Evaluation of in Vitro Serotonin-Induced Electrochemical Fouling Performance of Boron Doped Diamond Microelectrode using Fast-Scan Cyclic Voltammetry

1. Introduction

The central nervous system (CNS) uses chemical messengers called neurotransmitters to transmit information between neurons [1]. Many neurologic and neuropsychiatric disorders cause dysregulations of neurotransmitter function, and the neurological underpinnings of behavior can be uncovered by studying neurotransmitter dynamics [2,3,4,5,6,7,8,9]. Fast scan cyclic voltammetry (FSCV) is a powerful method to study the real-time release and reuptake of electroactive neurotransmitters. FSCV is an electrochemical method that typically uses a carbon fiber microelectrode (CFME) referenced to a Ag/AgCl electrode to detect redox currents from electroactive analytes by applying a cyclic potential waveform at high scan rates. This method utilizes the magnitude and voltage associated with the redox current peaks to estimate the concentrations and identity of an analyte, respectively [10]. FSCV is particularly useful for the detection of monoamine neurotransmitters, such as catecholamines (dopamine (DA), norepinephrine, and epinephrine) as well as indolamines (serotonin (5-HT) and histamine) [11,12,13,14,15,16].

Monoamine neurotransmitters can produce reaction products following oxidation. In particular, 5-HT induces the formation of polymerizing oxidative byproducts that have a higher adsorption affinity to the CFME than 5-HT itself, disrupting electron transfer and subsequent signal detection [15,17,18,19]. This phenomenon, known as “electrochemical fouling”, is separate from “biofouling” which arises from adsorption of other biological interferents to the electrode surface such as blood, damaged tissue, etc. [10,20,21,22]. Both types of fouling can interfere with in vivo FSCV signal detection, especially in the chronic setting where exposure to fouling agents may be prolonged. A variety of strategies have been explored to combat electrode fouling. Generally, these strategies treat both types of fouling as one phenomenon, but some groups, such as Weese et al. have shown that the two types of fouling may have unique requirements to best mitigate them. They found electrochemical fouling resistance can be achieved with electrodes that have a pristine surface, whereas biofouling resistance is achieved with functionalization with hydrophilic groups, indicating a different mechanism between the two fouling types [17]. Generally, to resist both types of fouling, researchers have explored new carbon materials, coatings, and alternative waveform parameters. Alternative carbon materials such as carbon nanotubes (CNTs) [17,23,24], nanodiamonds (NDs) [24,25,26], and boron-doped diamond (BDD) [21,27,28,29,30,31,32,33,34,35,36] have shown promise in reducing fouling by leveraging their unique surface characteristics, (e.g., polarity, defect sites, and carbon-bonding structures). Various coatings have been developed to reduce fouling, including Nafion, base-hydrolyzed cellulose acetate, and fibronectin [20]. Several studies have explored adjusting waveform parameters to mitigate both biofouling and electrochemical fouling through various mechanisms [14,21,37,38,39,40]. Oxidative etching driven by increased switching potentials (> 1.1 V) will constantly refresh the CFME surface, which can remove fouling agents of both types, but will degrade the electrode over time [41]. The “Jackson” N-Shaped waveform was designed to outrun and reject adsorption of the oxidative byproducts of 5-HT by altering the scan rate and potentials of the waveform to avoid the adsorption of byproducts while maintaining sensitivity and selectivity for 5-HT [15]. Likewise, our recent report indicated that when using the Jackson waveform, the effects of biofouling were less pronounced on boron-doped diamond microelectrodes (BDDMEs) in comparison to CFMEs [21].

Our group is developing a freestanding, all-diamond, batch fabricated BDDME (Figure 1) as an alternative to the CFME to potentially mitigate fouling while reducing etching [21,28,42]. The device is batch fabricated using a wafer-scalable process which can produce hundreds of identical microelectrodes simultaneously without the need for cumbersome manual assembly. The BDDME is a fiber-style electrode grown on silicon wafers using a combination of microwave chemical vapor deposition (CVD) to grow the BDD and hot-filament CVD to insulate the conductive core with polycrystalline diamond (PCD). This approach enables readily customizable geometries to be fabricated to meet the needs of various sensing configurations. Generally, BDD also provides several other purported benefits when used for electrochemical sensing such as a wider working potential window, stable background currents, and improved resistance to mass loss due to oxidative etching [27,31,42,43]. Many of these desirable characteristics arise from diamond’s sp³ hybridized carbon bonding structure, which can create a tradeoff in sensitivity from the sp² hybridized bonding of other carbon materials [44].

Our previous work with our BDDME includes an initial characterization [42] and explorations into FSCV waveform development and biofouling performance [21,39]. Given our confirmation that the BDDME does resist biofouling relative to the CFME, we wanted to test the electrochemical fouling performance of our electrodes. Concurrently, while exploring fabrication techniques to improve the sensitivity and consistency of our devices, we chose to characterize femtosecond laser cutting as an alternative method to expose the BDD electrode site at the fiber tip. In previous work, we completed this fabrication step by physically cleaving the tip of the electrode with a scalpel. Femtosecond lasers, often used in ophthalmic surgery, offer an extremely precise, focused laser which is known for its reduced propensity to damage unintended areas surrounding the focal point through minimized thermal flux [45]. We are also exploring the femtosecond laser for an “on-wafer” fabrication step, automating a portion of fabrication normally done by hand utilizing using motorized translation stages.

This study compared 5-HT electrochemical fouling performance between the CFME, the physically cleaved BDDME (C-BDDME), and the femtosecond laser-cut BDDME (FS-BDDME). After initially testing differences in sensitivity, the electrochemical fouling was explored for both DA and 5-HT with similar methods to previous work [14,17] using both a moderate concentration of 5-HT (5 µM) and a large concentration (50 µM). Additionally, we provide an analysis of these devices’ long-term stabilities to further explore the BDDME’s suitability as chronic FSCV sensor, where long term stability is especially important. We found the following: (1) The C-BDDME resists electrochemical fouling from 5-HT compared to the CFME and the FS-BDDME at both 5 and 50 µM 5-HT, (2) FS-BDDMEs have sensitivities comparable to CFMEs, but do not maintain the fouling resistance and stability of C-BDDMEs, (3) the C-BDDME has a stable background when exposed to the DA FSCV waveform at 60 Hz for 24 hours whereas the CMFE and FS-BDDME both experience changes. We intend this work to be additive to efforts toward viable, chronic neurochemical sensors.

2. Materials and Methods

2.1. Chemicals

All chemicals were purchased from Sigma-Aldrich, Inc. (St. Louis, MO, USA) and Fisher Scientific International, Inc. (Hampton, NH, USA). Stock solutions of 1 mM DA and 1 mM 5-HT were prepared in 1 mM perchloric acid to prevent degradation. Dilutions of DA and 5-HT were prepared in “Tris” artificial cerebrospinal fluid (aCSF) (pH 7.4; 25 mM Trizma Buffer, 126 mM NaCl, 2.5 mM KCl, 1.2 mM NaH₂PO₄, 2.4 mM CaCl₂, 1.2 mM MgCl₂) [22]. Solutions of 1 mM ferrocene carboxylic acid (FcCOOH), an electroactive compound with a symmetrical, well-defined redox response, were prepared in Tris aCSF and used to test for optimal placement of microelectrodes in the flow injection setup prior to measurements. All solutions were prepared with ultrapure water: 18.2 MΩ.cm, TOC < 5 ppb (Barnstead™ GenPure™ xCAD Plus Ultrapure Water Purification System, Thermo Scientific, Waltham, MA, USA).

2.2. Carbon Fiber Microelectrode (CFME) Fabrication

CFMEs were constructed similarly to previously reported methods [21]. Briefly, individual 7.4 μm diameter, unsized, AS4 carbon fibers (Hexel, Stamford, CT, USA) were aspirated into glass capillaries (World Precision Instruments, Sarasota, FL, USA) using a vacuum pump. The capillaries were pulled with a vertical micropipette puller (Stoelting Co., Wooddale, IL, USA). Electrical connections were made by coating 32 AWG silver-plated wire with PELCO conductive carbon-based glue (Ted Pella, Inc., Redding, CA, USA) and inserting the wire into the open end of the capillary. After 24 hours, the wire was epoxied in place. The carbon fibers were cut to an approximate 100–150 μm exposed length measured from the glass seal. All CFMEs were allowed to stabilize for 20–30 min using using FSCV by applying the standard cyclic waveform of −0.4 V to 1.3 V at 400 V s−1, 60 Hz frequency in Tris aCSF, and then allowed to finish stabilizing for 10 min with a reduced frequency of 10 Hz before being used for data collection.

2.3. Boron Doped Diamond Fabrication

The all-diamond boron doped microelectrodes were fabricated using a similar method to our previously published fabrication pathways [21]. Briefly, BDD was grown on a 4″ diameter, 500 µm thick single-side polished silicon wafer using a 915 MHz microwave chemical vapor deposition reactor. Diamond growth synthesis conditions include a 9 kW microwave power with a 900 °C stage temperature and a chamber pressure of 60 Torr with a gas chemistry of 2% methane. During growth, diborane was added to the diamond grown at a B/C ratio of 37,500 ppm to ensure high conductivity. After BDD growth, to structure the BDD, titanium and copper (Ti: 10 nm/ Cu: 500 nm) were deposited using electron-beam evaporation. Photolithography was used to pattern the metal hard mask (ABM-USA, Inc., Jan Jose, CA, USA), with wet chemical etching and the diamond was structured using reactive ion etching. The diamond electrodes were then released from the silicon wafer using an HNA etchant with an HF:HNO₃:CH₃COOH composition of 5:11:6, and fully insulated with polycrystalline microcrystalline diamond using hot filament chemical vapor deposition (HF-CVD), encasing the microelectrodes in > 10 µm intrinsic diamond. Microcrystalline diamond was grown using a base pressure of 35 Torr and 2% methane. After deposition, both ends of the electrodes were physically cleaved to expose the BDD core on both the tip and connection pad, and the electrical connection was made using the conductive carbon glue (Ted Pella, Inc., Redding, CA, USA). Electroactive areas for the diamond cores ranged from 100 to 200 µm² based on a 50 µm and 70 µm wide lithography pattern, and a BDD growth thickness of ~4 µm.

Femtosecond laser cutting of the tip was done using a similar setup as our previously fabricated carbon fiber microelectrodes [46]. Briefly, an Astrella-USP-1 K (Coherent Corp. Santa Clara, CA, USA) 800 nm, 1 kHz, 5 W system with a lab assembled 3-axis stage was used to slice the tips of the microelectrodes [47]. Laser power was attenuated to 400 mW and diamond was cleaved by adjusting the y-axis rapidly by hand, making 3-4 passes across the electrode tip.

2.4. Fast-Scan Cyclic Voltammetry (FSCV)

A two-electrode setup (a working electrode versus a combined quasi-Ag/AgCl reference/counter electrode) was utilized in a custom flow injection cell for FSCV experiments. A custom potentiostat with a variable gain headstage (50 nA/V, 100 nA/V, 200 nA/V, 500 nA/V, 1 μA/V) was connected to the electrodes for experimentation. Data were collected using a NI-6363 data acquisition card and High-Definition Cyclic Voltammetry (HDCV) software (Version 4, Department of Chemistry, University of North Carolina, Chapel Hill, NC, USA) [48]. The data were filtered using a Bessel 4th order lowpass filter at 2000 Hz. For all experiments, the flow injection system used a TTL voltage-controlled source to switch a six-way HPLC valve to introduce a bolus of test analyte. A flow rate of 750 µL min⁻¹ was used to deliver Tris aCSF buffer by a NE-1000 syringe pump (New Era Pump Systems, Inc., Farmingdale, NY, USA). All experiments were performed in a custom flow injection cell except for the 24-hour stability tests. The “standard” waveform (-0.4 V to 1.3 V to -0.4 V at 400 Vs^-1) was used for all experimentation applied at 10 Hz except for 24-hour testing where it was applied at 60 Hz. All data points are the average of 3 consecutive oxidation peaks unless otherwise noted. All analyte injections are captured in a 30 s window, with the flow-injection valve opening at 5 s and closing at 15 s. Data were background subtracted to remove the non-faradaic current component.

Electrodes were calibrated with 5-HT (CFME and FS-BDDME: 0.025 – 1 µM, C-BDDME: 1 -100 µM). A simple linear regression (Microsoft Excel LINEST function) was used to find the slope, linearity, and to calculate the limit of detection (LOD) of different device types by dividing the Y-variance by the slope of the best fit line using the equation:

$$\mathit{L}\mathit{O}\mathit{D}=\frac{3.3\mathit{\sigma }}{\mathit{m}}$$

(1)

Where $\mathit{\sigma }$ is the standard deviation of the y intercept of the best fit response and $\mathit{m}$ is the slope of the calibration curve. This method is as described in [49], the “Validation of Analytical Procedures: Text and Methodology (ICH Q2(R1))”. Smaller concentration ranges than the entire working ranges were used to maximize linearity (CFME and FS-BDDME: 0.025-0.2 µM 5-HT, C-BDDME: 2-20 µM 5-HT) [50].

2.5. Electrochemical Fouling Protocol

Serotonin fouling was carried out similarly to work published by Weese et al. in 2019 [17]. The working electrode was first exposed to 25 sequential injections (single data points) of 5 µM DA as a non-fouling control followed by 25 injections of 5 µM 5-HT. New 20 mL vials of analyte dilutions were prepared for every 5 injections to avoid degradation. The injection syringe was refilled as quickly as possible every 5 injections to avoid excessive time in between sets of 5 injections so that self-cleaning is minimized. Due to the C-BDDME’s signal in response to 5 µM 5-HT being on the same order as normal system noise (~1 nA), these experiments were repeated with 50 µM DA and 5-HT to generate larger signals with minimized noise interference. Current values were reported as a percentage of the first injection’s peak oxidation current. Electrodes were immediately removed after the 25^th 5-HT injection to attempt to preserve visible fouling agents during imaging.

2.6. Electrode Stability Analysis

To assess the response repeatability of the devices, 5 µM 5-HT boluses were injected every 10 minutes for 120 minutes, leaving the electrode in place with the waveform applied in between injections to allow for surface renewal [41]. To assess longer-term stability, electrodes were placed in a small beaker of Tris aCSF for 24 hours with the standard waveform applied at 60 Hz to simulate six days of constant waveform application at 10 Hz. The electrochemical background was collected once at the beginning and once at the end of the experiment.

2.7. SEM Imaging

Scanning electron microscopy (SEM) images were collected using a JSM-6610LV SEM (JEOL Ltd., Tokyo, Japan) in the Michigan State University Center for Advanced Microscopy.

2.8. Statistics

Raw data were extracted to Excel (Microsoft Inc., Redmond, WA, USA) from the HDCV software, and analysis of limits of detection (LODs), standard t-tests, and the linear mixed model ANOVA were performed using Excel (Microsoft Inc., Redmond, WA, USA), Graphpad Prism 10 (Graphpad Software Inc., La Jolla, CA, USA), and SPSS (IBM, Armonk, NY, USA), respectively. Statistical significance was defined at the p < 0.05 level. All error bars are derived from the standard error of the mean. Figures were produced using Graphpad Prism 10 and Microsoft Office.

2.9. Hot-Acid Boiling

BDDMEs (C-BDDMEs and FS-BDDMEs) were removed from their attached PCBs by soaking in methanol for several minutes. The detached BDDMEs were then added to a 500 mL beaker. A total volume of 15 mL of acids were then added to the beaker. 5 mL of ${\mathit{H}}_2\mathit{S}{\mathit{O}}_4$, 5 mL of $\mathit{H}\mathit{N}{\mathit{O}}_3$, and 5 mL of ${\mathit{H}\mathit{C}\mathit{l}\mathit{O}}_4$ were pipetted into the 500 mL beaker ensuring the BDDMEs were completely submerged. ${\mathit{H}}_2\mathit{S}{\mathit{O}}_4$ was added to the beaker first, then the oxidizers were added subsequently onto the inner sides of the beaker to prevent the exothermic reaction from overheating. A watch glass was placed on top of the beaker and slowly heated to around 230 °C over the course of 30 minutes. After 30 minutes, the solution was allowed to cool then slowly diluted with deionized water. The BDDME’s were then removed from the diluted acid with a small paint brush and reattached to PCBs with conductive carbon glue (Ted Pella, Inc., Redding, CA, USA). FSCV backgrounds were collected before and after acid boiling.

3. Results

3.1. Serotonin Response

5-HT is known to exhibit an attenuated response using FSCV with repeated exposure [15,17,19]. Figure 2A-2C depicts the oxidation-peak reduction during a 30 s recording window between the 1^st and 25^th injection of 5 µM 5-HT across the three device types tested. All three devices show marked reductions in peak current by the 25^th injection with the C-BDDME showing the lowest percent reduction (Figure 2B), the CFME with the second least (Figure 2A), and the FS-BDDME with the largest reduction (Figure 2C). Notably, the current magnitudes are highly variable between the three devices. The CFME shown here has a peak current of about 150 nA, the C-BDDME has a peak current of approximately 1.5 nA, and laser cutting brings the peak current magnitude of the BDDME to approximately 75 nA.

3.2. Electrode Calibrations

Electrodes were calibrated using increasing concentrations of 5-HT to characterize their individual responses. Figure 3 shows the individual calibration curves with their linear regression lines and equations. Due to sensitivity differences between the devices, the concentration ranges of 0.025 – 1 µM for CFMEs and FS-BDDMEs and 1 -100 µM for C-BDDMEs were chosen to reflect their working ranges. All three device types exhibited a highly linear concentration response, with R² values of 0.996 for the CFME (Figure 3A), 0.998 for the C-BDDME (Figure 3B), and 0.996 for the FS-BDDME (Figure 3C). The LODs are as follows: 0.019 µM for the CFME, 1.31 µM for the C-BDDME, and 0.021 µM for the FS-BDDME calculated using Eqn. (1). The three device types have variable sensitivities. The CFME has a sensitivity of 63.3 nAµM^-1, the C-BDDME has a much lower sensitivity of 0.118 nAµM^-1 , likely as a result of its much smaller electroactive surface area and less-adsorptive, higher sp³/sp² hybridized carbon ratio [21,27,28], and the FS-BDDME, functionalized by laser cutting, exhibits a sensitivity much closer to the CFME of 39.4 nAµM^-1.

3.3. Electrochemical Fouling with 5 µM DA and 5-HT

For all three electrode types, repeated injections showed minimal reductions in DA peak-oxidation currents by the 25^th injection and marked reductions in the 5-HT peak-oxidation currents (Figure S3, 4A, 4D, 4G). Slight increases in current following each fifth injection result from solution exchange (see Section 2.5). The CFMEs had an average 31.8 ± 2.4% drop from the first to the last injection (Figure 4B, 4C, n = 6), the C-BDDMEs exhibited a 25.6 ± 7.3% reduction (Figure 4E, 4F, n = 4), and the FS-BDDMEs had a 43.9 ± 3.4% drop (Figure 4H, 4I, n = 5). When comparing responses between device types, the 5-HT injection response curves were all significantly different from each other, with the FS-BDDMEs fouling the most, CFMEs second, and the C-BDDMEs fouling the least (P < 0.001, linear mixed model ANOVA with Bonferroni post-hoc test). The individual 5-HT percent drops from the 1^st to 25^th injections were found to be significant for all device types. The individual DA percent drop reached statistical significance only on the CFME. In Figure 4E, it can be observed that the C-BDDME data are more variable than data collected on the other two device types, particularly with DA. This is likely due to the response amplitude being of the same order as system noise [51,52]. While the 5-HT data are clear enough to determine a trend, we chose to repeat these experiments with 50 µM concentrations to better reveal electrochemical fouling responses for the C-BDDME.

3.4. Electrochemical Fouling with 50 µM DA and 5-HT

As expected, the C-BDDME displayed a less variable response to 50 µM neurotransmitter concentrations than it did to 5 µM (Figure 4E and 5E), allowing a fouling effect to revealed with 5-HT. All three device types exhibited exacerbated 5-HT fouling than with lower concentrations. The 5-HT CVs (Figure 5 A, D, G) show an almost complete absence of the redox peaks by the 25^th injection relative to the 1^st unlike the stable DA CVs (Figure S3). The CFME average 5-HT peak-oxidation current drops from 253 nA to 25.9 nA, the C-BDDME drops from 6.4 nA to 1.3 nA, and the FS-BDDME drops from 224 nA to 30.9 nA. Additionally, the rate of decay of the oxidative peak current is much more pronounced for all devices, particularly with the CFME. The CFMEs had an average 90.1 ± 1.1% peak 5-HT oxidation drop (Figure 5B, 5C, n = 4), the C-BDDMEs reduced by 75.5 ± 9.7% (Figure 5E, 5F, n = 4), and the FS-BDDMEs had an 82.1 ± 4.9% drop (Figure 5H, 5I, n = 2). Similarly to the 5 µM evaluation, there were significant differences in the serotonin trajectories between the three different devices (P < 0.001, linear mixed model ANOVA with Bonferroni post-hoc test). As for the 5 µM 5-HT test, the FS-BDDMEs had the largest peak oxidation current percent drop relative to the first injection, followed by CFMEs and the C-BDDMEs. The individual 5-HT percent drops from the first to last injections were all found to be significant (P < 0.05, paired, two-tailed t-test). No statistically significant effects were observed with DA results for 50 µM tests.

3.5. Electrode Stability Analysis

The long-term stability of the devices was explored through testing the 5-HT responses over the course of 2 hours. Peak oxidation currents were normalized to the 0-minute data point similarly to the 5-HT fouling experiments. Both the CFME and the C-BDDME exhibit a stable 2-hour response repeatability, with no significant difference found between the first and final data points (Figure 6A and 6B). The FS-BDDME’s (n=3) peak current percentage over the 120-minute experiment trended downward, with change in current of -0.0017 % min^-1 culminating in a final value that is approximately 80% of the original current (Figure 6C). In addition to this, 24-hour etching tests at 60 Hz with the DA waveform were completed to assess material longevity and the evolution of the electrochemical background, which was assessed here to reveal potential mechanisms related to the electroactive surface area of the device and its sensitivity [10,41,53]. Figure 6D shows the evolution of the average of three CFME backgrounds after a 24-hour, 60 Hz waveform application to simulate six days of constant use. The peak background current grew by 170.2 nA at the data point which corresponds to a portion of the background near the DA and 5-HT oxidation peaks on the CFME (0.6 V on the first half of the cyclic potential waveform) (Figure 6D). As demonstrated by Takmakov et al., the application of switching potentials of > 1.1 V grows the CFME background over time by creating defects as a result of etching that increase the number of adsorption sites [41]. The C-BDDME maintained a very stable background over the course of the experiment with only a very slight increase of ~3.9 nA (Figure 6E), which is to be expected given BDD’s inherent resistance to etching and wider working-potential window [27]. In alignment with the results of Figure 6C, FS-BDDMEs show a time-dependent decrease in sensitivity. Unlike the other two devices, the FS-BDDME background is reduced after 24 hours (~141.3 nA reduction, Figure 6F). It is also notable that the FS-BDDMEs recordings lost signal stability during testing, resulting in an unusable signal on all devices by the end of the 24-hour experiment (Figure S1). The other two electrode types remained stable.

4. Discussion

Neurochemical sensing can be a powerful way to improve the understanding and treatment of neurological behaviors and conditions. Using FSCV, real-time insight into neurochemical dynamics can be elucidated. However, there are limitations for chronic applications of this technique, including electrode fouling. Biofouling is considered as the interference of any unwanted biological material that adheres to the electrode sensing surface such as proteins, blood, soft tissue, etc. Electrochemical fouling is fouling induced by oxidative byproducts of analytes that outcompete for adsorption sites on the sensing surface. This is a well-known phenomenon affecting 5-HT detection when using the standard waveform with FSCV and CFMEs [15,17,19]. In order to reduce the effects of fouling, alternative carbon materials are being explored as replacements for CFMEs. Our group has been developing an all-diamond, batch fabricated BDDME as an alternative to the traditional CFME. As an extension of our work on biofouling [21], we tested the electrochemical fouling performance of the BDDME since it has been indicated that the two fouling types are facilitated through different mechanisms [17]. In part, we were motivated to explore this effect by our recent findings that the BDDME exhibits a more robust and consistent response to 5-HT than DA, which is practically significant due to serotonin’s roles in neuropsychiatric disease states (Figure 4D, S3) [54,55,56,57].

The ultimate goal of the development of our all-diamond, freestanding BDDME is to develop a chronically implantable sensor capable of neurotransmitter detection over long periods of time. However, several challenges remain, including achieving adequate sensitivity to detect physiologically relevant concentrations of neurotransmitters in vivo, reducing fouling, and maintaining electrode stability during long-term use. The sensitivity of the BDDME is much lower than the CFME due to its ~10x smaller electroactive surface area (CFME: ~1000-1600 µm², BDDME: ~100-200 µm²) as well as diamond’s less adsorptive material properties [28,30,41]. Femtosecond laser cutting can be used to functionalize the BDDME to a comparable sensitivity to the CFME, which has been used for numerous in vivo explorations [11,52,57,58,59,60,61]. Laser cutting converts sp³-bonded diamond to sp²-bonded graphitic content, which may come at the expense of some of diamond’s desirable properties, including fouling resistance and background stability [27,44,62,63,64]. With the FS-BDDME, this is supported by visible inspection of a dark appearance on the electrode surface in SEM images (Figure 1F, S2), as well as its similar electrochemical characteristics to that of the CFMEs (e.g. background sizes and shapes, sensitivities to DA and 5-HT, response kinetics, and LODs). Additionally, high temperature-acid oxidation can be used to etch sp² bonded carbon from diamond [68,69]. We soaked BDDMEs in a 1:1:1 ratio of concentrated acids (Section 2.9). A large reduction in the electrochemical backgrounds were observed with FS-BDDMEs (972.52 nA, n=2) as compared to a near zero (0.44 nA, n=3) reduction from C-BDDMEs, indicating the removal of a larger amount of sp²-bonded carbon from the FS-BDDME (Figure S5). Post-acid cleaned FS-BDDME backgrounds were still larger than the C-BDDME backgrounds. This may be attributed to an incomplete removal of sp²-bonded carbon or an increase in the electroactive surface area due to the addition of microchannels from the FS-laser pulsing or a combination of both. It also appears that the graphitic content covers a larger span than just the conductive BDD, since the darkening and laser pulses (vertical channels) can be observed across the entire face of the electrodes as seen in the SEM images (Figure 1F, S2). Given the large increase in background size and sensitivity, it is possible that laser cutting functionalizes more than just the BDD core, resulting in a substantial increase in the electroactive surface area. If so, these devices would be expected to exhibit fouling characteristics more like the CFME than the C-BDDME, aligning with the presented findings.

Electrochemical fouling is an important consideration in the development of a chronic 5-HT sensor, since the oxidative byproducts of 5-HT are known to foul the electrode surface. The FSCV potential waveform has been adapted to avoid some effects of 5-HT fouling as demonstrated by Jackson et al. (0.2 V to 1.0 V to -0.1 V to 0.2 V at 1000 Vs^-1 at 10 Hz) [15], and expanded upon by Dunham and Venton [14]. However, in the Dunham and Venton experiments [14], the dopamine waveform exhibits the least fouling after 25 injections of 1 µM 5-HT using CFMEs compared to the Jackson waveform and its tested variants. In this study, we observed that the C-BDDME exhibits improved electrochemical fouling performance over the CFME, with a significantly more stable peak oxidation current response to repeated injections of 5-HT. This may be explainable by low surface adsorption due to BDD’s sp³-bonded structure [29,34]. The FS-BDDME was more susceptible to 5-HT fouling than both the CFME and the C-BDDME. We observed a clear concentration-dependent fouling effect for 50 µM 5-HT compared to 5 µM 5-HT. Since 5-HT oxidation produces polymerizing radicals that form an electron-transfer-hindering film on the electrode surface [18], increased 5-HT oxidations should be expected to cause more fouling. This effect is further supported by absence of measurable fouling to 1 µM 5-HT shown by Dunham and Venton with CFMEs [14]. The C-BDDME maintains its resistance to fouling at higher concentrations and exhibits a slower rate of fouling than the CFME’s exponential decay. This reduced rate of decay could potentially reflect a contribution of a less absorptive surface of the C-BDDME [21,39,65]. The FS-BDDME fouled less than the CFME in the 50 µM experiments and had a rate of decay more closely matched to the C-BDDME than the CFME. We speculate that with large concentrations, the FS-BDDME’s highly adsorptive graphitic content is quickly fouled while the sp³-bonded BDD-portion retains some level of fouling resistance, leading to an “in-between” fouling performance to the CFME and C-BDDME. Optimization of the laser-cutting protocol may allow for a similar, but less pronounced “in-between” effect where the FS-BDDME maintains improvements in both sensitivity over the C-BDDME and fouling resistance compared to the CFME.

Signal stability is another important goal in chronic applications where recording is required over long periods of time. The CFME is known to be very stable over short windows (~ 90s) [51,52] and exhibited stable responses over a two hour window with 5-HT in our study. However, the etching behavior of the CFME [41,66] causes a self-cleaning, surface renewing effect and increases the number of adsorption sites on the electrode surface. In turn, this leads to increased sensitivity and a growing background current, which we observed on our 60Hz, 24-hour test. This evolution over time needs to be accounted for in the chronic setting particularly when estimating concentrations since there may be sensitivity changes [51,67]. Eventually, the carbon-fiber will erode away to an unusable level from consistent etching [31,41]. Part of the motivation to use BDD electrodes in place of the CFME is to reduce etching and maintain signal stability over longer periods of time [31]. Our data confirmed that the C-BDDME had both a stable two-hour response repeatability, and a very minimal background shift after the 24-hour etching test. The C-BDDMEs may better maintain a stable surface without sacrificing material to etching. On the other hand, stability experiments revealed a trend toward a reduction in sensitivity over time with the FS-BDDME, likely due to etching away of the sp²-bonded carbon created by the laser, revealing the sp³-bonded diamond beneath. Thus, although the FS-BDDME greatly improves the sensitivity of the electrode in comparison to the C-BDDME, this benefit comes at the expense of stability. Additionally, we plan to investigate micromachining strategies and laser-cutting parameters to tune the level of sp² conversion of the FS-BDDME, as FS machining has been used to form nanometer-width grooves that may provide an increased electrochemically active surface area while maintaining diamond composition [70].

As the neurochemical sensing field progresses toward improved chronic sensing capabilities, it is important to thoroughly characterize emerging technologies that intend to replace traditional devices. BDDMEs may deliver improved stability and reduced fouling, but at the expense of sensitivity. Surface-modifying techniques may recover sensitivity, but at the expense of stability. These trade-offs require further design iteration to fully leverage the benefits of the BDDME. However, it is also important to acknowledge that the BDDME presented here is not a one-to-one comparison with the CFME. The large disparity between the sensitivity of the BDDME and the CFME is affected by the electroactive surface area differences between the two devices [24,53]. For reference, example CVs from the CFME and the C-BDDME with currents normalized to estimated electroactive surface areas (CFME: 1250 µm² and C-BDDME: 150 µm² calculated from estimated exposed geometric surface area) are included (Figure S6). These surface-area adjusted plots result in an approximately 10x lower response from the C-BDDME. This is contrary to the 100x smaller currents from the non-corrected data, indicating the sensitivity differences are largely due to differing electroactive surface areas. However, a better comparison would still include comparable sensing surface sizes and geometries. As next steps, we recommend increasing the electroactive surface area to increase sensitivity, while maintaining a small device size in order to minimize glial encapsulation [71,72,73,74,75,76]. In addition to this, refinement of the BDDME through sensitivity-increasing methods, such as exploring coatings, overoxidation, waveform development, boron-doping level, and chemometric data analysis may be viable solutions to deliver an optimized, chronic BDDME sensor with customizable, batch-fabricated architectures.