Revisiting Silicalite-1 Nucleation in Clear Solution ... - ACS Publications

Article pubs.acs.org/Langmuir

Revisiting Silicalite‑1 Nucleation in Clear Solution by Electrochemical Impedance Spectroscopy G. Brabants,†,‡ M. Hubin,† E. K. Reichel,‡ B. Jakoby,‡ E. Breynaert,† F. Taulelle,† J. A. Martens,† and C. E. A. Kirschhock*,† †

Centre for Surface Chemistry and Catalysis, KU Leuven, Leuven 3000, Belgium Institute for Microelectronics and Microsensors, Johannes Kepler University, Linz 4040, Austria

‡

S Supporting Information *

ABSTRACT: Electrochemical impedance spectroscopy (EIS) was used to detect and investigate nucleation in silicalite-1 clear solutions. Although zeolite nucleation was previously assumed to be a step event, inducing a sharp discontinuity around a Si/OH− ratio of 1, complex bulk conductivity measurements at elevated temperatures reveal a gradual decay of conductivity with increased silicon concentrations. Inverse Laplace transformation of the complex conductivity allows the observation of the chemical exchange phenomena governing nanoaggregate formation. At low temperatures, the fast exchange between dissociated ions and ion pairs leads to a gradual decay of conductivity with an increasing silicon content. Upon heating, the exchange rate is slower and the residence time of ion pairs inside of the nanoaggregates is increasing, facilitating the crystallization process. This results in a bilinear chemical exchange and gives rise to the discontinuity at the Si/OH− ratio of 1, as observed by Fedeyko et al. EIS allows the observation of slow chemical exchange processes occurring in zeolite precursors. Until now, such processes could be observed only using techniques such as nuclear magnetic or electron paramagnetic resonance spectroscopy. In addition, EIS enables the quantification of interfacial processes via the double layer (DL) capacitance. The electrical DL thickness, derived from the DL capacitance, shows a similar gradual decay and confirms that the onset of nanoaggregate formation is indeed not narrowly defined.

■

INTRODUCTION Electrochemical impedance spectroscopy (EIS) has experienced a great increase in popularity in the recent decade. Applied initially for the determination of double layer (DL) capacitances, it nowadays finds application in various facets of materials research, ranging from interfacial processes to applications in fuel cells,1 coatings,2−4 and membranes.5,6 Until recently,7 EIS had never been used for monitoring early zeolite formation. Opposite to most other electrochemical techniques, EIS probes the frequency domain rather than the time domain. This has the major advantage in that the measured impedance spectra contain information about all aspects of the material under study, including dynamic processes occurring on different time scales and thus at different specific frequencies. The relatively quick and simple impedance measurement by consequence gives access to a wide range of electrical properties of the sample. Provided a physically relevant model is used to describe the electrical behavior of the system, complex, simultaneous processes can be monitored in an easy and noninvasive way. Classically, zeolite crystallization is achieved by aging a geltype hyper alkaline aluminosilicate synthesis mixture under hydrothermal conditions. Silicalite-1 often serves as a model © 2017 American Chemical Society

system for studying the zeolite nucleation and growth because its formation can also be studied in a synthesis starting from a clear precursor liquid, historically referred to as the “clear solution”. In a first stage of the synthesis (see Figure 1), a silica source, typically TEOS is hydrolyzed at room temperature in aqueous tetrapropyl ammonium hydroxide (TPAOH). Silicic acid forms by slow hydrolysis of TEOS (reaction 1) and is readily deprotonated in the highly alkaline solution (reaction 2). In the resultant silicate solution, multiple reversible reactions occur simultaneously and generate a score of interconverting dissolved silicate species (reactions 3 and 4). These silicate anions ion-pair with alkaline cations, in the present case TPA+. With increasing silicate concentration, and hence also with increasing silicon over hydroxide ion ratios (Si/OH−), hydrated oligomeric silicate anions and hydrated TPA+ cations form larger ion-pair assemblies (reaction 5). The state of the precursor liquid thus evolves from a true solution to a transparent suspension of nanoaggregates at Si/OH− ratios Received: November 16, 2016 Revised: February 1, 2017 Published: February 17, 2017 2581

DOI: 10.1021/acs.langmuir.6b04135 Langmuir 2017, 33, 2581−2589

Article

Langmuir

Figure 1. Mechanism of nanoaggregation from room temperature precursor formation, as proposed by van Tendeloo et al.8 Silicic acid is formed by the hydrolysis of tetraethyl orthosilicate (TEOS) (reaction 1). The silicic acid readily deprotonates and forms ion pairs with the alkaline cations (reaction 2). Dimerization and oligomerization (reaction 3) result in a score of interconverting (reaction 4) silicate species. Nanoaggregation (reaction 5) occurs at a Si/OH− ratio exceeding 1, for ethanol-containing sols. Further dehydration leads to the condensation of species in the nanoaggregates (reaction 6), slowly converting into nanoparticles and ultimately crystallizing into the final zeolite structure.

Figure 2. Technical drawing of the custom-built impedance cell (above) for EIS measurements on silicalite-1 precursors at elevated temperatures. The left and right views (below) of the assembled cell show the adjustable electrode gap and sample and air inlet and outlet. The electrodes are shown in brown, and the sample volume is shown in green. The cross section of the interchangeable spacer is indicated as the shaded area (bottom left) and determines the electrode gap.

2582


Article

Langmuir

Figure 3. (a) Bulk resistance in silicalite-1 precursors with Si/OH− = 2.50, measured at different electrode gap distances (filled circles) and the corrected values for a constant error in resistance (triangles) or in the electrode gap determination (open circles). (b) Corresponding resistivity values, calculated from the bulk resistance. The dotted line indicates the resistivity calculated from the linear correlation of bulk resistance vs electrode gap distance.

properties accessible by quick and simple, noninvasive impedance measurements.

exceeding 1. Nanoaggregates form at the expense of isolated oligomers through a reversible process. Consequently, the concentration of the isolated oligomers drops drastically as nanoaggregation starts.9 Both time and temperature lead to stronger aggregation by expulsion of hydration water from the nanoaggregates, resulting in an increased residence time of the silicate oligomers and cations in the aggregates. Finally, this leads to the condensation of species in the nanoaggregates (reaction 6), slowly converting into nanoparticles as the silicon connectivity increases, and ultimately crystallizing into the final zeolite structure.8,9 Given the highly ionic nature of the zeolite precursor liquid, the complex interplay between charged species proves to be a crucial aspect of unraveling the early phase of zeolite formation. It is here that EIS comes into play. EIS basically studies the response of a system to a periodic perturbation. In this case, the perturbation is a small-amplitude alternating-current signal over a wide frequency range. The impedance response is fundamentally determined by the electrical current conducted through the system following the path of least impedance. The interpretation of EIS data, so-called equivalent circuit modeling, is thus an attempt to describe the complex physical, chemical, electrical, and/or mechanical processes in the sample in purely electrical terms.10 Using prior knowledge from other characterization techniques, a physically relevant equivalent circuit is constructed and used to obtain information about, for example, electrolyte resistance, bulk conductivity, DL capacitance, charge-transfer resistance, and diffusion. In the present work, the interest goes out to bulk conductivity and DL effects in zeolite precursors. Initially, all species are in solution, which can be directly monitored by the bulk conductivity. Because nanoaggregates form at the expense of isolated oligomers in the bulk, changes to the bulk conductivity are directly related to the nanoaggregate formation. Increasing oligomerization modifies the dielectric constant of the medium, thereby influencing the stability of these colloidal nanoaggregates through changes in the electrical DL. This can be monitored by EIS via the DL capacitance. The potential of EIS to study the initial zeolite formation has previously been communicated.7 The current work explores it in depth and shows the wide range of

■

EXPERIMENTAL SECTION

The electric behavior of a series of silicalite-1 precursors was studied during the initial zeolite formation. The precursors were synthesized by room temperature hydrolysis of TEOS (>99%, Aldrich) in an aqueous TPAOH (40%, Aldrich) solution under vigorous stirring, resulting in molar compositions of TEOS/TPAOH/H2O = x:9.00:480, with x ranging from 0 to 45 mol. The Si/OH− ratio consequently varies from 0.0 to 5.0. Electrochemical impedance measurements were recorded using an Agilent E4990A precision impedance analyzer (Agilent Tech., Santa Clara, CA, USA) in combination with a custombuilt impedance cell. The impedance cell, depicted in Figure 2, was designed for the purpose of in situ EIS on the highly ionic and corrosive zeolite clear solutions. The zeolite precursor sample is contained between two parallel stainless steel electrodes (radius Re = 15 mm) and the Teflon housing. To avoid distortions of the electrical field near the electrode edge predominating the impedance response, the gap distance de is limited by de2 ≪ 2Re. An air inlet and outlet prevents the trapping of air bubbles in the sample chamber during filling and ensures the removal of ethanol bubbles forming at elevated temperatures and during in situ synthesis. An exchangeable spacer separates the upper and lower part of the cell and allows the variation in the electrode gap de, as indicated in Figure 2. The purpose of this is twofold: first, by changing the electrode gap distance, the bulk properties yield an associated change in the measured characteristics while not affecting the impact of the DL properties. Unlike for dc conductivity measurements, the bulk conductivity can hence be fully separated from processes occurring at the interface. Second, by measuring at different electrode gap distances and calculating the corresponding conductivity from the slope of their correlation, rather than from one individual measurement, more accurate conductivity data are obtained. The latter is demonstrated in Figure 3 for bulk resistance measurements: the bulk resistances obtained from the measured impedance spectra show a perfect linear relationship with increasing electrode gap distance, but the correlation does not cross the origin (Figure 3a). As the zero electrode gap distance must inevitably lead to short-circuiting and should thus, as a result of calibration, give zero resistance, the correlation of bulk resistance versus electrode gap distance needs to be corrected. After the correction for a constant error in resistance (triangular symbols in Figure 3a) and/or in the electrode gap determination (open circles), the bulk resistivity calculated from the bulk resistance is independent of the electrode gap distance. This correction is essential, as bulk 2583


Article

Langmuir resistivities calculated from individual uncorrected data points give rise to errors up to 23% (Figure 3b). Impedance spectra were recorded continuously by four-point measurements at 100 mV, in the frequency ranging from 100 Hz up to 10 MHz. Before the measurement, calibration of the experimental setup was performed in the same frequency range using a series of open, short, and load (50 Ω) calibrations. All silicalite-1 precursor solutions were measured in a climate chamber at temperatures ranging from 0 to 60 °C. Up to a temperature of 60 °C, nanoaggregate formation is reversible. At higher temperatures, the nanoaggregates are assumed to grow irreversibly via Ostwald ripening.11 Varying both silicon content (from Si/OH− = 0 to 5) and temperature (from 0 to 60 °C) results in a data set of 90 individual impedance spectra. Additionally, the nanoaggregate formation was monitored using multiangle dynamic light scattering (DLS) and small-angle X-ray scattering (SAXS) while simultaneously following the pH evolution. SAXS data were collected on an Anton Paar SAXSess instrument (Anton Paar, Graz, Austria) using line-collimated Cu Kα radiation (1.542 Å), and were modeled as a polydisperse population of spheres with a Gaussian distribution of radii. The Hayter−Penfold mean spherical approximation (MSA) was used as a structure factor. The combined result is a model of charged, polydisperse spheres, which describes the scattered data with high precision. DLS was carried out on an ALV CGS-3 compact goniometer (ALV, Langen, Germany) with a wavelength of 632.8 nm and at 8 scattering angles ranging from 45 to 150°. All precursor samples were filtered with Chromafil 200 nm polytetrafluoroethylene filters to facilitate the detection of the smallest species. The pH evolution was monitored with a S47-K SevenMulti dual meter (Mettler Toledo, Greifensee, Switzerland), in combination with a Metrohm LL Profitrode electrode (Metrohm, Herisau, Switzerland) for high pH measurements. The electrode was calibrated using buffer solutions of pH 4, 7, and 10 before measurements.

CPE behavior, but even minor factors such as surface roughness already lead to deviations from ideal capacitive behavior. The CPE, as the name suggests, displays an impedance with a constant, frequency-independent phase angle differing from the ideal 90°. ZCPE =

1 Q (iω)α

(2)

where ω (=2πf) is the angular frequency, Q is the pseudocapacitance, and α is the CPE exponent. The exponent α can range from 0 to 1. In case the exponent equals 1, the CPE describes a pure capacitor with a capacitance of Q. In the case of α = 0, eq 2 describes an ideal resistor with Q = 1/R. The individual parameter values are obtained by complex nonlinear least square error minimization between the measured impedance response Zmeas and the equivalent circuit model Zeq. Error =

′ − Zeq ′ )2 + (Zmeas ″ − Zeq ″ )2 ] ∑ [(Zmeas

(3)

Here, the single and double inverted commas refer to the real and imaginary parts of the impedances, respectively. The equivalent circuit model yields reasonable fits in the frequency range from 100 Hz to 10 MHz. By equivalent circuit modeling of the impedance spectra, both bulk parameters (via the bulk conductivity RB) and interfacial processes (via the DL capacitance CDL or CPE behavior) can be accessed. Conductivity. The bulk conductivity σ (and resistivity ρ) is calculated from the bulk resistance RB, obtained by fitting the impedance spectra based on the equivalent circuit

■

RESULTS AND DISCUSSION The obtained impedance spectra were modeled using the simplified Randles equivalent circuit.12 This is one of the most commonly used models, built up from a bulk resistance RB, and a charge-transfer resistance RCT in parallel with a DL capacitance CDL. The impedance response of the simplified Randles equivalent circuit (assuming time dependence eiωt) is given by

σ=

A 1 = e ·RB ρ de

(4)

The bulk resistance RB is directly proportional to the conductivity σ and inversely proportional to the resistivity ρ, by a geometrical factor which, for perpendicular electrodes, is determined by the distance de between the electrodes and the electrode surface area Ae. Conductivity is typically expressed in millisiemens per centimeter (mS/cm). Fedeyko et al.11,15 reported in a series of publications on dc conductivity measurements in dilute silicalite-1 clear solutions, where they observed a discontinuity in the conductivity when a critical silicon concentration is reached. For different initial molar compositions and different tetraalkylammonium cations, this limit was observed at the Si/OH− ratio of 1.15 From these observations, it was concluded that nanoaggregate formation is driven by a process similar to micellization and hence the distinct discontinuity observed at Si/OH− = 1. Only above this limit, which they termed the critical aggregation concentration, nanoaggregates are observed. Note however that Fedeyko made these observations based on dilute silicalite-1 precursors at ambient temperature. At first sight, this trend seems to be confirmed in concentrated silicalite-1 precursor solutions. However, the apparent discontinuity fades with decreasing temperature and the transition around the Si/OH− ratio of 1 occurs more gradually. The bulk conductivity decays nearly exponentially with an increasing Si/OH− ratio, with all coefficients of determination approaching unity (Raverage2 = 0.975). This suggests that the onset of nanoaggregation is in fact not so narrowly defined. The complete data set is displayed in Figure 5 in a logarithmic plot, both as a function of their Si/ OH− ratio and the temperature during the measurement. A

⎤ ⎡ ωR 2C ⎤ ⎡ R CT CT DL ⎥ ⎢ ⎥ Zeq = ⎢RB + i − 2 2 2 2 2 2 1 + ω R CT C DL ⎦ ⎣ ⎣ 1 + ω R CT C DL ⎦ (1)

Because the DL commonly does not behave as an ideal capacitance due to time-constant distributions, CDL was replaced here by a nonideal circuit element;13 a so-called constant phase element (CPE) (see Figure 4). Lasia14 investigated the effects of several experimental factors on the DL behavior, such as cell geometry, surface roughness, and adsorption and concluded that the adsorption of trace impurities on the electrode is one of the major reasons for

Figure 4. Simplified Randles equivalent circuit, with the ideal DL capacitance CDL replaced by the nonideal CPE. 2584


Article

Langmuir

Figure 5. Conductivity of the concentrated silicalite-1 precursors, calculated from the bulk resistance, as a function of temperature and Si/OH− ratio. A 3D surface is interpolated through the individual measurements (a). Both parameters are plotted separately below: (b) conductivity as a function of Si/OH− ratio and (c) as a function of temperature. Individual data points in (b) are connected as a guide to the eye. Linear interpolations are shown in gray in (c). The dotted line indicates the Si/OH− ratio of 1, where particle formation is observed to take off.

of ion pairs. The single ion pair acts as a reaction intermediate here.

three-dimensional surface is interpolated through the individual measurement data to indicate the trend. When TEOS hydrolyzes into silicic acid in the highly alkaline solution, it immediately deprotonates.8 The deprotonation, limited to approximately one proton per Si(OH)4, results in a silicate anion ion-paired to a TPA+ cation, Si(OH)3O−···TPA+. In the aqueous solutions, the lattice energy is dominated by solvation of both ions, rather than by the Coulombic interaction. This is described by the Debye−Hückel solvent model. Increasing silicon concentrations lead to an increasing degree of silicate oligomerization and the associated release of ethanol. The solubility of the TPA+ cations in the water− ethanol phase decreases and nanoaggregation occurs.8 The presence of these nanoaggregates has been confirmed by both DLS and SAXS observations (see Figure S2). Castro et al.9 monitored the interaction between TEA+ cations and the anionic silicate oligomers in zeolite beta precursors by 14N and 1 H NMR spectroscopy. Both resonances have been observed to show a strong discontinuity in their interactions at the onset of nanoaggregate formation. Because nanoaggregate formation is assumed to be reversible up to temperatures of 60 °C,11 the processes observed in the silicalite-1 precursor are essentially chemical exchanges between dissociated ions, single ion pairs, and nanoaggregates consisting

dissociated ions ⇌ ion pairs ⇌ nanoaggregates nSi(OH)3 OH2O− + nTPA H2O+ ⇌ nSi(OH)3 O− ··· TPA H2O+ ⇌ [Si(OH)3 O− ··· TPA+, pH 2O]n

The bulk conductivity is sensitive to the exchange between these species because this exchange affects the total number of mobile charges at any instance. Changes in the characteristic lifetime of all species in exchange can thus be monitored, with varying temperature. At low temperatures, the exchange between dissociated ions and ion pairs is fast compared with the measurement time scale, and the residence time in the nanoaggregates is consequently short. When the dielectric constant of the bulk phase increases upon heating, the residence time inside of the nanoaggregates increases. Hence, the exchange rate must be slower, facilitating further condensation within nanoaggregates.16,17 This might seem counterintuitive, but the underlying kinetics become easier to visualize by inverse Laplace transformation (ILT) of the 2585


Article

Langmuir

Rexponential2 = 0.90 versus Rbilinear2 = 0.98. Both curves intersect halfway. This exchange phenomenon between dissociated ions and nanoaggregates is very different from the exchange phenomenon observed using nuclear magnetic resonance (NMR), as reported by Daniel Shantz.19 The exchange between oligomers increases with temperature. The outside−inside nanoaggregate is a different phenomenon with a reversed temperature dependence. This temperature dependence is not observed in the pH behavior. The pH evolution shows a comparable sigmoidal trend for all temperatures up to 60 °C (see Figure 8). The

conductivity data at different temperatures (see Figure 6). Numerical ILT of the logarithmic conductivity was carried out using a non-negative, linearly regularized ILT.18

Figure 6. Numerical ILT of the log(σ) vs Si/OH− data, for temperatures ranging from 0 to 60 °C.

The ILTs show two distinct processes shifting further apart with increasing temperature. For increased temperatures, this will lead to two distinct single peaks, corresponding to dissociated ions and ion pairs, respectively. In the low temperature limit, intermediate and fast chemical exchange would coalesce at the weighted average of the resolved peaks observed at high temperatures. The discontinuity observed in the conductivity is then simply the result of the slow, bilinear chemical exchange at higher temperatures and should turn into a smooth transition upon cooling. This transition becomes apparent when comparing the quality of bilinear fitting (proposed by Fedeyko et al.15) with that of exponential fits. Figure 7 shows the R-squared values of both bilinear and exponential fittings, as a function of temperature. At low temperatures, the exponential model seems to describe the logarithmic conductivity best (Rexponential2 = 0.99 versus Rbilinear2 = 0.95). The opposite is observed at higher temperatures, with

Figure 8. pH evolution of silicalite-1 precursors with molar compositions of TEOS/TPAOH/H2O = x:9.00:480, where x ranges from 0 to 45 mol and at temperatures ranging from 0 to 60 °C. The dotted line indicates a sigmoid curve fitted to the pH data. The slope of the fitted sigmoid does not significantly vary with temperature.

transition occurring around Si/OH− = 1 is gradual, which seems to agree with the gradual decay in conductivity discussed earlier. However, within the accuracy of the pH measurements, the pH trend is similar for all temperatures from 0 to 60 °C, although the absolute pH values obviously differ. Opposite to the bulk conductivity measurements, no sharper transition of the pH around the Si/OH− ratio of 1 is observed at higher temperatures. DL Capacitance. The electrical DL effect plays an important role in the stabilization and aggregation of colloids. It is by consequence an important property during the early stages of zeolite formation. The DL thickness can be estimated by monitoring its capacitance through EIS. According to the Gouy−Chapman model for electrical DLs,20,21 ions accumulate next to an electrode in a diffuse layer whose thickness scales with the Debye length λD. The two ion layers, separated by a single layer of solvent molecules, ideally behave as a capacitor. Because in practice the DL does not behave as an ideal capacitor, its behavior can be modeled by a CPE. If the CPE exponent α equals 1, as described by eq 2, the CPE describes an ideal capacitor with a capacitance equal to the CPE coefficient Q. However, if the exponent approaches unity, in practice for 1 > α > 0.8, the CPE can be expressed as an “effective equivalent capacitance”, Ceff. In the case of the concentrated silicalite-1 precursor solutions, all CPE exponent values lie around 0.90, closely approaching the behavior of an ideal capacitor (see Figure S1). In such cases, Brug et al.22 proposed the use of an

Figure 7. Coefficients of determination (R-squared value) of both bilinear and exponential fittings of the logarithmic conductivity data shown in Figure 5b. 2586


Article

Langmuir “effective equivalent capacitance” Ceff that can be calculated via the Brug conversion: Ceff

α − 1⎤1/ α ⎡ ⎛ 1 1 ⎞ ⎥ ⎢ = 2 Q⎜ + ⎟ ⎢⎣ ⎝ RB R CT ⎠ ⎥⎦

effect of the temperature cannot be observed due to the measurement noise, although the Debye lengths from higher temperature spectra on average are shorter than those obtained from impedance spectra at lower temperatures. This is in full agreement with observations from both DLS and SAXS. SAXS modeling (see Figure 10) reveals a population of charged subnanometer species that is growing in size with

(5)

The capacity of the electrical DL can be related to the Debye length, using the Gouy−Chapman model for the electrical DL. The DL capacitance per unit area, according to Gouy− Chapman, is given by23 C DL = Ae

⎛ eψ ⎞ ⎛ eψ ⎞ ε 2e 2c0εbulk ·cosh⎜ 0 ⎟ = bulk ·cosh⎜ 0 ⎟ λD kBT ⎝ 2kBT ⎠ ⎝ 2kBT ⎠ (6)

with the electrode surface area Ae, elementary charge e, the ion concentration at the surface c0, the bulk liquid permittivity εbulk, the Boltzmann constant kB, temperature T, and surface potential ψ0. The hyperbolic cosine can be expanded into a series (cosh x = 1 + x2/2! + x4/4! + ...) and, for sufficiently small surface potentials, approximated by considering only its first term. This leaves the following approximation, where the DL capacitance CDL, and by approximation the effective equivalent capacitance Ceff obtained from EIS, is inversely proportional to the Debye length λD ε C DL ≈ bulk Ae λD

Figure 10. SAXS patterns of silicalite-1 precursors modeled with a polydisperse sphere form factor and a Hayter−Penfold MSA as a structure factor.

(7)

When compared with the capacitance of a parallel-plate capacitor (Cpp = εε0Ae/de), the DL capacitor basically behaves like a plate capacitor with a surface area Ae and an electrode distance de equal to the Debye length. Figure 9 shows the Debye lengths derived for the concentrated silicalite-1 precursors, as a function of Si/OH− ratio and at different temperatures. It can be observed that the Debye length drops with increasing silicon concentration and levels off at Si/OH− ratios exceeding 1, thus allowing ionic species in solution to sufficiently approach to form nanoaggregates. A significant

increasing Si content. These species are already visible from Si/ OH− = 0.50. For Si/OH− ratios above 2.0, a second and larger population of about 1.5 nm in radius appears. The hydrodynamic radii obtained from DLS are in agreement with these results. In DLS, the subnanometer population is observed from Si/OH− = 1, growing slightly larger in size with increasing silicon concentration. Taking into account that the particle size estimate obtained from DLS is a hydrodynamic radius, the slightly larger particle sizes are consistent with the smallest population observed in SAXS. From the Si/OH− ratio of 2, again a larger particle population of approximately 6−7 nm in radius is observed. Because scattering intensity for DLS scales with the sixth power of the particle radius (I ≈ r6), these larger particles effectively hide the smaller population from detection by DLS, although SAXS confirms their presence also at these higher Si/OH− ratios. The increasing difference between the DLS-observable hydrodynamic radii of the larger population and the radii obtained from SAXS, indicate a nanoaggregation behavior. This is also clear from the decreasing surface charge observed in the SAXS model fitting, similar to the decrease in the estimated Debye lengths. A summary of all particle size measurements can be found in Figure S2.

■

SUMMARY AND CONCLUSIONS The presented data explore and demonstrate the potential of EIS for studying the initial zeolite formation in more detail. The custom-built setup with adjustable spacer distance allows for complex conductivity measurements, separating bulk and interfacial properties by EIS. Complex conductivity measurements in silicalite-1 precursors, with varying Si/OH− ratios and over a wide temperature range, show that the onset of nanoaggregate formation is in fact not a sharply defined

Figure 9. Debye lengths, calculated according to the Gouy−Chapman model for the electrical DL, as a function of the Si/OH− ratio of the concentrated silicalite-1 precursors. Temperature does not seem to have a significant influence, although higher temperatures on average do show lower Debye lengths. The dotted lines are merely a guide for the eye, indicating the decreasing trend. 2587


Langmuir

■

transition. Instead of a discontinuity near the Si/OH− ratio of 1, a gradual decay is observed. The kinetics governing nanoaggregate formation can be observed via ILT of the complex conductivity. At low temperatures, the chemical exchange between dissociated ions and ion pairs is fast compared with the measurement time scales of EIS. This leads to a gradual change in slope for the conductivity as a function of the Si/ OH− ratio. At higher temperatures, the residence time of the ion pairs inside of the nanoaggregates increases, facilitating crystallization, and hence the exchange rate must be slower. This results in a bilinear exchange and gives rise to the discontinuity previously observed at the Si/OH− ratio of 1 by Fedeyko et al.15 The kinetics of zeolite formation thus slow down at higher temperatures, facilitating crystallization. EIS allows the observation of these slow chemical exchange processes. Although chemical exchange phenomena were until now limited to techniques such as NMR and electron paramagnetic resonance spectroscopy,16,17,24,25 the exchange between oligomers to nanoaggregates has never been measured using these techniques. This trend, and the onset of nanoaggregate formation around the Si/OH− ratio of 1, is confirmed by changing DL thickness with increasing silicon concentrations. The DL thickness is approximated by the Debye length, which in turn is calculated from the DL capacitance values derived from EIS. SAXS, DLS, and pH measurements support consistently the observed trend. The reported results show that EIS is a valuable technique to complement traditional methods used for studying the initial zeolite formation, by quick and simple, noninvasive and in situ impedance measurements, giving information on both bulk and interfacial processes. Moreover, the chemical exchange with nanoaggregates has been observed for the first time.

■

ASSOCIATED CONTENT

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.langmuir.6b04135. Schematic illustration of CPE exponents as a function of the Si/OH ratio and summary of the particle size measurements obtained by SAXS and multiangle DLS (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

G. Brabants: 0000-0002-6750-7100 E. Breynaert: 0000-0003-3499-0455 Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) Wagner, N.; Schnurnberger, W.; Mü ller, B.; Lang, M. Electrochemical impedance spectra of solid-oxide fuel cells and polymer membrane fuel cells. Electrochim. Acta 1998, 43, 3785−3793. (2) Jüttner, K. Electrochemical impedance spectroscopy (EIS) of corrosion processes on inhomogeneous surfaces. Electrochim. Acta 1990, 35, 1501−1508. (3) Mansfeld, F.; Jeanjaquet, S. L.; Kendig, M. W. An electrochemical impedance spectroscopy study of reactions at the metal/coating interface. Corros. Sci. 1986, 26, 735−742. (4) Van Westing, E. P. M.; Ferrari, G. M.; De Wit, J. H. W. The determination of coating performance using electrochemical impedance spectroscopy. Electrochim. Acta 1994, 39, 899−910. (5) Coster, H. G. L.; Chilcott, T. C.; Coster, A. C. F. Impedance spectroscopy of interfaces, membranes and ultrastructures. Bioelectrochem. Bioenerg. 1996, 40, 79−98. (6) Cai, X.; Zhang, Y.; Yin, L.; Ding, D.; Jing, W.; Gu, X. Electrochemical impedance spectroscopy for analyzing microstructure evolution of NaA zeolite membrane in acid water/ethanol solution. Chem. Eng. Sci. 2016, 153, 1−9. (7) Brabants, G.; Lieben, S.; Breynaert, E.; Reichel, E. K.; Taulelle, F.; Martens, J. A.; Jakoby, B.; Kirschhock, C. E. A. Monitoring early zeolite formation via in situ electrochemical impedance spectroscopy. Chem. Commun. 2016, 52, 5478−5481. (8) van Tendeloo, L.; Haouas, M.; Martens, J. A.; Kirschhock, C. E. A.; Breynaert, E.; Taulelle, F. Zeolite synthesis in hydrated silicate ionic liquids. Faraday Discuss. 2015, 179, 437−449. (9) Castro, M.; Haouas, M.; Lim, I.; Bongard, H. J.; Schüth, F.; Taulelle, F.; Karlsson, G.; Alfredsson, V.; Breyneart, E.; Kirschhock, C. E. A.; Schmidt, W. Zeolite Beta Formation from Clear Sols: Silicate Speciation, Particle Formation and Crystallization Monitored by Complementary Analysis Methods. Chem.Eur. J. 2016, 22, 15307− 15319. (10) Lvovich, V. F. Impedance SpectroscopyApplications to Electrochemical and Dielectric Phenomena; Wiley: NJ, 2012. (11) Rimer, J. D.; Lobo, R. F.; Vlachos, D. G. Physical Basis for the Formation and Stability of Silica Nanoparticles in Basic Solutions of Monovalent Cations. Langmuir 2005, 21, 8960−8971. (12) Randles, J. E. B. Kinetics of Rapid Electrode Reactions. Discuss. Faraday Soc. 1947, 1, 11−19. (13) Perevedentsev, A. Personal communication, July 15, 2016. (14) Lasia, A. Electrochemical Impedance Spectroscopy and Its Applications; Springer Verlag: Berlin, 2004. (15) Fedeyko, J. M.; Rimer, J. D.; Lobo, R. F.; Vlachos, D. G. Spontaneous Formation of Silica Nanoparticles in Basic Solutions of Small Tetraalkylammonium Cations. J. Phys. Chem. B 2004, 108, 12271−12275. (16) Taulelle, F. Clipping a network into a crystal. Solid State Sci. 2001, 3, 795−800. (17) Taulelle, F.; Pruski, M.; Amoureux, J. P.; Lang, D.; Bailly, A.; Huguenard, C.; Haouas, M.; Gérardin, C.; Loiseau, T.; Férey, G. Isomerization of the prenucleation building unit during crystallization of AlPO4-CJ2: An MQMAS, CP-MQMAS, and HETCOR NMR study. J. Am. Chem. Soc. 1999, 121, 12148−12153. (18) Rogers, S. Matlab code, 2005, http://www.wolfson.cam.ac.uk/ ~ssr24/invlap.html. Press, W. H.; Teukolsky, S. A.; Vetterling, W. T.; Flannery, B. P. Numerical Recipes in C; Cambridge University Press: Cambridge, 1996. (19) Cheng, C.-H.; Shantz, D. F. 29Si NMR Studies of Zeolite Precursor Solutions. J. Phys. Chem. B 2006, 110, 313−318. (20) Gouy, M. Sur la constitution de la charge électrique à la surface d’un électrolyte. J. Phys. Theor. Appl. 1910, 9, 457−468. (21) Chapman, D. L. LI. A contribution to the theory of electrocapillarity. Philos. Mag. A 1913, 25, 475−481. (22) Brug, G. J.; Van Den Eeden, A. L. G.; Sluyters-Rehbach, M.; Sluyters, J. H. The analysis of electrode impedances complicated by the presence of a constant phase element. J. Electroanal. Chem. 1984, 176, 275−295.

S Supporting Information *

■

Article

ACKNOWLEDGMENTS

G.B. and C.E.A.K. acknowledge support by the Belgian Prodex Office and ESA and long-term structural funding by the Flemish government (Methusalem grant of J.A.M.). This work was in part supported by the Austrian COMET program (Linz Center of Mechatronics). 2588


Article

Langmuir (23) Butt, H.-J.; Graf, K.; Kappl, M. Physics and Chemistry of Interfaces; Wiley: NJ, 2006. (24) Hore, P. J. Nuclear Magnetic Resonance; Oxford University Press: Oxford, 1995. (25) Clément, J.-L.; Ferré, N.; Siri, D.; Karoui, H.; Rockenbauer, A.; Tordo, P. Assignment of the EPR spectrum of 5,5-dimethyl-1pyrrolineN-oxide (DMPO) superoxide spin adduct. J. Org. Chem. 2005, 70, 1198−1203.

2589


Revisiting Silicalite-1 Nucleation in Clear Solution ... - ACS Publications

Recommend Documents