Anion Intercalation into Graphite Drives Surface Wetting

The unique layered structure of graphite with its tunable interlayer distance establishes almost ideal conditions for the accommodation of ions into its structure. The smooth and chemically inert nature of the graphite surface also means that it is an ideal substrate for electrowetting. Here, we combine these two unique properties of this material by demonstrating the significant effect of anion intercalation on the electrowetting response of graphitic surfaces in contact with concentrated aqueous and organic electrolytes as well as ionic liquids. The structural changes during intercalation/deintercalation were probed using in situ Raman spectroscopy, and the results were used to provide insights into the influence of intercalation staging on the rate and reversibility of electrowetting. We show, by tuning the size of the intercalant and the stage of intercalation, that a fully reversible electrowetting response can be attained. The approach is extended to the development of biphasic (oil/water) systems that exhibit a fully reproducible electrowetting response with a near-zero voltage threshold and unprecedented contact angle variations of more than 120° within a potential window of less than 2 V.


Preparation of the electrodes
HOPG served as the working electrode. Electrical connection was made by stripping an enameled Cu wire (RS components, UK) for about 1 cm at each end and adhering one side to the edge plane of the HOPG with silver conductive epoxy (RS components, UK). After curing for 24 h, the silver epoxy was covered by an insulating resin and left to dry for 3 h. The reference electrode used for the capacitance experiments was a custom-made Ag/AgCl (3M KCl) electrode with an agarose gel frit. Its detailed preparation procedure can be found in the supplementary material of 1 . A Pt wire was once again used as a counter electrode. In all cases, prior to each measurement the potential of the reference electrode was recorded with respect to a commercially available Ag/AgCl (3M KCl) electrode (from Sigma) in a saturated KCl solution to exclude the possibility of potential drifts among different measurements.

Electrowetting setup configuration
The setups used for the liquid|air and liquid|liquid electrowetting experiments are displayed in Figure 1b and c, respectively. Micropipettes were fabricated by pulling a borosilicate capillary (inner diameter 0.84 mm, outer diameter 1.5 mm, length 10.16 cm, from World Precision Instruments, UK) with a Sutter P-97 Flaming/Brown Micropipette puller. The inner diameter of the tip in the resultant micropipettes was ca. 5-6 μm. A microinjector (PV820 Pneumatic PicoPump, from World Precision Instruments, FL, US) was used to deposit a droplet on the surface of HOPG by controlled expulsion of the electrolyte. A platinum wire (99.99% purity, 0.05 mm diameter, from Advent, UK), carefully placed on the upper inner part of the micropipette was used as a counter and pseudo-reference electrode. The position of both the HOPG and the micropipette was controlled using manual micro-positioners (Thor Labs). The micropipette was brought close to the surface of the working electrode and the smoothest regions of the HOPG were targeted. In the case of the liquid|liquid electrowetting experiments, a quartz container was also filled with the surrounding (light) insulating phase. A Photron FASTCAM SA3 high speed camera controlled via Photron FASTCAM Viewer and a Storz Xenon Nova 300 light source were used in static and dynamic (stability tests) experiments. In the case of the stability tests, the frame rate was adjusted to 50 fps.

PTFE cell setup configuration
For the capacitance experiments the setup used is illustrated in Figure 1d The whole assembly was then transferred to an oven where it was kept at 90 o C for 2 h to allow the PTFE gel to cure, followed by cooling at room temperature for at least 3 h and removal of the cylinder.

Synthesis and NMR of 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl) imide
1-ethyl-3-methylimidazolium chloride (75 g, 0.5115 moles) and lithium bis(trifluoromethanesulfonyl) imide (154 g, 0.54 moles) were dissolved separately in water (100ml) using two different flasks. Subsequently, the solutions were slowly mixed and the resultant mixture was heated at 40 ℃ under stirring overnight. After full mixing, the final solution was added to a separating funnel until two distinct phases formed. The bottom phase, containing the ionic liquid and LiCl impurities, was collected, washed with water under stirring (water to ionic liquid volume ratio greater than two) and transferred again to the separating funnel. The process was repeated for at least 10 times to remove LiCl completely. Finally, the pure (free of LiCl impurities) ionic liquid was heated at 70 ℃ under vacuum (6 × 10 -2 mbar) for 3 days. Figure S1 shows the 1 H ( Figure S1a) and 13 C ( Figure S1b) NMR spectra of the synthesized ionic liquid electrolyte. The NMR measurements were recorded using a Bruker Avance II+ 500 MHz NMR spectrometer equipped with a 5 mm Bruker Prodigy Cryo probe. All NMR data were collected at 298 K. For the analysis of the electrolyte, NMR tubes equipped with a coaxial insert filled with a mixture of 10% TMS, 10% TFT and 80% deuterated-DMSO were used in order to lock the magnetic field. The observed 1 H and 13    μL, collected with a calibrated automatic micropipette.

Electrochemical measurements
All electrochemical experiments were performed on an Autolab PGSTAT302N potentiostat from Metrohm equipped with the FRA32 module and operated with Nova 1.11.2 software. Each measurement was conducted on a freshly cleaved HOPG surface. Prior to the experiment, the Pt wire, which served as a counter and pseudo-reference electrode, was flame cleaned with a blue butane flame. To avoid the contamination of the HOPG by the adsorption of air-bound hydrocarbons, the working solution was deposited on the working electrode within 1 min of cleaving the surface. Unless specified otherwise, the applied potential, , throughout the main text is referred vs. pseudo-Pt. Potential values recorded vs.
Ag/AgCl (3M KCl) reference electrode (used in capacitance studies) were converted to the pseudo-Pt potential scale (used in electrowetting experiments), by measuring the potential difference between the two electrodes in each electrolyte used, until equilibrium was reached (i.e., < 100 ). Five different / -1 measurements were performed, and the data were averaged to give the final value. The experimental protocol used for the static measurements involved the application of consecutive potential pulses from 0 V to the desired anodic and cathodic potential limit with a step of either 50 or 100 mV. The duration of the pulses was adjusted to 3 s; this time was found to be sufficient to attain equilibrium. A similar strategy was adopted for the stability tests, in which the potential was directly stepped in between the desired values.
The duration of the pulses was 0.5 s. Each repetition represents one cycle.
The AC electrowetting experiments were performed by applying consecutive DC potential pulses between −0.5 and +1.1 V vs. pseudo-Pt with a superimposed sinusoidal perturbation of 10 mV RMS peak-to-peak amplitude, , at various constant frequencies in the range of 100 kHz to 1 Hz. To probe the effect of the 0 imposed voltage amplitude on the electrowetting response the contact angle changes were monitored at a constant applied bias and frequency with varied . The experimental protocol used for the dynamic 0 measurements using different grades of HOPG, i.e., ZYA and ZYB was composed of a potential pulse from 0 to +1.1 V for the 1 M LiClO 4(PC) |Hexadecane system and from −0.5 to +1.5 V for the 20 m LiTFSI (aq) |Hexadecane system. The duration of the pulse was adjusted based on preliminary experiments to attain a steady state response for the electrolyte used, that is an equilibrium value for the contact angle at each applied bias. The frame rate used was chosen to successfully probe the timescales of the droplet's motion within the timeframe of the experiment, i.e., recording at least 30-50 points between the equilibrium plateaus for the wetting states. The characteristic time for the advancing motion for each electrolyte was estimated adopting the following strategy: the mathematical matrix corresponding to each image recorded during the experiment was subtracted by the matrix describing the image of the droplet at equilibrium exhibiting the highest contact angle (i.e., at 0 and -0.5 V for the non-aqueous and aqueous electrolyte respectively), denoted as the reference matrix. Subsequently, the sum of all elements in each of the resultant matrices was calculated and the data were plotted relative to the time of the experiment based on the number of frames and the corresponding frame rate. Finally, the characteristic times used for the determination of the timescales were taken to be 90% of the corresponding average steady state response.
The latter is shown in the y-axis of Figure S5 as a dimensionless number with no physical meaning. The overall approach is based on the fact that changes in contact angle as a consequence of the applied bias, will be depicted as contrast differences, i.e., a change in the value of the matrix element corresponding to each pixel (e.g., subtraction of two identical images will result in a zero matrix). The larger the contact angle changes, the larger the contrast differences and hence a higher sum is derived. On this basis, monitoring the changes in the sum of the pixel values in each image matrix can provide us with an estimation of the timescales (in complete analogy with monitoring the contact angle changes, see reference 45). Cyclic voltammetry (CV) experiments were carried out in the microdroplets performed in all cases at a scan rate of 100 mV s -1 .

Calculation of capacitance from electrochemical impedance measurements
Electrochemical impedance spectroscopy (EIS) measurements were performed in the frequency range between 20 kHz -1 Hz, using an imposed AC rms amplitude of 7 mV peak-to-peak. The EIS experimental data was evaluated for its compliance with Kramers-Kroning (KK) criteria by fitting the AC response of the system to the admittance representation of a theoretical circuit containing a ladder of n RC elements in series, with an additional capacitance and/or inductance in parallel to the ladder structure, using the S8 software developed by Boukamp. 3 The compliance with KK criteria was assured for all data by the values of the relative residuals, calculated to be less than 0.5 % for both the real and imaginary parts of the impedance and the chi-square parameter which was found to be on the order of 10 -7 for the complete data series. Capacitance was extracted from the EIS data by adopting the graphical approach developed by Orazem and co-workers for systems exhibiting frequency dispersion effects. 4 The value of the constant phase exponent, α, was calculated by performing a linear fit to the plot vs. , where and represent the imaginary part of the total impedance in and the applied frequency in Hz, respectively. The Ω effective capacitance, , was then calculated at each frequency using the following equation: The final capacitance values, , were determined by averaging the obtained values in the frequency range within which variations smaller than 0.2 μF cm -2 were recorded (linear portion of the vs. plot).

Contact angle measurements
Contact angle values were extracted from the recorded images of the droplets using a custom-made

Predicting the electrowetting response using the Young -Lippmann equation
In EWOD, the changes in the macroscopic contact angle with respect to the applied bias are described by  Table S1. This approach involved the following steps: (i) The difference of the cosines between at each applied potential and , , was calculated by inserting in the Y-L equation the corresponding , and values for the electrolytes studied (blue squares in Figures 1c and 4).  where and are the interfacial liquid|air surface tension and the equilibrium contact angle at , respectively. Figure S3: Contact angle variations for EMIM-TFSI and 20 m LiTFSI (aq) in air, after applying a potential pulse at +1.7 V and +0.8 V respectively, for 200 s and subsequently stepping the potential to 0 V. The applied positive potential values correspond to the potential region within which staging intercalation occurs (see Figure 4, Figure 6 and the relevant discussion in the main text; see also supplementary movies #2 and #3 in the SI). The error bars show the standard deviations for three consecutive cycles. Figure S4: Cyclic voltammogram recorded in a sessile droplet of 1 M LiClO 4(PC) deposited on HOPG in hexadecane, using a scan rate of 100 mV s -1 .

Interfacial surface tension and work of adhesion at the liquid|liquid interface
S13 where , are the interfacial liquid|air surface tensions for the liquid 1 and 2 (see Tables S1 and S2) 1 2 respectively and the interfacial liquid|liquid surface tension (see Table S2). Figure S5 presents the contact angle values determined during the forward and reverse scans by sweeping the applied bias from 0 to +1.5 V with varying scan rates. In the case of the 1 M LiClO 4(PC) |Hexadecane system, a fully reproducible response is observed during both forward and reverse scans, suggesting that the scan rate has a negligible effect on the electrowetting response. Considering that the scan rate used extends up to 1 V s -1 , the high degree of reproducibility indicates the fast intercalation/deintercalation kinetics of the ions into HOPG. On the contrary, for the 20 m LiTFSI (aq) |Hexadecane system a clear ClO -4 effect of the scan rate is seen already at the lowest scan rate used, i.e., 100 mV s -1 . It is noteworthy that at 1 V s -1 the change in contact angle is ca. 52 o , i.e., 2.4 times lower than that in the static measurements of Figure 7b, while reversing the scan direction has a negligible effect on the electrowetting response. We ascribe this finding to the surface reconstruction processes occurring during intercalation/deintercalation. It has been already proved that the large TFSIanions cause larger lattice expansion in the galleries of graphite and during advanced intercalation stages defects are also introduced (see Figure 6). The higher the extent of the step edges formation/reconstruction, the more energetically demanding the overall process will be and hence slower kinetics are expected. On this basis, the timeframe of the continuous potential pulse applied during the polarization experiments does not provide the time required for the completion of these processes, which is reflected in the observed contact angle changes with scan rate.  Figure S6 shows the indicative timescales of the advancing motions in the two electrolytes studied. From the recorded data it is evident that the characteristic time for the advancing motion of the droplet in both non-aqueous and aqueous electrolytes is significantly lower for the highest-grade quality HOPG. A closer look on the average timescales for each system reveals the estimated characteristic times to be ca. 67. forces holding together the graphene layers on graphite, in order for intercalation to occur. 11,12 In more detail, the higher number of stacked graphene layers on the larger step edges of the ZYB sample increases the binding forces between the adjacent graphene sheets compared to the ZYA HOPG 13 and therefore more energy is required for the intercalation to commence and subsequently proceed. This is further supported by the observed contact angle changes for the ZYB sample. As can be seen from the insets in  Figure 7 using as a substrate HOPG of (a, c) ZYA (0.4 ± 0.2 o ) and (b, d) ZYB (0.8 ± 0.2 o ) grade quality. The potential was stepped from 0 to +1.1 V vs. Pt pseudo-reference for the 1 M LiClO 4(PC) |Hexadecane system and from -0.5 to +1.5 V vs. Pt pseudo-reference for the 20 m LiTFSI (aq) |Hexadecane system (for details about the experimental protocol see the Experimental Section). The extracted timescales are found to be ca. 67. 2 10 -3 (± 11.03 10 -3 ) and 4.66 (± 0.94) s for the ZYA × × and ZYB HOPG, respectively in 20 m LiTFSI (aq) and ca. 43. 7 10 -3 (± 5.79 10 -3 ) and 688 10 -3 (± × × × 79. 2 10 -3 ) s for the ZYA and ZYB HOPG, respectively in 1 M LiClO 4(PC) . The derived average values × were determined based on three different measurements. Insets (b, d): Droplet images on the ZYB HOPG.

Electrowetting under AC in the biphasic systems
The electrowetting response of the biphasic systems was also investigated under AC conditions based on the experimental protocol described in the SI. Figure S7a shows the dependence of the apparent equilibrium contact angle, , at the HOPG|20 m LiTFSI (aq) interface in hexadecane on the applied AC voltage. From the recorded response it is evident that electrowetting occurs under AC bias, however the overall response is very close to the DC case. Interestingly, no shape oscillations of the droplet arising by the frequency variations on the applied AC voltage pulses were observed, which is in contrast to what is well-established for EWOD systems. 16 Furthermore, as can be seen in Figure S7b, there is no influence of the imposed voltage amplitude, , on the recorded . A possible explanation for these findings is the strong 0 effect of the applied potential on the electrode|electrolyte interfacial surface tension due to the occurrence of the underlying electrochemical reactions (in this case anion adsorption/intercalation). On this basis, the overall mechanism of the phenomenon is predominantly of electrochemical nature and hence any electromechanical contributions, e.g., the effect of Maxwell stress on the three-phase contact line, are expected to be less significant. An additional important factor needs to be considered is that based on the instrumentation used, the AC perturbation was superimposed to a constant DC voltage pulse, i.e., DC current flows through the cell. In other words, the faradaic processes, being responsible for the electrowetting response, still occur in a similar way to the static measurements of Figure 7.