Reproducibly Defining Electrode Area of Carbon Paper Electrodes via Machine Cutting and High-Throughput Waxing

Carbon paper is one of the most common carbon electrode materials employed in electrochemical research due to its low cost, disposability, and controllable dimensions and scaling. Carbon paper is usually hand cut and a variety of strategies are used to define electrode area. This procedure is tedious, imprecise, and inaccurate, yielding coefficients of variation in peak current output as high as 20%. Imprecision in hand-preparation translates directly to poor reproducibility in electrochemical data leading to challenges in directly comparing data across research groups and research fields. This work introduces an inexpensive and precise method to automatically cut and rapidly wax carbon paper electrodes, circumventing laborious traditional preparation and substantially improving precision in electrochemical data.

Electrodes composed of a variety of carbon allotropes have been used in electrochemistry for decades, with advantages over other electrode materials including low cost, wide potential window, inert electrochemistry, electrochemical stability, and simple chemical modification. 1 Rigid carbon electrode materials (CEM) include glassy carbon, pyrolytic graphite, and graphite, while less rigid variants from spun and resin-bound carbon fiber electrodes such as carbon felt, carbon cloth, and carbon paper (CP) electrodes have been introduced in recent decades. These provide unique advantages such as higher specific surface area, improved biocompatibility, low cost, disposability, and controllable dimensions, but with the challenge of commercial materials being available mainly as unprocessed sheets. [2][3][4][5] The consequence of this commercial reality is that electrodes are hand prepared by the researchers themselves, leading to poor reproducibility of electrochemical data and challenges in comparing data across different researchers, research groups, and research fields. Because CP can more easily be approximated as a two-dimensional surface compared to its thicker and more porous carbon fiber cousins, it is perceived as having greater precision in electrochemically assessable surface area (A E ), leading it to be the most widely employed carbon fiber electrode material for research in (bio)fuel cells, electrolyzers, (bio)sensors, and supercapacitors. [5][6][7] Fuel cell research exploring novel catalytic systems and employing CP electrodes focuses chiefly on surface-immobilization of catalysts aimed at maximizing current density, long-term stability, and energy conversion efficiency. CP based fuel cells have been developed with a wide range of catalyst species including: molecular, nanomaterial, microbial, organelle, enzymatic, and hybrid catalysts. [8][9][10][11][12] Like fuel cells, sensor research focuses on novel catalysts and immobilization architectures and resulting current densities, but with additional concern for precise correlation of current density and substrate concentration, as well as minimizing limit of detection (LOD), and improving analyte selectivity to specific metabolites or industrial contaminants. 5 The two fields, therefore, have uniquely overlapped interests and a capacity to crossinspire novel applications, with inter-field communication limited by factors including the time and resource investment of preparing fuel cell data with sensor-level precision.
In the typical procedure for preparing CP into electrodes, commercial pieces of carbon paper are hand-cut into pieces appropriate for a researcher's electrochemical assembly, with all but a 0.25 to 5 cm 2 functional region at one end covered or coated in a protecting agent such as wax to define the electrode area. In addition to reproducibility issues with these techniques, the handpreparation of electrodes is tedious, time-consuming, and skillintensive. Although laser-cutting machines can maximize precision, this comes with a monetary and temporal cost beyond the budget or priority of many laboratories, much less for proof-of-concept fuel cell or biosensor projects. [13][14][15][16] CP electrodes prepared using a craftcutting machine as described in this piece have also been described in literature, [17][18][19] but a standardized protocol for both cutting and defining electrode surface area has not. This work introduces a method using an inexpensive, table-top cutting machine to cut CP electrodes up to an order of magnitude faster and more precisely than the hand-cutting method. A high throughput protocol for precisely waxing electrodes as well as an executable that converts electrode measurement specifications into input images for cutting machines, AutoCutter, are also introduced. Precision and accuracy compared hand-cut electrodes are demonstrated through microscopy and electrochemical methods. A method to evaluate the quality of CP manufacturing methods is also described.

Experimental
Chemicals.-4-Amino-2,2,6,6-tetramethylpiperidine-1-oxyl (amino-TEMPO) was purchased from TCI. Sodium citrate and sodium phosphate was purchased from Sigma-Aldrich. AVCarb MGL 190 carbon paper was purchased from Fuel Cell Store. Household paraffin wax was purchased from Gulf Wax. The Silhouette Cameo cutter was employed in machine-cutting of electrodes and a VWR 89032-210 digital water bath was employed in the waxing of electrodes. All images for machine cutting were prepared using the software AutoCutter that can be accessed at github.com/MinteerLab/AutoCutter.
Approximations of electrode functional area.-The peak current (i p ) of a freely-diffusing redox species in a cyclic voltammetry experiment is described by the Randles-Ševčík equation: 20 The term A E requires special attention, because it is a significant source of imprecision in electrochemical data contributing to the gap between fuel cell and sensor research. The A E of an electrode, which is the net surface area of carbon fibers directly interacting with a redox species on the surface of-and within the volume of the electrode, is proportional to the electrode's geometric functional area (A g , the surface area of unprotected electrode which contains these fibers) and the CP's material factors including thickness, porosity, structural resin properties, etc. (comparison of A E vs A g illustrated in Fig. S1). It should be noted that in the case of CP electrodes, it is common to refer to only one side of the electrode when reporting its A g despite both sides of the electrode contributing to catalytic current and, therefore, its A E . To improve a dataset's direct comprehensibility, cyclic voltammograms are often reported not as current i (amps) but current density (j, amps cm −2 ), by dividing the output current by A g . It is, therefore, common practice to ignore these material factors and approximate the electrochemical area as the geometric area (A E ≈ A g ), since it is understood that an electrochemical measurement is specific to the electrode material, and consequently its material factors. An equally common but much more problematic assumption is that CP electrodes may be prepared by hand with sufficient precision and accuracy that an electrode's geometric area may be approximated by the target geometric functional area (A t ), which the electrode preparer intended to produce when cutting and waxing the electrode (A g ≈ A t , issues with this approximation Illustrated in Fig. S2), without precisely measuring the electrodes after fabrication either microscopically or electrochemically. This approximation is much more common in fuel cell research, but as will be demonstrated in this piece, neither of these approximations are necessarily valid.
Electrode preparation.-Hand-cut electrodes were prepared using an X-Acto knife to split 10 × 10 cm pieces of CP into 0.5 × 10 cm strips using a ruler, subsequently split into four 0.5 × 2.5 cm strips before marking the functional area with a notch 0.5 cm from one end. To wax these electrodes, paraffin wax was heated in a stainless-steel crucible on a hotplate at medium heat until the wax was melted, then cooled at low heat for approximately 30 min before electrodes were submerged in the melted wax until the wax front reached the functional area notch.
Machine cut electrode cut files were generated using the program AutoCutter, and electrodes were prepared using a Silhouette Cameo cutting machine and waxed with paraffin between titanium plates using an aluminum trough in a hot bath as described in detail in the supplementary material.
Microscopy.-Microscopic scans of prepared electrodes were performed using a Keyence VHX-5000 using the image-stitch function at the University of Utah Electron Microscopy and Surface Analysis Lab, or an in-house Renishaw InVia at 10× magnification using the snap montage function. The wax front in scanned images was labeled in Adobe Photoshop to distinguish the waxed from the functional region, which was then measured in ImageJ using the "convert to mask" option and counting pixels using the "measure" function. Repeating wax front delineation and measurement three times on nine randomly selected electrodes revealed an imprecision in wax front determination of 1.5 × 10 −3 cm 2 .
Cyclic voltammetry.-Cyclic voltammetry was performed using an Admiral Squidstat Prime using a standard three-electrode cell consisting of the CP working electrode, a saturated calomel electrode (SCE) reference electrode, and a Pt mesh counter Figure 1. example microscopic scans of hand cut (above) and machine cut (below) electrodes. Compare the wavy, gradual shaded region (wax front) above the functional area notch in the hand-cut electrode, vs the clean, linear wax front in the machine-cut electrode. electrode. Measurements comparing machine and hand-cut electrodes were conducted in citrate-phosphate buffer (25 mM each, pH 8.2) containing amino-TEMPO (1 mM), at a scan rate of 10 mV s −1 at 20°C. Hand cut electrodes were cut to 1.9 cm to match the machine-cut electrodes more closely and eliminate a thicker wax region, which develops in the hand-waxing process. Cyclic voltammetry for the determination of A E vs A g was conducted in the same conditions except at pH 6, as discussed in the discussion section. Calculated A E values were divided in half in accordance with the convention of referring to only one side of a CP electrode when determining its A g . A diffusion coefficient of 3.77 × 10 −6 cm 2 s −1 for amino-TEMPO was used in A E calculations. 21

Results and Discussion
Precision and accuracy in hand-cut electrodes.-A sample set of 77 electrodes with an A t of 0.25 cm 2 were hand-cut by 7 participants (9-12 electrodes each), including two undergraduate students, two graduate students, and three postdocs, and analyzed by microscopy to quantify A g values (Fig. 1). The mean A g of electrode sets prepared by individuals varied from 0.239 to 0.298 cm 2 within sets ( Fig. 2A) with standard deviations (SD) of 0.02 to 0.04 cm 2 (Coefficient of variation (CV) = 8.5% to 12.6%). The complete set of hand-cut electrodes showed a mean A g of 0.26 ± 0.03 cm 2 (CV = 12.6%).
The same set of 77 hand-cut electrodes was employed in cyclic voltammetry experiments (Fig. 3) of amino-TEMPO and the firstcycle peak current (Fig. 2B) was used to calculate A E values (Fig. 2C). Hand-cut electrodes demonstrated mean peak-current (i p ) values between 32 and 48 μA translating to mean A E values of 0.28 to 0.46 cm 2 with standard deviations of 0.03 to 0.05 cm 2 and coefficient of variation values from 9.1 to 14.4%. The mean values for the full set of 77 hand-cut electrodes were 39 ± 8 μA translating to 0.37 ± 0.08 cm 2 (CV = 20%).
Precision and accuracy in machine-cut electrodes.-A set of 46 electrodes prepared by 4 participants (10-12 electrodes each, three graduate students and one postdoc) using the machine-cutting and plate-waxing protocol (MC) from images generated using AutoCutter demonstrated mean A g of 0.239 to 0.253 cm 2 with SDs from 0.005 to 0.01, with an overall mean for the set of 0.244 ± 0.009 cm 2 (CV = 3.5%). Cyclic voltammetry of the MC electrodes demonstrated a mean i p of 24 ± 1 μA translating to a mean A E of 0.23 ± 0.01 cm 2 (CV = 4.8%).
Variation in geometric functional area.-Even macroscopic observation of traditionally hand-cut and waxed electrodes reveals significant variation in electrode width, while a close analysis of the wax front reveals significant, regular deviation from the intended  position at the functional area notch (Fig. 1). The MC waxing procedure reliably produces a visibly "clean" wax front if guidelines are closely followed. Furthermore, the use of a cutting machine eliminates tedious measuring and cutting procedures while producing electrodes both an order of magnitude faster and more precisely than the traditional procedure.
A common conception in laboratories is that the difference between the mean of repeated, targeted measurements and a target magnitude decreases with increasing replicates. However, the deviation of hand-cut sample means demonstrates that both the precision and accuracy of hand cut electrodes are extremely userdependent and do not necessarily improve with more replicates or preparers ( Fig. 2A). Employing the MC process in contrast, produces electrodes which are more precise, with mean A g deviating by ±0.03 cm 2 for hand cut vs ±0.009 cm 2 for MC electrodes, and more accurate, with average |A g -A t | = 0.01 cm 2 for the total mean of hand-cut, up to 0.05 cm 2 for the means of individual sets, vs 0.009 cm 2 for the set of MC electrodes, up to 0.01 cm 2 for individual preparers (Table I). One may be tempted to blame poor hand-cutting performance on the diversity of skill levels in the dataset, but two of the three most geometrically accurate sets were prepared by undergraduate students with no prior cutting experience, while experience showed no notable relationship with precision.
Variation in electrochemically active surface area.-The electrochemical determination of A E reveals even more concerning unreliability in the traditional hand-preparation procedure, and the 20% CV in hand-cut A E is particularly haunting (Table I). This CV indicates that two researchers can prepare electrodes from the same brand of CP and produce mean output i p over 40% apart in the same electrolyte solution. Indeed, the lowest and highest mean i p output from hand-cut 0.25 cm 2 electrode sets (9-12 replicates) were 30 and 48 μA, a difference of 60%.
An initial attempt was made to determine A E via double layer capacitance (C dl ), but it was determined that although the wax layer prevents the faradaic transfer of electrons to the electrolyte, it does not appear to prevent longer range non-faradaic interactions contributing to C dl (Fig. S3). Critically, this reveals that the A E of CP electrodes cannot be determined electrochemically without potential contamination with redox-active species, emphasizing the importance of a CP electrode fabrication method with high accuracy and minimal variation. Due to confounding interactions between pH and i p in TEMPO described in the following section, the absolute A E values calculated in this experiment may be universally larger or smaller than the actual values for these electrodes. However, because this pH effect can be expected to modify all i p values to a similar magnitude, it would have little effect on the variation calculated within and between these electrode sets, or the demonstration that the MC protocol yields significantly more precise A E and A g values within and between preparers when compared to the traditional procedure.
Electrochemically active surface area as a function of geometric functional area.-To determine the precise relationship between A g and A E for this batch of CP, 6 sets of 4 electrodes each at A t between 0.3 and 0.9 cm 2 were prepared using AutoCutter and had their A g values determined microscopically before being employed in cyclic voltammetry experiments with A E values (Fig. 4) calculated from the peak height of the first cycle i p . Cyclic voltammetry was performed in the presence of amino-TEMPO (1 mM) in citrate-phosphate buffer (25 mM each, pH 6), at 10 mV s −1 , and 20°C. The line of best fit A E = (1.19 ± 0.02) A g -(0.02 ± 0.02).
Several points should be noted regarding the determination A E /A g. First, the A E /A g of hand-cut electrodes cannot be determined below 23% uncertainty due to the compounding contribution of deviation in A E and A g to the term, so the MC process enables a new way for manufacturers to evaluate critical material parameters of a given CP production process. A E /A g was determined at a lower pH than in experiments comparing A E in MC vs hand-cut electrodes (pH 6 and 8.2, respectively) to avoid confounding interactions between amino-TEMPO, CP, and pH. A pH of 6 was chosen as higher pHs led to lower i p output and decreased inter-cycle stability of amino-TEMPO, while lower pHs led to the apparent loss of CP stability (Fig. S4). These findings are in line with reports of the decay of TEMPO derivatives in acidic and basic conditions. 22,23 The lack of universality of this A E /A g value between different pHs does not hamper the precision of its determination, however (CV of A E /A g = 6.3%), or the capacity of the MC process to serve as an invaluable tool for researchers to compare A E /A g values emerging from different CP manufacturing processes.

Conclusions
The accuracy and versatility of the machine-cutting plate-waxing process for carbon paper electrode preparation was demonstrated through microscopic and electrochemical methods. The MC process produces electrodes with lower coefficient of variation in both A g (3.5% vs 12.6% for hand-cut electrodes) and A E (4.8% vs 20% for hand-cut electrodes), while minimizing demonstrated user-specific bias in hand-cutting. The MC process is also significantly faster and less tedious than traditional hand-cutting, producing 64 electrodes with a 0.25 cm 2 functional area in about 20 min, plus 10-20 min for waxing. The precision of the MC process enables the determination of the ratio of geometric to electrochemically active surface area, serving as a tool for carbon paper manufacturing. The superior convenience, minor cost, and significantly improved reliability of resulting electrochemical data strongly suggest that this method should be adopted anywhere laser-level precision in functional area is not necessary or is too costly. The superior precision and accuracy of electrodes prepared with this method will aid communication between different areas of electrochemistry leading to growth in the field overall.