Physical and Electrochemical Properties of Iodine-Modified Activated Carbons

Transcription

1 Journal of The Electrochemical Society, A467-A /2007/154 5 /A467/10/$20.00 The Electrochemical Society Physical and Electrochemical Properties of Iodine-Modified Activated Carbons P. Barpanda,*,z G. Fanchini, and G. G. Amatucci** Energy Storage Research Group, Department of Materials Science and Engineering, Rutgers, The State University of New Jersey, North Brunswick, New Jersey 08902, USA A467 Carbon and activated carbons have been halide modified through the mechanochemical incorporation of iodine. The effect of the processing and chemical modifications on the physical and electrochemical properties was examined. Distinct evidence was obtained to support significant structural and electrochemical modification of iodine-modified carbon nanostructures vs unmodified nanostructures. The resulting materials show an interesting combination of lower surface area coupled with markedly higher volumetric and gravimetric electrochemical capacity. The incorporation of halide leads to a % increase in volumetric capacitance of carbon materials. This modification is discussed in the context of electronic modification of the carbon and the impact on non-faradaic and also pseudocapacitive reactions for use as positive electrodes in nonaqueous symmetric and asymmetric electrochemical capacitors The Electrochemical Society. DOI: / All rights reserved. Manuscript submitted October 30, 2006; revised manuscript received January 17, Available electronically March 21, In the past two decades, electrochemical double-layer capacitors have emerged as a promising energy storage device for applications requiring high power and exceptional storage and cycle life. 1 These electrochemical capacitors are commercially available with various trade names such as supercapacitors NEC-Tokin, Japan or ultracapacitors Maxwell, USA. The electrochemical double-layer capacitors hereafter referred to as EDLC, when containing pseudocapacitive reactions, show the distinctive ability of possessing times greater capacitance than conventional capacitors. Generally, the EDLC employs high-surface-area electrode materials coupled with very thin electrolytic dielectric layers as a result of the Helmholtz double layer formed by the electrolyte and carbon surface to achieve a magnified capacitance. 2,3 This leads to a unique combination of high energy density along with the characteristic high power density in EDLCs. Hence, EDLCs occupy an intermediate state of power and energy between capacitors and batteries. Even more importantly, EDLCs exhibit outstanding robustness, chemical stability, and high cycle life exceeding 1,000,000 cycles. Recently improved capacitors referred to as nonaqueous asymmetric hybrid capacitors NAH containing one electrode of a faradaic intercalation compound and another of a non-faradaic capacitive compound has shown the potential for improved performance. 4,5 An asymmetric hybrid system consisting of a positive electrode of activated carbon and negative electrode of nanostructured lithium titanate has been shown by our group to have three times the energy density of the traditional nonaqueous symmetric EDLCs while still showing the ability to cycle greater than 750,000 times over a 5 year period. 6 Generally, the EDLC employs the distributed interface of conducting materials with large surface area such as activated carbon, 4 carbon cloths, and conducting metal oxides like RuO 2, 7 nitrides, 8 or conducting polymers. Among these positive electrode materials, activated carbon occupies a large share due to its very high surface area and economic production. The activated carbon can be produced very economically with varied surface area, porosity, pore structure, particle size, and amorphosity. In addition, these carbon materials can be easily activated by various surface reactions and mechanochemical modifications to engineer the electrochemical properties The activated-carbon-based EDLC operates on the basis of the non-faradaic double-layer electrochemical reaction at the surface of the high-surface-area mesoporous activated carbons, thereby triggering much research work on the optimization of the mesoporous architecture of activated carbon. However, the true capacitance of the activated carbon electrode is a collective mechanism consisting of * Electrochemical Society Student Member. ** Electrochemical Society Active Member. z prabeer@eden.rutgers.edu the ionic Helmholtz double layer on the electrode-electrolyte interface and the space-charge capacitance of the solid electrode. Though the electrochemical double layer at electrode-electrolyte interface has been viewed as the limiting factor to overall capacitance, the space charge of solid electrode has been proposed to play a dominating role toward improved carbons. For example, Hahn et al. 13 has proposed the dominating limiting effect of the electronic side of the double layer of activated carbon on its capacitance. The current work is a first step in our attempt to decipher the relative effect of the electrochemical double layer and space-charge of the carbon electrode. While the former is directly related to the effective surface area and meso/micro porosity of the electrodeelectrolyte interface, the latter is a function of electronic properties of the solid electrode. Herein we examine the effect of halide modification of various bulk carbons and their resultant physical, electronic, and electrochemical properties through a combination of techniques including X-ray diffraction XRD, X-ray fluorescence spectroscopy XRF, differential scanning calorimetry DSC, the Brunauer Emmett Teller BET method, Raman spectroscopy, and electrochemical characterization. Experimental Materials synthesis. Different commercially available carbon materials, ASupra from Norit, Inc., ASP from Timcal, Inc., and SP from MMM Carbon, Inc., were taken as precursor carbons. SP is a relatively low surface area m 2 /g carbon black, which is networked to give extraordinary conductivity and is typically utilized as a carbon additive in Li-ion battery electrode materials. ASP is a high-surface-area m 2 /g version of SP that has been activated. ASupra is an ultrahigh purity activated carbon of very high surface area m 2 /g. Here onward, these materials are mentioned as SP, ASP, and ASupra, respectively. These carbon materials were high energy milled for time ranging from 0 to 80 min using a SPEX 1000 high-energy shock-type milling machine. Spherical steel milling media of different sizes were used throughout. Further, these activated carbon precursors were mechanochemically modified by with the addition of 0 25 wt % of iodine during high-energy milling. All materials loading and collecting modified carbon from milling cells were performed inside a He-gas filled glove box with better than 80 C dew point to ensure no external contamination or hydrolysis. Physical characterization. XRD was conducted using a Scintag X2 powder diffractometer equipped with a copper target X-ray tube 40 kv, 35 ma, Cu K radiation. The carbon powder samples were placed inside a square-shaped mm deep sample holder. Each sample was scanned between 10 and 60 at a fixed rate of 0.8 /min. Following, peak fitting was used to identify the peak angles, d spacing, and peak maximum of all 002 and 100 peaks.

2 A468 Journal of The Electrochemical Society, A467-A Besides fitting and calculation of lattice parameters, the diffractograms were analyzed for calculating the L a and L c parameters. The 002 peaks were used to determine L c, which measures the dimension of crystallites normal to the graphene sheets. The 100 peaks were used for calculating L a, showing the linear dimension of graphene sheet plates. These L a and L c parameters were calculated by using the full width and half maximum fwhm data at peak angle of the 100 and 002 peaks, respectively The calculation was carried out through the use of the Debye Scherrer equation Eq. 1 t = / B cos 1 where t is the graphene crystallite dimension, is the X-ray wavelength, is the scattering angle in radians, and B is the fwhm in radians of. Thermal analysis TA Instruments, Inc. was conducted in nitrogen atmosphere using DSC. For DSC analysis, the carbon powder samples were sealed in aluminum pans inside a He-filled glove box. X-ray elemental analysis was performed using a Philips MiniPal XRF spectroscopy system. The carbon powders were mixed with silicon Si powder in 1:1 ratio. Using silicon as a reference, the XRF tests were conducted for 100 s by applying 10 kv and 100 ma. The surface morphology and micropore structure were determined by surface adsorption of N 2 at 77 K using a BET Micromeritics ASAP 2000 surface area analyzer. Prior to surface area analysis, the powder samples were pre-evacuated at 383 K and 10 4 Pa for 4 h. Raman spectroscopy was performed with a Renishaw in Via Raman spectrometer having 2 cm 1 instrumental resolution. Only 5% 15 mw of the laser power at 785 nm wavelength was used to avoid possible laser-induced damage. For resistivity measurement, a special piston and cylinder cell design was used. 17 The activated carbon powder was gradually compressed between two copper electrodes, while intermittently measuring its volume and electrical resistance R. By using a constant weight of 0.5 g activated carbon, the density was calculated as the inverse of the cylinder volume until the maximum compression was achieved. During compression, the carbon resistivity was calculated at several intermediate points by using the formula Resistivity = r = R S /h where S is the surface of the cylinder base and h is the height of the sample. Electrochemical characterization coin-type cells were fabricated for electrochemical characterization of activated-carbonbased EDLC. The cell consisted of various activated carbons as positive electrode, Li metal as negative electrode, and 1 M LiBF 4 dissolved in a mixture of propylene carbonate PC as electrolyte solution. The positive electrode composition was prepared by dispersing a mixture of 57 wt % active carbon material, 6 wt % Super P, and 15 wt % poly vinylidene difluoride hexafluoropropylene copolymer and 23% dibutyl phthalate plasticizer into acetone solvent. The slurry was cast. The resulting electrodes were soaked for 20 min in dimethyl ether to remove the dibutyl phthalate plasticizer and then dried at 105 C for 2 h. The positive electrode loading was 5 6 mg/cm 2. The cells were assembled in a He-filled glovebox to ensure no contamination. These cells were cycled in either the voltage range of or V using an Arbin and a Maccor galvanostat cycler. The same voltage range was scanned at a rate of 5 mv/100 s in potentiostatic testing for cyclic voltammetry CV study. Results and Discussion Figure 1. XRDs showing the amorphous nature along with 002 and 100 broad peaks in iodinated, high-energy milled carbons a ASupra, b ASP, and c SP. BM stands for ballmilling high-energy milling and the subsequent numbers present the duration of the milling process in minutes. XRD phase analysis. Normally, electrochemical capacitors employ high-surface-area, mesoporous carbon materials which are usually considered to be X-ray amorphous in nature. The X-ray diffractograms of three families of precursor and modified carbon materials, namely, ASupra, ASP, and SP, are shown in Fig. 1. The corresponding X-ray analysis data are summarized in Table I. The broad spectra in all cases confirm the existence of amorphous carbon. However, a careful study reveals the evolution and variation of 002 and 100 broad peaks at around 25 and 44. These peaks indicate the presence of localized regions of graphitic order in either an amorphous carbon matrix or nanographene domains interconnected in a disordered fashion. The broad peak at around 18 is a system peak originating from the sample holder. In all materials, longer milling of carbon leads to a narrowing and increase in intensity of the Bragg reflections, especially those originating from 100. Also, the d spacing of all carbons gradually decreases with longer milling time. The d 002 is shown to decrease from approximately 3.6 Å for the precursor carbons to the 3.5 Å range for the milled carbons. This is consistent with the trend toward graphitic-like do-

3 Journal of The Electrochemical Society, A467-A A469 Table I. The summary of XRD of carbon materials. d spacing Å Materials ASupra Raw ASupra BM ASupra BM ASupra BM ASupra 15%I BM ASupra 25%I BM ASP Raw ASP BM ASP BM ASP BM ASP 15%I BM ASP 25%I BM SP Raw SP BM SP BM SP BM SP 15%I BM SP 25%I BM L a Å L c Å Capacity mah/g main formation 3.35 A for d 002. As discussed in previous literature, longer shock-type milling of amorphous carbon leads to the formation of graphite nanoplatelets, which grow both longitudinally and laterally. From Fig. 1, it is clearly seen that a higher amount of iodine doping to pristine carbon leads to destruction of any crystallinity, hence decreasing the 002 and 100 peaks compared to both the milled and unmilled pure carbon standards. The fwhm values for 002 and 100 peaks were used to measure the crystallite size L a and crystallite thickness L c of graphitic regions inside carbon. The 100 peaks for milled samples can be deconvoluted to a broad peak at around 43 followed by a sharp peak. In the current investigation, the sharp peak arising due to milling was used to calculate the crystallite size. The variation of L a and L c parameters of carbon with milling time and iodine doping are summarized in Table I. As these carbon samples are quite amorphous in nature, the L a and L c parameters were calculated to an approximate value. It indicates the longer milling time leads to graphitic crystallite growth, especially with respect to L a. This is in agreement with the XRD on carbon powders showing evolution of 002 and 100 peaks with longer milling time. In sharp contrast, the doping of iodine decreases the crystallinity of carbons to a state of X-ray amorphosity. XRF analysis. The high-energy milling may have led to sublimation of iodine during processing, although no overt physical signs of this existed. XRF was utilized as a qualitative technique to characterize the iodine content as a function of milling time. The XRF analysis data of carbon samples are summarized in Table II. A powder mixture of modified carbon materials with a silicon internal standard were used in 1:1 weight ratio. The data confirms the presence of iodine in all iodine-doped, high-energy milled carbon samples. Further, the ratio of intensities of K peaks of iodine at kev and silicon at kev affirms the increasing amount of iodine content in the iodinated sample with higher starting iodine content. Also, the iodine content is almost unaffected by longer milling time. This holds good for all three classes of carbons tried in the current investigation. Moreover, the XRF data shows no metallic contamination in iodated carbons due to high-energy milling. Thermal analysis (DSC). Thermal analysis of various iodinated high-energy milled carbon powders was conducted. The DSC curves are shown in Fig. 2. As shown in the figure, free iodine gives rise to an endothermic peak at around 110 C due to the sublimation of iodine. The exothermic peak immediately following it marked with an asterisk may be due to the reaction of the sublimated iodine with aluminium pan or the failure of the pan itself under the pressure of sublimed I 2. All pristine carbon materials show neglible reaction. Upon doping these carbons with iodine, the DSC does not show any endothermic peak in the range C, indicating the absence of any free iodine. The iodinated carbon shows a sharp exothermic peak at around C, suggesting that the high-energy milling leads to some degree of chemical bonding between carbon and iodine, thereby leaving no free iodine. Resistivity of modified activated carbons. The general physicochemical properties of these carbon precursors are given in Table Table II. XRF data (I/Si ratio) of carbon materials. Materials system ASupra ASP SP BM % I BM % I BM % I BM % I BM % I BM Figure 2. DSC curves of iodinated carbons.

4 A470 Journal of The Electrochemical Society, A467-A Table III. Physicochemical properties of activated carbon precursors. Manufacturer Morphology Resistivity cm Capacitance F/g Surface area m 2 /g Norit-ASupra Spherical Timcal-ASP Porous interconnected Timcal-SP Interconnected III. Figure 3 shows the variation of resistivity as a function of density of different carbons, indicating the effect of compression on resistivity, clearly showing a marked increase in density for the milled and especially the iodine-reacted samples. The high-energy milling leads to breaking of the interconnected morphology and large macropores of carbon raw materials, which leads to a closer packing of carbon particles hence resulting in higher density. The iodine-treated sample shows the highest density due to the combined effect of high-energy milling and inclusion of highly dense iodine in carbon. The high-energy milling significantly destroys the spherical Figure 4. Raman spectra of a ASupra- and b SP-based materials. particle morphology of ASupra and the interconnected morphology of ASP- and SP-based materials. This in turn enhances the resistivity of the carbon materials. The iodinated carbons show a slight increase in resistivity for the ASP and SP samples vs the noniodated milled and is similar to the noniodated milled ASupra sample. However, there is very little increase in resistivity with high-energy milling and iodine doping. Upon complete densification, the resistivity of all samples narrows down to almost the same value. A more detailed study of the change in underlying electronic structure of iodine-treated carbons will be presented elsewhere. Figure 3. Resistivity of iodinated carbon with density of a ASupra, b ASP, and c SP materials. Raman spectroscopy. The Raman spectra of mechanochemically modified carbons are given in Fig. 4. All carbon materials show two distinct peaks at around 1300 cm 1 D peak, A 1g symmetry and 1600 cm 1 G peak, E 2g symmetry. These D and G peaks correspond to the presence of breathing mode of the A 1g symmetry and the individual carbons involved in the sp 2 planar hybridization states in the carbon structure, respectively. Thus, the occurrence and change in these D and G peaks can be utilized to know any internal structural modification that is quite difficult to notice by techniques like XRD. Overall, the D peak 1300 cm 1 of SP-based materials shows higher intensity relative to the G peak than the corresponding ASupra materials. For amorphous materials the increase of the D peak vs G peak is consistent with an increase in the probability of finding a sixfold carbon ring in a cluster of carbon as opposed to much longer range L a development which would enhance the G band over the D band. The latter phenomena is due to the confinement of the breathing mode of the 6 carbon as crystallinity progresses that is characteristic of the D peak. 18 The maximum in D/G ratio has been suggested to occur at approximately L a = 20 Å. XRD revealed that upon milling the L a is much greater than 20 Å; in this region the D/G ratio should therefore increase with crystallinity. The Raman analysis data of carbon materials are summarized in Table IV. From Table IV, it is indicated that the I D /I G ratio decreases with milling of pristine carbons. This indicates the formation of a much more crystalline graphitic region in the milled carbon consistent with the XRD characterization. In sharp contrast, the presence of iodine in

5 Journal of The Electrochemical Society, A467-A A471 Table IV. Summary of Raman spectroscopy data of ASupra and SP materials. Materials I D /I G ratio D Peak cm 1 G Peak cm 1 D-FWHM cm 1 ASupra ASupra BM ASupra 25% I BM SP SP BM SP 25% I BM G-FWHM cm 1 pristine carbon leads to a higher I D /I G ratio, consistent with the destruction of any crystalline graphitic regions for L a 20 Å. However, the XRD results indicate complete X-ray amorphosity; therefore in these region L a 20 Å the D/G should actually decrease with decrease in crystallite size. This paradox may indicate the sur- Figure 5. BET surface area of high-energy milled and iodinated carbon materials: a ASupra, b ASP, and c SP. Figure 6. Average pore size of a ASupra, b ASP, and c SP materials.

6 A472 Journal of The Electrochemical Society, A467-A Figure 7. Pore size distribution of a ASupra-, b ASP-, and c SP-based materials. vival of a significant number of six-membered rings which are isolated from crystal development due to the presence of I n -. This interesting possibility will be investigated in more detail in subsequent research. Figure 8. Discharge capacity of 2nd cycle of a ASupra-, b ASP-, and c SP-based materials cycled between 4 and 3.2 V. Surface area and morphology analysis (BET). The nonfaradaic capacitance of any capacitor is directly proportional to its mesoporous surface area of the carbon. In the current work, the surface area of carbon precursors was modified in two ways: by high-energy milling of carbon for different times and by the chemical treatment of this milled carbon with different amounts of iodine doping. Figure 5 shows the variation of surface area of carbon materials. The high-energy milling leads to smaller surface area in the case of ASupra and ASP carbons, whereas the surface area increases for SP carbon. The shock-type high-energy milling transfers an appreciable amount of impact energy to the carbon particles, resulting in a continuous local fracture of carbon materials. As these highsurface-area carbon precursors contain small particles even before any treatment, the application of milling reduces the particle size below a limit where they become thermodynamically unstable due to very high surface energy. As a result, these particles tend to agglomerate together, forming larger particles. 19 In this case, even if there are many small particles, the overall surface area decreases due to the agglomeration of fine particles, making them inactive. Further, it is noted that the addition of iodine into carbon leads to even smaller surface area Fig. 5. The possible formation of nanocomposites due to iodine doping may envelop the mesoporosity in the carbon making part of their surface area inactive. Also, the inherently high density of iodine leads to lower surface area. Thus, the surface area decreases both by high-energy milling and iodine doping. The mechanochemical treatment of carbon not only changes the overall surface area but also affects the internal pore size and pore structure distribution. Figure 6 shows the change in overall pore size of carbon materials. While for ASupra, pore size increases with longer milling time and decreases with iodine doping, the reverse phenomena occur with ASP and SP. The initial pore size is relatively larger for ASP- and SP-based materials. The high-energy milling destroys the interconnected particle morphology of the intrinsically mesoporous ASP and SP materials. This in turn leads to lower mesoporosity and smaller pore size. In contrast, it leads to higher mesoporosity and increased pore size in the initially smaller pore size ASupra materials. Although the overall surface area is responsible for the electrode performance, on a microscale, the pore size distribution is an equally important factor. Even with the same surface area, the relative degree of mesoporosity leads to better electrochemical performance as microporosity 1.2 nm sterically inhibits the adsorption of the required electrolyte, specifically the solvated ion. The porosity distribution in the modified-carbon-based samples is presented in Fig.

7 Journal of The Electrochemical Society, A467-A A473 Figure 9. Specific capacity V of a ASupra, b ASP, and c SP materials. Specific capacity V of d ASupra, e ASP, and f SP materials. 7. In the case of ASupra materials, the initial ASupra possesses no mesoporosity, but high-energy milling of ASupra leads to smaller particles with larger pores, hence inducing some degree of mesoporosity, especially at around 2.5 nm. However, this mesoporosity is destroyed upon addition of iodine. In the case of ASP and SP, the precursor itself possesses high mesoporosity for nm. This mesoporosity is destroyed by high- energy milling and more so by iodine doping. From Fig. 7, it is inferred that high-energy milling may lead to favorable pore structure, while the iodine doping often negates these modified pore structures. Electrochemical characterization. Galvanostatic. The specific capacity of the initial precursor as well as modified carbon electrodes are summarized in Fig. 8a-c for ASupra, ASP, and SP materials, respectively, for cells cycled between 3.2 and 4. V. Capacity was calculated in the non-faradaic reaction region between 3.2 and 4 V for all the materials. Here, capacity is utilized as opposed to capacitance, because if one looks closely, most activated carbons have discharge curves that are not completely linear, resulting in skewed data if given in F/cm 3 based on a narrow range. As we see this is a region of electrochemical activity that is defined by non-faradaic behavior. It is clearly marked that for untreated carbon materials, the capacity decreases with longer milling time. This can be solely assigned to the previously noted decrease in surface area with longer milling. However, upon doping the carbons with iodine via high-energy milling, the capacity systematically increases for all three systems, even though a significant portion of the weight of the electrode is iodine and not carbon. Of particular interest is the fact that there is a very large disparity in surface area normalized capacitance between the iodine-doped and undoped carbon samples Fig. 9a-c. These specific capacitance values were normalized per surface area by dividing the specific capacitance F/g of carbon materials to their corresponding BET surface area m 2 /g values. Theoretically all materials should give Figure 10. The comparison of second discharge voltage profile showing the effect of iodine addition: a ASupra, b ASP, and c SP carbon materials. approximately the same value; however, a tremendous increase in surface area normalized capacitance is developed for the iodinedoped samples, where normalized capacitance is in excess of % of the raw materials. The surface area normalized capacitance data for carbon electrodes were obtained using the slope of linear portion of corresponding voltage discharge profiles. Upon halide doping, the normalized capacitance increased from 5.9 to 15.7 F/cm 2 in the case of ASupra 150% increase and from 6.8 to 32.8 F/cm 2 in the case of ASP 380% increase. From the BET tests, the relative amount of micropore area of ASupra materials was seen to increase from 24.9 to 60.4% upon halide doping. Even though the amount of inaccessible micropores increases significantly upon high-energy milling of iodinated carbon, the resultant capacitance increases instead of decreasing. The very large increase in surface specific capacitance, among the highest reported, suggests an internal electronic modification to the carbon. Figure 9d-f plots the volumetric capacity of the modified carbons vs the raw carbons. The density of the carbons was based on the

8 A474 Journal of The Electrochemical Society, A467-A Figure 11. Volumetric capacity of a ASupra, b ASP, and c SP materials cycled between 4 and 3.2 V. Figure 12. CV plot showing the effect of iodine doping on the voltage profile of ballmilled carbons: a ASupra, b ASP, and c SP materials. density of carbon pellets pressed at 7000 psi. Although we have previously shown that the specific capacity of the doped activated carbons is less in the non-faradaic region V, the volumetric capacity is superior to the raw material by almost 100%. For the best raw materials, ASupra, the volumetric capacity raises from 12 to 24 mah/cm 3 with mechanochemically induced iodine addition. The comparative discharge voltage profiles to lower voltages from 4 to 2.8 V of pristine and modified carbons are shown in Fig. 10. In the case of ASupra- and ASP-based materials, the voltage profiles indicate a decrease in capacity with milling, which can be correlated to the decreasing surface area, but an incremental increase in capacity occurs in SP materials as high-energy milling leads to higher surface area. Further, the addition of iodine results in a significant modification of the voltage profile in all three classes of carbon. The incorporation of iodine gives rise to a large plateau at around 3.1 V, making the voltage profiles nonlinear. This voltage is well above the traditional Li + I 2 reaction potential of 2.7 V and preliminary evidence points to a faradaic reduction reaction of lithium with a CI x complex. The nature and degree of covalency of this bond will be discussed in a future paper. This C-I reduction increases the capacity of the carbon materials significantly. Figure 9b shows that the incorporation of this faradaic reaction increases the volumetric capacity of the iodine-reacted samples vs the doped samples by 500%. For example, the volumetric capacity of ASupra materials increases from 11 to 56 mah/cm 3 upon iodine doping. This increase in volumetric capacity is even more significant in case of ASP- and SP-based materials. Figure 11 summarizes the effect of milling time and iodine doping on volumetric capacity of modified carbons. In all cases, the incorporation of iodine into carbon leads to higher volumetric efficiency, the maximum increase being in the case of SP materials.

9 Journal of The Electrochemical Society, A467-A A475 Figure 13. Voltage profile of iodinated carbon from 4 to 3 V. From a practical application point of view, unlike batteries, the electrochemical capacitors are often limited by volume rather than by weight. Thus, these iodine-doped carbons showing high volumetric capacity can be useful candidates for electrode materials. Also, the higher volumetric energy density makes the electrodes thinner with less porous electrode diffusion required, resulting in higher performance. Cyclic voltammetry. In order to closely follow the change in electronic properties of carbon upon modification, the potentiodynamic cycling was conducted on undoped and doped samples. The potentiodynamic cycling curves of ASupra-, ASP-, and SP-based materials are presented in Fig. 12. They clearly illustrate the change in the specific current profiles of active carbon materials upon iodine doping. While the undoped carbons show a smooth curve indicative of a non-faradaic reaction through most of the voltage region, the doping of iodine leads to peaks in current around V. These spikes Figure 14. Cycle life of some key carbon samples a from 4 to 3.2 V at 15 ma/g and b from 4 to 2.8 V at 100 ma/g. Figure 15. The comparative discharge voltage profiles of iodinated ASP samples at different cycles, cycled between 4 and 2.8 V at 100 ma/g. The numbers in the figure indicate the cycle index numbers. are reversible in nature, indicating no permanent formation of any particular compound, and also are a strong indication of a faradaic process. Also, over the non-faradaic region, the specific current is higher for iodinated carbons, consistent with the galvanostatic results. To get a better understanding of the underlying electrochemistry, a sample voltage discharging profile of an iodinated carbon cycled between 4 and 3 V is shown in Fig. 13. It shows three distinct regions marked I, II, III corresponding to three proposed electrochemical reactions. Region I involves the movement of adsorbed - BF 4 ions of electrolyte from the mesoporous carbon. The change in slope at about 3.5 V is due to the p- to n-type transition point of zero charge, PZC. After PZC, the adsorption of Li + ions within the Helmholtz double layer occurs. Further, a large plateau is marked at around 3.1 V, starting with region III, which is likely due to the cathodic reaction of C-I compound. The transition from I to II and II to III regions yields corresponding reversible peaks in the CV curves of the iodinated sample. Cycle life. The volumetric discharge capacities of six key carbonbased samples are shown as a function of cycle number in Fig. 14. The charge/discharge cycle is shown for more than 200 cycles from 3 to 4.2 V cycling at 15 ma/g Fig. 14a and from 2.8 to 4.2 V cycling at 100 ma/g Fig. 14b. The high capacity of carbon modified with iodine is fairly retained even after long cycling. The capacity profile is marked to be completely flat in the case of SP and ASP materials and there is a slight decrease in the case of ASupra materials. Even though the samples are cycling through the faradaic region Fig. 14b, the cycling on this plateau is very reversible, as evidenced by the flat cycling profile. The retention of this faradaic region indicates the chemical stability of iodinated carbon upon cycling. The very efficient cycling of the iodine-doped ASP in both the non-faradaic region and faradaic regions makes it a superior choice in volumetric energy density 20 mah/cm 3 relative to ASupra 12 mah/cm 3 by almost 100%. Figure 15 compares the voltage discharge profiles of iodinated ASP at different cycles from 1 to 600, cycled at 100 ma/g from 4 to 2.8 V. It shows an excellent cyclic stability after the drop in initial cycles. A detail study of cycle life and the effect of long cycling on carbon-iodine chemistry is currently under investigation. Conclusions We have shown that reaction of carbon with iodine under mechanochemical conditions leads to significant modification of the physical and electrochemical properties of the carbons. Relative to nondoped carbons under the same mechanical treatments, the iodated samples revealed a distinct crystallographic amorphization. The

10 A476 Journal of The Electrochemical Society, A467-A electrochemical properties showed drastic increases in surface specific and volumetric capacitance in the non-faradaic region and the development of a faradaic charge transfer at around 3.1 V vs Li/Li +. The latter is most likely a charge-transfer reaction involving a semiionic iodine species and Li + and will be discussed at a later time. The modification to the non-faradaic region may be directly related to the modification of the carbon s electronic structure, as we see a positive shift in the carbon s PZC most likely brought about by a degree of p doping. This effect would theoretically have a significant effect on the density of states of the carbon. In the EDLC, the electrochemical capacitance is a parallel mechanism of the Helmholtz double layer on the electrode-electrolyte interface and the space-charge effect of the solid electrode. Though the non-faradaic double layer is believed to be a controlling mechanism, the spacecharge effect has been proposed to be limiting. 13 No doubt, a mechanism as described above would induce a modification of the space charge. A detailed study of iodine-induced modification of electronic states and electrochemical performance will be reported in a future publication. The reversible capacity of three groups of carbons was investigated. The carbon materials were mechanochemically modified with high-energy shock-type milling and iodine doping. The relative effect of surface morphology due to high-energy milling and electronic properties due to iodine doping on the reversible capacity of carbon was examined. Longer milling time resulted in lower surface area and less capacity. However, the incorporation of iodine into the carbon leads to an exceptional enhanced capacity in spite of smaller surface area. This enhancement was observed in both the nonfaradaic and faradaic regions of the voltage curve. Due to the halide modification of carbon, a remarkable % improvement in surface specific capacitance from 5.94 to F/cm 2 in the case of ASupra-type carbon and % increase in volumetric capacity from4to20mah/cm 3 in the case of ASP-type carbon was observed. Although early in the research, this new approach to activated carbons with high volumetric capacity may offer promise toward a significant energy density improvement for future generations of nonaqueous symmetric and asymmetric electrochemical capacitors. Acknowledgments The authors would like to thank the Office of Naval Research for support under contract no. N The technical assistance of F. Badway, N. Pereira, I. Plitz, J. Gural, A. DuPasquier, and P. Smith is greatly appreciated. Rutgers, The State University of New Jersey, assisted in meeting the publication costs of this article. References 1. B. E. Conway, Electrochemical Supercapacitors: Scientific Fundamentals and Technological Applications, Kluwer-Plenum Publications Co., New York R. Kotz and M. Carlen, Electrochim. Acta, 45, A. Burke, J. Power Sources, 91, G. G. Amatucci, F. Badway, A. DuPasquier, and T. Zheng, J. Electrochem. Soc., 148, A A. DuPasquier, I. Plitz, J. Gural, S. Menocal, and G. Amatucci, J. Power Sources, 113, I. Plitz, A. DuPasquier, F. Badway. J. Gural, N. Pereira, A. Gmitter, and G. G. Amatucci, Appl. Phys. A, 82, T. C. Liu, W. G. Pell, and B. E. Conway, Electrochim. Acta, 42, T. Liu, W. Pell, B. Conway, and S. Roberson, J. Electrochem. Soc., 145, W. Xing, J. S. Xue, T. Zheng, A. Gibaud, and J. R. Dahn, J. Electrochem. Soc., 143, F. Disma, L. Aymard, L. Dupont, and J.-M. Tarascon, J. Electrochem. Soc., 143, F. Salver-Disma, C. Lenain, B. Beaudoin, L. Aymard, and J.-M. Tarascon, Solid State Ionics, 98, F. Salver-Disma, A. Du Pasquier, J.-C. Lassegues, J.-N. Rouzaud, and J.-M. Tarascon, J. Power Sources, 81-82, M. Hahn, M. Baertschi, O. Barbieri, J.-C. Sauter, R. Kotz, and R. Gallay, Electrochem. Solid-State Lett., 7, A Y. Liu, J. Xue, T. Zheng, and J. R. Dahn, Carbon, 34, B. E. Warren, Phys. Rev., 59, A. K. Kercher and D. C. Nagle, Carbon, 41, A. DuPasquier, I. Plitz, J. Gural, F. Badway, and G. Amatucci, J. Power Sources, 136, A. C. Ferrari and J. Robertson, Phys. Rev. B, 61, A. P. Weber, M. Seipenbusch, C. Thanner, and G. Kasper, J. Nanopart. Res., 1,