithout knowledge of C from the normalized chrono-amperometric response according to the following equa-tion, where id(t) is the chronoamperometric current duringthe potential step, iss is the steady-state current, and r isthe electrode radius (13.3 µm).35

The chronoamperometric response was obtained by step-ping the potential ∼ (250 mV beyond the redox formalpotential. This current is normalized by the steady-statelimiting current obtained from the steady-state CVsacquired at 2 mV/s and then fitted to the above equationto obtain Dapp independent of C. Only data that yieldedregression coefficients greater than 0.96 were used.

Figure 2 shows how the diffusion coefficient of theentrapped reagent changes as the gel dries in a controlledhumidity environment. The mean and standard deviations(error bars) from individual measurements in three to sixdifferent gels prepared at different times are shown. ForFe(CN)6

3- trapped in a gel prepared from TMOS, Dappdoes not change and is nearly identical to that measuredin solution. As previously described, this is consistent withthe dopant residing in the center of solvent filled pores.26

Significant changes are observed, however, when arelatively small amount of the positively charged precursoris incorporated in the gel. For gels prepared with a higheramount of QAPS, the steady-state current drops tooquickly to accurately measure Dapp. For gels containinga smaller amount (i.e., 0.75 g of QAPS), it was possibleto calculate Dapp. As can be seen in Figure 2, Dapp dropsover an order of magnitude in just a few days. Similarresults were observed when a small amount of the cationicpolymer (polyamine-Si) was incorporated in the matrix.Under these conditions, gel-entrapped Fe(CN)6

3- diffusesat a slower rate due to electrostatic interactions with thepositively charged groups on the surface. As the sol turnsinto a gel and the porous network starts to form, therewill be a greater interaction between the dopant and themodified silica walls. In addition, as solvent evaporates,there may also be a greater attraction between chargedsites due to reduced screening. Hence, Dapp drops as thegel is slowly dried.

Just the opposite is observed for gel-entrappedCo(bpy)3

2+ (Figure 2). In gels prepared solely from TMOS,Dapp drops an order of magnitude in a very short periodof time (similar to that for Fe(CN)6

3- in the QAPS gel). Aspreviously described, the decreased diffusion rate is dueto electrostatic interactions between the positively chargeddopant and the negatively charged surface.23 Co(bpy)3

2+

can, however, be made to diffuse at rates similar to whatit does in solution by changing the charge on the porousnetwork. Figure 2 shows relative changes in Dapp as afunction of drying time for Co(bpy)3

2+ trapped in gelscontaining different amounts of QAPS or the cationicpolymer (polyamine-Si). As the amount of QAPS copoly-merized with TMOS increases, the rate of change of Dappbecomes less. Eventually, Dapp stays approximately con-stant and at a value similar to that measured forCo(bpy)3

2+ in the sol.The diffusion coefficients of two other dopants (FcN+

and FcCH2OH) trapped in the materials prepared fromQAPS have also been measured during gel formation andslow drying. For both compounds, the drop in Dapp is notas large during the early stages of drying (<1 week) forthe QAPS gels (1.1 g) compared to the TMOS-only gels.After ∼1 week, the rate of change in Dapp is nearly thesame and the normalized diffusion coefficient is also thesame. The extent of intermolecular interactions betweenthe host and these two guests would obviously be lessrelative to that expected for Co(bpy)3

2+ and Fe(CN)63-.

The data support this statement.In summary, this work has shown that the diffusion

rates of charged molecules trapped in a silicate gel can beeasily changed and tuned by modifying the charge on thematrix walls. Anionic molecules can be made to diffuse atrates similar to what they do in solution by retainingthe native structure of the gel. Cationic molecules can bemade to diffuse at rates similar to that in solution byincorporating a sufficient number of positively chargedfunctionalities in the gel. The ability to control therates of diffusion of different species in sol-gel materialswill enable their performance to be enhanced in sensor,catalysis, and solid-state applications. An interestingexample of this is to attach a receptor to a silicate matrixthat has been appropriately modified so that an analytespecies will be able to more “rapidly” move through thematrix to react with it.

Acknowledgment. We gratefully acknowledge sup-port of this work by the National Science Foundation(CHE) and the Office of Naval Research.

LA047518R(35) Shoup, D.; Szabo, A. J. Electroanal. Chem. 1982, 140, 237-245.

Figure 2. Normalized diffusion coefficients (D/D0) as a functionof drying time for gel-encapsulated Fe(CN)6

3- (top) or Co(bpy)32+

(bottom). Gels doped with Fe(CN)63- were prepared from ([)

TMOS only, (O) 0.75 g of QAPS/TMOS, or (1) polymer-Si/TMOS.Gels doped with Co(bpy)3

2+ were prepared from (b) TMOS only,(1) 1.1 g of QAPS/TMOS, (9) 1.8 g of QAPS/TMOS, or (O)polymer-Si/TMOS. D0 is the diffusion coefficient of the redoxprobe obtained within the first hour after the sol was made.The error bars represent the standard deviations in Dapp forthree to six gels. Lines have been added to guide the eye.

id(t)/iss ) 0.7854 + 0.4431(Dt/r2)-1/2 +

0.2146 exp(-0.3911(Dt/r2)-1/2) (1)

Letters Langmuir, Vol. 21, No. 3, 2005 829