Langmuir 2005, 21, 10615-10623
10615
Enhanced Anion Electroadsorption into Carbon Molecular Sieve Electrodes in Acidic Media Linoam Eliad, Gregory Salitra, Elad Pollak, Abraham Soffer,* and Doron Aurbach Department of Chemistry, Bar-Ilan University, Ramat-Gan 52900, Israel Received February 28, 2005. In Final Form: July 24, 2005 We previously showed that, for neutral electrolytes of small cations and relatively larger anions, it is possible to design certain pore sizes in active carbons that are large enough to electroadsorb cations but too small to allow anion electroadsorption. This situation leads to an electrical double-layer (EDL) capacitance that is significant only at potentials that are negative to the potential of zero charge (PZC); hence, much smaller capacitance is measured at potentials positive to the PZC. It was found that when the electrolyte is a strong acid (e.g., H2SO4, HCl), a considerable capacitance is observed at positive potentials, even when the average pore size is too small to allow the insertion of large anions in neutral electrolyte solutions. This effect disappears when the pore size becomes considerably larger than the size of the ions. In this case, the EDL capacitance at positive potentials for both neutral and acidic solutions is comparable. The following four-step mechanism was found to comply best with the experimental data: (1) By acid catalysis, the protons form carbonium species within the conjugated carbon network. (2) The anions react with the carbonium ions, providing uncharged species in an activated state, which are chemibound as surface groups to the walls of the pores. (3) Because these surface groups are effectively much smaller in size than are the charged ions, they can migrate by chemical bond exchange within the carbon skeleton via constrictions (known to exist in microporous and molecular sieving carbons), which are too narrow to accommodate hydrated charged species. (4) Upon reaching wider spaces, the uncharged species are reionized and solvated by water molecules, which can fill small pores. The justification for the above mechanism is thoroughly discussed and demonstrated by the experimental results.
Introduction Microporous (active) carbons, as high-surface-area adsorbents, are extensively used in gas- and liquid-phase separations. The pore size of active carbons can be modified over a range of 0.3-0.7 nm, which is approximately the size of small, simple molecules, by controlling the activation process (i.e., temperature and chemical additives). Well-controlled activation processes yield adjustable molecular sieve carbons, which, by the end of the 1970s, had enabled the development of carbon molecular sieve gas adsorbents1 and molecular sieve membranes,2,3 which are suitable for both gas separation and for purification processes for liquids. Taking advantage of their electrical conductivity, microporous carbons exhibit high-capacity electroadsorption in electrolyte solutions. Thus, ions can be efficiently adsorbed on electrified carbon surfaces, leading to the build-up of the pronounced electrical doublelayer (EDL) capacity of porous electrodes.4 By adding molecular sieving properties to the porous carbon electrodes, electroadsorption stereoselective or molecular sieve carbon electrodes (MSCEs)5 are obtained. With pore-size adjustable MSCEs, employed as a molecular yardstick, the effective dimensions of Na+ and * E-mail:
[email protected]. (1) Knoblauch, K.; Tarnow, F.; Heimbach, H. Process for Producing Carbon Molecular Sieves. U.S. Patent #272694, 1989. (2) Soffer, A.; Saguee, S.; Golub, D.; Azaria, M.; Hassid, M.; Cohen, H.; Method of Improving the Selectivity of Carbon Membranes by Chemical Vapor Deposition. U.S. Patent 5,695,818; Eur. Patent EP 0617997A1. (3) (a) Koresh, J. E.; Soffer, A. Sep. Sci. Technol. 1983, 18 (8), 723. (b) Soffer, A.; Saguee, S.; Koresh, J. Carbon Membrane, A Device for Gas Mixture Separation. U.S. Patent #4,685,940, 1987; Can. Patent #1236660, 1989. (4) Soffer, A.; Folman, M. J. Electroanal. Chem. 1972, 38, 25. (5) Koresh, J.; Soffer, A. J. Electroanal. Chem. Interfacial Electrochem. 1983, 147, 223.
Cl- in aqueous solutions and the elective size of BF4-, (C2H5)4N+, and Li+ in propylene carbonate were assessed in previous systems.6 The last three ions are well-known components in Li batteries and in novel EDL capacitors. It was concluded from these studies that electroadsorption into pores is only possible when these pores are as large as the hydrated or solvated ions. In a subsequent communication,7 the assessment of the dimensions of a larger variety of single and doubly charged cations and anions in aqueous solutions was reported, verifying again that these ions enter the pores only while being hydrated. The most interesting observations were associated with acidic media.8 Protons behave within the pores of the MSCE as if they were the smallest of all of the ions. However, water molecules were found to enter pores that were even smaller than hydrated protons. The latter conclusion was derived from studies of nonactivated carbons, which showed no EDL capacitance at negative potentials in acidic media but were nevertheless found to contain significant amounts of water molecules in their pore system. Hence, water molecules that are adsorbed into nanoporous carbon matrixes can hydrate protons, thus forming protium H3O+ ions within the very small pores of the MSCEs.9 In previous studies of these unique molecular sieve carbon electrodes, it was frequently observed that when the degree of pore opening was too low, the EDL (6) Salitra, G.; Soffer, A.; Eliad, L.; Aurbach, D. J. Electrochem. Soc. 2000, 147, 2486. (7) Eliad, L.; Salitra, G.; Soffer, A.; Aurbach, D. J. Phys. Chem. B. 2001, 105, 6880. (8) Eliad, L.; Salitra, G.; Soffer, A.; Aurbach, D. J. Phys. Chem. B. 2002, 106, 10128. (9) Eliad, L.; Salitra, G.; Soffer, A.; Aurbach, D. Langmuir 2005, 21, 3198.
10.1021/la0505317 CCC: $30.25 © 2005 American Chemical Society Published on Web 09/28/2005
10616
Langmuir, Vol. 21, No. 23, 2005
capacitance measured in aqueous, neutral salt solutions was very small at potentials positive to the potential of zero charge (PZC). However, the capacitance measured with the same electrodes having potentials positive to the PZC was very significant when acidic solutions with the same anions were used. In this paper, we present experimental evidence of this effect and show that it is of a general character, independent of the type and origin of the carbon, as long as the proper low degree of pore opening is maintained. Several possible mechanisms for this phenomenon are examined herein, one of which agrees very well with all the experimental results presented in this paper. Experimental Section Carbon Preparation. Six types of carbons were prepared by temperature-programmed pyrolysis under a 99.999% nitrogen flow, including four types of cellulose-based carbon, and carbons originating from Kapton polyimide films (50 µm thick; 200NH; Dupont, USA), and poly(vinylidene chloride) (PVDC; provided by Prof. David Quinn, Royal Military College, Kingston, Canada). The four cellulose-based carbons were: ashless cellulose filter paper (grade 541, Whatman, New Jersey, USA) with and without NH4Cl as a carbonization promoter and raw, unpainted Jeans cloth (local retailer) with and without NH4Cl. The use of this salt, as well as other electrolytes,10 is known to dramatically change the chemical course of the pyrolysis, leading, finally, to a much greater carbon yield. Thus, four different types of carbons were obtained, all of which can be considered hard carbons, which, if carefully activated, can yield molecular sieve carbons. The carbonization temperature programming of Kapton was the same as that of cellulose, which is described in detail elsewhere.6 The carbonization temperature programming of PVDC was different because PVDC melts and starts to decompose at approximately the same temperature, 190 °C. Therefore, a sufficiently slow thermal ramp to this temperature was used. It is important to note that the fibrous paper or cloth precursors employed in this research as well as in our previous studies1,3,5-9 contain no pores larger than 7 Å. Thus, there is practically no or negligible contribution from meso- or macropores to the total surface area and to the EDL capacity of these carbons. Activation. The activation (pore enlargement and surface area development) of the carbon samples was carried out with carbon dioxide at 900 °C under ambient pressure and with a concentrated nitric acid solution at room temperature (RT) or at 80 °C. The activation time was the only variable parameter employed to achieve different pore openings. In some cases, activation with nitric acid solutions was followed by heating the samples up to 1000 °C under an inert atmosphere. The specific activation time for each sample is given in the corresponding figure captions. It was previously shown that increasing the activation time not only increases the surface area6,11 but also increases the average pore size (in the molecular sieving range of porosity). Therefore, activation time was employed as a measure of the extent of the pore enlargement.6,7,12 Surface and Electrochemical Characterizations. Adsorption isotherms were performed by a Gemini 2375 surface analyzer (Micromeritics, Inc.) using a nitrogen adsorbate at 77K. The X-ray photoelectron spectrometry (XPS) analysis of the carbon surfaces was obtained with an HX Axis system from Kratos, Inc. Cyclic voltammetric measurements of the carbon electrodes were performed using a PGSTAT Autolab electrochemical measuring system from Ecco Chemie, Inc. (Holland). Standard three-electrode cells, with deaeration through or above the solution, were used. Assessment of the PZC. Because changes in the PZC may be thought of as being responsible for changes in the potentials (10) Dae-Young, K.; Yoshiharu, N.; Masahisa, W.; Shigenori, K. Cellulose 2001, 8, 29. (11) Gregg, S. J.; Sing, K. S. Adsorption, Surface Area and Porosity; Academic Press: London, 1967; Chapter 4. (12) Koresh, J.; Soffer, A. J. Chem. Soc., Faraday Trans. 1 1980, 76, 2457.
Eliad et al. of the adsorption of anions into the porous carbons, a brief description of this parameter is given below. The PZC separates the anodic (positive) potentials, at which anions are predominantly adsorbed,4 from the cathodic (negative) potentials, at which cations are predominantly adsorbed. Therefore, assessing the PZC is very relevant to this work because it indicates which ions occupy the carbon pore systems at given potentials. There are several methods of assessing the PZC,13,14 and there are at least three associated physical properties that provide methods for its measurement: the maximum in surface tension (and related magnitudes),4,15,16 the immersion potential,13 and the potential of equal cation and anion adsorption.17 The immersion potential is useful for assessing the PZC in neutral (i.e., pH ≈ 7) solutions. However, identifying it as the PZC may be questionable in acidic solutions because of the specific adsorption of protons, as is discussed below. Therefore, the measurement of the PZC, which is the point of equal cation and anion adsorption,4,16 was adopted. This becomes easy in the case of molecular sieving electrodes in which one ion is too large to be electroadsorbed, leading to a triangular shape of the relevant cyclic voltammograms (CVs), as shown in Figure 1. In this case, the PZC falls at the potentials where the differential capacity drops to very low values (see below).
Results Experimental Evidence of the Enhanced Adsorption of Anions in Acidic (Protic) Media. Examples of the high anodic capacity of electroadsorption into nanoporous carbon electrodes in acidic media are shown below. Some of the features of this phenomenon are described herein, and mechanisms that explain it are proposed. In Figure 1,18 a sequence of CV curves of the carbon electrodes obtained from type 541 Whatman filter paper (high-purity cellulose) is presented. The precursor filter paper of these electrodes was carbonized and then activated under a CO2 atmosphere at 900 °C for different periods of time, as indicated. Figure 1a shows that, initially, there is a negligible capacity. After a short activation step, a normal molecular sieving effect is obtained (Figure 1b) in which protons are the only ions that show an appreciable electroadsorption capacity. Enhanced electroadsorption at the positive branch of the EDL potential window appears, because of the presence of protons, when a moderate activation is applied, as seen in Figure 1c, and it practically disappears with the electrodes that underwent extended activation, for example, when the pores become sufficiently large to accommodate all the ions. In that case, the anion electroadsorption no longer depends on the presence of protons in solution (Figure 2d). Recalling that the extent of pore widening increases upon increasing the activation time, one concludes that the enhancement of anionic adsorption due to the presence of protons occurs within pores that are smaller than the sulfate ion, as can be judged from the behavior of the sulfate ions in the neutral (pH ≈ 7) solutions. A similar behavior is shown in Figure 2 for solutions containing chloride as the common anion, which is considerably smaller than sulfate.7 Therefore, shorter activation periods of the carbon electrodes were needed to achieve an enhanced anionic adsorption with Cl(13) Tobias, A. H.; Soffer, A. J. Electroanal. Chem. Interfacial Electrochem. 1983, 148, 221. (14) Argade, S. D.; Gileadi, E.; in Electrosorption; Gileadi, E., Ed.; Plenum Press: New York, 1967; p 87. (15) Oren, Y.; Glatt, I.; Livnat, A.; Kafri, O.; Soffer, A. J. Electroanal. Chem. 1985, 187, 59. (16) Oren, Y.; Soffer, A. J. Electroanal. Chem. 1986, 206, 101. (17) Frumkin, A. N.; Balashova, N. A.; Kazarinov, V. E. J. Electrochem. Soc. 1966, 113, 1011. (18) Eliad, L.; Salitra, G.; Soffer, A.; Aurbach, D. Langmuir 2005, 21, 3198 (Figure 3).
Anion Electroadsorption into Carbon Electrodes
Langmuir, Vol. 21, No. 23, 2005 10617
Figure 1. CV curves (1 mV/s) of cellulose-based carbon paper electrodes that were gradually activated under CO2 at 900 °C for different durations: (a) 10 min, (b) 35 min, (c) 90 min, and (d) 135 min. Carbonization was carried out without NH4Cl serving as a pyrolysis promoter.
Figure 2. CVs (1 mV/s) of raw cellulosic carbon-cloth electrodes. Electrodes were activated by CO2 at 900 °C during different periods of time, up to 30 min: (a) pristine, (b) activation for 5 min, (c) activation for 15 min, and (d) activation for 30 min. The enhanced anionic adsorption due to the presence of protons is tested here with Cl- serving as a common anion.
solutions because of the presence of protons in the solutions. The CVs related to sulfuric acid solutions were also presented so that the adsorption or lack of adsorption of the large sulfate anions would provide a rough estimate of the upper limit of the pore’s size. The enhanced adsorption effect for Cl- ions is observed for electrodes that underwent 5 min of activation. It becomes less pronounced with electrodes that underwent 15 min of activation and almost disappears with electrodes that underwent 30 min of activation. It is evident from Figure 2b that, although the enhanced anion adsorption by proton
assistance is effective for chloride anions, there is hardly any EDL capacity at positive potentials for the sulfuric acid solution. This provides strong evidence that the enhanced adsorption (by proton assistance) is a molecular sieving phenomenon; namely, it depends on the anion size, and it disappears almost entirely (in the case of chloride anions) when the activation periods of the electrodes are increased further. Figure 3a relates to carbon electrodes comprising PVDCbased carbons. The pore system of these electrodes is already well-developed after carbonization (i.e., carbon-
10618
Langmuir, Vol. 21, No. 23, 2005
Eliad et al.
Figure 3. CVs (1 mV/s) of PVDC carbon electrodes that were activated by an HNO3 solution for 15 min at RT, recorded (a) after being heated in a vacuum at 900 °C for 1 h and (b) without vacuum heat treatment. Table 1. Relative Atom Concentrations of the Various Carbons as Obtained from XPS Measurements
ization forms activated carbons). The carbon samples were further treated with an HNO3 solution at RT for 15 min, followed by a vacuum treatment for 1 h at 900 °C. This treatment led to a further opening of the pores but left little or no surface groups that contain oxygen. The common anion in these experiments, as seen in this figure, was sulfate. When changing the electrolyte from a salt to an acid, one would expect an increase in the EDL capacitance at the negative potentials because of both molecular sieving and the selective electroadsorption of the protons.8 The increase in capacity at the more positive potentials, where only anions can be electroadsorbed in the acidic solution, demonstrates again the unique phenomenon described herein, namely, the electroadsorption of anions by proton assistance, abbreviated as EAPA. Furthermore, clear evidence of the EAPA phenomenon was encountered previously19 with electrodes comprising cellulose-based carbons, but it was not discussed at that time. The EAPA Phenomenon in Highly Oxidized Carbons (Containing Surface Groups). Figure 3b relates to the same precursor and electrolytes as those shown in Figure 3a, but the carbon was activated with an HNO3 solution with no subsequent heat treatment. This activation results in pore walls that are rich in surface groups and contain oxygen dipoles in which the negative side faces the solution. This leads, in turn, to ion-dipole repulsion, which has been repeatedly observed in previous studies.6,7,8,20 Therefore, for such carbon electrodes, it is expected that a much weaker EAPA effect will be seen. Indeed, it is clearly demonstrated in Figure 3b that the CV curves of a carbon rich in these surface groups in a H2SO4 solution show a relatively lower capacity in the anodic branch, compared to those of other carbons. (Note that the relevant CV is not as rectangular as, for example, the parallel voltammogram in Figure 3a). Does the Presence of Nitrogen Play a Role in the EAPA Phenomenon? The data shown so far have been related to carbon electrodes that did not contain incorporated nitrogen. The results presented below deal with carbons that, because of the precursor composition or the use of additives containing nitrogen (e.g., NH4Cl) during the carbonization process, acquire thermally stable, chemibound nitrogen. The contribution of nitrogen to the EAPA phenomenon might be the result of the strong specific adsorption of protons onto the lone pairs of the surface nitrogen atoms. This electroadsorption can be at least partially reversed at potentials that are positive to the PZC, leading to an EAPA effect in addition to that found with carbons that do not contain nitrogen (Figures
1-3). To examine this possibility, two different carbons having thermally stable nitrogens were prepared, as confirmed by XPS. One carbon was a cellulose-based material that was treated with ammonium chloride during carbonization. The second one was a Kapton polyimidebased carbon. Both types were first activated by nitric acid and then vacuum-degassed at 900 °C. The atom percentage of nitrogen, oxygen, and carbon, as measured by XPS analysis for different carbons, is summarized in Table 1. The presence of oxygen in samples a, c, d, and f-i is most probably the result of incomplete degassing during the thermal treatment and exposure to air during transfer from the carbonization oven to the XPS apparatus. The three samples formed by pyrolysis in the presence of NH4Cl (d-f) clearly show that they contain thermally stable, chemibound nitrogen that withstands the further oxidizing (activation) process by CO2 at 900 °C. XPS analysis showed that the nitrogen atoms in this carbon are quaternary (402 eV) and ternary (398 eV) and are most likely pyridinic.21,22 The excessive oxygen in samples b and e is the result of oxidation by the HNO3 solution, and most of it is removed by further thermal treatment. Figure 4 presents a series of CV curves of electrodes comprising the NH4Cl-treated carbons activated by CO2 at 900 °C during different periods of time, as indicated. All of these carbons acquire thermally stable nitrogen atoms in the carbon matrix. The EAPA effect is clearly seen with electrode b for the solutions containing chloride anions, and it is weakened with electrodes that have pores that are opened more because of a more extensive activation. As expected, the electroadsorption of sulfate ions is possible only for the highly activated carbons (6 h of activation time, Figure 4e).
(19) Eliad, L.; Salitra, G.; Soffer, A.; Aurbach, D. J. Phys. Chem. B. 2002, 106, 10128 (Figures 8-9). (20) Koresh, J.; Soffer, A. J. Electroanal. Chem. Interfacial Electrochem. 1983, 147, 223.
(21) Yang, C. M.; El-Merraoui, M.; Seki, H.; Kaneno, K. Langmuir 2001, 17, 675. (22) Ang, A. K. S.; Kang, E. T.; Neoh, K. G.; Tah, K. L.; Cui, C. Q.; Lim, T. B. Polymer 2000, 41, 489.
Oa
Na
Ca
2.26 9.2 2.7
ND 1.33 ND
96.05 88.34 95.89
carbon paper + NH4Clb (d) pristine carbon 3.32 (e) activation by HNO3, 15 min, RT 8.22 (f) sample e heated at 900 °C in a vacuum 2.52
3.18 4.09 3.08
92.97 86.48 92.9
Kapton-based carbon (g) Kapton-based carbon, pristine (h) Kapton carbon, 1 h 900 °C, CO2 (i) Kapton carbon, 3 h 900 °C, CO2
2.29 2.43 2.68
92.09 93.75 93.0
carbon paper (a) pristine carbon (b) activation by HNO3, 15 min, RT (c) sample b heated at 900 °C in a vacuum
5.62 3.82 4.62
a Percentage of atomic concentration (%). b NH Cl added for the 4 carbonization step.
Anion Electroadsorption into Carbon Electrodes
Langmuir, Vol. 21, No. 23, 2005 10619
Figure 4. CVs (1 mV/s) of a basic carbon cloth that was carbonized in the presence of NH4Cl. The electrodes were activated by CO2 at 900 °C during different periods, from 5 to 60 min: (a) pristine, (b) 5 min, (c) 15 min, (d) 30 min, and (e) 60 min.
In Figure 5, the voltammetric behavior of carbon electrodes made of Kapton films and activated with CO2 (at 900 °C) for different periods is displayed. The CVs in this figure were measured with solutions containing sulfate ions. The EAPA effect is most evident in Figure 5b and is progressively less pronounced as the pores of the electrodes become large enough to accommodate the sulfate ions, even without the assistance of the protic environment. The results presented above clearly show that the EAPA phenomenon is observed with activated carbons containing chemibound nitrogen atoms as well. However, the presence of nitrogen has no special effect on this because the EAPA occurs on carbons that do not contain nitrogen (Figures 1-3). It is possible, however, that carbons combining greater amounts of stable nitrogen atoms and high EDL capacities23 could show a significant increase in the EAPA. A comparison of the behavior of some of the porous carbons in HCl, NaCl, and H2SO4 solutions, as demonstrated in Figures 2b and 5b, clearly shows the pronounced (23) ) Hulicova, D.; Yamashita, J.; Soneda, Y.; Hatori, H.; Kodama, M. Chem. Mater. 2005, 17, 1241.
effect of the size of the anions on the ability of the protons to assist anionic adsorption, at positive potentials, into the relevant porous carbons (which have pores that are sufficiently narrow to show clear differences between the adsorption of anions in neutral pH and in acidic solutions). Discussion Below, the main features of the EAPA phenomenon are summarized, and mechanisms to account for it are suggested. The Main Features of the EAPA Phenomenon. (1) When carbons with a specific average pore size (adjusted by the activation process) are used, changing the electrolyte solution from a neutral pH to an acidic one leads to a surprising increase in the EDL capacity at positive potentials. However, on the basis of the lack of adsorption of the anions in neutral (pH ≈ 7) solutions, it is clear that the anion is too large to be accommodated within the pores of the carbon. The increased EDL capacity that is measured in acidic solutions at negative potentials is expected because a selective adsorption of protons into nanoporous carbon electrodes is favorable, as already demonstrated.9
10620
Langmuir, Vol. 21, No. 23, 2005
Eliad et al.
Figure 5. CVs (1 mV/s) of Kapton-based carbon electrodes that were activated in CO2 at 900 °C for different periods: (a) pristine, (b) activation for 1 h, (c) activation for 2 h, (d) activation for 3 h, (e) activation for 4 h, and (f) activation for 6 h.
(2) The anion size plays a crucial role in the EAPA phenomenon: the smaller the anion is, the smaller the pores that exhibit the EAPA phenomenon are. (3) Carbons that were activated by a nitric acid solution (i.e., rich in polar surface groups that contain oxygen) show little or no EAPA effect. This is most evident in the case of highly oxidized carbons, as demonstrated in Figure 7 of ref 8. Below, two mechanisms for the EAPA are suggested and examined on the basis of the available experimental data. Several other mechanisms were also considered, but they were rejected because they could not account for the molecular sieving character of the EAPA nor for the role of the size of the anions. The Possibility of Explaining the Enhanced EDL Capacitance in Acidic Media at Positive Potentials as a Shift of the PZC to Very Positive Values. A positive shift in the PZC may arise from a strong adsorption of protons, so that the potential regions that are considered positively charged, that is, suitable for anion adsorption, fall within the potential range of cation adsorption, as is described in Figure 7 of ref 4 (for the shape of the electroadsorption curves of anions and cations). As shown above, the EAPA effect occurs only under molecular sieving conditions. Hence, if the above suggestion is valid, the PZC shift, if it plays a role in this effect, should also depend on molecular sieving conditions and should disappear without them. This is, in fact, possible if the carbon contains crevices and shallow pores (on the molecular scale) that are approximately of the size of the hydronium H3O+ ion. Such pores may strongly interact with the hydronium ions, but they do not allow the adsorption of larger ions.
Because of the electroneutrality condition, adsorbed hydronium ions should interact with nearby anions (
