Surface charge dependent structure of ionic liquid/alkali halide interfaces investigated by atomic force microscopy

A new class of ionic salts, ionic liquids (ILs), have been proven to possess considerable potential in electrochemistry, catalysis, solar cell, lubrication applications and also used as green alternative solvents to organic solvents. ^1–5) Furthermore, they exhibit several advantages over conventional solvents because they have remarkable properties such as high thermal and chemical stability, low vapor pressure, and high ionic conductivity. ⁶⁾ Therefore, ILs have attracted great interest in the scientific community. In general, when a liquid is in contact with a solid substrate, the properties of the liquid at the interface are different from those of the bulk liquid. ⁷⁾ The interfaces between the solids and ILs play a crucial role for some of the applications mentioned above. To enhance the efficiency and selectivity of a particular process, detailed knowledge of the solid/IL interface at the molecular level is required for not only a fundamental understanding point of view but also the industrial applications.

Numerous approaches, such as X-ray reflectivity, resonance shear measurements, neutron reflectometry, sum-frequency generation spectroscopy, and scanning probe microscopy, including scanning tunneling microscopy and atomic force microscopy (AFM), have been applied to study the IL/solid interfaces. ^8–11) In particular, AFM is capable of investigating the density distribution of ions composing the IL near the substrate by AFM-based force curve measurement, which revealed the existence of layered structure of ILs, so-called solvation layers. ^11–15) However, there is limited knowledge regarding the interfacial structure between ILs/solid and the effects of the nature of solid substrates on the formation of solvation layers. Thus, this subject still has a demand for discussion.

Frequency modulation (FM) AFM, where an AFM force sensor vibrates at its resonance frequency and the force acting to the sensor is detected as its resonance frequency shift (Δf), has a higher resolution imaging capability than the classical AFM mentioned above. Thus, it is gaining particular interest to achieve true atomic-resolution imaging on various materials in different environmental conditions such as in vacuum, ambient conditions, and in liquids. ^16–19) It can also characterize the density distribution of liquid molecules on liquid/solid interfaces, and the presence of solvation layers, has been revealed. ^20–22) These studies revealed that the FM-AFM technique significantly impacts various research fields like surface science and nanotechnology. However, FM-AFM imaging of the substrates and interfaces in ILs is considered more difficult than in other types of liquid because the quality factor (Q) of the force sensors (Si cantilevers are widely used in FM-AFM) is heavily suppressed due to the high viscosity of ILs. Yokota et al. decreased the sensor noise in FM-AFM by improving the optical beam deflection system and achieved molecular-resolution imaging in IL using a Si cantilever. ¹⁹⁾ In contrast, Ichii et al. took another approach; a quartz tuning fork with a sharpened metal probe, the so-called qPlus sensor, was used as a force sensor. Just a tip apex was immersed into the IL to keep the Q factor high, and atomic/molecular-resolution imaging was successfully achieved. ^23,24)

Ichii et al. also demonstrated the atomic-resolution imaging and structural analysis of an interface between a KCl(100) surface and 1-butyl-1-methylpyrrolidinium tris(pentafluoroethyl)trifluorophosphate (Py_1,4-FAP) IL. ²⁵⁾ The two-dimensional Δf mapping revealed a different interfacial structure from other IL/solid interfaces. This difference was explained from the viewpoint of surface charge, i.e. a KCl(100) surface is electrically neutral, whereas charged substrates, such as mica, were used in other reports. ^23,26)

In this work, we extended our study with this background to understand the effects of surface nature on the interfacial structure. The KBr(100) and (111) crystal surfaces were investigated in a hydrophobic IL, 1-butyl-3-methylimidazolium hexafluorophosphate (BMI-PF₆), using FM-AFM, in which the (100) surface is electrically neutral, and the (111) surface is highly charged. Both the surface structures in BMI-PF₆ IL were characterized on the atomic scale. Since (100) surfaces of alkali halide crystals can be easily obtained by cleavage, many AFM studies have been carried out on these surfaces. ^27–32) On the other hand, their (111) surfaces are highly charged and have high surface energy, which is difficult to prepare. However, Y. Matsumoto's group has recently developed a method to prepare (111)-oriented alkali halide crystals via IL flux and found that BMI-PF₆ is suitable for this process. ^33,34) In addition, the hydrophobicity of the BMI-PF₆ IL minimizes the effect of water molecules even in atmospheric experiments, which is beneficial for AFM analysis. For these reasons, BMI-PF₆ IL was chosen in this study. There have been few examples of AFM observation of alkali halide (111) surfaces, and thus, their atomic-scale structural analysis is both challenging and crucially important in the field of surface science. Furthermore, we carried out Δf versus distance curve measurements on these interfaces and discussed the influence of surface charges on their solvation structures.

We used an FM-AFM instrument based on a commercial AFM (JEOL JSPM-5200), and the original AFM head was replaced by a home-built AFM head for a qPlus sensor. ³⁵⁾ The qPlus sensor was prepared by attaching one prong of the commercially available quartz tuning fork (STATEK Co. TFW-1165) to the substrate with the help of epoxy adhesive. The probe was made from a tungsten wire with a diameter of 0.1 mm (Nilaco Co.), which was electrochemically etched in a potassium hydroxide solution (1.2 mol L⁻¹) and then glued at the other prong of the tuning fork. The resonance frequency and spring constant of the tuning fork before the probe attachment were 32.768 kHz and 1884 N m⁻¹ respectively. After attaching the tungsten probe, the resonance frequency was typically decreased to 13–16 kHz. The qPlus sensor was mechanically vibrated by a piezoelectric zirconate titanate (PZT) plate. The deflection of the qPlus sensors was detected by a transimpedance amplifier with a 100 MΩ feedback register. Then, the deflection signal was amplified by an inverting amplifier with a gain of −10. These electronic circuits were embedded in the AFM head. Δf of the qPlus sensor was detected by a commercial FM demodulator (Kyoto Instruments KI-2001) with some modifications. The vibrating amplitude was kept constant using an amplitude feedback system. Because the vibrating amplitude was detected using root-mean-square (RMS)-DC converter with a bandwidth of more than a few hundred kHz, a band-pass filter was inserted between the transimpedance amplifier and the RMS-DC converter for precise amplitude detection.

All the AFM experiments were performed at room temperature. The AFM topographic images were obtained as the two-dimensional tip trajectories during the tip scanning parallel to the sample surface plane (xy plane), where Δf was kept constant, and were processed by WSxM software. ³⁶⁾ Δf versus distance curves were obtained by changing the tip-to-sample distance without the Δf feedback at a constant xy position. The atomic models of KBr crystals [Figs. 1 and 2(e), 2(f)] were processed by VESTA software. ³⁷⁾

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

BMI-PF₆ IL (Tokyo Chemical Industry Co., LTD, >98%) was purchased and used without further purification. First, we prepared the saturated solution of KBr in BMI-PF₆ IL by adding 1 wt % KBr powder (Nacalai Tesque Inc. >99%). After that, the KBr(100) crystal (Furuuchi Chemical Co.) was cleaved using a sharp knife, and the solution droplet (volume: 0.2 μL) was immediately placed on the cleaved KBr(100) sample with a syringe. This sample containing the IL droplet was kept in a drying chamber (Daikin Industries, ltd., dew point < −50 °C) for 1 day. In contrast, the KBr(111) sample cannot be obtained by cleavage. It was fabricated by Y. Matsumoto's group in Tohoku University using the IL-assisted vacuum deposition method. ^33,34) Then, the KBr(111) sample was transferred to H. Sugimura's group in Kyoto University by post. Note that the surface of the KBr(111) sample was covered with BMI-PF₆ IL, which protected the surface from contaminants. The ion arrangement on the (100) and (111) crystal surfaces of KBr is shown in Fig. 1. Alternate K⁺ and Br⁻ ions are present on the KBr(100) surface, while on the KBr(111) surface either K⁺ or Br⁻ ions are present. The KBr-saturated BMI-PF₆ droplet (0.5 μL) was placed on the KBr(111) sample and left in the drying chamber for several weeks. Finally, both the samples were used for topographic imaging and Δf versus distance curve measurement with the FM-AFM instrument.

BMI-PF₆ IL possesses a much higher viscosity (308 cP) than water (0.89 cP). We measured the Q factor of a qPlus sensor in air and BMI-PF₆, which were found to be 1700 and 55, respectively. Although the measured Q factor was heavily suppressed in the IL compared to air, it was still much higher than that of Si cantilevers in viscous ILs (∼1). ^38,39) The low Q factor of the force sensor increases the frequency noise and reduces the force sensitivity. ⁴⁰⁾ In this study, only the tip apex was inserted into the IL so that the Q factor was relatively kept high and high force sensitivity was achieved. ²⁵⁾ When the tip apex was sufficiently close to the sample surface, the resonance frequency of the qPlus sensor generally changed due to tip-to-sample interaction. The electrostatic and van der Waals forces are suppressed in conductive liquids, and hence positive Δf was usually observed. The Δf of the sensor was responsible for obtaining the topographic images in FM-AFM.

The KBr(100) is known as a natural cleavage plane because it has zero net charges. Figure 2(a) shows a topographic FM-AFM image of (100) surface obtained in BMI-PF₆, in which atomic steps were clearly seen with a step height of ∼0.32 nm [Fig. 2(b)]. This value is in good agreement with the half of the lattice constant a₀/2, where a₀ = 0.66 nm, as shown in the atomic model of KBr(100) [Fig. 2(e)] and also consistent with the previously reported values by several researchers using AFM. ^27,28) As we used the KBr-saturated solution of the IL, the possibility of further dissolution of the KBr(100) crystal in IL was highly suppressed, and hence the stable atomic steps were imaged. In contrast, the KBr(111) surface is electrostatically unstable due to high surface energy and becomes quite difficult to prepare. Y. Matsumoto's group fabricated this sample by growing the KBr(111) microcrystal on α-sapphire(0001) substrate via IL thin film through vacuum deposition method. ³⁴⁾ In our study, we used this KBr(111) sample for the topographic imaging and the interfacial study. We kept the sample in a drying chamber for several weeks because the IL/KBr(111) interface was unstable soon after placing the IL droplet. Figures 2(c) and 2(d) show the topographic image of the (111) surface and the cross-sectional plots obtained along line C–D in Fig. 2(c), respectively. A step structure with a height of ∼0.4 nm was imaged, which was in good agreement with twice of interplanar spacing in the KBr(111) crystal surface, as shown in the atomic model Fig. 2(f). Although we cannot conclude which type of ion (K⁺ or Br⁻) was exposed to the surface, this result indicates that only one type of ion was exposed. From these low-resolution images of the KBr surface, we obtained a preliminary understanding of the surface topography and atomic arrangement.

Furthermore, we have carried out high-resolution imaging on these surfaces in the IL. Figure 3(a) shows an atomic-resolution topographic image of the KBr(100) surface obtained in KBr-saturated BMI-PF₆. The square lattice with a spacing of 0.46 nm was clearly seen, corresponding to the distance between the nearest equally charged ions (theoretically 0.47 nm). The previous FM-AFM studies on ionic crystals, such as NaCl, KCl, and CaF₂, carried out in UHV and aqueous solution demonstrated that either cations or anions are visible in most cases. ^29–32,41) Our previous study on KCl(100) in Py_1,4-FAP IL and KBr(100) in BMI-PF₆ showed the same result. ^25,42) Therefore, the bright spots in the atomic-resolution images could be assigned to K⁺ or Br⁻ ions. Figure 3(b) shows a high-resolution topographic image of the (111) surface of KBr. As shown in Fig. 1(c), on KBr(111), either K⁺ or Br⁻ ions form a triangular lattice with a lattice constant of 0.47 nm. However, the obtained atomic distance was approximately 0.4 nm, revealing the high distortion of atomic arrangement. Figure 3(c) shows another topographic image obtained on a different area on the KBr(111) in BMI-PF₆ IL. It also shows the atomic-scale contrast. The lattice was highly distorted, and the distance between the nearest equally charged ions was ∼0.51 nm, obviously not matching the theoretical value (0.47 nm). The (111) surface consists of alternate anions and cations layers, and hence this surface possesses high surface energy, resulting in a distorted surface by surface rearrangement or reconstruction as schematically shown in Fig. 3(d). This may be one of the reasons why the distance between the closest bright spots did not match the theoretical value. It should be noted that from this experimental result, we cannot deny the possibility that the counterions derived from the IL strongly adsorbed on the KBr(111) were imaged, rather than the outermost ions of the KBr(111) surface. However, even if this were the case, it would be due to the high surface charge density of the KBr(111) surface. Thus, it is reasonable to conclude that the difference in the interfacial structure owing to the difference in the physical properties of the (100) and (111) surfaces was imaged on the atomic or molecular scale. At such interfaces where the surface atom reconstruction occurs, the interfacial solvation structure is also likely to be strongly influenced by the surface. Therefore, we analyzed the behavior of the BMI-PF₆ IL at the IL/KBr(100) and IL/KBr(111) interfaces by Δf versus tip-to-sample distance curve measurement.

Zoom In Zoom Out Reset image size

Figures 4(a) and 4(b) present the Δf versus tip-to-sample distance curves which were obtained at the BMI-PF₆/KBr(100) and BMI-PF₆/KBr(111) interfaces, respectively. At the longer tip-sample distance, no Δf was observed on both the interfaces. On the BMI-PF₆/KBr(100) interface, Δf was positively and monotonically increased without oscillatory profile as the tip approached the surface. In contrast, on the BMI-PF₆/KBr(111) interface, an oscillatory Δf profile was observed. The period of oscillation was 0.65–0.73 nm, which agreed well with the theoretical ion pair diameter of BMI-PF₆ (∼0.7 nm). The previous AFM studies also reported that the detected thickness of the layered structure at solid/IL interface matched well with the theoretical ion-pair diameter of the ILs. ^{11,13–15,26)} This is often explained by a model in which the cation and anion layers of the IL are alternately distributed at the interface. Furthermore, the theoretical study on AFM force curves, including Δf versus distance curves in FM-AFM, also supports this model. ⁴³⁾ Thus, our results indicate that the layered distribution of the ions was present only on the BMI-PF₆/KBr(111) interface. The difference in Δf profiles between BMI-PF₆/KBr(100) and BMI-PF₆/KBr(111) interfaces would be due to the different surface charges. The (100) surface is a natural cleavage plane with zero net charges since the number of cations and anions present on the surface is equal. In contrast, the (111) surface of KBr possesses either cations or anions, which leads to a highly charged surface. This would be one of the reasons for the layered solvation structure formation at this interface. When obtaining topographic images at an interface Δf distance curve oscillates, the probe may scan on the solvation layer instead of the outermost surface of the sample because there exist multiple z-positions corresponding to a certain frequency shift. In other words, the KBr(111) surface may have been scanned in Fig. 3(c), which was imaged with the larger Δf, and the solvation layer may have been scanned in Fig. 3(b), which was imaged with the smaller Δf. However, we cannot conclude this at this stage because Figs. 3(b), 3(c), and 4(b) were all obtained with different probes. Topographic imaging at different Δf and the Δf versus distance curve measurements with the same probe and the same amplitude would be a promising way to evaluate the imaging position.

Zoom In Zoom Out Reset image size

We previously reported two-dimensional Δf mapping on an interface between KCl(100) and Py_1,4-FAP IL, which were composed of 128 Δf versus distance curves. ²⁵⁾ In that experiment, some of them showed a monotonical increase of Δf as the tip approached the KCl(100) surface, which was similar to the BMI-PF₆/KBr(100) interfacial study (this study). It should be noted that some of Δf profiles on the Py_1,4-FAP/KCl(100) interfaces showed saw-tooth like oscillations with a period of ∼0.3 nm, which agreed well to the height of the atomic step of KCl(100) surface, and obviously smaller than the ion-pair diameter of Py_1,4-FAP IL. These oscillations were explained by spontaneous and/or tip-induced dissolution of the KCl(100) surface and were considered unrelated to the interfacial solvation structures. Thus, the results on Py_1,4-FAP/KCl(100) and BMI-PF₆/KBr(100) interfaces both revealed the absence of a layered and alternating cation-anion density distribution at IL/alkali halide(100) interfaces. Recently, FM-AFM studies on water/alkali halide(100) crystal surfaces were reported. Interestingly, the presence of hydration layers at the interfaces was recognized. ^31,32) Moreover, the molecular dynamic simulation study also showed the presence of hydration structure at the interface. ³²⁾ Our experimental results in this study revealed that the mechanism of IL solvation layer formation is completely different from that of water hydration layers on alkali halide(100) surfaces, suggesting that surface charge is the critical factor for the formation of IL solvation layers.

In conclusion, we performed atomic-scale topographic imaging on (100) and (111) surfaces of ionic crystal KBr in viscous BMI-PF₆ IL by FM-AFM utilizing a qPlus sensor. The (100) surface showed a periodic square lattice corresponding well to its atomic arrangement model, whereas the (111) surface showed a strongly distorted atomic-scale contrast. This difference would be attributed to the higher surface charge density and higher surface energy of the (111) surface than the (100) surface. Also, the BMI-PF₆ IL/KBr interfaces for both the surfaces were studied by Δf versus distance curve measurement. The oscillatory Δf profile was only observed on the IL/(111) interface, while the monotonical increase of Δf was observed on the IL/(100) interface. That is, the layered density distribution of the ions was only found on the IL(111) interface, which was also attributed to the high surface charge density of the (111) surface. To the best of our knowledge, this is the first report on atomic-scale imaging on a polar (111) surface of KBr in BMI-PF₆ IL and on surface charge effects on the solvation layers at the interface. The selection of particular solids and liquids and the understanding of their interfacial properties are essential for improving technological processes such as catalysis, lubrication, and electrochemistry. Therefore, we consider the interpretation of the influence of substrate surface nature on interfacial properties to be of great importance, and further knowledge is urgently needed for academic and development of technological processes.

This work was supported by a Grant-in-Aid for Scientific Research B (No. 17H02787) from the Japan Society for Promotion of Science (JSPS). The author H.P.M. is grateful to JSPS for a postdoctoral research fellowship (P 16075).