Early Stage Anodic Instability of Glassy Carbon Electrodes in

Subscriber access provided by RYERSON UNIVERSITY

Article

Early Stage Anodic Instability of Glassy Carbon Electrodes in Propylene Carbonate Solvent Containing Lithium Hexafluorophosphate Emily V Carino, Daniel J Newman, Justin G. Connell, Chaerin Kim, and Fikile R. Brushett Langmuir, Just Accepted Manuscript • DOI: 10.1021/acs.langmuir.7b02243 • Publication Date (Web): 19 Sep 2017 Downloaded from http://pubs.acs.org on September 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.

Langmuir 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 28

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

Langmuir

Prepared as an Article for publication in Langmuir

Early Stage Anodic Instability of Glassy Carbon Electrodes in Propylene Carbonate Solvent Containing Lithium Hexafluorophosphate Emily V. Carinoa,c,*, Daniel J. Newmanb, Justin G. Connella,c, Chaerin Kimb, Fikile R. Brushett a,b,* a

Joint Center of Energy Storage Research, Argonne National Laboratory, Argonne IL 60439 Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge MA, 02139 c Materials Science Division, Argonne National Laboratory, Argonne IL 60439 * to whom correspondence should be addressed: [email protected], [email protected] b

Abstract

Irreversible changes to the morphology of glassy carbon (GC) electrodes at potentials between 3.5 - 4.5 V vs. Li/Li+ in propylene carbonate (PC) solvent containing lithium hexafluorophosphate (LiPF6) are reported. Analysis of cyclic voltammetry (CV) experiments in the range of 3.0 V to 6.0 V show that the capacitance of the electrochemical double-layer increased irreversibly beginning at potentials as low as 3.5 V. These changes resulted from nonfaradaic interactions, and were not due to oxidative electrochemical decomposition of the electrode and electrolyte, anion intercalation, nor caused by presence of water, a common impurity in organic electrolyte solutions. Atomic Force Microscopy (AFM) images revealed that increasing the potential of a bare GC surface from 3.0 V to 4.5 V resulted in a 6× increase in roughness, in good agreement with the changes in double-layer capacitance. Treating the GC surface via exposure to trichloromethylsilane vapors resulted in a stable double-layer capacitance between 3.0 and 4.5 V, and this treatment also correlated with less roughening. These results inform future efforts aimed at controlling surface composition and morphology of carbon electrodes.

ACS Paragon Plus Environment

Langmuir

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 This paper describes the evolution and control of the electrochemical double-layer capacitance (CEDL) and underlying morphology of glassy carbon (GC) electrodes at positive potentials in several O2-free nonaqueous electrolytes. We observed significant and irreversible changes to the CEDL and surface structure of GC due to microscopic roughening after poising the electrode at potentials between 3.0 V and 5.0 V vs. Li/Li+ within a solution of propylene carbonate (PC) containing 1.0 M lithium hexafluorophosphate (LiPF6), a common Li-ion battery salt. The evolution of GC surface structure exhibited electrolyte dependence, and roughening was attenuated by treating the GC surface with silanes. Whereas the oxidative electrochemical stability limit of carbon electrode materials is typically noted at potentials more positive than 4.5 V vs. Li/Li+ in both nonaqueous and aqueous solutions,1 the morphological and interfacial chemical evolution reported herein became evident at potentials as low as 3.5 V, and in the absence of bulk solution decomposition, or other detectable faradaic reactions. Conductive carbon materials are commonly employed as current collectors, active electrodes, and conducting supports or additives in a broad range of electrochemical energy conversion and storage devices including fuel cells, alkali ion batteries, metal-air batteries, redox flow batteries, and electrochemical capacitors. For these applications, in general, it is desired that the carbon materials remain electrochemically and chemically inert and promote or otherwise do not hinder electrochemical reaction(s) or interest. Accordingly, methods to stabilize carbon


Page 2 of 28

Page 3 of 28

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

Langmuir

electrode materials within relevant electrochemical environments are of scientific technological and importance. Coordinatively unsaturated carbon atoms on the surface of a variety of electrode materials react readily with oxygen and water in ambient and electrochemical conditions, but the nature of these reactions depends on the specific physiochemical properties of the carbon electrode, such as the surface morphology and the chemical composition of the electrodeelectrolyte interface.1–4 Although the standard electrode potential for the electrochemical oxidation of carbon to CO2 is only ca. 0.2 V vs. RHE,5 the practical potential limit of electrooxidative stability is higher, typically due to slow activation kinetics or self-limiting surface reactions which attenuate electro-oxidative decomposition of the electrode and preserve electrochemical activity.1,6 For carbon electrodes in aqueous electrolyte, the electro-oxidative stability limit is typically noted at potentials positive of 1.5 V vs. RHE (approximately equivalent to 4.5 V vs. Li/Li+).1,3,4,7,8 Electro-oxidizing GC at 1.8 V (vs. RHE) in sulfuric acid (H2SO4) results in corrosion via electrochemical oxidation of carbon to CO2, and commensurate formation of a graphitic oxide surface layer.7 Moreover, electro-oxidizing GC electrodes in the presence of oxygenated organic species can result in increased microscopic surface area due to chemisorption and functionalization of the GC surface.7 The quantity of graphitic oxides generated during the electro-oxidation of pyrolyzed photoresist electrodes, a material with electronic and morphological characteristics similar to GC, also increased with alkali cation size, and graphitic oxide production may be related to anion oxidation.8 In addition to effects on apparent electrochemical reactivity of the carbon electrode-electrolyte interface, the electrolyte composition can also impact the nature of the surface composition and the formation of stable oxide layers. For example, electrochemically generated graphitic oxides, produced during


Langmuir

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

electro-oxidation at potentials above 1.5 V vs. RHE, are shown to be more stable in acidic aqueous solutions,1,7,9 where they form an intact oxide layer, yet these oxide layers may exhibit greater solubility in basic solutions, resulting in dissolution.2,7 The electrochemical stability of carbonaceous electrode materials in nonaqueous electrolytes is particularly important for use in energy conversion and storage devices. Carbon black, an amorphous carbon powder, is often used as a conductive support in the positive electrode in Li-ion batteries, and instability due to unwanted reactions contributes to reduced battery performance and lifetime. In particular, electro-oxidation of both the carbon electrode material and the organic solvent into CO and CO2 gases,10 and anion intercalation into graphitized carbon black have been observed at potentials above ca. 4.0 V vs Li/Li+,11,12 (approximately equivalent to 1.0 V vs. RHE). Alkoxy functional groups can form on GC surfaces during electrochemical anodization in various environments.9,13 Increases in GC surface functionalization at anodic potentials can lead to changes in the capacitance of the electrochemical double-layer due to changes in microscopic roughness as a result of organic and electrolyte adsorption.7 Subsequently, these surface oxides can participate in additional electrochemical reactions, as well as adsorption / desorption reactions, resulting in uncontrolled changes to the composition and properties of the electrode, electrolyte, and the interface. This, in turn, may convolute the electrochemical response and complicate interpretation of data. Removing surface oxides from GC electrodes has been shown to improve the stability of the electrode surface, preventing functionalization during some electrochemical reactions in electro-oxidative conditions, but this requires heating the electrode to temperatures above 500 °C.14 Moreover, the reduction of these GC surface oxides can alter the electrochemical properties of the electrode, for example, slowing reaction kinetics of some classes of reactions.1,3 Chemical treatment of GC electrodes with organosilanes, resulting in


Page 4 of 28

Page 5 of 28

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

Langmuir

surface oxide bound silicon (Si) species that are resistant to removal via hydrolysis, has also been explored as a method for controlling electrode surface composition and reactivity for applications in electrochemical sensing.15,16 While silanes are commonly employed to reduce corrosion of metal oxides, their utility in preventing oxidative deterioration of electrochemically active carbon materials has yet to be explored. The purpose of this study is to determine the nature of the earliest stages of instability of GC electrodes in air-free, nonaqueous environments. Our results show that the electrochemical double-layer capacitance of bare GC electrodes increases irreversibly at potentials between 3.5 V and 4.5 V in solutions of PC containing 1.0 M LiPF6, due to increased microscopic roughening. This finding is especially interesting because the change in morphology does not appear related to the presence of oxygen or water, and does not follow any obvious chemical transformations, such as carbon oxidation, anion intercalation, or electrochemical reactions, as no such evidence of these reactions was detected electrochemically or via surface characterization using X-ray photoelectron spectroscopy (XPS). In contrast, GC electrodes treated with silanes appeared significantly more stable than bare GC electrodes, while remaining electrochemically active.

Experimental Methods Chemicals and materials. LiPF6 (99.95%) and lithium bis(trifluoromethane)sulfonimide (LiTFSI, 99.95%) were purchased from BASF (USA). Tetrabutylammonium hexafluorophosphate (TBAPF6, electrochemical grade), PC (99.7% anhydrous), sodium hydroxide (NaOH, 97.0%) and trichloromethylsilane (TMS, 99%) were purchased from Sigma Aldrich (Missouri, USA). Electrolyte salts were dried for 48 h at 80 °C under vacuum and


Langmuir

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

immediately transferred to the glovebox without exposure to air prior to use. PC was dried using activated molecular sieves (4A, Sigma-Aldrich, USA). Lithium foil (Li, 99.9% metals basis, Alfa Aesar, USA) was packaged under Argon and opened inside of the glove box. All materials were stored and prepared for analysis in a positive-pressure, Ar-filled glove box (MBraun, Exeter, New Hampshire) with oxygen and moisture content below 2 ppm. The water content of nonaqueous electrolyte solutions used for experiments was < 25 ppm as measured by KarlFischer coulometry (Mettler-Toledo C20 coulometric titrator). All solutions were prepared by mixing electrolyte salts with solvent immediately before electrochemical measurements. Electrochemistry. Electrochemical measurements were made in the glovebox using a CH760E potentiostat (CH Instruments, Austin, Texas). The glovebox temperature was about 29 °C. Where noted, a correction to the uncompensated resistance was applied during the measurement. GC electrodes (3 mm diameter) were purchased from CH Instruments and used as working electrodes for cyclic voltammetry (CV) and capacitance measurements. Removable GC disks (5 mm diameter, Pine Instruments) enclosed within a Teflon collet (Pine Instruments) served as the working electrodes for electrochemical studies, XPS, and AFM studies. Prior to performing electrochemical studies, the GC electrodes were cleaned by sequentially polishing with 3 sizes of alumna grit (1.0, 0.3, and 0.05 µm, Buehler), followed by potentiostatic anodizing at 1.2 V (vs Hg/Hg2SO4) for 30 s in 0.1 M NaOH to remove polishing impurities. Details of the anodizing treatment have been described elsewhere.2 Freshly polished and anodized electrodes were rinsed with Millipore water (Millipore, Billerica, MA), dried at 80 °C for 2 h, and transferred immediately to the glovebox. For some experiments, the electrodes were silanized immediately after drying by exposing them to TMS vapors for 30 s. This was accomplished by suspending the electrode above a solution of TMS liquid while the TMS evaporates at room temperature in the


Page 6 of 28

Page 7 of 28

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

Langmuir

fume hood. A Hg/Hg2SO4 reference electrode (CH Instruments) was used for the anodizing treatment, whereas a Vycor®-fritted reference electrode consisting of a Li metal foil immersed in a solution of 1.0 M LiTFSI in PC was used in nonaqueous electrochemical experiments. A platinum (Pt) wire served as the counter electrode in all electrochemical experiments. All voltammetry was carried out in glass vials in which the electrodes were inserted through a Teflon cap. The reference potential scale for all measurements reported herein was calibrated to a ferrocene standard (standard electrode potential = 0.40 vs. RHE) and all potentials in this study, from here onward, are reported versus the standard Li/Li+ redox couple (-3.04 vs. RHE). Characterization. Prior to surface characterization with XPS and AFM, the removable GC disks were rinsed with fresh PC, and the residual PC was removed via sublimation in a vacuum atmosphere. The GC disks were transferred to the XPS stage without exposing them to air. Identically prepared GC disks were analyzed for surface roughness using AFM. These GC disks were exposed to air during AFM analysis, following rinsing and vacuum treatment to remove residual solvent. XPS measurements were performed using a Specs PHOIBOS 150 hemispherical energy analyzer using a monochromated Al Kα X-ray source. The load-lock of the analytical UHV system is connected directly to a glovebox, enabling loading of samples without any exposure to ambient atmosphere. Survey spectra were measured using a pass energy of 40 eV at a resolution of 0.2 eV/step and a total integration time of 0.4 sec/point. Core level spectra were measured using a pass energy of 20 eV at a resolution of 0.05 eV/step and a total integration time of 0.5 sec/point. Deconvolution was performed using CasaXPS software with a Shirley-type background and 70-30 Gaussian-Lorentzian peak shapes, except for sp2 carbon, which was fitted with an asymmetric 70-30 Gaussian-Lorentzian function. Peaks were charge referenced using the


Langmuir

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

position of the sp2 glassy carbon C 1s peak at 284.4 eV, whose position was independently calibrated by the Au 4f7/2 at 84.0 eV via the use of a thin Au-coated GC electrode in a separate experiment. AFM measurements were performed on an Agilent 5500 AFM (Agilent Technologies, Inc., US) with a Pt-coated Si cantilever (~ 20 nm tip radius, Bruker) to map the morphology of the electrode surface and to quantify surface roughness. The surface morphology images were recorded using acoustic AC tapping mode on an area of 10 × 10 µm. Image acquisition and processing was performed with WsxM software (Version 5.0).

Results Prior to measuring the CEDL, the GC electrodes were activated and cleaned via an anodizing procedure described previously.2 Following anodization and drying, the CEDL and microscopic surface area of the GC electrodes were measured with 2 electrochemical techniques: first, from CVs recorded at different scan rates, and second from chronocoulombetry (Supporting Information, Figure S1). The measured CEDL is comparable to the expected value of the Helmholtz capacitance of GC, 13 µF/cm2,17 assuming a roughness factor of 3.3, which is within the expected range for GC electrodes, 1.3 – 3.5, for similarly prepared GC surfaces.2,18 Figure 1a displays the voltammetry of a GC electrode cycled within two potential ranges in a solution of 1.0 M LiPF6 in PC: first from 3.0 V to 5.0 V, and then from 3.0 V to 6.0 V. The CV recorded between 3.0 V and 5.0 V (Figure 1a, blue line) displays a featureless capacitive (non-faradaic) current. The current starts to rise at potentials above 5.0 V due to the onset of electrode and electrolyte oxidation.10 CVs recorded between 3.0 V and 6.0 V (Figure 1a, red


Page 8 of 28

Page 9 of 28

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

Langmuir

line) more clearly display electrolyte decomposition as evidenced by the increasing irreversible oxidative currents at potentials positive of 5.0 V. Figure 1b displays CVs of the double-layer capacitance region, related to the CEDL of the GC electrode before and after it was used to record the CVs shown in Figure 1a. Three successive CVs were recorded between 3.0 V and 3.6 V (the final cycle is displayed), prior to cycling the GC electrode (Figure 1b, black line), after cycling between 3.0 V and 5.0 V (Figure 1b, blue line), and after cycling between 3.0 V and 6.0 V (Figure 1b, red line). Overall, these data show that cycling the GC electrode between 3.0 V and 6.0 V leads to increased CEDL (Figure 1b, blue line and red line) relative to the CEDL recorded prior to cycling (Figure 1b, black line). Interestingly, the CEDL increased the most following cycling between 3.0 V and 5.0 V, with only a small successive increase in CEDL after the electrode was cycled between 3.0 V and 6.0 V. This result is worth noting because, at first glance, the cycling data displays a featureless current between 3.0 V and 5.0 V (Figure 1a, blue line). However, upon closer examination, the current begins to deviate upwards from a purely capacitive current at potentials positive of 4.0 V (Figure 1a, inset), suggesting that the underlying cause of the increased CEDL could take place at less positive potentials than 4.0 V in 1.0 M LiPF6, preceding electro-oxidative electrode and electrolyte decomposition. To probe the origin of the increased CEDL observed in Figure 1b, we tracked the CEDL as a function of applied potential and electrolyte composition. First, the GC electrode was inserted into the electrolyte solution and held at a potential of 3.0 V for 20 min. Following the potential hold, the CEDL of the GC electrode was measured by cycling it successively between 3.0 V and 3.6 V, as described in the Experimental Methods. After cycling, the potential of the GC electrode was stepped from 3.0 V to 3.25 V, and again held for 20 min. The cycling procedure was performed again, to obtain the CEDL following the 3.25 V potential hold. This pattern was


Langmuir

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

repeated, with the held potential increasing in 0.25 V increments, up to 5.0 V. The experiment was performed in triplicate, in solutions of PC containing TBAPF6, LiPF6, and LiTFSI, all at a concentration of 1.0 M. TBAPF6 was selected for comparison to LiPF6 as a control for the effects of the cation, and LiTFSI was selected as a control for the effects of the anion. Figure 2 illustrates how the CEDL changes with potential, expressed as a dimensionless parameter based on the CEDL at a particular potential step over the CEDL at 3.0 V (Cstep /Cinitial), thus enabling all changes in the CEDL to be normalized to the initial measurement at 3.0 V. Normalization of CEDL in this way was necessary because the individual GC electrodes displayed significant differences in their baseline CEDL measurements, and therefore comparing nonnormalized measurements of CEDL across the different electrodes introduced large errors that obscured any trends. Normalizing to the initial CEDL measurement taken after the potential step at 3.0 V enabled clearly resolving trends in the CEDL. Following the potential hold at 3.25 V, Cstep /Cinitial increased by a factor of 1.3 ± 0.3 in the LiPF6 electrolyte (Figure 2a, blue), 0.7 ± 0.3 in the LiTFSI electrolyte (Figure 2a, red), and 0.9 ± 0.2 in the TBAPF6 electrolyte (Figure 2a, black). After holding the potential at 3.5 V, the change in Cstep /Cinitial was a factor of 2.1 ± 0.3 in LiPF6, while the change in Cstep /Cinitial measured in the LiTFSI and TBAPF6 solutions was insignificant from the previous measurements at 3.0 and 3.25 V (0.9 ± 0.2 and 1.0 ± 0.1, respectively). Cstep /Cinitial increased with increasing potential up to 4.25 V in the LiPF6 electrolyte (Figure 2a, blue). Although Cstep /Cinitial for GC in the LiTFSI and TBAPF6 electrolytes increased slightly at potentials below 4.0 V, the change in Cstep /Cinitial was not significant until after holding the potential at 4.0 V, at which point Cstep /Cinitial for the GC electrode tested in LiTFSI and TBAPF6 electrolyte each increased by a factor of 1.5 ± 0.2. Cstep /Cinitial for the GC electrode tested in LiPF6 exibited a maximum increase of a factor of 6.6 ± 4.6 following holding the potential at


Page 10 of 28

Page 11 of 28

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

Langmuir

4.25 V. Cstep /Cinitial of the GC electrodes tested in LiTFSI and TBAPF6 exhibited maximum increases of factors of 1.8 ± 0.2 and 2.4 ± 1.1, at potentials of 4.25 V and 4.5 V, respectively. At potentials greater than 4.5 V, Cstep /Cinitial decreased slightly from the maximum values. All solutions tested indicated Cstep /Cinitial increases at potentials below 4.5 V, however, in the LiTFSI and TBAPF6 solutions, Cstep /Cinitial did not increase by more than a factor of ca. 1.8 – 2.4, whereas Cstep /Cinitial of the GC electrode tested in LiPF6 increased by more than a factor of 6. Interestingly, the standard deviation appeared to increase with increasing Cstep /Cinitial. We are unsure of the cause of this phenomenon. We observed the high standard deviations in preliminary data sets (not included here) for the GC electrode tested in LiPF6, and subsequently repeated the entire triplicate study in LiPF6 using all-new GC electrodes, however the standard deviation remained high (Figure 2a). Because freshly prepared electrolytes were used for each test, the only variable was the GC electrode. We hypothesize that uncontrollable inhomogeneities in individual electrodes, such as impurities from precursors, or other micro- or nano-scopic defects, may enhance roughening along defect sites containing high amounts of dangling bonds.9 The GC electrode remains electrochemically active following treatment with TMS and can be compared to electrodes which were not exposed to TMS vapors, based on CVs of ferrocene (Supporting Information, Figure S2). Figure 2b shows how Cstep /Cinitial changes with potential in LiPF6 electrolyte for GC electrodes that were anodized only (blue), as well as those that were anodized and silanized (red). LiPF6 electrolyte was selected because the GC electrode displayed the greatest degree of instability with increasing potential, based on changes to Cstep /Cinitial. In this comparison, the data for the anodized-only electrode is the same as displayed in Figure 2a (blue). In contrast to the anodized-only GC electrodes, which exhibit a significant


Langmuir

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

increase in Cstep /Cinitial with potential (Figure 2a and Figure 2b, blue), Cstep /Cinitial of the anodized and silanized electrode remains unchanged up to 4.5 V, afterwhich it decreases slightly. To determine that the changes in Cstep /Cinitial are the result of changes in electrode potential, and are not due to other processes taking place successively in the solution, we compared changes in Cstep /Cinitial for GC electrode immersed in electrolyte for a total of 120 min at OCV, 3.0 V, and 4.5 V. A freshly polished and anodized GC electrode was used for each treatment. The GC electrodes were initially immersed at 3.0 V and the CEDL was measured from CVs recorded between 3.0 and 3.6 V, as was done for the previous experiments (Figure 1 and 2). The potential was then set at OCP, or stepped to 3.0 V, or 4.5 V. The Cstep /Cinitial of each electrode was measured, via CV, 3 times throughout the treatment, at 40 min intervals. For the GC electrodes held at OCP and 3.0 V, Cstep /Cinitial increased by a factor of 1.2 and 1.3, respectively, relative to the initial measurements, after the first 40 min interval, and did not change further when measured at the second and third intervals of 80 min and 120 min, respectively (Supporting Information, Figure S3). Cstep /Cinitial of the GC electrode held at 4.5 V increased by a factor of 5.9 after the first 40 min interval. Following the second 40 min interval, or a total of 80 min, at 4.5 V, Cstep /Cinitial was 6.5, and did not increase any further between the second and third 40 min intervals at 4.5 V. Returning the electrode to 3.0 V, rinsing with fresh PC, and recording CVs in a freshly prepared LiPF6 electrolyte did not reduce or change the capacitive currents (Supporting Information, Figure S4). Prior work has shown that the presence of water enhances the corrosion of amorphous conductive carbon electrodes.10 To determine if the increase in Cstep /Cinitial was caused by trace amounts of water in the electrolyte, we intentionally added water to the LiPF6 electrolyte and measured the Cstep /Cinitial as a function of potential as shown before (Figure 3). Surprisingly, we found that the presence of water


Page 12 of 28

Page 13 of 28

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

Langmuir

suppresses, rather than promotes, the increase in Cstep /Cinitial at potentials < 5.0 V. We will discuss the possible protective effect of water later on. These controls show that the changes in Cstep /Cinitial are indeed due to increasing potential, are not caused by the presence of water, and are not reversible upon returning the potential to the initial value or rinsing with PC, and therefore are likely due to a morphological or chemical change on the electrode surface. XPS analysis of electrodes polarized at low (3.0 V) and high (4.5 V) potentials reveal that in all cases, a small amount of LiF is adsorbed on the GC surface (Figure 4a). The total amount of F, in the form of LiF and PF6, accounts for less than 2% for each electrode (Supporting Information, Figure S5). On the basis of the low faradaic currents displayed in the CVs in Figure 1, such a low percentage of salt decomposition products is reasonable and we do not expect to see evidence of large amounts of electrolyte decomposition. XPS spectra also reveal some changes in carbon speciation (Figure 4b), notably, increased graphitization, given by the greater contribution from sp2 carbon, and decreased O content for the electrodes that were not silanized, following polarization at 4.5 V. Others have reported increased graphitization in similar electrolyte at potentials > 4.5 V). For the silanized electrodes, we observed an increased O content, weaker contribution from graphitic sp2 carbon and stronger contribution from C-O species on the electrode polarized to 4.5 V. These changes appear in conjunction with a higher % coverage of Si. Silanes are expected to bind to hydroxyl-terminated carbons, resulting in a C-OSi-R bond.16 The higher overall percentage of Si, along with increased C-O signal, could result from selective protection and subsequent preservation of the C-O-Si-R bonds on the surface at higher potentials while changes in morphology and surface chemistry could attenuate other chemical species. As the silanes will bind selectively to C-OH surface groups, differences in GC homogeneity likely lead to the variation in Si coverage detected between different samples with


Langmuir

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 14 of 28

XPS (11.8 and 26.8 at% surface concentration for the low and high V samples, respectively, Supporting Information, Figure S5). These variations in Si coverage track almost directly with the percentage of C-O bonding in the C 1s on the GC surface (28.4 and 49.5 at% on the low and high V samples, respectively, Supporting Information, Figure S5), with approximately double the relative amount of C-O bonding resulting in approximately double the silane coverage. Given that we expect a spatially inhomogeneous silane layer to form, the use of XPS photoelectron escape depth to define a film “thickness” would be inappropriate, as such calculations require the presence of a compact, homogeneous layer to be meaningful. Overall, however, XPS analysis does not reveal any clear, systematic differences in chemical speciation between electrodes polarized to 3.0 and 4.5 V, regardless of silane treatment, suggesting that the differences in CEDL observed electrochemically are largely structural, rather than chemical, in origin. Figure 5 provides tapping mode AFM images, topography histograms, sample line scans, and root mean squared (RMS) roughness measurements corresponding to bare and silanized GC electrodes, before immersion in the 1.0 M LiPF6 electrolyte, and following immersion at 3.0 V then stepping the potential to 4.5 V for 40 min. The topography histograms depict the distribution of the Z-value differences, where the “number of events” is defined as the difference in measured Z-value from the mean Z-value. The RMS roughness is calculated as the standard deviation of Z-values. The initial RMS roughness of the freshly prepared bare GC electrode was 5.8 (Figure 5a). Following immersion and potentiostatic polarization at 4.5 V, the RMS roughness was 31.9 (Figure 5b). The initial RMS roughness of the silanized GC electrode was 10.8 (Figure 5c). Following immersion and potentiostatic polarization at 4.5 V, the RMS roughness was 11.0 (Figure 5d). The RMS roughness of AFM images cannot be directly compared to the roughness factors found from electrochemical measurements of microscopic


Page 15 of 28

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

Langmuir

surface area (Supporting Information, Figure S1), because these methods are sensitive to changes in topography at different scales. The RMS roughness measured by scanning probe methods depend on the lateral resolution of the microscope given by the cantilever tip radius (~20 nm) while the electrochemical measurements are sensitive to changes at the atomic scale. For this reason, we compared trends in the RMS roughness with the trends in surface area measured from CEDL. This comparison should be considered while bearing in mind the caveat of the differing spatial sensitivities between the electrochemical and AFM measurements. The increases in roughness for the bare and silanized GC electrodes following potentiostatic polarization at 4.5 V, relative to the initial roughness measurements, are comparable to the changes in the normalized CEDL (Cstep /Cinitial in Figure 2 and Supporting Information, Figure S3). Specifically, the RMS roughness for bare GC increased by a factor of 5.5, and this can be compared to the increased Cstep /Cinitial following potentiostatic polarization at 4.5 V, which was about a factor of 6. For silanized GC, neither the RMS roughness nor Cstep /Cinitial changed significantly. This indicates that increasing microscopic roughness contributes significantly to the increasing Cstep /Cinitial observed for bare GC electrodes, and the silane treatment employed here stymies these changes. We note that the AFM images of both the bare and silanized GC electrodes show numerous islands, which persisted even after multiple rinsings with pure solvent. These islands are >100 nm in height, however, the topography histograms show that they have a neglible impact on the RMS roughness. Specifically, the incidence of occurrence of islands, represents an insignificantly low number of events compared to tens of thousands of events in the range of 1030 nm, as can be seen on the histograms. Sample line scans through the areas of lower contrast depict the height variation in the < 30 nm regime. Despite the visual prominence of the islands due to their high contrast in the AFM images, the roughness measurements are clearly dominated


Langmuir

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

by events < 30 nm in height. This is exemplified by the data in Figure 5c-d, which show that the RMS roughness of the silanized GC electrode (10.8) remained almost entirely unchanged (11.0) following polarization at 4.5 V, regardless of the persistence of the islands. Several previous studies found that roughening of conductive carbon materials in electrochemical environments, accompanied by graphitization, occurred at potentials above 1.5 V (vs. RHE) in aqueous solutions.8,9 The mechanism of roughening remains unclear, but could relate to disruption of the carbon bonding structure by alkali cations, which form near-surface intercalation complexes with graphitic oxides. Others have suggested anion intercalation may take place in some types of graphitic carbon black powders in nonaqueous solutions of LiPF6 in mixed carbonate solvents.11,12 Those authors proposed that these redox reactions are due to various organic functional groups present on the electrode surface, and could be related to changes in specific capacitance of the carbon. We do not observe similar results with GC electrodes in LiPF6/PC electrolyte. Specifically, the CV studies in this paper do not resemble other examples of pseudocapacitance and anion intercalation,11,12 in the potential range of 3.0 V to 4.5 V. However, irreversible adsorption with charge transfer may take place above 3.9 V, based on the appearance of a small increase in faradaic current (Figure 1a, inset). The most likely mechanism of instability and subsequent electrode roughening follows closesly the idea proposed by Walker, Stevenson, and others,8,9 that formation of a cationgraphitic oxide complex is the first step that later results in disruption of the carbon bonding and electrode roughening at higher potentials. First, a cation exchange reaction could take place between surface hydroxyl groups and Li+, resulting in cation-surface oxide complex that serves as the seed for further morphological instability as the electrolyte stability window is approached at more positive potentials. Because silanes are also likely to adsorb and form covalent bonds


Page 16 of 28

Page 17 of 28

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

Langmuir

with surface hydroxides,15,16 and such oxides are often formed at defect sites, where anodic degradation of the carbon surface via roughening is more facile.9 The silanizing treatment employed here may therefore effectively hinder reactivity of certain types of defects, such as those having surface oxides, via steric or chemical effects (e.g., the silanes are either passivating a reactive center and lowering its reactivity, or blocking a reactant present in the solution at the interface, resulting in reduced roughening as observed in the AFM analysis). The solvation of the electrolyte is likely to play a role in this degradation mechanism In aqueous solution, the the cation-graphitic oxide complex and subsequent electrode roughening was observed only after water electrolysis commenced, while no changes were noted at potentials comparable to those where “early-stage” instability is seen in the current study.8 Upon adding water, we noted that the CEDL remained stable (Figure 3). These findings suggest that the presence water in the solvation shell of the cation may prevent the early-stage degradation reaction, in this case, preventing the cation-graphitic oxide complexes from forming by strongly solvating the cation. However, this is merely speculation at this point, and requires further study of well-defined carbon electrode surfaces to resolve the exact mechanism of roughening. Overall, the impact of silanizing treatment and polarization at 4.5 V on the chemical bonding of the GC are unclear from the XPS data, but changes in chemical bonding resulting from polarization cannot be ruled out. However, in light of the AFM analysis showing that changes in roughness are well-correlated with changes in the CEDL, we can confidently attribute the increased CEDL to surface roughening. Furthermore, treatment with silanes as described here could be an effective strategy to stabilize carbon electrodes against these early stages of instability.


Langmuir

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

Summary and Conclusions Despite many known surface reactions of GC electrodes, which suggest that the electrode surface and its interface with solution is highly susceptible to a wide variety of chemical and morphological changes, there is a general tendency to avoid discussion of these changes unless they are strikingly obvious, such as significant fouling and subsequent dimishing of electrochemical activity or other obvious variations in faradaic reactivity. However, we may also define the earliest stages of instability as an irreversible change in the electrode surface composition or morphology, which subsequently results in changes to electrochemical properties, such as the increased CEDL observed in this report. In seeking to understand the electrochemical and chemical processes that defined the electro-oxidative instability of GC electrodes in several PC-based electrolytes, we determined first that the earliest stages of instability, indicated by irreversible changes in the CEDL, occurred in the absence of faradaic reactions at potentials far less positive than the electro-oxidative stability limit given by electrode and electrolyte electro-oxidation and decomposition in PC solutions containing LiPF6. Furthermore, these changes could be prevented by treating GC electrodes with silane vapors. Though the exact mechanism of roughening is yet unclear, these studies unearth some unexpected changes in GC morphology and stability in PC containing LiPF6 , and furthermore describe the conditions that lead to the changes, as well as a possible route for stabilizing the GC surface. Continued studies targeting the mechanisms of degradation will provide scientific insight useful for improving the stability, safety, and operating lifetime of carbon-containing electrochemical energy systems.


Page 18 of 28

Page 19 of 28

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

Langmuir

Acknowledgments EVC, CK, and FRB sincerely thank Jeff Kowalski for assisting with electrode preparation. DJN and FRB thank the MIT Energy Initiative for providing summer student support. XPS measurements were conducted at the Electrochemical Discovery Laboratory at Argonne National Laboratory. This work was supported as part of the Joint Center for Energy Storage Research, an Energy Innovation Hub funded by the U.S. Department of Energy, Office of Science, Basic Energy Sciences. Argonne, a U.S. Department of Energy Office of Science Laboratory, is operated under Contract No. DE-AC02-06CH11357.

Supporting Information Available: additional electrochemical analysis and XPS are provided as Supporting Information.


Langmuir

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

References (1)

McCreery, R. L. Advanced Carbon Electrode Materials for Molecular Electrochemistry. Chem. Rev. 2008, 108, 2646–2687.

(2)

McDermott, M. T.; Kiema, G. K.; Mirwais, A. Preparation of Reproducible Glassy Carbon Electrodes by Removal of Polishing Impurities. J. Electroanal. Chem. 2003, 540, 7–15.

(3)

Menéndez, J. A.; Phillips, J.; Xia, B.; Radovic, L. R. On the Modification and Characterization of Chemical Surface Properties of Activated Carbon: In the Search of Carbons with Stable Basic Properties. Langmuir 1996, 12, 4404–4410.

(4)

Gallagher, K. G.; Fuller, T. F. Kinetic Model of the Electrochemical Oxidation of Graphitic Carbon in Acidic Environments. Phys. Chem. Chem. Phys. 2009, 11, 11557.

(5)

Zana, A.; Speder, J.; Reeler, N. E. A.; Vosch, T.; Arenz, M. Investigating the Corrosion of High Surface Area Carbons during Start/stop Fuel Cell Conditions: A Raman Study. Electrochim. Acta 2013, 114, 455–461.

(6)

Dey, A. N.; Sullivan, B. P. The Electrochemical Decomposition of Propylene Carbonate on Graphite. J. Electrochem. Soc. 1970, 117, 222–224.

(7)

Kepley, L. J.; Bard, A. J. Ellipsometric, Electrochemical, and Elemental Characterization of the Surface Phase Produced on Glassy Carbon Electrodes by Electrochemical Activation. Anal. Chem. 1988, 60, 1459–1467.

(8)

Walker, E. K.; Vanden Bout, D. A.; Stevenson, K. J. Spectroelectrochemical Investigation of an Electrogenerated Graphitic Oxide Solid–Electrolyte Interphase. Anal. Chem. 2012, 84, 8190–8197.


Page 20 of 28

Page 21 of 28

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

Langmuir

(9)

Gewirth, A. A.; Bard, A. J. In Situ Scanning Tunneling Microscopy of the Anodic Oxidation of Highly Oriented Pyrolytic Graphite Surfaces. J. Phys. Chem. 1988, 92, 5563–5566.

(10) Metzger, M.; Marino, C.; Sicklinger, J.; Haering, D.; Gasteiger, H. A. Anodic Oxidation of Conductive Carbon and Ethylene Carbonate in High-Voltage Li-Ion Batteries Quantified by on-Line Electrochemical Mass Spectrometry. J. Electrochem. Soc. 2015, 162, A1123–A1134. (11) Seel, J. A.; Dahn, J. R. Electrochemical Intercalation of PF6 into Graphite. J. Electrochem. Soc. 2000, 147, 892–898. (12) Qi, X.; Blizanac, B.; DuPasquier, A.; Meister, P.; Placke, T.; Oljaca, M.; Li, J.; Winter, M. Investigation of PF6 and TFSI Anion Intercalation into Graphitized Carbon Blacks and Its Influence on High Voltage Lithium Ion Batteries. Phys. Chem. Chem. Phys. 2014, 16, 25306–25313. (13) Sullivan, M. G.; Schnyder, B.; Bärtsch, M.; Alliata, D.; Barbero, C.; Imhof, R.; Kötz, R. Electrochemically Modified Glassy Carbon for Capacitor Electrodes Characterization of Thick Anodic Layers by Cyclic Voltammetry, Differential Electrochemical Mass Spectrometry, Spectroscopic Ellipsometry, X-Ray Photoelectron Spectroscopy, FTIR, and AFM. J. Electrochem. Soc. 2000, 147, 2636–2643. (14) Chen, P.; McCreery, R. L. Control of Electron Transfer Kinetics at Glassy Carbon Electrodes by Specific Surface Modification. Anal. Chem. 1996, 68, 3958–3965. (15) Elliott, C. M.; Murray, R. W. Chemically Modified Carbon Electrodes. Anal. Chem. 1976, 48, 1247–1254.


Langmuir

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

(16) Hao, C.; Yan, F.; Ding, L.; Xue, Y.; Ju, H. A Self-Assembled Monolayer Based Electrochemical Immunosensor for Detection of Leukemia K562A Cells. Electrochem. Commun. 2007, 9, 1359–1364. (17) Pandolfo, A. G.; Hollenkamp, A. F. Carbon Properties and Their Role in Supercapacitors. J. Power Sources 2006, 157, 11–27. (18) Ye, H.; Crooks, J. A.; Crooks, R. M. Effect of Particle Size on the Kinetics of the Electrocatalytic Oxygen Reduction Reaction Catalyzed by Pt Dendrimer-Encapsulated Nanoparticles. Langmuir 2007, 23, 11901–11906.


Page 22 of 28

Page 23 of 28

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

Langmuir

Figures

Figure 1 a. Three consecutive CVs between 3.0 V and 5.0 V (blue line), followed by three consecutive cycles between 3.0 V and 6.0 V (red line). The inset is a zoomed-in view of the blue line, showing deviation from non-faradaic behavior at potentials above 4.0 V b. CVs showing the CEDL following immersion at 3.0 V without cycling (black line), cycling between 3.0 V and 5.0 V (blue line), and cycling between 3.0 V and 6.0 V (red line). All CVs began at 3.0 V and were swept positively at 20 mV/s.


Langmuir

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

Figure 2a. GC electrodes tested in the presence of three different salts in PC: LiPF6 (blue), LiTFSI (red), and TBAPF6 (black). The salt concentration was 1.0 M. Error bars represent the standard deviation from 3 replicates. b. Comparison of differently treated GC electrodes in 1.0 M LiPF6 in PC: anodized only (blue), and anodized and silanized (red).


Page 24 of 28

Page 25 of 28

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

Langmuir

Figure 3. Control for the effect of water on Cstep/Cinitial as a function of potential for GC electrodes that were anodized only, or anodized and silanized prior to the experiment.


Langmuir

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

Figure 4 XPS spectra of GC electrodes with and without silanes after potentiostatic polarization for 40 min at 3.0 V and 4.5 V. a. F 1s, and b. C 1s


Page 26 of 28

Page 27 of 28

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

Langmuir

Figure 5 AFM microscopy of a. GC before and b. after potentiostatic polarization for 40 min at 4.5 V in LiPF6 electrolyte. c. Silanized GC before and d. after potentiostatic polarization for 40 min at 4.5 V in LiPF6 electrolyte.


Langmuir

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

TOC graphic


Page 28 of 28

Early Stage Anodic Instability of Glassy Carbon Electrodes in

Recommend Documents