Technical Notes Anal. Chem. 1994,66, 1204-1207
Carbonized Microcellular Foam-Based Porous Flow-Through Electrodes with Unit Coulometric Efficiency Brian K. Davis and Stephen G. Weber' Department of Chemistty, University of Pittsburgh, Pittsburgh, Pennsylvania 15260
The carbon foam used was pyrolyzed from PAN at 1100 "C. Its specific surface area is 21 000 cm-l and its porosity is 0.97; thus, it is ideally suited for coulometric cells. Although the material is fragile, it can be bored with steel or glass tubing. The most effective cells consisted of cylindrical segments of foam which were from 0.5 to 3.0 mm long inside of a 1.0 mm in diameter glass tube. In the smallest cell, 0.5 mm long, the electrode volume was 0.4 rL,yet it yielded unit coulometric efficiency at 1.0 mL/min. The pressure required for higher flow rates caused electrode failure. Longer electrodes yielded cells with unit coulometric efficiency up to the system limits near 3 mL/min. Flow-through porous electrodes have been studied extensively for their in use in coulometric cells for continuous electrolysis in a flowing stream. Attention has been given not only to the theory associated with their ~perationl-~ but also to their practical appli~ation.~J-l~ For all their promise, porous electrodes are plagued with complications arising from uneven potential and current distributions within the interior of the electrode in regions remote from the reference electrode. Considerable theoretical work has been done on the influence of cell design on these parameters and the resulting electrode f u n c t i ~ n . l For , ~ ~instance, several researchers have reported that careful placement of the auxiliary electrode upstream of the working electrode is imperative for efficient cell electroly~is.~*~J In spite of their morecomplicated nature, porous electrodes have been applied sucessfullyto HPLC applications, operating both in the amperometric and coulometric m ~ d e s . ~ -One l ~ of the more promising electrode materials has been reticulated vitreous carbon (RVC). The attributes that lend RVC to (1) Sioda, R. E.; Keating, K. B. In EIectroanulytical Chemistry; Bard, A. J., Ed.; Marcel Dekker, Inc.: New York,1981; Vol. 1, p 1. (2) Newman, J. S.;Tiedman, W. Adu. Electrochem. Eng. 1978, 125, 58. (3) Alkire, R.; Gracon, B. J. Electrochem. Soc. 1975, 122, 1594. (4) Sioda, R. E. Electrochim. Acfa 1970, 15, 783. (5) Fedkin, P. S.J. Electrochem. Soc. 1981, 128, 831. (6) Newman, J. S.;Trainham, J. A. J. Electrochem. SOC.1978, 125, 58. (7) Delanghe, B.; Tellier, S.; Astruc, M. Electrochim. Acta 1990, 1369. (8) Newman, J. S.; Trainham, J. A. J. Electrochem. Soc. 1977, 124, 1528. (9) Wang, J. Electrochim. Acta 1981, 26, 1721. (10) Strohl, A. N.; Curran, D. J. Anal. Chem. 1979.51, 1045. (11) Strohl, A. N.; Curran, D. J. Anal. Chem. 1979, 51, 1050. (12) Curran, D. J.; Tougas, T. P. A w l . Chem. 1984,56, 672. (13) Schieffer, G. W. Anal. Chem. 1980, 52, 1994. (14) Blaedel, W. J.; Wang, J. Anal. Chem. 1979, 51, 799. (15) Sioda, R. E. J. Electroanal. Chem. 1971, 34, 411.
1204 Analytikal Chemistty, Vol. 68, No. 7, April 1, 1994
porous electrode applications are its porosity (0.97),specific surfqce area (66 cm-l),16 and low resistance to flow. In addition, carbon has become the material of choice in many electrochemical applications. Cells incorporating RVC have been e f f e c t i ~ e . ~ ~The J ~ Jmost ~ refined detector ce1112 shows unit coulometric efficiency for five compounds at 0.5 mL/ min, with nearly unit (95%) coulometric efficiency at 1.OmL/ min. Electrode lengths were greater than 1 cm, while the cell diameters were 0.2-0.55 cm. While small by some standards, a cell with a smaller diameter more closely matched to HPLC systems should lessen band broadening. We report here on a coulometric cell based on carbonized polyacrylonitrile (PAN) f o a m ~ . l ~ -These ~ l novel foams have been used successfullyas microelectrode ensemble~.~~-z~ They are similar to RVC in that they are continuous, reticulated materials of pyrolyzed polymer with low resistance to solution flow. While their porosity is identical in magnitude to that of RVC (0.97), their specific surfaces are 3 orders of magnitude larger, 21 000 cm-l. Such a high surface area should result in considerable improvements in the efficienciesof electrolysis at higher flow rates using an electrode of reduced length and diameter. The need for high efficiency is most acute in applications in which the porous electrode is used synthetically or as a scrubber electrode. The objective of this study was to take advantage of the unique attributes of the carbon foam and to design a cell that supported high conversion coefficiencies at reasonable flow rates with smaller dimensions than those used for RVC. EXPERIMENTAL SECTION Materials. Solutions used were either ferrocene (Aldrich, Milwaukee, WI) in 0.1 M tetraethylammonium perchlorate (TEAP; GFS Chemical, Columbus, OH, polarographic grade) in acetonitrile (Fisher, Pittsburgh, PA, LC grade) or potassium ferrocyanide (Fisher) in 0.1 M potassium chloride (EM (16) Kinoshita, K. Carbon: Electrochemical and Physiochemical Properties; Wiley: New York, 1988. (17) Sylwester, A. P.; Aubcrt, J. H.; Rang, P. B.; Arnold, C., Jr.; Clough, L. R. Polym. Mater. Sci. Eng. 1987, 57, 113. (18) Renschler, C. L.; Sylwester, A. P. Mater. Sei. Forum 1990, 52-53, 301. (19) Sylwester, A. P.; Clough, R. L. Synth. Mer. 1989, 29, F53. (20) US.Patent 4,832,870, 1989. (21) US.Patent 4,832,881, 1988. (22) Davis, B. K.; Weber,S.G.; Sylwester, A. P. AMI. Chem. 1990, 62, 1OOO. (23) Wang, J.; Brennsteiner, A.; Sylwester, A. P. AMI. Chem. 1990, 62, 1002. (24) Wang, J.; Brennsteincr, A.; Angnes, L.; Sylwester, A. P.; LaGasse, R. R.; Bitsch, N. Anal. Chem. 1992, 64, 151.
0003-2700/94/03661204$04.50/0
0 1994 American Chemical Soclety
h working electrode
Working elyrode
I
n ,R ~ ~ H B I C ~
electrode
I \"
U Flgute 1. Design of cell used for aqueous solutions.
Science, Cherry Hill, NJ) in pH 7.0 aqueous phosphate concentrated buffer. The carbonized PAN foams were manufactured at Sandia National Laboratories (Albuquerque, NM) through a procedure described previously. The process involved a thermally induced phase separation of PAN from a solution. Controlling the temperature and pressure can produce materials in a wide variety of pore sizes and morphologies. The resulting PAN foam is carbonized at temperatures of >1000 "C under an inert atmosphere to produce the final product. The carbon foam used in this study had been carbonized at 1100 "C and possessed an average pore size of five pm and a density of 60 mg/cm3. Instrumentation. A DTK 486 personal computer was used both for instrument control and data storage. A National Instruments NI-488.2 GPIB board controlled both a LeCroy 9410 digital oscilloscope and a Stanford Applied Research SynthesizedFunction Generator (Model DS345). Waveforms were synthesized using the manufacturer-supplied Arbitrary Waveform Composer (AWC) software. Customized waveforms were applied to the cell through the external input of a Princeton Applied Research (PAR) 173 potentiostat/ galvanostat. Data were collected using the LeCroy Oscilloscope and were stored using the National Instruments Labwindows 2.1 software. The flow in the system was controlled using a ConstaMetric 3 pump that had been fitted with a Rheodyne injector. Flow rates were confirmed daily by running timed trials. The pump has been retrofitted so that the flow rate range was 0-3.33 mL rather than 0-10.0 mL. Cell Design. See Figure 1. The auxiliary electrode in all cases was a platinum wire grid while the reference electrode was either a silver wire pseudoreference electrode or a Ag/ AgC1/3M NaCl reference electrode. All cells were designed with the auxiliary and reference electrodes on opposite sides of the porous foam electrode,with the auxiliaryon the upstream side of the cell. Products from the reaction at the auxiliary electrodewere isolated from the main flow stream by placing the electrode in an external electrolyte reservoir. Contact to the flow stream was established through a Vycor frit surrounded by heat-shrink tubing leading to a LC T. The referenceelectrode was placed close to the exit of the working electrodeeither secured in a T or clamped in the waste reservoir. 18y20v21
Two variations in working- electrodehardware are discussed herein that represent stages in the cell engineering process. The difference between the two lies in the type of tubing material used to house the carbon foam electrode. The first design utilized a thin walled stainless steel tube having an i.d. of 1.5 mm. The steel served not only as the housing but also as the electrical contact. A glass capillary tube having an i.d. of 1.0 mm was the housing for the second design. Contact was made to the electrode through a platinum wire threaded through the glass tube's exit. In both cases, the housing also served as the cutting tool to fashion the carbon foam cylinder. Although it is fragile, the foam can be fashioned into a cylindrical shape by operating its eventual housing as a drill bit. THEORY The efficiency, R, in a porous electrode has been described by the expression1y4
R = 1 - exp(-ukl/y)
(1)
where a is the specific surface area, k is the mass-transfer coefficient, L is the electrode length, and y is the linear flow velocity defined as the volume flow velocity divided by the cross-sectionalarea. The mass-transfer coefficient,k is given by
where j and x are experimentally determined constants. To predict a cell's performance, it is necessary to assume values for these constants, thus rendering the prediction an estimate. Another approach to predicting the cell's performance is to consider the two competing processes within the cell, diffusion from the bulk solution to the cell wall and the movement in the linear flow direction which eventually leads to exit from the cell. An approximate relationship for efficiency has been developed25based on the diffusion time, td, and the residence time, tp25
Uis the volume flow velocity, Vis the electrode's interstitial volume, A is the surface area, D is the diffusion coefficient, and n is the number of dimensions in which travel by diffusion can be rewarded by electrolysis (three for this case). While this expressiondoes not take into considerationiR drops within the cell or kinetic limitations on the electron transfer, it is useful in determiningcell design or predictingthe performance of an existing system under mass-transfer control. It predicts high conversion from an electrode of high surface area in a small volume and/or having a long length and/or operating at a low flow rate. Applying this logic to the present system, several things become apparent. First, the objectives of this study include developing a coulometric cell having a cross(25) Weber, S.G. Detection Basedon Electricaland ElectrochemicalMeasurements. In Detectorsfor Liquid Chromatography;Yeung, E., Ed. Chem. Anal. 1986, 89, 214.
AnalyticalChemistry, Vol. 66, No. 7, April 1, 1994
1205
sectional area more compatible with HPLC (- 1 mm) which can also operate at reasonable flow rates used in chromatography (several milliliters per minute). This imposes constraints on the physical parameters used in eq 4. However, the specific surface area, u of the carbon foams is very high, 21 000 cm-I. It certainly satisfies the high surface area within a small volume criterion. For comparison, the specific surface area of 100 pores/in. reticulated vitreous carbon is 66 cm-1 and that of graphite felt is 133 cm-l.16 For a given material porosity, it follows that the only parameter that can be varied to any appreciable extent is the length, keeping in mind that its variability cannot be infinite because as the length of a porous electrode increases, control of the potential distribution within the cell becomes more difficult due to iR drop.l4 With this in mind, the minimum electrode length required for coulometric function is the optimum. Equation 4 can be used to predict this value. The surface area, A, of the cell is simply the product of the specific surface area, u and the cell volume, V. Substitution into eq 4 and rearranging gives V = RU/(24a2D(1- R ) )
(5)
1.2
.6
a Z
4
I
i
0 I
450
550
740
640
eQ
POTENTIAL, millivolts Figure 2. Hydrodynamic voltammogram for 0.1 11 mM ferrocene in 0.1 M TEAP/acetonltrile: flow rate 2.33 mL/mln; silver wlre pseudoreference electrode; potentlal ramp 1 mV/s. 600
co n
E 500 0 (0
1 -
L
u '2 400 -
The interstitial volume of a cylindrical porous electrode is
v = ar2Lt
H
300
(6)
H
where r is the radius and e is the electrode porosity. Substituting into eq 5 and solving for L gives
a
-
II
a f 200 0
L = RU/24scr2a2D(1- R )
(7)
0
0 100 t
-
I
0 I
0
I
1
I
2
I
3
FLOW RATE, mL/min
Using u = 21 000 cm-I, c = 0.97, r = 0.05 cm, D = 5 X 1od cm2/s, U = 2 mL/min or 0.0333 mL/s, and an R of 0.99 leads to a length of 0.008 1 cm. Unfortunately, we have been unable to manufacture electrodes this small. The physical fragility of the material has precluded experiments which test the theory. However, electrodes of practical utility have been made.
RESULTS AND DISCUSSION The first cell design studied was the one in which the porous electrode was enclosed by the stainless steel tube. The tube served to cut, enclose, and make electricalcontact to the carbon foam. Hydrodynamic voltammograms of ferrocene (Figure 2) were gathered at several flow rates. Any unwanted characteristics of a cell, such as uncompensated iR drop or a passivated surface, often are manifested as apparent deviations from ferrocene's normally reversible behavior. Regression of log(Z/(Zlim- I)) vs E, where Z1im is the steadystate current, yields a slope of 64 f 1.72 mV (r2= 0.9936), close to the theoretical value of 59.1 mV. This agreement means that iR drop is small within the volume where electrolyses occurs. Avoidance of iR drop is important to maintain the selectivityinherent in electrochemica1detection.l2 This cell design allowed us to determine that the electrode placement minimizes iR drop. In this sense, it proved useful. However, when the conversion efficiency for the voltammogram is considered, a value of 0.47 is obtained. Figure 3 1206
Analytical Chemisby, Vol. 66, No. 7, April 1, 1994
Figure 9. Hydrodynamiccurrent vs Row rate for 0.11 1 mM ferrocene, 0.1 MTEAP inacetonitrile,E= 1.OV vs silver wire pseudoreference electrode.
shows a plot of hydrodynamic current vs flow rate for the cell. At low flow rates, the experimental data approach the theoretical values. At high flow rates, the current levels off and becomes independent of flow rate. The reason for this phenomenon is unclear. It is possible that the high flow rates are inducing some channeling through or around the porous electrode. Aside from this unexpected behavior, the hydrodynamic current often oscillated slowly following an electrical or hydrodynamic perturbation. These oscillations could be electrical in nature and may result from using the large stainless steel tube as the electrical contact. Additionally, this cell did not respond well in aqueous solutions because of the electrochemical activity of the steel. For these reasons a new cell design was needed. In this design, the housing was a 1-mm glass tube, and contact to the electrode was made with a Pt wire. A similar analysis was done using the second cell design with the glass capillary tubing as the porous electrode housing. Regression of data from a hydrodynamic voltammogram of Fe(CN)& yields a slope of 69 f 2.09 mV (9 = 0.9964) for log(Z/(Zlim- I))vs E. While deviation from the theoretical is more here than in the ferrocene trials, this value is acceptable. It may be that the point contact in this cell causes more iR drop than there was in the steel-based cell.
. 1
L: .E 5000
Table 1
0
R
m
a
4
4000
I-
$
3000
-
2ooo
-
1000
-
a a
flow rate, mL/min
3.0 mm
1.0mm
0.5 mm
1.0 2.0 3.0
1.04 1.04 1.07
1.05 1.05 1.02
1.01
3 0
zz 4
z
& 0
a
D 2.
=
,
0-
I
0
.5
I
1
I
1.5
I
2 FLOW RATE, mL/min
I
2.5
I
3
Figure 4. Hydrodynamic current/concentratlonvs flow rate for three porous electrodes, all 3 mm In length. Solutions were 0.5 mM K,Fe(CN)e, 0.1 M KCI in pH 7.0 phosphate buffer. ,Eaw = 0.65 V vs Ag/AgCI. The theoretical slope Is Just nF.
To determine the influence of flow rate on conversion efficiency,a study was conducted in which the analyte solution was switched from static to flowing. The resulting hydrodynamic steady-state current was recorded at several flow rates, typically 0.33-3.0 mL/min. In theory, the ratio of current to concentration is just nFU, where U is flow rate. Figure 4 shows steady-state current/concentration vs flow rate for three porous electrodes having lengths of 3.0 mm. The solid line is the predicted value for R = 1. The analyte was 0.5 mM K4Fe(CN)6. The applied potential was 0.65 V, a value selected to be well into the diffusion-controlled regime. Over the entire range studied, the efficiency was 1.03 f 0.02 (SEM). This result is consistent with the efficiency predicted earlier. Careful attention to cell design, especially electrode placement, has made high coqversion possible. A comprehensive characterization of the cell would include pushing the limits of flow rate and cell length to explore a regime where R < 1. Such a study would make it possible to compare the cell response in a classical way to the more quantitative expression for R presented by Sioda1q4and would be intellectually satisfying. However, the Omnifit LC fittings needed in the cell were low-pressure fittings, and they could not withstand the pressure of the system when operated above 3.0 mL/min. Subsequent designs of the system will address this matter. Cells with shorter electrodes were made and the results are shown in Table 1. In each case the same 3-mm electrode was consecutively cut to shorter lengths. The conversion efficiencies remained high for all electrodes. Recall from the earlier discussion that the minimum length required for R = 1 at 2.0 mL/min is less than 0.1 mm. Given the semiquantitative nature of this approach, it was conceivable that the R < 1 regime may be obtained at L = 0.5 mm. At 1.0 mL/ min, this was not the case. At higher flow rates, the thin foam
disk began to fracture, making it impossible to obtain meaningful data. Therefore, limitations in the foam's physical properties as well as in the cell hardware make tests of theory impossible now. In light of the failure to determine R as a function of flow rate, the increase in the efficiency of this material over RVC can be determined by comparing the exposure or residence time of a solution in the cell at a flow rate corresponding to R = 1. Curran's cellI2had a physical volume of 149 pL (0.14 cm radius X 2.42 cm long). At 1.0 mL/min it had an R of 0.95. The exposure time, cell volume divided by flow rate, is 8.9 s. The physical volume of the shortest electrodes used here was 0.4 pL (0.05 cm radius X 0.05 cm long), and this was coulometric at 1.0 mL/min also. The residence time is 2.4 X l e 2 s. This is -370 times shorter than the residence time in the RVC cell. We haveinvestigated thecell in flow injection experiments, and as expected, the coulometric efficiency does not suffer. We have not yet built a cell with a low dead volume; however, some discussion of that aspect is warranted. The RVC cell with unit efficiency at 0.5 mL/min12 discussed above demonstrated a mixing volume lower than the physical volume of the electrode. The band spreading induced by this cell was 8 pL. One factor that contributed to the low dead volume/ physical volume ratio was the exponential decrease of the concentrations along the flow axis of the cell.12 Another was the good engineering of connections between the cell and the rest of the apparatus. Finally, as it is not surprising that a chromatography column with a physical volume on the order of 10 mL or more has a peak volume of 100 p L or less, by analogy we should not expect, in a porous electrode, the physical volume and the dead volume to be equivalent. Thus, with appropriate engineering of connecting tubing, unit coulometric efficiency with submicroliter dead volume should be possible with the microcellular carbon foam.
-
ACKNOWLEDGMENT We are grateful to Dr. A. Sylwester and Dr. Robert LaGasse for providing the foam. We thank NIH for financial support through Grant GM-44842. Received for review July 6, 1993. Accepted January 19, 1994.' Abstract published in Aduonce ACS Absrracts, March 1, 1994.
AnalytkalChemistty, Vol. 66, No. 7, April 1, 1994
1207
