Angelika
Holzinger
a,
Gregor
Neusser
a,
Benjamin J. J.
Austen
b,
Alonso
Gamero-Quijano
c,
Grégoire
Herzog
c,
Damien W. M.
Arrigan
b,
Andreas
Ziegler
d,
Paul
Walther
d and
Christine
Kranz‡
*a
aInstitute of Analytical and Bioanalytical Chemistry, Ulm University, Albert-Einstein-Allee 11, 89081 Ulm, Germany. E-mail: christine.kranz@uni-ulm.de
bCurtin Institute for Functional Molecules and Interfaces, Curtin University, Perth, Western Australia 6845, Australia
cLaboratoire de Chimie Physique et Microbiologie pour les Matériaux et l’Environment (LCPME), UMR 7564, CNRS-Université de Lorraine, 405 Rue de Vandoeuvre, 54600 Villers-les-Nancy, France
dZentrale Einrichtung Elektronenmikroskopie, Ulm University, Albert-Einstein-Allee 11, 89081 Ulm, Germany
First published on 23rd February 2018
The investigation of electrochemical processes at the interface of two immiscible electrolyte solutions (ITIES) is of great interest for sensing applications, and serves as a surrogate to the study of biological transport phenomena, e.g. ion channels. Alongside e-beam lithography, focused ion beam (FIB) milling is an attractive method to prototype and fabricate nanopore arrays that support nanoITIES. Within this contribution, we explore the capability of FIB/scanning electron microscopy (SEM) tomography to visualize the actual pore structure and interfaces at silica-modified nanoporous membranes. The nanopores were also characterized by atomic force microscopy (AFM) using ultra-sharp AFM probes to determine the pore diameter, and using scanning transmission electron microscopy (STEM) and energy dispersive X-ray (EDX) spectroscopy, providing additional information on the elemental composition of deposits within the pores. Si-rich particles could be identified within the pores as well as at the orifice that had faced the organic electrolyte solution during electrochemical deposition. The prospects of the used techniques for investigating the interface at or within FIB-milled nanopores will be discussed.
Besides standard microfabrication techniques i.e. deep reactive ion etching (DRIE)8 for microporous membranes used in ITIES measurements, e-beam lithography with reactive ion etching27 and focused ion beam (FIB) techniques are highly attractive to fabricate solid-state nanopores and nanopore arrays in various 2D materials.28–30 FIB has been used as fabrication route for pore arrays supporting micro- and nanointerfaces milled in thin silicon nitride (SiN) membranes,31 as the pore-to-pore distance can be readily adapted. FIB prototyping has further advantages such as obtaining tunable nanopore array geometries, omitting multiple fabrication steps and modifying commercially available silicon nitride membranes with varying thicknesses. FIB milling is not limited to SiN membranes and other substrates, e.g., porous alumina32 have also been structured. Pores with diameters of 10 nm and less can be fabricated by additional treatments such as low pressure chemical vapor deposition of additional silica to FIB-milled nanoporous SiN membranes33 or by (cold) ion beam sculpting.28,30 FIB milling results in a truncated cone-shaped geometry of the pores due to re-deposition of milled material.28,34,35
Previous results and calculations for ITIES in nanopore arrays produced using e-beam lithography predicted that the pores are filled with the organic phase.36 Hence, the pore walls should be hydrophobic in nature. Experimental evidence is based on cyclic voltammetry, which provides information on the diffusional behavior, and on contact angle27 measurements, which determine the hydrophilicity of both sides of the porous membrane. However, contact angle measurements fail on the nanopore arrays like those investigated in the current study, as this method cannot be used for characterization of the inner pore walls. For alumina membranes, electrochemical studies at the water/DCE interface indicated that the pores are filled with the aqueous phase, through the location of deposited nanoparticles at the interface.37 Similarly, the pores of silicalite membranes were shown to be filled with the aqueous phase, using a facilitated ion transfer process with size-dependent exclusion of a ligand molecule due to the pore dimensions.38 For FIB-milled membranes, implanted Ga+ ions39 may influence the hydrophobicity of the pores’ inner walls, which in turn may influence the location of the liquid/liquid interface within the pore. Hence, exploring the characteristics of arrayed nanointerfaces produced using FIB milling is essential in order to understand the ion transfer at such interfaces. The truncated cone shaped geometry may be used to investigate the location of the interface when either the larger or smaller orifice of the pores is facing the organic electrolyte or aqueous electrolyte, respectively. Hence, pore arrays were FIB-milled from the front-side or back-side of the membrane within this contribution. The ion transfer at the back-side milled arrays with the smaller pore orifices facing the aqueous phase should be decreased due to a smaller interface than the ion transfer at the front-side arrays with larger orifices facing the aqueous phase, as long as the diffusion zones at adjacent pores are not overlapping. Recently, Liu et al. have shown that the diffusional behavior of nanopore arrays with different pore-to-pore spacing can be visualized via the deposition of silica and using AFM-scanning electrochemical microscope measurements.40 Depending on the pore spacing, overlapped diffusion profiles at arrays with smaller pore–pore distance or independent diffusion profiles at arrays with larger pore center-to-center separations lead to a large silica deposit covering the whole array, or individual deposits at the pores, respectively. However, no information on the behavior within the pores nor on the location of the interfaces was retrieved from these studies.
To visualize the pore shape and inner pore space of SiN membrane-supported nanopores, high-resolution transmission electron microscopy (TEM) tomography has been employed.41,42 In contrast to TEM, FIB/SEM tomography43 enables the reconstruction of large sample sections with a resolution down to several nanometers.44,45 FIB-milled cross-sections of nanopores have been published by Liu et al. in order to investigate the shape of the nanopores, especially with respect to electron- or ion beam-induced shrinking.34 Using repeated FIB sectioning and SEM imaging of the freshly prepared face, not only the morphology but also the pore volume and interconnectivity of porous material can be reconstructed, providing three-dimensional information in the micro- and nano-range. 3D FIB/SEM investigations have been demonstrated for biological samples, e.g. cells, providing structural information on membranes,46 for molecular imprinted polymers, for investigating pore connectivity47 and for materials related studies of the inner structure of Al–Si alloys48 or arrays of metal nanotubes.49
The aim of this work was to investigate the location of the interface of FIB-milled nanopores using high-resolution inspection of deposited silica. These studies have revealed that processes occur also within the pores. Within this contribution, we employ FIB/SEM tomography as an innovative method to study the modification of nanopores and specifically the interior of the pores in an ex situ, post-deposition approach. This provides 3D information on the electrochemically deposited silica at nanoporous arrays as solid-state supports for ITIES measurements. Information is also gained about the truncated geometry of the nanopores by 3D reconstructions. For silica deposition, tetraethoxysilane (TEOS) is used as the silica precursor, which is located in the aqueous phase, while cetyltrimethylammonium (CTA+) is the template molecule dissolved in the organic phase.50 The ion transfer of the template CTA+ from the organic to the aqueous phase leads to the condensation of silica on the aqueous side of the interface. The location of the silica might give information about the location of the ITIES during deposition. By forming a solid phase at the interface of the immiscible solutions, the location of this interface can be visualized, as has been shown for Pd deposition at alumina membranes.37 For the silica deposits studied here, the pores’ interior is analyzed using EDX and STEM measurements, an example of which is shown in Fig. 1.
The average pore diameter measured using AFM is 183 nm ± 29 nm (n = 3) for the front-side milled and 72 nm ± 12 nm (n = 3) for the back-side milled arrays, respectively. The pore diameter determined using SEM is on average 151 nm ± 14 nm (n = 100, front-side milled). As the membrane is located on a frame with a 300 μm thickness, FIB milling and SEM imaging of the back-side of the SiN membrane is less accurate, which results in a slightly distorted pore diameter as determined using SEM. Therefore, the diameter determined using SEM for the back-side milled array is not taken into account. In addition, the diameter of the small orifice could not be determined via SEM without destroying the membrane. The overall small discrepancy in average diameter (determined using AFM and SEM) observed for the front-side milled pores may be explained by the limited number of measurements for AFM and, more importantly, by the limitations of SEM in providing pure surface information.44 In addition, TEM tomography also reveals the conical pore shape (see ESI, Movie 1†).
The dimensions of the silica deposits at these two SiN membranes, milled from either the front-side or the back-side of the membrane, should give information about the influence of the FIB milling on the hydrophobicity of the pores. In general, the location of the interface, and hence the hydrophobicity influences the diffusion and the formation of the silica deposits. If the interface is inlaid on the aqueous side (organic phase fills the pore), the diffusion should be more rapid (radial diffusion) compared to a fully recessed interface (aqueous phase fills the pore), leading to linear diffusion. As a result, the silica deposits should be larger for the inlaid case or smaller for the recessed case, given that the pore sizes are the same and the pores are cylindrical. If the pore orifices are different, then the effects will be altered and the diffusion at conical pores is governed by three pore-related factors, which in turn influence the silica deposition: (1) interface location – recessed or inlaid; (2) interface size (radius), and (3) pore shape (cylinder, cone, or inverted cone). In the case, the FIB-milled conical pores are filled with the organic phase as shown for SiN membranes fabricated by e-beam lithography27 and assumed for FIB-milled membranes,31 the interface is located at the orifice on the aqueous side of the membrane. As the orifice facing the aqueous electrolyte differs in size for the front-side and back-side milled membranes (Fig. 2B), the silica deposition should be larger at the front milled pore array compared to the back milled array. In Fig. 3A, the results for the silica deposition at two arrays of 100 nanopores (front and back-side milled) show different sizes of the silica deposits for both approaches, which are randomly distributed. By comparison of the different diameters of the silica deposits, the back-side milled sample interestingly shows deposits with larger diameters of up to 10 μm (Fig. 3B), which is indicative of enhanced ion transport.
In the first step, a paired two-tailed t-test with a 95% confidence level was applied, which revealed that the variation of the silica deposits for front-side and back-side milled arrays is significant. Table 1 shows the diameters and respective standard deviations determined within the individual arrays.
Pore diameter determined using AFM (n = 3; mean value ± SD) | Pore diameter determined using SEM (n = 100; mean value ± SD) | Diameter of silica deposits determined using SEM (n = 100; mean value ± SD) | |
---|---|---|---|
front-side milled | 183 nm ± 29 nm (large orifice) | 151 nm ± 14 nm, SD equals 17% of the mean value | 5.3 μm ± 3.0 μm, SD equals 57% of the mean value |
back-side milled | 72 nm ± 12 nm (small orifice) | 170 nm ± 14 nm, SD equals 16% of the mean value | 7.4 μm ± 3.2 μm, SD equals 43% of the mean value |
The standard deviation of the silica deposits (43% and 57% of the mean value for diameters, respectively) within one array is evidently larger than the difference between the silica deposits at the two arrays. Interestingly, the initial pore sizes of the two arrays had only small standard deviations (16 and 17% of the mean value, respectively). As a result, an unambiguous interpretation of the location of the interface is challenging. It has been shown that osmotic pressures or double layers within the nanopores influence the ion transfer at truncated pores, even for miscible solutions. This also leads to an ion current rectification depending on the ion flow direction within the truncated cone-shaped geometry.56 In the next step, FIB/SEM tomography was evaluated as method to obtain information about silica deposits. In principle, silica is formed in the aqueous phase,50 following the electrochemical ion transfer of the template CTA+ from the organic to the aqueous phase. TEOS, which is the precursor of silica via hydrolysis and condensation, is only present in the aqueous phase.
An example 3D reconstruction of a FIB/SEM tomography is depicted in Fig. 5 with a corresponding series of SEM images from the tomography stack (Fig. 5A, I–IV; the complete tomography stack is given in Movie 2†). The 3D reconstruction (Fig. 5B) reflects the residue, which is located around the pore orifice facing the organic electrolyte during electrochemical deposition, and an inhomogeneous structure inside the pore partially connected to the silica deposit. The drop-like structure below the membrane shows a high porosity, which seems also to be encapsulated by a thin layer (colored in yellow in the 3D reconstruction shown in Fig. 5B). This layer is visible across the entire membrane on the organic facing side, whereas the drop-shaped residue is just located close to the nanopore. Differentiation of independent structures or particles in the SE contrast is difficult. Also, discrimination between the in-plane area and the sample volume is not possible at these dimensions in the nanometer range due to the penetration depth of the electron beam.44 However, the conical pore shape is clearly evident within the 3D reconstruction. Further information about the elemental composition of the deposit at the organic electrolyte facing side was obtained using EDX mapping from a thin TEM foil (for details see Fig. S2 and S3†).
Fig. 5 Extracted SEM images of a FIB/SEM tomography stack showing the deposit on the nanopore and the deposit that was formed at the orifice facing the organic side. The distance between single slices is 30 nm, the acceleration voltage is 5 kV and the current is 86 pA, depicted at a 38° tilt (A, I–IV). A 3D reconstructed pore showing the deposits located at the organic facing side and within the nanopore (B) according to the entire stack, which is given in ESI Movie 2.† |
Fig. 6 TEM images of a nanopore (A) and a magnified view of the pore. A SEM image of the TEM foil (B), vertically flipped by 180°; and EDX mapping of the area; false color image showing carbon in green and silicon in red (B, right). Complete data for the EDX mapping is given in the ESI, Fig S3.† |
It is necessary to discriminate between signals of the observed in-plane area exposed from FIB milling and the sample volume, as EDX and SEM are sensitive towards the subsurface composition with excitation depths larger than the thickness of the investigated TEM foils.44 As shown in the corresponding SEM image (Fig. 6B), the nanopore is not visible but it is contained within the sample volume of the TEM foil. Therefore, the EDX signals within the pore are overlaid by the signal of the SiN membrane and no information is gained about the elemental composition of particles and residues located within the pore, which are visible in the TEM image (Fig. 6A).
In case of small pore diameters (72 nm ± 12 nm), such as for the pore shown in Fig. 6, it is rather challenging to prepare a TEM lamella that contains a single pore, which is open from both sides, as illustrated in Fig. 7. The achievable thickness of the TEM foil is limited57,58 and the minimum thickness of the TEM samples fabricated within this study was approximately 150–200 nm. To overcome these limitations, nanopore arrays with pore diameters of 250–350 nm were FIB-milled, which is sufficient to prepare a TEM foil containing a single pore that avoids the problem of convolution of the EDX data of the pore content and the SiN membrane. These nanopore arrays were FIB-milled from the front-side of the membrane (thickness of the SiN membrane: 50 nm); this front-side was subsequently in contact with the aqueous electrolyte solution during electrochemical deposition. Two example nanopores of the investigated arrays are shown in Fig. 8 after silica deposition. The electrochemical formation of silica was performed at pH 9 (Fig. 8A and B) and also tested at pH 3 (Fig. 8C) using different deposition times at the two pH values.
Fig. 8 SEM images of the deposits formed at pH 9 (A, and magnified view shown in B, applied potential for deposition is −0.1 V for 60 s) and at pH 3 (C; applied potential is −0.1 V for 30 s), with an acceleration voltage of 5 kV and current of 86 pA, depicted at a 38° tilt. A 3D reconstruction of the FIB/SEM tomography of the silica formed at pH 9 (SEM images shown in A and B are single slices of this stack) is depicted in (D). The entire stack is given in the ESI Movie 3.† |
At pH 9, the transport of CTA+ is strongly facilitated and highest as the polynuclear species (Si4O6(OH)62−) and (Si4O8(OH)44−) are predominant in the aqueous phase.59 Hydrolyzed TEOS polycondensates around CTA+ ensembles to form the silica. At pH 3, mononuclear species of Si(OH)4 are dominant, which may not facilitate the transfer of CTA+ to the aqueous phase and therefore silica deposits should be absent at or within the nanopores.52 To exclude the idea that the observed Si-rich residue within the pore or at the organic electrolyte facing side is related to silica removed from the deposits during rinsing, cleaning or drying, experiments were performed at low pH. As expected, the electrochemical deposition at pH 3 (Fig. 8C) does not lead to silica formation as observed at pH 9 (Fig. 8A and B). However, some pores are filled with solid particles (2 of 3 measured pores contained solid material) embedded in the residue filling the nanopore (Fig. 8C). In Fig. 8A and B, a randomly mixed phase obviously composed of different materials, according to the contrast of the SEM image, is visible. Single particles are labeled as green in the 3D reconstruction (Fig. 8D and Movie 3 in the ESI†). The black areas in the reconstruction reflect “holes” within the residue. The entire residue, which cannot be identified as holes or particles due to insufficient contrast in the SE image, is depicted in yellow.
Also, a nanoporous array, which was used in experiments with an acidic aqueous phase, was investigated. Although, it is not expected that silica deposits will be formed at pH 3, a modification of the pores was observed. After preparing a TEM foil with a pore as schematically shown in Fig. 9A–C, EDX analysis was performed.
The EDX spectra shown in Fig. 9 clearly reveal that the bright spot recognizable in the STEM image (Fig. 9D, marked with a blue square) consists of Si. Other areas, e.g. that marked with a yellow square, mainly consist of oxygen and carbon. In both spectra, the detected Cu signal is related to the copper TEM grid. As no silica formation should be observed at pH 3, these Si-rich particles might be remains of the precursor TEOS. Another reason for these particles within the pores might be related to pre-concentration effects of the non-condensed Si(OH)4 during the removal of the electrolyte solution or drying, leading to a reduced volume and the formation of silica. If we assume that these particles are related to the aqueous phase with dissolved non-condensed Si(OH)4, the question arises as to why the Si(OH)4 containing phase is obviously also below the pore separated from the silica deposit and not condensed to silica, indicated by the Si-rich phase in the EDX map shown in Fig. 6. In general, the condensation of silica is faster in the presence of a template, hence Si(OH)4 should be condensed to silica directly in the ITIES at a suitable pH value. In close proximity of the silica deposits, these single Si-rich particles can be residues flushed or moved inside the pore while drying or cleaning. However, this would not explain the large residues observed at the organic electrolyte facing side, which are visible at most investigated pores and whose dimensions were significantly larger in the sample with small pore diameters of 72 nm ± 12 nm (n = 3) (Fig. 10A and B). It is unlikely that such particles were accidentally moved through the pore to the organic facing side of the membrane during cleaning or rinsing. Hence, it is hypothesized that these particles were formed either during electrochemical deposition or due to a pre-concentration of TEOS in the post-deposition steps as described above. These residues are also evident at pore arrays with large diameters of 250–350 nm, as determined using AFM and SEM (Fig. 10C, AFM results not shown).
These observed particles may also be related to the silica forming chemistry. The hydrolysis of TEOS to silica forms ethanol as a side product.60 Ethanol is soluble in both water and DCE, which may lead to a mixed layer. To avoid the formation of that mixed layer at the interface, ethanol resulting from the hydrolysis was evaporated. Hence, these particles can be related to the aqueous phase located within the nanopores during electrochemical measurements or, less likely, be related to displaced particles moved into the pores during cleaning and drying.
In general, direct characterization of the liquid/liquid interface is a challenging task, and FIB/SEM tomography appears to be a suitable technique to visualize nanopores, and therefore to fundamentally understand the processes that may lead to the modification of such pores. Evidently, FIB/SEM is an ex-situ approach, which requires that the samples need to be transferred into a vacuum. Hence, for the visualization of immiscible solutions it is anticipated that future studies will take advantage of cryogenic FIB/SEM tomography at such pores.61 In the presence of an electrolyte or an analyte only in one phase (e.g. metal ions dissolved in one of the immiscible electrolyte solutions), it is anticipated that sufficient contrast will be generated for direct visualization of the interface.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8fd00019k |
‡ Presenting author. |
This journal is © The Royal Society of Chemistry 2018 |