Ultraslow Dynamics at a Charged SiliconâIonic Liquid Interface

Subscriber access provided by UB + Fachbibliothek Chemie | (FU-Bibliothekssystem)

Article

Ultra-slow Dynamics at a Charged Silicon-Ionic Liquid Interface Revealed by X-ray Reflectivity Miaoqi Chu, Mitchell Miller, Travis Douglas, and Pulak Dutta J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.6b10443 • Publication Date (Web): 26 Jan 2017 Downloaded from http://pubs.acs.org on January 26, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry C is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 16

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Ultra-slow Dynamics at a Charged Silicon-Ionic Liquid Interface Revealed by X-ray Reflectivity

Miaoqi Chu, Mitchell Miller, Travis Douglas and Pulak Dutta* Department of Physics and Astronomy, Northwestern University, Evanston, Illinois 60208

Abstract

The interfacial structure of the room temperature ionic liquid methyltrioctylammonium bis (trifluoromethylsulfonyl)-imide ([MTOA]+[NTF2]-) near a silicon electrode was investigated using specular X-ray reflectivity. Using this technique, we have previously observed “crowding”, i.e. formation of a thick anion layer on a positively charged electrode. We now report that this layer develops over timescales in the range of ~400 sec -1100 sec. This is different from the timescales reported in other experiments, and is inconsistent with most theoretical predictions. A tentative explanation is proposed which assumes that the formation and dispersion of the crowding layer requires collective re-ordering of anions/cations through the electrochemical cell. We suggest that because of the presence of multiple timescales in these systems, the observed timescales will vary depending on the timescale of the measurement.

E-mail: [email protected]

ACS Paragon Plus Environment

1


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Introduction

The next generation of electrochemical energy storage devices, including batteries and supercapacitors, will ideally require not only high energy density but also high power density and fast charging rate1. Room temperature ionic liquids (RTILs) have been considered as novel electrolytes for such applications, due to their high charge density, wide electrochemical windows (~5V wide compared to 1.23V for water)2-3, high thermal stability, low flammability and non-volatility. However, they are handicapped by their slow dynamics, including high viscosity, low electrical conductivity and small diffusion coefficient4-5. An understanding of the nanoscale factors leading to these slow processes may lead to ways of selecting or designing molecular ions to optimize the electrochemical behavior of RTILs. A number of theoretical papers

6-9

have addressed the origins of the observed slow

dynamics in RTILs. The charging time scale for a capacitor C with resistance R is τ = RC, which is of the order of a millisecond in these systems. However, this timescale does not generally fit the observed response, and in those cases the limiting factor is the motion of ions within the RTIL. It is not clear how these should be described; for example, are there only local rearrangements of the ions, or do ions travel over distances comparable to the size of the electrochemical cell? Several theoretical predictions will be discussed and compared to our experimental data later in this paper. There have also been a number of experimental studies of interfacial dynamics in RTILs. Uysal and coworkers performed an X-ray reflectivity study on a [C9mim]+[NTF2]- –graphene interface

10-11

, and discovered two processes associated with double layer formation, with time

constants 1 sec and 10 sec respectively. Both of these are much longer than RC. Roling and


2

Page 2 of 16

Page 3 of 16

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


coworkers studied the [BMPyrr]+[FAP]- –gold (111) electrode interface with electrochemical impedance spectroscopy. The complex capacitance plot shows two distinct arcs with frequency of 1MHz-20Hz and 20Hz – 0.1Hz12. The fast process is of the order of RC, while the slow process is order of 1s. Similar results were also reported 13 in [Py14]+[FAP]- and [EMIM]+[FAP]-. Nishi and coworkers studied the [TOMA]+[C4C4N]- –gold interface with surface plasmon resonance (SPR)14. When a potential was applied to the gold electrode, the SPR signal took ~100 sec to stabilize. They also found that the response to increasing and decreasing the potential had slightly different time scales, which was attributed to the asymmetrical anion/cation. In a study of [BMIM]+[NTF2]- at passivated silver surface with no applied voltage, the second harmonic generation technique revealed that the surface kept changing for 20-50mins15. During our own previous study16, we noticed that the crowded layer of anions took time to form, but once formed the layer was stable. The data we reported were for the time-independent state, collected at least ~1200 sec after a voltage was applied or changed. It has also been reported that the length scale of RTIL ordering next to a surface can be 10-60nm15, 17-18, which is far beyond the width of a double layer. The large time scales and length scales indicate that the RTIL-electrode system cannot be simply treated like a traditional electric double-layer. Instead, its response to electric fields involves complex dynamics that requires careful study. However, measurements such as capacitance or SPR, while detecting time dependence, do not identify the nanoscale origins of the time dependence. In the present study, we measured the time dependence of the ‘crowded’ RTIL layer near an electrode surface. Materials and Methods Methyltrioctylammonium

bis(trifluoromethylsulfonyl)-imide


3

([MTOA]+[NTF2]-)

was


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

purchased from IoLitech (purity 99%). To remove water and other dissolved gases, it was placed in a vacuum oven for 3 days at room-temperature. Conductive silicon chips were processed with rapid thermal treatment at 1100 °C for 300s, which resulting a layer of silicon oxide about 460Å, measured by ellipsometry. This thermal oxide layer was removed using buffered oxide etching

Fig. 1. (a) Schematic diagram of the transmission cell used in this study. The cell body and window frames are made of Kel-F. The two yellow films are Kapton windows. The air-tight seal is achieved by pressing the frames with o-rings against the cell body. (b) A schematic diagram of the experimental geometry. The conductive silicon (red) is the working electrode while gold wires (yellow) are the counter-electrode and pseudo-reference electrode. X-rays are reflected from the silicon-RTIL interface. solution (BOE) to expose a fresh silicon layer. The silicon chip was then mounted to a transmission cell, reported previously16. Gold wire twisted into a coil with effective surface area of 1.5 times the silicon chip was used as the counter electrode, which was placed about 3mm away (a distance that is 6 times the X-ray beam size). The interfacial structure caused by the external electrical field (data shown later) is less than 100Å, many orders smaller than the X-ray beam size as well as the gold wire radius. Thus the electric field introduced by our setup can be safely considered as uniform. A schematic diagram is shown in Figure 1. Specular X-ray reflectivity measurements were conducted at Beam Line 12BM-B, Advanced Photon Source, Argonne National Laboratory. The incoming X-ray beam had energy 19.5 KeV


4

Page 4 of 16

Page 5 of 16

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


and was focused to 0.5mm x 0.5mm, reflected from the RTIL-silicon interface, and collected by a Pilatus 100K detector. Corrections for variable footprint, attenuation and variable incoming flux, as well as subtraction of off-specular background, were performed to obtain the specular reflectivity data. Slab model modeling of the RTIL-silicon interface was calculated with Parratt’s recursive method and fitted to the experimental data 19-20.

Results and Discussion 1. Cyclic voltammogram (CV)

Fig. 2. Cyclic voltammogram for [MTOA][NTF2] recorded at sweep rate10mV/s. The potential was measured against the Au wire pseudo-reference electrode (see Fig. 1).

We first determined the electrochemical window, i.e. the voltage range in which the generation of electrolysis products can be ignored. This window is also a qualitative test of the existence of impurities in the RTIL. The CV curve (Fig. 2) is typical for an RTIL, essentially flat at low voltages and increasing sharply when the potential becomes either too high or too low. If the threshold current density is chosen to be 0.025mA/cm2 (which is below the thresholds commonly used21-22, 0.1-1.0 mA/cm2) the cathode and anode limit are -3.2V and +2.4V


5


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

respectively, or 5.6V for the width of the electrochemical window. This is roughly in agreement with number reported by Ueda and coworkers23 in their study of the same RTIL. The potential range during our experiment was -2.20V to +2.35V, i.e. within the electrochemical window.

2. Time dependent structure of the crowding layer for an increment in potential. When a high potential is applied, the anions will come close to the surface and form a crowded layer. This process requires the reorganization of anions/cations within a distance from the electrode surface of several times the molecular dimensions. We previously reported that the data were stable after 20 mins for [TDTHP]+[NTF2]-, but the time dependence was not studied in detail at the time16. We have now recorded the XRR data at various stages during the formation of the crowded layer: these are shown in Figure 3. Each XRR measurement takes about 5mins, thus much faster dynamics are not captured. However, the time resolution of the XRR measurement is somewhat less than 5 mins because the higher-q data take the most time to collect; the fitting of XRR data is more sensitive to the dips and peaks, and that region of each scan is collected in less than 2 min. The XRR scan for the initial state at V=0 is featureless in the observable momentum transfer range (0-0.3Å-1), meaning that there is no observable structure normal to the interface within the liquid near the interface. When a voltage of 1.55V is applied (Fig. 3a), an oscillating pattern in the reflectivity develops in the second scan, increases in amplitude and shifts towards lower q at the next scan. This indicates that a layer at the interface is growing in thickness. The XRR patterns of the 4th and 5th scans are almost identical, indicating the formation of the crowded layer has been completed. This oscillation in the reflectivity curve is due to the formation of a crowded layer on the


6

Page 6 of 16

Page 7 of 16

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Fig. 3. In each of the three data sets (a)-(c), the left panels show reflectivity data (solid dots) at different times and best fits with one-slab model (lines through data). The times given in the labels are the starting times of each scan. The right panels are the normalized electron density profiles corresponding to the best fit. The labels at top give the change of voltage after which the time-dependent data were collected.

surface of silicon that is enriched in anions (which have higher electron density) and thus create the electron density contrast seen by X-rays. The simplest model for fitting this pattern is a oneslab model in which we assume that the interfacial layer is uniform, with electron density (ρ), thickness (W) and roughness for the two interfaces, while the silicon substrate and bulk RTIL are modeled as semi-infinite media with constant electron density. The crowded layer is expected to have a uniform density due to the 'lattice saturation' effect24. The roughness at the interface of the interfacial layer and bulk RTIL characterizes the diffusion properties. Fitting the data with a oneslab model (Figure 3a) shows the interfacial layer increases from 25Å for the first scan to 33Å for the last scan. It should be noted that a one-slab model, like any model, is an approximation to reality. More complex models could be used, but the use of models with more adjustable parameters cannot be justified when our simpler model with fewer parameters gives a very good fit to the data (Occam’s Razor). Uniform density slabs are analogous to pixels in digital images; in this case, the spatial resolution of our technique, while better than any other applicable technique, justifies only a single-pixel ‘image’. The time dependence of W can be fitted with an exponential decay function, =


7


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

A exp t/τ. However, the initial value, 0, is below the resolution limit (as represented by the large error bar in Fig. 3a). 0 = 0Å if the RTIL near the surface is no different from the bulk liquid at t=0. On the other hand, perhaps 0 = 10Å in the unlikely event that a monolayer of anions (too thin to be seen given the range of our reflectivity scans) has formed on the silicon surface even before the voltage is applied. In Figure 4(a), fitting using the two extreme cases yields time constant τ of 4. 5 10 sec for 0 = 0 Å and 6.0 10 sec for 0 = 10Å . These numbers, while different, are of the same order of magnitude, the significance of which will be discussed below. The electron density enhancement above the bulk liquid electron density may be increasing slowly with time in Fig. 4(a), but the error bars are large and it could also be constant. In fact, the density enhancement is almost the same even at

Fig. 4. Fitting parameters of the interfacial layer as functions of time, for the three cases in Fig. 3 (a)-(c). Upper graphs show layer width; lower graphs show electron density enhancement above the bulk RTIL electron density. The layer width data are fitted with exponential functions or other functions as described in the text. In Fig. 4(c), the dashed blue line is a fit to exponential decay, while the solid red line is a fit to a logistic function (see text).

higher potentials (Fig. 4(b)-(c)), suggesting that ~25% is the saturated density enhancement of the crowded layer.


8

Page 8 of 16

Page 9 of 16

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Time dependent XRR with a much smaller potential increment, a change from 2.05V to 2.20V, shows similar behavior (Fig. 3b). The stabilized interfacial layer width at 2.05V was obtained as before by repeating the XRR scan multiple times until the data stopped changing. It takes 620 sec to add 4Å and another 3000 sec to add another 4Å. The fitting yields a somewhat longer τ≈10 sec (Fig. 4(b)).

3. Time dependent structure of the crowding layer for a decrement in potential. To exclude the possibility that the interfacial layer is just an irreversible electrolysis product that develops with time, XRR measurements were performed when the potential was decreased from 2.35V to 1.55V, shown in Figure 3(c). The reflectivity oscillation pattern shifts towards higher q, indicating thinning of the crowding layer with time. The W-t relationship can be fitted with an exponentially decaying function, which gives us a time constant of ~2500 sec. However, as shown in Fig. 4(c), it is much better fitted 25 with =

!"# $$% /&

+(

(1)

A and B are the initial and final width while defines the value of t at the center of the logistic function. The best fit parameters are A=66Å, B=37Å, = 2 10 sec, * = 6.0 10 sec. This function was selected because its shape is a good match to the observed data (Fig. 4c). This may imply that that the decrement and increment in potential, corresponding to the thinning and thickening of the crowding layer respectively, have different kinetics. For example, thinning may require the anions to detach from the crowding layer first.

Discussion There is no generally accepted theory of the ion dynamics in an RTIL, but as noted earlier,


9


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 16

there are several theoretical predictions. In the following, +, is the Debye length, L is the size of the electrochemical cell (distance between anode and cathode) and D is the diffusion constant. Two obvious timescales are *, = +, /- and *. = / /-. These are the time constants associated with diffusion over very different length scales: *. is the time constant for diffusion across the macroscopic length of the cell, whereas *, is associated with local charge redistribution. In the dilute electrolyte case, Bazant et al6 argue that the harmonic mean * = +, //- (which equals RC, see Ref. 6) is the primary timescale for diffuse-charge dynamics. (This limit does not apply to RTILs unless they are weakly dissociated, as has recently been claimed.26) For a typical RTIL system, +, ≈ 1Å, and - ≈ 10 μ2 3 . Taking L=3mm, as in our own cell, we get as order-ofmagnitude estimates * ≈ 323 and *. ≈ 9 106 378. Of course *, is far smaller. Comparing to our own results, the differences in the time scales observed in different measurements, or the differences resulting from different fitting assumptions, are small in relation to the order-ofmagnitude inconsistency: * and *, are far too small to fit our data, while *. is too large. Zhao8 proposed a modification to account for the strong electrostatic correlation in RTILs, /

/

leading to a time constant *9 = * :+, ⁄+ = +, //-+ , where + is the typical length of charge correlations. Setting + to be the size of anion/cation of RTIL (~10Å, this would mean that we estimate *9 ~0.323, also inconsistent with our results. If + is larger, the timescale is even smaller. Lee et al.7 have developed a lattice liquid model for ionic liquids, and used it to perform ‘numerical experiments’ yielding interface charge vs time. They see a rapid initial rise in the integrated charge for the half-electrochemical cell, followed by a slow decay. The simulations when performed for different + and +, are consistent with a decay time constant


10

Page 11 of 16

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


*.9

=

B

> C *. = >? A @

=

B

.C >? C = A , >@

(2)

This expression follows from the scaling behavior of the model results of Lee et al. However, our order-of-magnitude estimates, again taking + to be 10 Å, gives us *.9 ≈ 28003. In other words, with these estimated numbers, τD

Ultraslow Dynamics at a Charged SiliconâIonic Liquid Interface

Recommend Documents

Ultraslow Dynamics at a Charged SiliconâIonic Liquid Interface