Faradaic Fenton Pixel: Reactive Oxygen Species Delivery Using Au/Cr Electrochemistry

Reactive oxygen species (ROS) are an integral part of many anticancer therapies. Fenton‐like processes involving reactions of peroxides with transition metal ions are a particularly potent and tunable subset of ROS approaches. Precise on‐demand dosing of the Fenton reaction is an area of great interest. Herein, we present a concept of an electrochemical faradaic pixel that produces controlled amounts of ROS via a Fenton‐like process. The pixel comprises a cathode and anode, where the cathode reduces dissolved oxygen to hydrogen peroxide. The anode is made of chromium, which is electrochemically corroded to yield chromium ions. Peroxide and chromium interact to form a highly oxidizing mixture of hydroxyl radicals and hexavalent Cr ions. After benchmarking the electrochemical properties of this type of device, we demonstrate how it can be used under in vitro conditions with a cancer cell line. The faradaic Fenton pixel is a general and scalable concept that can be used for on‐demand delivery of redox‐active products for controlling a physiological outcome.


Introduction
According to the World Health Organization, cancer is one of the leading causes of death, in the majority of countries, for persons below the age of 70. [1] Considerable research effort focuses on optimizing treatment options in terms of effectiveness, efficiency, affordability, and reduction of unwanted side effects. [2] Reactive oxygen species (ROS), namely singlet oxygen, hydrogen peroxide, superoxide radicals, and hydroxyl radicals, are a common feature of many anticancer therapies. ROS are, to varying degrees, toxic to cells. If delivered locally and in high enough concentration, they can lead to selective destruction of cancerous tissues. In moderate quantities, on the other hand, ROS are essential chemical species for maintaining redox homeostasis. [3] However, if ROS exceed a certain concentration in cells, they can lead to harmful oxidative stress and further to programmed cell death or DNA damage. As they are found in elevated concentrations in the tumor microenvironment, ROS are considered as oncogenic. [4,5] Nevertheless, antitumor thera-pies like radiation therapy, photodynamic therapy, or Fenton nanoparticle-based chemotherapy are utilizing the principle of further elevation of ROS levels for triggering cell death to eliminate cancer tissue. [2][3][4]6,7] As ROS are nonspecific and highly reactive, the molecules' lifetime and transport distance in the body are limited. [8] This results in a local and confined reaction area of ROS, and therefore a limited area of targeted cells. [2,9,10] This spatiotemporal control of ROS is a great advantage, however one has to figure out how to create relevant concentrations of ROS at the area of interest. One basic principle is the Fenton reaction, a familiar concept from synthetic chemistry and industrial oxidation processes. Fenton and Fenton-like systems, which follow the Haber-Weiss reaction mechanism, are a common way to locally form hydroxyl radicals. [6,11,12] Hydroxyl radicals are the strongest known oxidant in aqueous systems and they can be formed via a reaction of hydrogen peroxide with transition metal ions, the best known of which is Fe 2 + . In this classic Fenton reaction, Fe 2 + ions react with H 2 O 2 to form hydroxyl radicals, hydroxy anions, and Fe 3 + [Eq. (1)]. Fe 2 + can be regained for the catalytic cycle by the reduction of Fe 3 + with H 2 O 2 [Eq. (2)].
However, the Fenton reaction [Eq. (1)] works efficiently only at acidic pH. At near-neutral pH (physiological) the reaction kinetics are sluggish and Fe 2 + tends to form oxide compounds which precipitate and is therefore no longer available for the catalytic cycle. [13] Moreover, Fenton-based therapies as well as photodynamic therapy are either limited by the penetration depth of the light, lack efficiency due to constrained availability of reactants, or suffer from particle aggregation and instability. [6] To our knowledge, electro-Fenton-based techniques where H 2 O 2 is electrochemically generated by oxygen reduction on a cathode, and chromium ions are released due to an anodic oxidation have not been investigated for antitumor treatments so far. Therefore, we present an electrochemical device for direct current anti-tumor treatment with local ROS generation and evaluated it on the ability to generate cytotoxic species to target cancer cells in near neutral pH. The thin film device consists of a gold layer that electrochemically generates H 2 O 2 via 2e À -oxygen reduction reaction (2e À -ORR). Concurrently, a thin chromium metal film, deposited on an indium tin oxide (ITO) back electrode, acts as a sacrificial counter electrode that dissolves and releases soluble ions. Chromium ions react with H 2 O 2 in a Fenton-like way to form hydroxyl radicals and highly oxidizing Cr VI (Figure 1a). Chromium is chosen because it 1) is much more stable in thin films than iron; and 2) forms ions which are resistant to precipitation at neutral pH. We characterized the device in regards of H 2 O 2 generation, metal dissolution, and hydroxyl radical formation. We next tested the effects of such electrodes on the A375 cell line in vitro. We found that even with a relatively low current (10-30 μA cm À 2 ) treatment it was possible to induce cell death due to the formation of cytotoxic species. This current can be regarded as low relative to previously explored direct-current electrochemical anticancer approaches, which rely on electrocatalytic water splitting and thus the formation of pH gradients at current densities in excess of 1 mA cm À 2 .

Electrochemical H 2 O 2 evolution on gold
Gold is a well-known oxygen reduction reaction (ORR) catalyst, which can, depending on the exposed surface plane, the electrolyte composition, applied potential and pH, either convert molecular oxygen to H 2 O [Eq. (3), or (4), and (5)] or to H 2 O 2 [Eq. (3)], but can also, at more negative potentials, catalyze H 2 evolution [Eq. (6)]. [14] Under neutral conditions, gold shows electrochemical activity towards 2-electron ORR to form H 2 O 2 depending on the applied potential. Simulated physiological electrolyte is afforded by phosphate buffer saline, PBS. At first molecular O 2 is reduced via equation 3. Increasing negative potential leads to further peroxide reduction to water via equation 5. Further negative potential increase will lead to water-splitting and thus H 2 evolution [Eq. (6). [14,15] Cyclic voltammetric measurements in Figure 1. The electrochemical thin-film device generates H 2 O 2 and Cr VI . a) A schematic of the device with the proposed reaction mechanism for radical formation. b) Cyclic voltammogram of Au (70 nm) on Ti (10 nm)/ glass vs Pt in PBS, 50 mV s À 1 at ambient, O 2 saturated, and deoxygenated (N 2 ) conditions. c) Recorded current densities of Au (70 nm) on Ti (10 nm)/glass at different applied potentials in PBS with respective spectrophotometrically determined H 2 O 2 concentration; potentials were applied vs Ag/AgCl. d) Recorded potentials at different applied current densities of Au (70 nm) on Ti (10 nm)/glass in PBS with respective spectrophotometrically determined H 2 O 2 concentration. e) Cyclic voltammogram of chromium (40 nm) on ITO vs. carbon mesh in PBS, 50 mV s À 1 . f) Spectrophotometrically determined concentration of dissolved Cr ions after different potential application of chromium (40 nm) on ITO for 10 min each. PBS at ambient conditions (Figure 1b) of thermally evaporated gold thin films on glass with Ti as adhesive layer show two reductive peaks at À 0.29 and À 0.71 V vs Ag/AgCl. When the electrolyte solution was saturated with O 2 , the measured current density increased and the reductive peaks shifted slightly to À 0.35 and À 0.73 V versus Ag/AgCl whereas when deoxygenated with N 2 these peaks disappear, which indicates that these peaks correspond to ORR. Moreover, in oxygen-free electrolyte the onset potential for H 2 evolution was found to be at À 0.79 V vs Ag/AgCl. In order to prove the electrochemical generation of H 2 O 2 , we performed amperometric measurements at potentials between À 0.3 and À 0.6 V and quantified the average generated hydrogen peroxide with the spectrophotometric-based horseradish peroxidase/3,3',5,5'-tetramethylbenzidine (HRP/TMB) assay (Figure 1c). [16] This confirmed that at potentials between À 0.3 V and À 0.6 V H 2 O 2 evolution was favored with a maximum of concentration reached at À 0.4 V vs Ag/AgCl. This represents the potential at which reaction 3 is maximum and reaction 5 is not competitive. Potentiometric measurements ( Figure 1d) confirm that at À 20 μA cm À 2 for 10 minutes a stable potential of À 0.4 V with a peak concentration of 41.1 μM H 2 O 2 was achieved, whereas at a current density of À 30 μA cm À 2 the potential as well as the peroxide concentration dropped. Moreover, we found that the electrocatalytic activity is independent of the thickness of gold between 40 and 100 nm ( Figure S1 and Table S1 in the Supporting Information). Hence, we decided to proceed with a gold film thickness of 70 nm.

Chromium as a sacrificial electrode
To complete the electrochemical cycle, we chose chromium as a sacrificial anode electrode. Chromium is one of the transition metals which can, in ionic form, initiate Fenton-like reactions at neutral pH. [13] Bokare et al. employed this hydroxyl radical generating mechanism for Cr VI waste water treatment of organic pollutants. In their study they show 4-Chlorophenol, Rhodamine B, Orange II, phenol, and nitrobenzene degradation at a pH range between 3-11 in the presence of hexavalent Cr and H 2 O 2 . According to their observations H 2 O 2 can activate Cr VI by reducing it to Cr V . The Cr V then further reacts with H 2 O 2 to form hydroxyl radicals and Cr VI again [Eqs. (7) and (8)]. [17] Cr VI þ reductant ! Cr V ! Cr IV ! Cr III ! Cr II (7) Additionally, this cascade of reactions is considered as one reason for the toxicity of hexavalent chromium itself. In the biological system either low molecular-weight species or cellular reductants can trigger the activation of Cr VI . These species thus become oxidized in the process. Further, the induced ROS formation can lead to DNA damage and mutation, lipid peroxidation, 8-OHdG formation, and the activation of nuclear transcription factors followed by oncogene expression. [18] We also found that chromium is one of the few thin-film metals, which is stable as zero-valent metal in physiological conditions. Iron, copper, and manganese were not stable in PBS and either corroded or dissolved from the substrate without any potential application, making their use in an "on-demand" faradaic Fenton pixel impossible.
As the cyclic voltammogram in Figure 1e indicates, chromium is inert between À 1 to 0.5 V vs Ag/AgCl. At potentials more positive than 0.5 V, corrosion begins, and the current density starts to rise. This anodic current appearance correlated with a color change in solution resulting from soluble chromium salts. At higher potentials, dissolution competes with oxygen evolution and chlorine formation which results in an increase of the current density to 1.8 mA cm À 2 at 1 V vs Ag/AgCl. [19] We then spectrophotometrically investigated the amount of released Cr ions ( Figure 1f). Therefore, we applied different potentials to the film and determined the metal ion concentration with direct UV-Vis spectrometry, as described elsewhere [20] (for potential curves and spectra of metal ion solutions, see Figure S2). Therefore, we concluded that chromium only dissolves under positive potential application from 0.5 V vs Ag/AgCl in a linear manner. The higher the applied potential the more chromium will dissolve and due to the pH of 7.4 result in hexavalent chromium ions. [21]

Hydroxyl radical formation with thin film device
To ensure that H 2 O 2 and the released Cr ions are in proximity and can react with each other, an interdigitated device consisting of Au on Ti and Cr on ITO was fabricated. The electrode fingers are 1 mm broad, 20.9 mm long and have a distance of 500 μm between each other (Figure 2a). The dimensions were chosen due to practical reasons for further experiments with cells. Ti was applied as an adhesive layer for Au since it does not diffuse into the gold layer and therefore does not alter the electrocatalytic behavior of gold. [22] To guarantee uniform and continuous current delivery to chromium, commercial ITO was chosen as a conducting back electrode. Detailed device fabrication steps can be found in the Experimental Section and Figure S3. Figure 2b shows the device design with the applied experimental set up. To interface the device properly with the potentiostat we fixated 2 mm banana cables on the electrodes. The copper cable was soldered on gold or glued with conductive silver paste to the surface of chromium, since soldering did not lead to acceptable results in contact. To protect the contacts from humidity we applied an epoxy glue as isolation. Moreover, to create a well on the device a poly(methyl methacrylate) (PMMA) ring with an inner diameter of 16 mm was glued on top of the area of interest.
Based on the above-mentioned information (Figure 1b-d) the optimum potential for the device was determined to be À 0.4 V vs Ag/AgCl, when gold was used as a working electrode. However, for initial test with an A375 cell line we found a slightly more negative potential of À 0.65 V was optimal. This is due to electrode overpotential ( Figure S4) and underscores the importance of testing the "peroxide generation sweet spot" for every specific electrode setup. Moreover, due to spatial limitations in the set-up we needed to use a different reference electrode, which could influence the system as well. We observed a stabilization of current in the range of À 20 μA and therefore decided to run all following experiments with a protocol of 10 minutes of À 20 μA followed by 10 minutes at open circuit potential and alternating this sequence for four hours. The steps with no potential application were introduced to increase the possibility of O 2 diffusion to the electrode and H 2 O 2 diffusion away from the electrode.
We then tested the performance of the device under the proposed current treatment for the formation of H 2 O 2 , the release of chromium ions, and further the formation of hydroxyl radicals. Additionally, we conducted similar measurements with configurations with only gold and only chromium as exposed electrodes with the same dimensions as in the original device configuration. The concentrations of released chromium ions amounts to an average of 25.6 mg L À 1 for the combined device and 26.8 mg L À 1 for chromium only with an error less than � 1 mg L À 1 (Figure 2e). However, the generated H 2 O 2 differs significantly depending on the respective sample. It ranges from 19 to 105 μM H 2 O 2 for the device and from 48 to 142 μM H 2 O 2 for only gold in PBS (Figure 2e). This might be related to the electrode behavior itself. As seen in Figure 2c the measured potential of gold (top: device; middle: only gold) did not stabilize over time but dropped to À 0.8 and À 0.9 V vs Ag/AgCl for some samples. When examining the detected final concentrations and the recorded potentials of the samples separately, we can correlate the low final H 2 O 2 concentrations with the samples with a higher negative potential drop (Figures S5 and  S6, Table S2). We therefore assume, as shown in the cyclic voltammogram in Figure 1b, that the increase in negative potential resulted in the consumption of the generated H 2 O 2 and further to H 2 generation. To further test this hypothesis, we performed measurements with an amperometric peroxide sensor, which can probe the local concentration of peroxide in real time. The sensor was placed 200 μm above the gold electrode. We recorded the hydrogen peroxide generation during current treatment of two different samples with different preparation dates, whereas one was recently prepared while the second sample was fabricated 70 days prior to it. The newly prepared sample showed stable, reproducible recorded potentials for each current pulse and resulted in a peak peroxide concentration during every negative current pulse, which declined again during no current application. From the third negative current pulse, a slight decline of H 2 O 2 concentration appeared already during the pulse, which might correlate with oxygen depletion due to diffusion limitation ( Figure 2d). [23] However, the second sample ( Figure S7) showed a constant increase of negative potential during the negative current pulse and peaked after 120 min at almost À 1 V vs Ag/AgCl. This resulted in a maximum of 350 μM H 2 O 2 measured by the sensor, however we assume that this peak is not correlated to H 2 O 2 but rather to H 2 , since with the enzymatic based, spectrophotometric HRP/TMB assays we cannot detect H 2 O 2 at these applied potentials. Moreover, Ehlich et al. reported that this specific amperometric sensor is also cross-sensitive to H 2 , thus the peroxide concentration from the sensor cannot be considered quantitative at higher potentials. [24] In the following pulses the potential dropped again and led again to a peroxide peak generated at approximately À 0.5 V followed by a constant decline of concentration correlated to negative potential increase.
To find out if hydroxyl radicals are formed, we added hydroxyphenyl fluorescein (HPF), a fluorescent reagent specific to hydroxyl radicals and reactive intermediates of peroxidase, [25] to the electrolyte after the current treatment and measured the fluorescence intensity directly after addition and 24 hours after addition. Only the combined device showed a significant increase of almost 100 % in intensity, whereas the solutions only containing electrochemical generated H 2 O 2, or electrodissolved Cr ions show only an increase of approximately 20 % (Figure 2f). This evidences that the combination of the generated Cr ions and H 2 O 2 indeed leads to hydroxyl radical formation. The increase of intensity due to electrochemical generated H 2 O 2 by gold might be due to different reactive intermediate species during the electrochemical conversion. Fluorescence spectra for each electrode treatment can be found in Figure S8.

In vitro cell experiments
Having validated the generation of peroxide/Cr VI /OH radicals, we hypothesized that applying such a device to a cancerous cell line may lead to cell death. As a model, we tested electrochemical versus direct ROS injection on the human malignant melanoma cell line A375. During control conditions cells were seeded directly on glass substrates or the devices without applying any current. Therefore, we used the same configuration as described above and shown in Figure 2b. To investigate the effect of the current treatment and the respective generated species we performed time-lapse digital holographic microscopy (DHM). Compared to commonly applied fluorescence-based techniques, DHM has been recently emerging as a label free and non-phototoxic imaging method with minimal invasiveness and the possibility to screen cells over time under incubated conditions. DHM allows a wide range of live-cell assays, for example, single cell tracking and cell morphology. [26][27][28] The cells were seeded on top of the samples inside the glued well and incubated for 24 h in Dulbecco's modified Eagle medium (DMEM). After media exchange, we placed the sample slide into the DHM, which itself was placed into an incubator ( Figure S9). After connecting the sample with the potentiostat and prior to the current treatment we selected at least three locations on the devices to be monitored during the experiment. The motorized stage of the DHM allows accurate imaging at predefined locations. The first image was recorded directly before the current treatment which was followed by a sequence of images taken every 10 min during current treatment for 4 h. Images were also recorded for an additional hour post-treatment. After completion of the current treatment, we continued incubating the cells with the device and recorded images of the cells again the next day after approx. 24 h to control if the cells had recovered. In addition to the current measurements with the combined device (CrÀ Au), gold (Au) only and chromium (Cr) only, we conducted time-match control recordings of the cells on the device (CrÀ Au control), gold (Au control) only and chromium (Cr control) only without current application (together labeled as "electrode controls"). A second type of time-match control experiments were performed with cells seeded in a PMMA well on glass where we injected either 100 μM of H 2 O 2 , and/or 25 mg L À 1 of potassium dichromate to determine if these species effect the cells in the same way as electrochemically generated species do.
When compared with current treatments conducted in the model system PBS solution, the recorded current/potential traces from the experiments with cells show a slightly different behavior for CrÀ Au (Figure 3a). The average potential is stabilizing at approximately À 0.7 V vs Ag/AgCl after 1 h, whereas CrÀ Au measured in PBS stabilizes at À 0.55 V after 2 h ( Figure 2c). A similar trend is observed for the potential traces of single gold electrodes exposed to the cells (Figure 3b), the average potential stabilizes at À 0.65 V after 1 h though. This might be either related to O 2 depletion, since cells block the diffusion of O 2 to the electrode surface, or because of potential changes at the reference electrode. Due to space limitations, we needed to use a modified reference electrode (Ag/AgCl wire in 2 % agar in 3 M KCl). After completion of the experiments, we observed that the tip of the used electrodes turned black and an increase in the potential difference between 100-300 mV vs an Ag/AgCl electrode immersed in 3 M KCl, protected by a glass frit, whereas prior to experiments the potential difference was below 50 mV. This stresses the importance of reliable reference electrode systems for long-term biological experiments.
When cells were exposed solely to chromium electrodes (Figure 3c), which were connected to a Pt wire as working electrode in a separate chamber, the potential traces behaved similarly to the traces measured during control experiments in PBS (Figure 2c). Moreover, chromium electrodes in single and combined configuration were more transparent after the treatment, which indicates that chromium did indeed dissolve.
Live-cell imaging of morphological changes under incubated conditions allows us to draw conclusions about the effects of H 2 O 2 , Cr ions and consecutively formed hydroxyl radicals on the viability of adherent cancer cells. Apart from the recorded images, the subsequent numerical analysis gives us the possibility to calculate and compare different parameters derived from the change in phase shift that are related to the morphology, such as average roughness, area, and maximum optical thickness. These parameters are well suited to assess cell death, though distinguishing between different forms of cell death such as apoptosis or necrosis can be challenging. [27,29] The change in surface roughness of a cell is related to blebbing, which is an observed feature of apoptosis. [27,30,31] Moreover, a decrease of cell area and an increase in cell thickness is either an indication for cell death or cell division. [27,31] We followed cells using DHM under the respective conditions at 0, 1, and 5 hours and after approximately one day of incubation following the current treatment or injection (Figure 4). These images can be compared with time-match control samples ( Figure 5) which were followed over the same time but without application of current or injection. The color coding of the legends indicates the cell thickness with pale pink as the background and white corresponding to a thickness of 13.2 μm.
In all examples of current-treated cells, there was evidence of extensive cell death. Electrochemically generated H 2 O 2 (Figure 4, Au) resulted in the strongest change to the cell morphology of A375 cells. Cell shrinkage with an increase of cell thickness and in some cases swelling with consecutive bursting and detaching of cells was observed.
Cr and CrÀ Au with current treatment (Figure 4, CrÀ Au and Cr) induced very similar morphological changes. Initially the area of the cells increased, followed by a gradual cell shrinking. Concurrently, the thickness increased however no bursting of the cell membranes took place. This missing bursting effect is reflected in the linear decrease of the area as well as in the gradual increase of the maximum optical thickness. Notably, the thickness of cells treated with CrÀ Au is significantly higher compared to Cr after 24 h. In comparison cells treated with injected Cr VI and injected Cr VI + H 2 O 2 (Figure 4 injected Cr VI and injected Cr VI + H 2 O 2 ) altered their morphology only gradually with a decrease in area and a concurrent shrinkage, swelling and thickness increase. After 24 h, they did not recover from the treatment though. Control samples, without any applied current or injection ( Figure 5), featured cells which not only did not change their morphology but after additional 18-24 h of incubation an increase in cell density could be observed. This is in stark contrast to samples which are treated with current, or injection, where cell death is dominant. These differences can be convincingly tracked in time-lapse videos 1-3 in the Supporting Information.
Additionally to the visual observation of the morphological changes, it is possible to extract various numerical values from DHM images to provide statistical data. Figure 6 shows the calculated changes in average roughness, area, and maximum optical thickness under various conditions over time as well as the related paired t-tests, which give an indication if the observed changes at 0 and 5 h are significantly correlated or not. Cells treated in any electrode configuration as well as with injected H 2 O 2 , showed a significant increase in surface roughness within five hours (Figure 6a). Injection of Cr VI and Cr VI in combination with H 2 O 2 did not induce a change in roughness. Same is valid for electrode controls. Notably, in a study by Makdasi et al. the roughness of HeLa cells treated with the toxic compound ricin did not significantly increase during their monitored time span and concluded that increased roughness might be only visible at a late state of cell death. [29] Moreover, depending on the type of cell line, it differs when roughness increases due to blebbing setting in. [31] Injected H 2 O 2 led to a decrease in area 1 h after starting electrical treatment whereas electrogenerated H 2 O 2 induced an immediate decrease in cell area (Figure 6b). Cells treated solely with hexavalent Cr ions or with a combination of Cr ions and H 2 O 2 injected as well as cells under current treatment with CrÀ Au electrode configuration behaved in a similar manner. The decrease in area follows a linear manner with almost similar slopes. However, the area of Cr-current treated cells increased slightly before decreasing. The electrode controls without current treatment show almost no decrease in area, therefore we suggest that no cell division occurred during the course of 5 h and that the observed change in area was solely due to cell death, which is also confirmed by the images after 24 h (Figures 4 and 5).
The change in the maximum optical thickness follows a similar trend as the areal shift, with an increasing tendency, (Figure 6c). Injected H 2 O 2 induced an increase in cell thickness 1 h after starting the recording whereas electrogenerated H 2 O 2 caused an immediate linear increase in thickness which then stabilized and further decreased slowly. Injected Cr VI and the injected Cr VI -H 2 O 2 combination led to a linear increase in maximum optical thickness during the course of 5 h, whereas Cr and CrÀ Au initially caused a decrease but after 60-90 min the thickness started rising. Controls of cells seeded on electro-

Discussion and Conclusions
These results show that the applied low direct-current treatment (20 μA or ca. 28.5 μA cm À 2 ) for all electrode configurations induced irreversible changes in the morphology of A375 cells. The Au and Cr electrodes can be applied separately or in a combined mode, leading to different morphological changes. Electrogenerated H 2 O 2 solely (from an Au cathode alone) resulted in the fastest cell damage where some cells even showed such a strong swelling effect that ultimately led to bursting of the cell membrane. On the other hand, released chromium ions and the combination of H 2 O 2 and Cr ions led to a steady increase of the cell optical thickness, which implies controlled cell death (likely apoptosis). In all cases, cells did not recover from their morphological changes after approximately 24 h incubation post-treatment and died eventually. A rise in roughness due to blebbing, shrinkage in area and swelling which is depicted in the thickness increase are distinctive features observed during cell death. [27,30,31] Control samples clearly do not show this kind of behavior. In control experiments, the cells appear to be healthy and dividing. The important distinction between Au cathodic peroxide evolution alone versus the combined Fenton device is the type of cell death, though it is instructive and surprising to consider that cathodic current alone is potent in killing cells.  To gain a better understanding of the electrochemically induced cell death we compared direct injection of the chemical species with the respective electrochemically generated reactive species. It became clear that electrochemical and non-electrochemical treatments influence the cells in a different manner. At the vicinity of the electrode, we expect higher concentrations of electrogenerated species. Such concentration gradients of electrogenerated H 2 O 2 above an electrode surface have been demonstrated previously by our group. [23] Therefore, we hypothesize that these higher concentrations at the electrode-electrolyte-cell interface have a higher disruptive impact on the integrity of the cell. Additionally, due to the electrochemical conversion, a change in pH is induced which might contribute to cell death, as this mechanism is believed to have a major contribution to induced cell death during electrolytic ablation, an electricity-based anti-tumor treatment technique similar to our proposed concept. Direct comparison is not possible since the current densities used here are at least 10 times lower than these examples. Nevertheless, pH gradients may be important. [32,33] During the characterization of the current treatments without cells in PBS we also determined the pH change in the bulk solution before and after. During ORR the pH of the solution above gold increased slightly whereas chromium dissolution led to a decrease in the pH. However, when the electrochemical processes were investigated in a combined mode (CrÀ Au) the pH change induced by the single electrode reactions equalized themselves and only a slight overall pH change occurred (Table S3). Future measurements with a pH microsensor will be conducted to give a better insight into the local pH changes above the electrode surfaces.
The question remains -are hydroxyl radicals involved in triggering cell death? The comparison between the injected Cr VI and H 2 O 2 control, which should in combination lead to hydroxyl radicals, and the sole injected Cr VI control reveals only minor observed differences in the morphology over the course of 5 h. Similar is valid when we compare the morphological changes inflicted on the cells via current application using Cr and CrÀ Au. However, CrÀ Au did indeed lead to a more advanced state of swelling after 24 h compared to Cr and the injected controls ( Figure 4). Reasons for this difference in injected control and CrÀ Au might be again the vicinity of the cells to the electrode which we expect to result in a higher diffusion rate of the electrogenerated species across the cell membrane into the cell. With injection mode there is the possibility that radical formation is mainly taking place in the bulk solution where certain components of the cell medium might function as radical scavengers. An important part of our future work will be to elucidate the molecular mechanisms leading up to cell death. Although the application of low direct currents and accompanying peroxide generation seem to have the most obvious and quickest effect in sponsoring cell death the slower cell death observed with CrÀ Au hints towards a different form of cell death.
The concept of a faradaic Fenton pixel that we present here with its capability to locally deliver hydroxyl radicals through the concerted in-situ electrochemical generation of Cr ions and H 2 O 2 is a promising approach towards the development of new local antitumor treatments with minimal systemic side effects. It is also well-suited to in vitro experiments. Using low direct current treatment with consecutive ROS delivery may offer the advantage of local application without major impact to the surrounding healthy tissue. In addition, the nature of electricity implies that it might be possible to control the treatment by toggling the current on and off, and thus allowing a controlled ROS delivery. The configuration, form, and size of the electrodes as well as the length of the current treatment can be adapted and optimized for specific needs.
In contrast to Fenton nanoparticle-based antitumor treatments, we might have solved the problem of sufficient exogenous H 2 O 2 delivery. Future investigations will focus on testing less cytotoxic Fenton like materials with faster kinetic characteristics regarding radical formation. Electrochemical methods can also lower costs due to its ease of preparation and handling.
Our measurements show that Cr VI is forming during electrochemical oxidation of chromium electrodes. Chromium is remarkable in that thin films of the metal are stable under physiological conditions, unlike other Fenton metals like Fe and Cu which spontaneously corrode. This allows electrical triggering of ion release. The released chromium ions not only sponsor the Fenton reaction, they are also strongly oxidizing in their own right. This is an interesting effect which merits further exploration. Hexavalent chromium is known as a toxic and carcinogenic species, which is due to the oxidative cascade it produces. Nevertheless, as Cr VI acts as an oxidizer, chromium becomes reduced until the Cr II state, which is no longer an oxidizer and is a soluble and benign form of chromium. Thus in vivo, one can expect that local chromium anodic release will produce damaging oxidation, but only locally. Reduced Cr II should be the byproduct. In a next in vivo embodiment of the concept disclosed in this manuscript, monitoring chromium species diffusion in the body as well as eventual excretion are critical.
Research around electrical antitumor therapy has already resulted in many forms of treatment modalities, such as irreversible ablation, [34] electrochemotherapy [35] or electrochemical treatment. [36] Irreversible ablation and electrochemotherapy apply electric fields with different strength to either irreversibly electroporate cells or to reversibly electroporate cells with a consecutive injection of chemotherapeutic agents such as bleomycin or cisplatin. [34,35] Electrochemical treatment, also known as electrolytic ablation or direct current treatment, applies direct electric current which leads to cell rupture and cell death mainly caused by pH changes close to the inserted electrodes. [32] Current densities starting from 0.2 mA cm À 2 have been reported in electrochemical tumor treatment, (HL-60 cell line, apoptosis visible after 90 minutes) [37] in contrast to the tenfold lower current density that we required to induce cell death after 60 minutes. ROS formation, however, is only marginally discussed in electrolytic ablation. [38] Our work suggests that careful consideration of the faradaic reactions of oxygen reduction and metal corrosion might offer directcurrent tissue ablation with low, and thus possibly safer, currents. Next steps should include complex assays to resolve cell death pathway, and in vivo experiments on tumor models with appropriate histological characterization.
Substrate cleaning: Glass slides and commercial indium tin oxide (ITO) slides (76x26 mm) were cleaned via ultrasonication for 10 min at elevated temperature with acetone, isopropanol, 2 % Hellmanex III and DI water consecutively. Afterwards the substrates were rinsed with DI water and dried under a stream of N 2 .
Device fabrication: A thin film of Cr (40 nm, < 2 × 10 À 6 Torr, 1.5-2.5 Å s À 1 ) was thermally evaporated on cleaned ITO slides with a Moorfield T090 M evaporation system. Afterwards the positive photoresist S1818 was spin coated on top and baked at 110°C on a vacuum hot plate. Then the desired area of Cr/ITO was masked, UVcured (Karl Süss MA6/BA6) and developed with MF-319 followed by another baking step at 110°C. The unmasked area was first treated with Cr etchant (Sigma Aldrich) and then with Zn powder /6 M HCl to remove Cr and ITO. Afterwards the photoresist was removed with acetone and the samples were cleaned with acetone and DI water and then dried with compressed air followed by a plasma treatment (Diener Zepto-W6, 100 W, 5 min). Another layer of S1818 was spin coated on top and baked at 110°C. Then everything, besides the area where Ti and Au should be evaporated, was masked, UV-cured and developed with MF319. After another baking step at 110°C and plasma cleaning (10 min, 50 W) the slides were transferred to a thermal evaporation system (Balzer BA510 or Edwards 306 A), where 10 nm Ti followed by 70 nm Au were deposited (< 1 × 10 À 6 Torr, 1.5-2.5 Å s À 1 ). After thermal evaporation, metal lift off with acetone was performed. After further cleaning with acetone and DI water the devices were dried in N 2 flow followed by plasma treatment (100 W, 5 min). For interdigitated electrodes with only ITO/Cr or only Ti/Au (on glass substrates) the material specific steps were carried out. Cut banana connectors were either soldered on the gold surface or glued with silver conductive paste on the Cr surface and insulated with an epoxy glue (Epoxy Universal 335).
Electrochemical characterizations: All electrochemical measurements were performed and recorded with an Ivium Vertex.One potentiostat. All photometric spectra were recorded with a Synergy H1 Microplate reader (BioTek® Instruments, Inc.) in a 96-well plate. pH measurements were performed with an Orion ROSS pH electrode (Thermo Scientific) connected to an Orion Star A211 pH meter (Thermo Scientific).
Cyclic voltammetry: The electrode of interest was placed (exposed area: 1 cm 2 ) in a 2-chamber cell (MM PEC H-Cell, redox.me) containing an Ag/AgCl in 3 M KCl reference electrode (redox.me) connected via a Nafion-membrane with a counter compartment chamber containing Pt as CE. The chambers were filled with 1x PBS. For Au electrodes additional purging of N 2 (1 h) or O 2 (15 min) prior to the measurements was performed. Cyclic voltammograms with a scan speed of 50 mV s À 1 were recorded until stable cycles were achieved.
Au on Ti -H 2 O 2 evolution: A desired area of Au/Ti on glass was masked with PVC tape and a droplet of 300 μL of 1x PBS was placed on it. An Ag/AgCl reference electrode was dipped into the solution. A 2 % agarose in 0.1 M KCl salt bridge connected the working electrode/reference electrode droplet with a counter electrode compartment containing a Pt CE and 1x PBS. The thin film was contacted with a metal probe and potentials between À 0.3 and À 1 V or current densities between À 10 to À 30 μA cm À 2 were applied for 10 min, and the solution analyzed for H 2 O 2 .
Cr on ITO -Cr ion dissolution: A desired area of Cr on ITO was masked with PVC tape and a droplet of 300 μL of 1x PBS was placed on it. An Ag/AgCl reference electrode was dipped into the solution. A 2 % agarose in 0.1 M KCl salt bridge connected the working electrode/reference electrode droplet with a counter electrode compartment containing a carbon mesh and 1x PBS. The Cr thin film was contacted with a metal probe and different potentials between 0.5 and 1 V were applied for 10 minutes and the solution analyzed for Cr ions.
Device preparation of PMMA well for measurements with and without cells: A FLUX Beamo laser (30 W CO 2 laser, FLUX Europe) equipped with an autofocus module was used to cut 5 mm thick PMMA sheets into the shape of a ring with an inner diameter of 14 mm that were eventually attached to the devices or glass substrates (for injection controls) with biocompatible 250 μm thick Flexdym TM adhesives which were also cut into the shape of a ring using the FLUX Beamo laser. Afterwards, the combined PMMA-adhesivesubstrate devices were treated for 1 h at 80°C to accelerate the adhesion of the Flexdym TM film. All devices and glass substrates with a PMMA well, which were used for experiments with cells, were sterilized with UV light as well as 70 % ethanol before cell seeding.
Au -current treatment in PBS: Gold on Ti was connected as working electrode and an Ag/AgCl electrode (for PBS measurements Ag/ AgCl in 3 M KCl (redox.me) was placed close to the WE. The well was connected to a second well containing a Pt counter electrode via a 2 % Agar in 3 M KCl salt bridge. 1.5 mL 1 × PBS was filled into the WE well; the CE chamber was filled with 1 × PBS as well. When connected, the open circuit voltage was recorded for 2 minutes. Then the current treatment was applied (À 20 μA for 10 min, open circuit potential for 10 min; sequence repeated for 4 h). The reaction solution was mixed, 3 times 50 μL of the solution were taken for H 2 O 2 quantification, 499 μL were taken for radical detection and the remaining liquid was used for pH measurement.
Cr -current treatment in PBS: Cr on ITO was connected as counter electrode. The well was connected to a second well containing a Pt working electrode via a 2 % Agar in 3 M KCl salt bridge. An Ag/AgCl electrode (redox.me) was placed close to the WE. 1.5 mL 1 × PBS was filled into the CE chamber, the WE chamber was filled with PBS as well. When connected, the open circuit voltage was recorded for 2 min. Then the current treatment was applied (À 20 μA for 10 min, open circuit potential for 10 min; sequence repeated for 4 h). The reaction solution was mixed, 250 μL were taken for Cr-spectra analysis, 499 μL were taken for radical detection and the remaining liquid was used for pH quantification.
Cr-Au -current treatment in PBS: The gold electrode was connected as working electrode, the chromium electrode was connected as counter electrode. A Ag/AgCl electrode (for PBS measurements Ag/ AgCl in 3 M KCl redox.me, for cell measurements Ag/AgCl wire in 2 % Agar in 3 M KCl) was placed close to the WE. 1.5 mL 1 × PBS was filled into the chamber. When connected, the open circuit voltage was recorded for 2 min. Then the current treatment was applied (À 20 μA for 10 min, open circuit potential for 10 min; sequence repeated for 4 h). The reaction solution was mixed, 3 × 50 μL of the solution were taken for H 2 O 2 quantification, 250 μL were taken for Cr-spectra analysis, 499 μL were taken for radical detection and the remaining liquid was used for pH quantification.
Determination of dissolved Cr concentration: 250 μL of solution were placed in a 96-well plate and an absorbance spectrum was recorded. The measured absorbance at 372 nm was extracted and with the help of a calibration line the concentration of hexavalent chromium ions was determined. Calibration was done as followed: 0.02 M K 2 Cr 2 O 7 solution was diluted with 1x PBS to the respective concentrations of 2.5, 25, 50 and 100 mg L À 1 and absorbance spectra were recorded. The maximum absorbance at 372 nm of each measurement gives a linear correlation.
Determination of H 2 O 2 concentration with the HRP/TMB assay: 50 μL of the reaction solution were mixed in a 96-well plate well with 250 μL of freshly prepared HRP/TMB assay mixture consisting of phosphate-citric acid buffer solution (500 mL 0.4 M Na 2 HPO 4 + 340 mL 0.2 M citric acid, diluted 1 : 1), 50 mg L À 1 3,3',5,5'-tetramethylbenzidine (TMB) and 90 U mg À 1 horseradish peroxidase (HRP). The mixture was spectrophotometrically measured at 653 nm. With a previously determined calibration line the concentration of the sample was calculated based on the measured absorbance. Calibration was done as followed. To the HRP/TMB assay mixture the respective amount of H 2 O 2 was added, to achieve calibration concentrations of 40, 30, 20, 10 and 0 μM.
Determination of OH-radical formation with HPF assay: 250 μL of reaction solution containing 10 μM HPF were placed in a multi well plate and a fluorescence spectrum was recorded. An excitation wavelength of 480 nm and an emission wavelength of 515 nm were applied. A gain of 80 was chosen.
Determination of H 2 O 2 evolution with amperometric sensing: With a distance of approximately 200 μm above the connected gold electrode (as described in the section Au) an amperometric H 2 O 2 sensor (ISO-HPO-2 polarized at 450 mV, connected with a TBR 4100 free radical analyzer; WPI) was placed. The distance was adjusted with a stereotaxic system. While the current treatment was applied (À 20 μA for 10 min, open circuit potential for 10 min; sequence repeated for 4 h) the current response of the sensor was monitored and with a previously obtained calibration line converted to H 2 O 2 concentration.
Cell culture: Human melanoma cell line A375 was purchased from CLS Cell Lines Service GmbH. Cells were cultured in DMEM high glucose supplemented with 10 % FBS (CLS Cell Lines Service GmbH) as well as 1 % Penicillin-Streptomycin (Corning, Thermo Fisher Scientific) and incubated at 37°C in a 5 % CO 2 incubator. Cells were subcultured according to the recommendations by CLS Cell Lines Service GmbH. First, cell medium was removed and the cells washed with PBS without calcium and magnesium (CLS Cell Lines Service GmbH) followed by a 10 min treatment with Accutase at ambient temperature (CLS Cell Lines Service GmbH). Accutase treatment was stopped by adding fresh DMEM High Glucose (10 % FBS, 1 % Penicillin-Streptomycin). The cell suspension was sub-sequently centrifugated for 3 min at 300 g, the cell pellet resuspended in fresh cell medium and dispensed into new cell culture flasks containing fresh medium. Resuspended cells were also immediately seeded onto the devices in the area confined by the PMMA ring at a density of 10 4 cells in 1.5 mL cell medium.

Current treatment of cells:
The respective electrodes with cells were placed into a microscope slideholder which was inserted onto the motorized stage of the digital holographic microscope HoloMonitor (Phase Holographic Imaging PHI AB). All cell experiments were conducted in an incubator and the electrodes connected as described in the section on current treatments in PBS. Due to spatial limitations a self-made reference electrode consisting of an Ag/AgCl wire in 2 % Agar in 3 M KCl in a plastic pipette tip was used. For each device at least three transparent locations were predefined in the electrode gaps, which should be monitored during the course of current treatment., Prior to the current treatment a first set of images was recorded followed by recordings every 10 min during and until 1 h after the current treatment (À 20 μA for 10 min, open circuit potential for 10 min; sequence repeated for 4 h). Incubation of the cells continued after current treatment for another 18-24 h followed by a final recording of the predefined locations. Data analysis: DHM images of the cells were analyzed using the HoloMonitor App Suite cell imaging software (Phase Holographic Imaging PHI AB) with additional single cell tracking assay. Numerical values for average roughness, area and maximum optical thickness were directly obtained from the analysis conducted in the single cell tracking assay and normalized with OriginPro 2021. Pair-Sample t-Test was also done in OriginPro2021

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author. 07432S, by funding from the Brno City Municipality, and by funding from the Knut and Alice Wallenberg foundation. We would like to thank Prof. Fredrik Nikolajeff (head of biomedical engineering at Luleå University of Technology) most sincerely for his support towards this project and Kempestiftelserna as well as LTU labbfond for funding the equipment in the newly established cell culture lab at Luleå University of Technology.