ARTICLE https://doi.org/10.1038/s41467-020-20546-w OPEN Graphene active sensor arrays for long-term and wireless mapping of wide frequency band epicortical brain activity R. Garcia-Cortadella1,9, G. Schwesig2,9, C. Jeschke3, X. Illa 4,5, Anna L. Gray6, S. Savage6, E. Stamatidou6, I. Schiessl 7, E. Masvidal-Codina4,5, K. Kostarelos 1,6, A. Guimerà-Brunet4,5, A. Sirota2 ✉ & J. A. Garrido 1,8 ✉ 1234567890():,; Graphene active sensors have demonstrated promising capabilities for the detection of electrophysiological signals in the brain. Their functional properties, together with their flexibility as well as their expected stability and biocompatibility have raised them as a promising building block for large-scale sensing neural interfaces. However, in order to provide reliable tools for neuroscience and biomedical engineering applications, the maturity of this technology must be thoroughly studied. Here, we evaluate the performance of 64- channel graphene sensor arrays in terms of homogeneity, sensitivity and stability using a wireless, quasi-commercial headstage and demonstrate the biocompatibility of epicortical graphene chronic implants. Furthermore, to illustrate the potential of the technology to detect cortical signals from infra-slow to high-gamma frequency bands, we perform proof-of- concept long-term wireless recording in a freely behaving rodent. Our work demonstrates the maturity of the graphene-based technology, which represents a promising candidate for chronic, wide frequency band neural sensing interfaces. 1 Catalan Institute of Nanoscience and Nanotechnology (ICN2), CSIC and BIST, Campus UAB, Bellaterra, 08193 Barcelona, Spain. 2 Bernstein Center for Computational Neuroscience Munich, Faculty of Medicine, Ludwig-Maximilians Universität München, Planegg-Martinsried, Germany. 3 Multi Channel Systems (MCS) GmbH, Reutlingen, Germany. 4 Instituto de Microelectrónica de Barcelona, IMB-CNM (CSIC), Esfera UAB, Bellaterra, Spain. 5 Centro de Investigación Biomédica en Red en Bioingeniería, Biomateriales y Nanomedicina (CIBER-BBN), Madrid, Spain. 6 Nanomedicine Lab, National Graphene Institute and Faculty of Biology, Medicine & Health, University of Manchester, Manchester, UK. 7 Division of Neuroscience and Experimental Psychology, School of Biological Sciences, Faculty of Biology, Medicine and Health, University of Manchester, Manchester M13 9PT, UK. 8 ICREA, Pg. Lluís Companys 23, 08010 Barcelona, Spain. 9These authors contributed equally: R. Garcia-Cortadella, G. Schwesig. ✉email: sirota@biologie.uni-muenchen.de; joseantonio. garrido@icn2.cat NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w I ncreasing the bandwidth of neuroelectronic interfaces in terms physiological role with an improved spatial resolution. However, of spatial resolution and sensitivity in a wide frequency range in order to advance in the actual application of g-SGFET arrays, is a major and ongoing challenge in neural engineering. In the several technical aspects remain to be thoroughly evaluated. last decades, large efforts have been dedicated to the development In this article, we present a sensing system composed of a of neural sensing interfaces with high sensor-count on conformal flexible 64-channel g-SGFET array and a wireless headstage substrates1–10, which are required for highly biocompatible (Fig. 1c–f and supplementary information S1), which we use to intracranial neural probes11–14. In this line, active sensors have demonstrate the maturity of this technology in terms of long- emerged as a promising building block for high-bandwidth neural term and wide frequency band recording capabilities in freely interfaces4,6,15–19 because they can be arranged in a multiplexed moving animals from a system perspective. First, the focus is array2,4,6–9 enabling high sensor-count probes. The detection placed on the assessment of the in vitro characteristics of the principle of active sensors is typically based on the modulation of system; including the yield and homogeneity of the graphene the conductivity of a transistor channel, which is electrically sensors, their intrinsic noise and the impact of the data- coupled with the biological environment through its acquisition (DAQ) system on the sensitivity of these devices. gate8,9,15,18,20–22, producing a local signal pre-amplification. Second, critical aspects for their chronic application in vivo have Although active sensing technologies present substantial advan- been resolved; including the stability of the graphene doping, the tages over conventional micro-electrode arrays, their imple- long-term stability of the g-SGFETs sensitivity and their acute, as mentation is currently limited by the demanding material well as chronic biocompatibility. Finally, we have applied this properties required. In order to achieve long-term and highly methodology to monitor the epicortical local field potentials sensitive neural recordings, materials for active sensing are (LFP) in a freely moving rat model simultaneously with its three- expected to exhibit semiconducting or semimetallic properties, a dimensional (3D)-position during long sessions of up to ∼24 h. high electrical mobility and low intrinsic noise, in addition to a The combination of behavioral and electrophysiological data has high stability, easy integration in flexible substrates and bio- been used to assess the capabilities of the wireless recording compatibility. Some active sensors based on organic semi- system to monitor brain dynamics across unperturbed alternation conductors and thin Si nanomembranes have exhibited of brain states and validate its sensitivity to detection of high- promising performance, with novel transistor architectures17,22 frequency oscillations associated with sparse behavioral events. As and insulating technologies4,14 improving their performance in an illustration of unique features of the g-SGFET recording we some typically constrained aspects such as their frequency provide a first case demonstration of infra-slow topographically response or their long-term stability. Graphene-based active specific and brain-state invariant pattern associated with high- sensors are another promising candidate to meet these require- voltage spindles (HVS). Furthermore, we find changes in infra- ments due to the flexibility of graphene23,24, its high expected slow signal power between slow wave sleep (SWS) and rapid eye stability25 and biocompatibility26,27, as well as its electronic movement (REM) sleep and identify the modulation of theta properties, including a high mobility of charge carriers28,29. oscillations and sleep spindles by the phase of the DC-signal Graphene solution-gated field-effect transistors (g-SGFETs) have infra-slow dynamics during REM and SWS, respectively. The demonstrated a high sensitivity for the detection of local field results reported here demonstrate that neural probes based on potentials15 (LFP), as well as a high performance in multiplexed graphene active sensor arrays represent a mature technology, with operation6,7. In addition, g-SGFETs have recently demonstrated a a high sensitivity, stability and biocompatibility, which allows to high sensitivity for the mapping of infra-slow (<0.5 Hz) brain chronically map wide frequency band epicortical brain dynamics activity (ISA)30–32 with high spatial resolution6,7,33,34. in freely behaving animals. ISA has recently attracted increasing attention due to its unique neurophysiological basis30 and its relation to resting state net- works31,35–37 and to brain states36,38–40. To date, ISA has been Results typically studied using full-band electroencephalography (fb- Homogeneity and sensitivity of graphene active sensor tech- EEG)41,42. However, increasing spatial resolution of ISA mon- nology. For the implementation of graphene active sensor arrays itoring by using small size electrodes is ultimately limited by the as a readily available tool for neuroscientific research, the dependence of the amplifier gain on the impedance of the elec- maturity of large-scale, flexible graphene electronics is critical. trodes used. This dependence leads to signal-to-noise loss and Two of the main challenges in the development of these tech- signal distortion43 at low frequencies. For this reason, studies of nologies have typically been the production of high-quality sin- ISA with high spatial resolution have been typically restricted to gle-layer graphene (SLG) and its transfer onto the required indirect measurement methods such as functional magnetic substrate. The development of wafer-scale methods to produce resonance imaging31,37, optical methods44 or the analysis of infra- SLG has concentrated many efforts and investment in the last slow changes of signal power at higher frequencies45. G-SGFETs, decade50, recently leading to important progress on the growth of as active sensors, transduce the electrochemical potential signals graphene by chemical-vapor deposition (CVD)51. Here, we show in the brain (Vsig) into drain-to-source current (Ids) signals (see that the quality of commercially available single-layer graphene Fig. 1a). The amplitude of the transduced signals is proportional produced by CVD and transferred on a flexible polymeric sub- to the transconductance (gm), defined as the slope of the Ids–Vgs strate (spin-coated on a 4-inch Si wafer) is high enough to enable curves divided by Vds (see Fig. 1b). gm is proportional to the gate the fabrication of g-SGFET arrays with a homogeneous good capacitance per unit area (intensive property) and to the W/L performance both in terms of gm and electrical low- ratio of transistor, but not to its active area17,46–49. Signal frequency noise. detection based on the field-effect mechanism, therefore, allows to Figure 2a shows the boxplot for the gm of nine neural probes, prevent the signal distortion and gain loss observed for small each of them containing 64 g-SGFETs (with a size of 100 × 100 passive sensors in the infra-slow frequency band. This advantage μm2 chosen for mesoscale epicortical LFP analysis10). These is expected to be valid for all FET-based sensor technologies with probes were randomly selected from three wafers, all of them stable transfer characteristics, however, experimental proof has processed in independent batches (see “Methods” for fabrication been only shown for g-SGFETs, which present a particularly high details). It is possible to observe a high homogeneity and yield in chemical inertness25,33. The properties of g-SGFETs represent a terms of gm, with 99% of channels working (defined as transistors qualitative change in the study of ISA, allowing to explore its having a gm above 0.7 times the median). The measured median 2 NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w ARTICLE Fig. 1 Graphene active sensor arrays for chronic, wireless monitoring of wide frequency band epicortical neural activity. a Schematic of a g-SGFET and its equivalent circuit. The small-signal transduction from voltage to current is modeled by the current source GmVsig, where Gm ≡ dIds/dVgs. The DC current is modeled by the Rds element. b Average stationary transfer characteristics of 8 g-SGFETs (left axis) and the gm of 64 g-SGFETs (right axis). The filled area indicates the standard deviation. c Illustration of the rat with the untethered recording system implanted. The headstage and the 3D-printed frame to hold it are covered by a 3D-printed enclosure. On top, the position markers of the motion capture (Mocap) system are fixed, which reflect light back to the Mocap cameras placed in the room. The neural signals transduced by the graphene sensors are digitized and transmitted wirelessly to the signal receiver, which is connected to a computer for signal recording. d g-SGFET array placed on the rat cortex; the position of the reference electrode in contact with the cerebellum and two Pt-Ir electrodes at either side of the g-SGFET array are marked with arrows. e Photograph of the wireless headstage designed for these experiments. f Photograph of the 64 g-SGFET array mounted on a customized connector (left) and zoomed image of the probe active area (right). The red squares indicate the g-SGFETs on the array, which are connected to the headstage inputs with DC capabilities. gm, 1.9 mS/V, is relatively high with respect to flexible silicon a mean of 4.13 μVrms and a standard deviation of 1.14 μVrms FETs8 and comparable with typical organic transistor values17,52 (excluding the outliers shown in Fig. 2b). These results show that due to the high electrical mobility and gate capacitance of graphene-based neural probes prepared using a 4-inch wafer- g-SGFETs. In Fig. 2b the equivalent noise at the gate (Vgs-rms) of scale fabrication process can be obtained with a high homo- the same devices is shown (see supplementary information S2). geneity and sensitivity. Additionally, upscaling of the fabrication Vgs-rms is an important figure of merit to evaluate the sensitivity process to an industrial scale is expected to further improve the of the sensors47, which is defined as the ratio between the homogeneity of g-SGFETs characteristics, especially in terms of integrated current noise (Ids-rms) of the transistor and its the contamination-dependent charge noise47,53. transconductance. Although this parameter presents a larger dispersion than gm, it is possible to identify 3 out of 9 probes with Wireless headstage design and characteristics. Another aspect 96% of the g-SGFETs showing a Vgs-rms below 10 µVrms, that contributes to the sensitivity of the recording system is the suggesting that the measured noise is not directly related to gm. noise introduced by the headstage in the amplification and digi- In fact, low-frequency noise in graphene has been reported to tization process. The amplification of the wide frequency band originate from charge trapping-detrapping events53, which makes activity requires a DC-coupled system, which implies the digiti- noise directly proportional to the density of traps and thus, zation of signals with large DC-offsets. In order to digitize signals sensitive to impurities in the environment of graphene. Figures 2c, with such a large dynamic range and minimize quantization d show the distribution of gm and Vgs-rms, respectively, for the noise, a two-stage transimpedance amplifier has been imple- probe #3 labeled with an asterisk in Fig. 2a. The dispersion in the mented (see schematic in Fig. 2e). The first stage converts the Ids transconductance of the g-SGFETs can be taken into account in currents from the g-SGFETs into voltage, which contains a wide the calibration of the neural signals, correcting dispersion in the frequency band signal, including the infra-slow frequency com- signal amplification. Therefore, the truly limiting factor in terms ponents of Ids. In the second amplification stage (see Fig. 2e), the of homogeneity of the g-SGFETs performance is the equivalent signal is high-pass filtered to remove the DC offset and fill the full noise at the gate. Vgs-rms presents a log-normal distribution54 with scale of the analog-to-digital converter (ADC). In order to NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w Fig. 2 Evaluation of system sensitivity. a Boxplot of gm for nine randomly selected probes from three different wafers produced in independent batches. The yield in terms of gm above 70% of the median is indicated. b Boxplot for Vgs-rms measured in the 1–10 Hz frequency range, plotted for the same neural probes evaluated in part a. All probes consisting of 64 g-SGFETs. The boxes extend from the lower to the upper quartiles, with a line at the median. The whiskers extend 1.5 times the inter-quartile range and the data points beyond the whiskers are indicated by a dot. c Histogram of gm for the 64 g-SGFETs of probe #3 (labeled in panel a) and Gaussian fit of the histogram excluding the outliers shown in panel a. d Histogram of Vgs-rms for the 64 transistors in probe #3 (see panel b) and log-normal fit of the histogram excluding the outliers shown in panel b. e Equivalent circuit of the wireless headstage. f Power spectral density (PSD) of the noise from DC channels (black) and AC channels (orange) in probe #3. The 1/f dependence is represented by the solid red line. The vertical orange line indicates the hardware high-pass filter applied to AC channels at 0.15 Hz. The quantization noise of the DC and AC channels is indicated by the horizontal dashed red lines. g Representation of the Vgs-rms for all g-SGFETs in probe #3 shown for different bandwidths; 0.05-0.5 Hz band for the DC-channels (left), 1–10 Hz band (middle), and 20–200 Hz (right). The position of the g-SGFETs on the array connected to DC-channels of the headstage is indicated by the red squares. h Time domain representation of the noise spectra shown in part f and g (DC-channels filtered in the 0.05–0.5 Hz band and AC channels in the 20–200 Hz band). Signal from eight channels are overlapped. dynamically choose between a DC or AC coupling for each (SVgs(f)), defined as the PSD of the current noise over the channel, a multiplexer has been added to switch between the transconductance (see Fig. 2f). The central part of the spectrum, output of the first and the second stages, see Fig. 2e. Multiplexers from roughly 0.05 Hz to 10 Hz, is dominated by the 1/f intrinsic have been implemented in only 8 of the 64 channels in order to noise of the graphene transistors53. For frequencies below 0.05 minimize power consumption and, therefore, battery weight and Hz, the DC-channels show a slight increase above the 1/f noise, volume of the dedicated wireless headstage. which is attributed to the contribution of additional noise sources Having a relatively high transconductance is important in in the amplification chain, leading to slightly larger Vgs-rms values order to pre-amplify the signals above the noise floor of the in the 0.005–0.05 Hz band (see supplementary information S2). transimpedance amplifiers. However, active sensors typically Above 10 Hz the noise spectra present a significant increase above present an intrinsic 1/f noise, which scales with the drain-to- the 1/f noise, caused by the quantization noise of the headstage source current47. Therefore, Vgs-rms is a more suitable figure of amplifiers, which is more pronounced in DC-channels. The merit to evaluate the sensitivity of active sensors. In order to SVgs(f) integrated in different frequency bands is shown in Fig. 2g validate that the sensitivity of the recording system is limited by for all channels on the neural probe #3. The three maps the intrinsic noise of the active sensors, it is paramount to demonstrate the similarity of the sensitivity of the system in the evaluate the impact of the amplification electronics on the different ranges, with only a significant increase in the 20–200 Hz sensitivity of the system in a wide frequency band. The noise level band. In this band, the noise of the DC-channels exceeds for DC and AC channels can be evaluated from the power the noise of the AC channels; however all the graphene spectral density (PSD) of the equivalent voltage noise at the gate sensors (except for an outlier) keep Vgs-rms values below 15 µV. 4 NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w ARTICLE The digitization noise for AC channels might be decreased by information S3). However, from these results it is not possible further optimizing the gain of the second amplification stage. to distinguish among all different contributions. The accumu- However, the intrinsic noise of the amplifier is expected to lated drift in the CNP measured during the first 24 h of dominate for large amplification gains. In order to better illustrate recording reaches approximately 50 mV, with a maximum the constant sensitivity over frequency, Fig. 2h shows the time change rate of ∼20 mV/h in the first hour (see supplementary domain representation of the noise signal filtered in the ISA band information S3). Figure 3b shows the measured signal in two (0.05–0.5 Hz) and the high-frequency band (20–200 Hz). The DC-coupled channels (high-pass filtered at 1 mHz) during the histogram plotted next to the time-domain representation of both first 2 h of recording. Figure 3c shows the amplitude-phase signals shows their probability density distribution, which relationship between these two DC-coupled channels in the demonstrates the similarity of their variance, as expected from 0.005–0.05 Hz band (see “Methods” section). The left panel the integration of a 1/f spectrum in these frequency bands. Note shows the amplitude-phase coupling measured in PBS, while that the apparently lower amplitude in the time-domain the right panel shows the equivalent results in vivo. The in vivo representation of the infra-slow noise is due to the different signals exhibit fluctuations with a much larger amplitude than timescales of 1/f noise in both frequency bands, but not due to a the signals recorded in PBS, ruling out the transistor 1/f noise different signal variance. and the headstage noise as the origin of these infra-slow These results show the high sensitivity of the system in a wide oscillations. Further, the in vivo signals recorded in the frequency band, with Vgs-rms below 5 μV in the infra-slow 0.005–0.05 Hz band present fluctuations in anti-phase, which frequency band. In the design of the headstage, we have confirm that neither instabilities in the reference electrode nor considered the compromise between reaching maximum sensi- adsorption/desorption of chemical species on graphene are tivity in the high-frequency range and minimizing the power responsible for these fluctuations. To conclude this discussion, consumption of the DC-coupled recording system with a Fig. 3d shows the effect of drifts in the graphene doping on the relatively high channel count. Smaller g-SGFETs are expected Vgs-rms of graphene sensors. These results demonstrate that to present a higher intrinsic noise (see supplementary informa- their sensitivity does not change significantly due to the tion S2), as expected for any active or passive sensor. Therefore, accumulated drifts during up to 24 h if the initial bias is our results indicate that the sensitivity of g-SGFETs in the infra- selected properly. Therefore, daily tracking of the CNP and slow frequency is not affected by the amplification electronics for readjustment of the Vgs overpotential back to optimum values is sensor areas below 100 × 100 μm. This is in strong contrast with enough to keep a constant sensitivity over long-term monitor- ISA detection using passive electrodes, for which the gain loss and ing of the brain dynamics. signal distortion is expected to increase for smaller sensor In addition to changes in the doping of graphene, the dimensions. These results demonstrate the limits and the transconductance and noise of the g-SGFETs might vary over scalability of the g-SGFET technology towards higher density time due, for instance, to the creation of defects in the graphene arrays with ISA detection capabilities. lattice. Pristine graphene has shown excellent chemical stability due to its sp2 hybridization25. However, dangling bonds at edges, grain boundaries, atomic vacancies or reconstructions in the Signal stability and sensitivity over time. Once the performance atomic lattice increase the reactivity of graphene, which might of the graphene transistors and the headstage is properly assessed, lead to the creation of defects over time25. In addition, there the stability of the g-SGFETs in an in vivo chronic setting needs might be mechanical causes of performance degradation such as to be evaluated in order to ensure the reliability of the recording detachment of graphene from the substrate or bending-induced system. strain on the graphene lattice and metal-graphene contacts. The Ids–Vgs curves of the g-SGFETs describe the relationship Another possible cause of sensitivity degradation could be the between the measured drain-to-source current and the electro- encapsulation of the device by glial scar tissue58. This layer of chemical potential at the graphene-electrolyte interface. The tissue can be modeled as an electrical impedance in series with the minimum in Ids occurs at a particular gate voltage, referred to as graphene-electrolyte interface59, which can eventually lead to a the charge neutrality point (CNP), which is also related to a degraded frequency response of the g-SGFETs. minimum in the sensitivity of the device (see Fig. 1a). The CNP In order to track changes in the sensitivity over time in a corresponds to the bias conditions for which the Fermi energy in chronic implant, the gm extracted from the Ids–Vgs curves and the the graphene channel is, on average, closest to the energy with a Vgs-rms were periodically measured for the 8 DC-coupled minimum density of states (i.e., the Dirac point)55. The Vgs channels over 4 weeks. Figure 3e shows that gm remained overpotential required to reach this energy depends on the approximately constant, suggesting that there are no major doping56 of the graphene channel, as well as on the electro- creation of defects in the graphene channel in the in vivo chemical potential of the reference electrode. Therefore, instabil- environment. Similarly, the Vgs-rms shows only a slight increase in ities in any of these two parameters will produce a shift of the the last days. Figure 3f shows the current noise (Ids-rms) for all the transfer characteristics in the Vgs axis. In turn, this shift implies 64 channels measured at 200 Hz over 4 weeks. At this frequency, that Ids will present a drift and that the sensitivity of the g- it is possible to estimate changes in the sensitivity of the recording SGFETs might vary over time for a constant Vgs overpotential. system due to the low average power of high-frequency neural Having a controllable doping of the g-SGFET and a homogeneous signals (see supplementary information S4). The numeric values CNP among sensors is, therefore, of paramount importance to displayed in Fig. 3f indicate the percentage of g-SGFETs working maintain a good sensitivity of the sensor array. (see supplementary information S4). The frequency response of Figure 3a shows the evolution of the transfer characteristics the transconductance (gm(f)) has also been measured in vivo over over 4 weeks after implantation of the neural probe (see 4 weeks after implantation. For this purpose, two Pt-Ir electrodes “Methods” for implantation details). The observed shift in the were implanted on both sides of the g-SGFET array (see inset in CNP is presumably due to a combination of factors, including Fig. 3g) and 1 µA amplitude pure tone signals of different desorption of contaminants by electrochemical cleaning of frequencies were applied using a current source. Figure 3g shows the graphene-electrolyte interface57, adsorption of charged the magnitude of gm(f) for different days after implantation of chemical species present in the environment or changes in the neural probe normalized by the mean magnitude at 1 Hz; the the reference electrode potential (see supplementary phase of gm(f) is shown in the supplementary information S5. The NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w Fig. 3 g-SGFET stability in-vivo. a CNP vs. time over 4 weeks. The inset shows the Ids–Vgs curves. Mean and standard deviation for n = 8 g-SGFETs (1 outlier excluded). b Signal from two DC-coupled channels. Indicated positions corresponding to map in panel h. The spectrogram of channel (5,7) is shown (bottom). c The phase-amplitude relation between the channels in panel b, for the noise measured in the beaker (left) and for the signals measured in vivo (right). d Boxplot of Vgs-rms vs. shifts in the effective gating (Vgs–VCNP) of the 64 g-SGFETs. The colored area indicates the measured drift in the CNP referred to a Ag/AgCl electrode during the first 24 h of recording. The initial bias and the CNP are indicated by the red and green vertical lines respectively. e gm (top) and Vgs-rms (bottom) measured over 4 weeks post implantation; gm was obtained from the Ids–Vgs curves of the DC-coupled channels (n = 8 g- SGFETs, 1 outlier excluded). f Current noise over 4 weeks after implantation (n = 64 g-SGFETs). The numeric values indicate the yield of working devices (see supplementary information S4). The boxes in panels a and d–f extend from the lower to the upper quartiles, with a line at the median. The whiskers extend 1.5 times the inter-quartile range and the data points beyond the whiskers are indicated by a dot. g Average and standard deviation of the frequency-dependent transconductance (|gm | (f)) shown for different days after the implantation (n = 10 g-SGFETs). The inset shows the approximate position of the Pt-Ir electrode close to the array, the simulated equipotential contour lines in a conductive plane and the relative signal amplitude measured by each of the g-SGFETs in the array (see supplementary information S5). h Signals measured by all g-SGFETs on the array during a state of increased theta activity on day 1 and day 6 after implantation. approximately constant slope (in a log-log scale) follows a greater detail, by studying a large animal cohort, the stability of fractional order attenuation (i.e., approximately ∝1/f0.1), which the biological signal over time, a critical aspect in electrophysiol- has been recently attributed to the non-ideal capacitive response ogy research and for the long-term performance of brain- of the graphene-electrolyte interface34. A calibration method to computer interfaces61,62. Furthermore, the polymers used as a correct such transconductance variation has also been pro- substrate and passivation layers could be modified to reduce the posed34. The evolution of the frequency response does not show moisture absorption14,63 and displace the neutral plane of the major changes in the slope of the gm attenuation, indicating that device at the position of the graphene channel (see “Methods” there is not a significant increase in the electrical impedance in section). Yet, the results presented in this section reveal a series with the graphene-electrolyte interface due to device promisingly stable performance over time, which sets a lower encapsulation60. To conclude, Fig. 3g shows the recorded neural bound for the stability of g-SGFETs in a chronic implant activity in a state of increased theta power in day 1 and day 6 after environment. Besides, from a system perspective, we show that g- implantation, illustrating the good homogeneity and stability of SGFET arrays can measure very slow biological signals (high-pass the g-SGFETs performance. Future studies could address in filtered above 1 mHz). 6 NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w ARTICLE Fig. 4 Biocompatibility testing of non-functional g-SGFET vs. control devices. a The timeline describes the procedures carried out on animals during the biocompatibility study. b Schematic of the high-surface area g-SGFET prototype developed for biocompatibility testing in-vivo. c Discrimination ratio from NOR test over different days after implantation (see “Methods” section). For all five groups tested, the discrimination ratio was above 0.5 at all timepoints. Evaluated for n = 7 animals per group at all timepoints, except 12 weeks which had n = 3 (sham), n = 4 (platinum and naive) and n = 7 (blank). The boxes from the lower to the upper quartile, while whiskers represent minimum and maximum values. d Inflammatory marker IL-17a in the brain tissue for all groups and timepoints. Evaluated for n = 4 animals after 2 and 12 weeks, and n = 3 animals at 12 weeks. e Microglial activation state, expressed as a percentage of total microglial presence in the site surrounding the electrodes. n = 3 animals at 2 and 12 weeks, n = 2 animals (or 3 for the contralateral hemisphere) at 6 weeks. Bars in panels d and e indicate the mean and range of data point. f Iba-1 immunoflourescent staining to assess activation status of microglia at the surgical site obtained from 40 sections per animal. Scale bar equals 500 µm (50 µm at the insets). g Haemotoxylin and Eosin staining at 2 weeks post implantation shows there is no structural damage to the cortical layers directly at the device implantation site. Forty sections at 25 µm per animal were imaged. Scale bar equals 1 mm (top) and 200 µm (bottom). In panels d and e two-way ANOVA test with Dunnett’s multiple comparison to the naive control within each timepoint with n = 3 or larger: *, **, ***, and **** indicate p = 0.015, p = 0.007, p = 0.0016, and p < 0.0001, respectively. Biocompatibility of graphene devices following subacute and significant differences in the discrimination ratio were found in chronic implantation. In order to assess the applicability of g- animals implanted with any device at any of the timepoints SGFET arrays for the long-term monitoring of brain activity under (Fig. 4c). The inflammatory response of the tissue was evaluated natural behavior we have also investigated the biocompatibility of using two main techniques: ELISA of blood or brain tissue for a graphene-based epicortical devices. For this purpose, animals were panel of inflammatory cytokines, and immunohistochemical implanted with one of three devices onto the parietal cortex of the analysis of brain tissue for cells associated with inflammation. brain, or had the full surgery without the implantation of any device ELISA was performed for four cytokines: interleukin-6 (IL-6), (sham control). A cohort of naive animals who had no intervention interleukin-17a (IL-17a), interferon gamma (IFN-γ), and tumor were used as a control. Three time points were chosen to assess necrosis factor alpha (TNF-a). In blood serum, there were no tissue response: 2 weeks, 6 weeks, and 12 weeks post implantation significant differences between any of the materials at any (Fig. 4a). Non-functional devices were custom-designed with an timepoints (see Fig. S6). For cytokine expression in brain tissue, enlarged surface area of CVD graphene, in order to maximize significantly higher levels of all four cytokines in both graphene exposure of the material to the brain tissue (see Fig. 4b for device and platinum devices were found at the 2 weeks timepoint, when dimensions). The experiments were designed following the gui- compared with the contralateral hemisphere control. Whilst dance from the ISO 10993 standard, which details the biological release of these factors is typically intended to prevent further evaluation of medical devices. Ethylene oxide sterilization was damage to the CNS tissue, prolonged expression can be applied prior to implantation64. After implantation, the immuno- detrimental. By 6 weeks post implantation, there was still a histochemical response of the tissue and potential effects on the significant elevation of both IL17a and IFN-γ for graphene and behavior were investigated. platinum devices vs. control expression, and by week 12, there Behavior was assessed with the novel object recognition (NOR) was no significant expression of any cytokine for any treatment test, used to assess impairment of cognition and memory65. No group (Fig. 4d and Fig. S7). These findings demonstrate that the NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w adverse tissue response to graphene is transient in nature, the LFP dynamics related to spontaneous behavior. The analysis comparable to the current clinical standard and specific to the of the relationships between epicortical brain activity and freely implantation site, with no observed systemic complications. moving behavior was performed over timescales enabled by the To confirm the ELISA data, manual counting of the activation wireless recording system. This capability is critical for the study state of microglial cells was also performed to assess the of sparsely occurring behavior events, as well as ISA patterns over inflammatory state within the brain. Microglial cells are always distinct brain states. present within the brain, but their morphology serves as an Brain states were classified through a combination of spectral indicator of the inflammatory state within the brain62. Expression features in the epicortical LFP signal and motor data. In this way of activated microglia was increased at both the 2 weeks and we distinguished the following classes: slow wave sleep (SWS), 6 weeks post implantation timepoints, and this activation was REM sleep (REM), Awake Theta (AwT), and Awake Non-Theta present to a significant level at 2 weeks post implantation in all (AwNT). Figure 5b illustrates the criteria for the brain states four treatment groups when compared with the contralateral classification, described in detail in the “Methods” section. First, hemisphere. However, similar to the ELISA, the activation of slow wave (SW) states, showing increased power in the 1–25 Hz microglia had returned to baseline levels by 12 weeks, indicating band, and Theta states were identified (see Fig. 5b). The behavior no prolonged inflammatory reaction to the devices (Fig. 4e, f and of the animal was then classified in either active or inactive Fig. S8). TUNEL cell counting was also performed, to assess any periods from the motion tracking data (see Fig. 5b and “Methods” cell death within the tissue as a result of the implantation of section). During inactive motor behavior, SW states were devices. At 2 weeks post implantation, there was a significant classified as SWS except in direct proximity to HVS events, increase in the number of TUNEL-positive cells for both while Theta states were categorized as REM if directly preceded graphene and sham surgery groups. However, by 6 weeks there by SWS. On the other hand, Theta states occurring during active was no evidence of cell death, which was also true at 12 weeks behavior were classified as AwT and SW states as AwNT. Finally, (Fig. S9). Finally, there was no obvious morphological changes periods not assigned to Theta or SW states were classified as seen with haemotoxylin and eosin staining. There was an AwNT regardless of the behavior of the animal. During the appearance of sunken cortex in some brains, however, this was majority of recording hours all four sleep/wake states were due to perfusion fixation with the glass window in place, and expressed at least once in line with the polyphasic nature of rat there was no effect on the thickness of the cortical layers below sleep66–68. Their relative prominence however varied substan- the implantation site, as shown in Fig. 4g. tially over the course of the recording day paralleling the changes Overall, both cytokine expression and histological analysis of observed in motor states (Fig. 5c), in line with circadian the brain area at the implantation site showed an acute reaction to rhythmicity. the implantation of devices. However, this was not specific to the Classification of brain states is typically based on the delta, graphene devices, even though an enlarged surface area of alpha-beta and theta frequency bands (see “Methods” section), graphene was used in order to maximize the material specific reflecting fast-time scale state-specific network dynamics. How- response. By 6 weeks, the reaction showed clear signs of ever, some recent research highlighted the role of infra-slow amelioration, and by 12 weeks, there was no detectable reaction dynamics in the regulation of brain sub-states40, via modulation to the devices using any technique. Graphene and Pt devices of higher LFP frequency bands during sleep39,45,69 and dynamic showed a similar level of microglia activation compared to coordination and segregation of the resting state35,70. These “blank” devices, while the latter shows much smaller presence of results show the potential importance of ISA for a complete inflammatory markers than for Graphene or Pt. These results classification and study of brain-states. The graphene-based suggest that microglia activation is more strongly associated with recording system presented here represents an ideal tool for the surgical procedure and probe insertion, while inflammation is study of cortical ISA signals with a high accuracy and spatial primarily affected by the device material. In this way, functional resolution in freely behaving animals. The spectrogram in Fig. 5b sensor arrays, which present a much lower graphene area, are illustrates changes of the spectral power for frequencies between expected to cause an inflammation closer to that caused by 0.015 and 4 Hz over the transition between SWS and REM. It is “blank” devices. In addition, according to NOR test, graphene possible to observe clear increase in the ISA-band power devices did not affect significantly the animal behavior neither in following the transition from SWS to REM, even at the single acute nor chronic timepoints. Based on these results, graphene- trial level (see Fig. 5b). Taking advantage of the long-term based devices presented adequate biocompatibility for chronic recording capabilities of our system, we could sample 44 of such implantation, comparable to the equivalent platinum-based sparsely occurring SWS-REM (REM duration longer than 40 s) devices. state transitions within a 24 h period. Besides, the spatial mapping of ISA enabled by the g-SGFET technology allows to resolve the topographic region-specific modulation of ISA at the SWS-REM Long-term monitoring of wide frequency band epicortical state transition (see supporting information S12). Interestingly, brain activity during natural behavior. During the longitudinal delta-band power, associated with slow oscillations, and infra- in vivo assessment of the g-SGFET sensitivity, we recorded epi- slow power showed changes in opposite directions between SWS cortical brain activity in a freely behaving rat for up to 24 h. and REM sleep. While delta band power expectedly decreases Throughout the recording period, the 3D-motion of the animal from SWS to REM, associated with desynchronized cortical state, was tracked with a motion capture (Mocap) system (see “Meth- infra-slow power increases in REM (see Figs. 5d, e and Fig. S12 ods” section and Fig. 5a). The conjunctive recording of animal and statistical analysis in “Methods”). motion and wide frequency band epicortical signals was used to In order to further illustrate the wide frequency band classify brain states and behavioral states over the recording sensitivity of the recording system, we quantified the strength period. In turn, this classification was used to support two main of modulation of LFP power in the slow frequency range (1–15 purposes. First, to validate the ability of the graphene-based Hz) by the phase of the ISA activity during REM and wireless recording system to perform long-term stable recordings SWS. Interestingly, ISA phase significantly modulated theta in freely moving rat across multiple brain states, and test its power (8–9 Hz) during REM sleep (Fig. 5f-left) and spindle band suitability to study infra-slow epicortical LFP dynamics. Second, power (9–13 Hz) during SWS (Fig. 5f-right). The strength of to assess the g-SGFET sensitivity in the high-frequency range of ISA phase modulation was tenfold higher during REM compared 8 NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w ARTICLE Fig. 5 Infra-slow to high-gamma correlates of sleep and behavioral states. a 3D trajectories of the head position of the rat. The inset shows a scheme of the position of the Mocap. b The spectrogram and raw LFP signal of an illustrative channel is displayed for distinct brain states (top); slow-wave (SW), high-voltage spindles (HVS) and Theta. Movement speed is displayed along with classification of motor state (middle) and brain states (bottom). c Top: percentage of time in the active vs. inactive state (interruptions to replace the battery not included). Bottom: percentage of time the rat was in each main brain state. d Average 0.015–4 Hz spectrogram for one DC channel triggered on REM episode onsets (n = 44). e Median PSD across SWS-REM transition episodes (n = 44) for 30 s periods pre and post REM onset. Shaded area marks frequency bins with significant difference (p < 0.05, permutation test). f Color-coded strength of modulation of LFP power across slow frequency range (y-axis) for one channel by the phase of ISA across 0.05–0.2 Hz range derived from one DC channel (see panel g) during REM sleep (left) and SWS (right) states. Gray color indicates nonsignificant modulation. Insets show circular plot of LFP power in theta/spindle band as a function of ISA phase. g Color-coded topographic maps of ISA phase modulation of LFP power in theta band during REM (left) and spindle band during SWS (right). ISA phase derived from DC channel marked with a red square. h Color-coded strength of modulation of LFP power across gamma frequency range (y-axis) for one channel by the phase of ISA across 0.05–0.2 Hz range derived from one DC channel (left) and by the phase of LFP in the slow frequency range rhythm (right) during REM. Inset, circular plot of LFP gamma power with respect to respective (ISA or theta) phase. i Average spectrogram for high- frequency range of LFP on posterior channel triggered on rear onset (n = 162). j Head elevation (bottom) and high gamma power (top, same channel as in i) color-coded and centered on rearing onset shown for all events sorted by duration of rear event. to SWS, and the ISA phase of maximal LFP power differed of LFP power across cortex, with both theta power during REM between states being close to the peak (~340°) in REM and and spindle power during SWS showing strongest modulation in ascending phase (~300°) in SWS. Taking advantage of the posterior part of the array (Fig. 5g). While theta oscillations coverage of a significant section of dorsal cortical mantle by our measured on the cortical surface are generated by volume array, we assessed the spatial extent of the ISA phase modulation conduction of multiple theta-rhythmic current generators of NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w entorhino-hippocampal circuits71,72, sleep spindles are generated on more frontal channels of our array (see supplementary by rhythmic currents of thalamo-cortical projections to granular information S11). In strong contrast no such suppression was cortical layers73. The fact that power of hippocampal theta and observed in the gamma band between 60 and 70 Hz (see Fig. 5i cortical spindle band is modulated by the phase of ISA derived and supplementary information S11). from cortical surface likely reflects global infra-slow dynamics Subsequently, we took advantage of the infra-slow recording that co-modulates both limbic and cortical circuits. While the capability of g-SGFET arrays to characterize topographic infra- topographic profile of theta power (Fig. 5g) modulation by ISA slow and spectral AC epicortical signals associated with rare phase is consistent with anatomical localization of underlying highly synchronous high-voltage spindle (HVS) oscillations78,79, hippocampal theta current generators, stronger modulation of the as the most likely cortical dynamics associated with large infra- spindle power on posterior cortical areas might reflect anatomical slow currents, similar to those shown during epileptiform activity thalamo-cortical subcircuits that are more strongly co-modulated in development80. Consistent with previous studies78,79, HVS by ISA dynamics than derived from epicortical DC signal. Finally, occurrence was associated primarily with alert immobility states we tested whether g-SGFETs SNR is sufficient to detect (IMM) (566 events, Fig. 6a), where IMM is defined as the fluctuations in the high-frequency LFP dynamics at different intersection between inactive and awake states. Benefiting from time scales and to this end quantified the strength of modulation the long-term unperturbed recording allowed by our system we of broad range gamma power (30–200 Hz) by both ISA phase and could also sample significant number of HVS events during REM theta rhythm phase during REM sleep. Gamma power in in the sleep (92 events), where they coexisted with hippocampal theta range of 60-120 Hz was modulated by the ISA phase reaching oscillations visible on posterior derivations (Fig. 6b). While maximum power at the peak of the ISA (~10°) (Fig. 5h-left) and, median duration of HVS events was comparable in immobility consistently with published work based on intracranial record- and REM sleep (~5 s, Fig. 6c-top), the rate of detected HVS events ings73, high gamma (120–150 Hz) power was modulated by theta varied across the recorded 24 h period, (Fig. 6c-bottom). phase (Fig. 5h-right). Interestingly, HVS during both brain states were associated with Having established that we can record state-selective transient infra-slow fluctuations as visible in single examples epicortical signals with g-SGFETs across a range of brain/ (Fig. 6a, b) and average profiles (Fig. 6d, e). Specifically, positive motor states, we subsequently went on to demonstrate (putative source) and negative (putative sink) infra-slow tran- applicability of the technique for linking behavior and cortical sients in, respectively, posterior and frontal positions on the array physiology. To this end we focused on a specific and sparsely coincided in duration (median 5 s) with oscillatory dynamics of occurring spontaneous behavior, rearing on the hindlimbs. HVS (Fig. 6c, d, e). Topographic profiles of spectral peak power of Rearing is a exploratory behavior in rodents, which is context- HVS were comparable for both states and showed maximal and stress-sensitive74,75, has been hypothesized to support increase to baseline in frontal derivations overlaying sensory- sampling of distal landmarks for construction of a cognitive motor cortex (Fig. 6f). REM-associated HVSs were, on average, model of the surrounding environment75,76 and is implicated slower and larger in power than immobility-associated ones in modulation of cortico-hippocampal interactions in theta and (Wilcoxon ranksum test, p < 1e-19 for power and p < 1e-8 for gamma frequencies76,77. In general, due to the sporadic and frequency, Fig. 6g). In contrast, spatial structure and magnitude spontaneous occurrence of rearing events their neural physiol- of infra-slow fluctuations associated with HVS as expressed by ogy has been less widely investigated with conventional positive fluctuations on posterior and negative fluctuation on recording methods compared to task-specific trained motor frontal DC channels, were comparable for both states (Wilcoxon actions. Technologies that combine long-term recording ranksum test between IMM and REM, p = 0.5 for posterior peak stability, high spatial resolution, wireless methodology and magnitude and p = 0.9 for frontal trough magnitude, Fig. 6h). precise 3D-tracking of animal behavior, as presented here, Although future work is required to replicate these observa- open the door to investigating this class of phenomena with a tions in a large cohort of animals, our long-term recordings great level of detail. Therefore, we took advantage of the enabled detailed quantitative analysis of rare physiological presented technology to collect a large number of individual patterns and illustrate the power of this technology. Because of spontaneous rearing events during a full 24 h period. Evaluat- their sparse, strongly state dependent occurrence and distribution ing the signature of rearing on the gamma epicortical activity over a wide frequency band these events illustrate the class of band is of additional interest for our study, since it can be used phenomena whose functional study necessitates the integration of to illustrate the capabilities of the g-SGFETs in the high- neural-behavioiral measurements with both high spatio-temporal frequency LFP range. In order to robustly detect rearing resolution and large spatio-temporal span in freely moving episodes we took advantage of the continuous 3D tracking, animals, as enabled by our wireless electrophysiology system. detecting rearing events (n = 163) based on head elevation above ground (see “Methods” and supplementary informa- tion S6). While rearing events occurred throughout the Discussion recording period, expression of rearing activity was highly Graphene active-sensor arrays represent an emerging technology variable across the day, as with overall motor activity, ranging in neural engineering, which has recently demonstrated a strong from 250 s to 0 s spent rearing per hour (mean 43.8 ± 12.1 s, potential for the production of high-count sensor arrays6,7, as supplementary information S10). Additionally, rearing events well as for wide frequency band neural sensing33. In this article, expressed variability in terms of height (mean 250,6 ± 2.7 mm, we have presented a detailed characterization of various technical see Fig. 5j and Fig. S10d) and duration (mean 5.7 ± 2.8 s, see aspects required for their actual application, such as the homo- Fig. S10e). geneity in the performance of the graphene sensors, the specifi- Having detected this set of spontaneous rearing episodes we cations required for a dedicated headstage, the limits in the proceeded to analyze the power spectra of epicortical LFP, which sensitivity of g-SGFETs or their chronic stability and bio- showed distinct rearing-associated changes in brain signals for compatibility in vivo and demonstrated lines of investigation specific frequency ranges. Rearing was associated with the enabled by their technical characteristics. suppression of epicortical high-frequency (90–200 Hz) activity In the first place, we have demonstrated the high yield and (Fig. 5i), which was also observable on the single trial level across homogeneity of the g-SGFETs produced in a wafer-scale process the range of rear heights (Fig. 5j) and most prominently observed using commercially available CVD graphene. This demonstration 10 NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w ARTICLE Fig. 6 Topography of AC and DC dynamics associated with high-voltage spindles. a, b examples of HVS events during immobility (a) and REM sleep (b). Spectrograms at posterior (top) and anterior (middle) positions on the array visualize dynamics of HVS (vertical lines at onsets) and ongoing oscillatory dynamics (low-frequency synchrony in IMM, theta oscillation in REM). Bottom two panels show DC LFP signals from posterior (green) and anterior (brown) positions on the array and head speed. c Distribution of duration of single HVS events for immobility and REM (top) and total hourly duration of HVS across recorded period (bottom). d Average DC LFP traces triggered on HVS onset during immobility and REM sleep (channels arranged as in Fig. 2g- left, inset showing anatomical localization of DC ECoG recording sites). e Average AC LFP spectrograms triggered on HVS onset during immobility (left) and REM sleep (right). The spectrograms at posterior (top) and anterior (bottom) positions on the array contrasting spectral content associated, respectively, with theta (posterior) and HVS (anterior) oscillations. Pseudo-color, spectral power (a.u.). f Topographic maps of change, relative to baseline, in HVS power at peak oscillation frequency during immobility (left) and REM sleep (right). g Distributions of power and peak frequency of individual HVS during immobility (blue) and REM sleep (red). h Distribution of DC fluctuation magnitude at posterior lateral and frontal medial derivations for single HVS during immobility and REM (color code as in g). represents an important milestone towards the production of process on the sensitivity of the recording system. In this way, we graphene-based neural probes at an industrial scale. On another have identified the challenges in the design of full-band amplifiers front, the application of graphene transistors for neural sensing given specific energy and weight constrains for wireless applica- also requires the development of specific electronic equipment for tions. Furthermore, using a dedicated headstage we have biasing the sensors and converting the measured drain-to-source demonstrated that the sensitivity of graphene active sensors in the currents into equivalent voltage signals at the gate. In this study, ISA band is not significantly affected by the rest of the data- we have presented a dedicated wireless headstage, which allowed acquisition system components. Yet, we observed slow drifts in describing the impact of the signal amplification and digitization the measured signals in vivo, which were presumably attributed NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w to adsorption of charged molecules on the graphene channel or large statistical sample of events. The rate of occurrence during drifts in the electrode reference potential. These drifts were easily different brain states could be determined over a 24 h period, eliminated with a high pass filter at 1 mHz; however, future highlighting their occurrence also outside the awake resting state experiments could explore the functionalization of the graphene in which they are commonly reported to be selectively occurring. channel and use of alternative reference electrodes to maintain a Specifically, we found them prominently expressed during REM more stable doping level. sleep83. Having established this bimodal state specificity, we Moreover, we have demonstrated the stability of the graphene analyzed HVS events separately for awake immobility and REM sensors in vivo by characterizing their sensitivity over 4 weeks. In sleep. Topographic analysis of the HVS peak power demonstrated order to determine the stability of the signal quality we have also that sensory-motor cortical preponderance for these oscillations evaluated the signals induced by bipolar stimulation in the LFP is comparable for these two states. In contrast, the large sample frequency band, demonstrating a rather stable frequency response size of events in both states allowed to identify a significant dif- over time. These results suggest that the impedance of glial tissue ference in both frequency content and power of the HVS between surrounding the implant does not significantly affect the sensi- these two brain states. Furthermore, the capabilities to simulta- tivity of the g-SGFETs in vivo. To conclude the characterization neously map ISA patterns allowed to determine the distinct of the device stability, we have evaluated the biocompatibility of topographic structure of spatially specific infra-slow frequency graphene devices over 12 weeks via behavioral and histological components associated with HVS, showing phase-reversal across markers. These results demonstrate an acute foreign body anterio-posterior axis. Importantly, these ISA features were con- response comparable to platinum-based devices, which returns to served between REM-sleep and awake immobility, highlighting values close to those of control animals 12 weeks after implan- that the underlying origin of the DC signal is related to the HVS tation. The demonstration of the graphene biocompatibility and event itself and is independent of the brain state. long-term stability in a chronic implant represents another Future work is required to replicate the presented results in a important turning point towards the large-scale production of large cohort of animals, follow them longitudinally and relate the graphene-based neural probes. surface pattern to intra-laminar and subcortical theta, spindle, The experimental validation of this technology opens up many gamma77 and HVS84 generators. Although we have focused the opportunities for electrophysiology studies in which having access attention on epicortical ISA patterns, analyzing their correlation to unconstrained behavior and multichannel recordings with with ISA dynamics across cortical laminae could provide sensitivity in a wide frequency band is relevant. In this study, we important insights into the origin and implications of ISA. A have shown quasi-continuous monitoring of brain activity in long promising strategy in this direction is to combine the epicortical recording sessions of up to 24 h, which allowed us to acquire large arrays presented here with graphene-based depth probes33. samples of neural activity across unperturbed behavioral and Future chronic recordings of depth and large-scale LFP signals brain states. In combination with 3D-tracking, the wireless across behaviors and behavioral states in freely moving uncon- technology allowed us to explore the relation between neural strained animals will lay the foundation for a new qualitative step activity patterns and behavior events, which occur sparsely over in the brain dynamics investigation from infra-slow to very fast time, with sufficient statistical power. In particular, analyzing frequencies72, contributing to our understanding of the origins of epicortical LFP signals with respect to rearing behavior we ISA dynamics in the context of resting states and default mode observed differential modulation of the 60–70 Hz gamma networks and its links to faster brain dynamics. and 90–200 Hz high gamma range. While high-frequency In summary, the thorough in vitro and in vivo evaluation of the activity between 90 and 200 Hz was consistently suppressed in sensing and long-term recording capabilities of graphene active a topographic manner, no such suppression was observed for the sensors from a system perspective demonstrates the maturity of 60–70 Hz range, pointing towards distinct behavioral selectivity this technology and supports its application for the study of ISA of underlying circuit mechanisms. The determination of without sacrificing high-frequency LFP components. In this frequency-specific power modulation in the gamma range, even direction, we have successfully evaluated ISA patterns during for single rearing events, demonstrates the high sensitivity of the distinct brain states and the correlation of high-frequency oscil- system in the high-frequency spectrum of the LFP. lations with specific sparsely occurring behaviors. Our results In the low-frequency range, we found that infra-slow power represent an important step towards the broad implementation of <0.2 Hz increased significantly across DC sites during REM sleep graphene active-sensor arrays for neuroscience research, offering episodes compared to SWS and thus showed the opposite state a stable and biocompatible sensing technology for long-term dependence than power in the slow oscillation band (1–4 Hz). mapping of wide frequency band epicortical brain activity during Interestingly, infra-slow dynamics modulated power of theta and spontaneous behavior. gamma rhythm during REM and with lower strength, power of sleep spindles during SWS. While modulation of LFP power in Methods theta, beta and gamma bands by ISA phase derived from BOLD Fabrication of g-SGFET arrays. Arrays of g-SGFETs and devices for bio- signal and DC EEG has been shown in humans81,82 and recently compatibility were fabricated on a 10 μm thick polyimide (PI-2611, HD Micro- in anesthetized rats70, the present result is the first demonstration Systems) film spin coated on a Si/SiO2 4” wafer and baked at 350 °C. Polyimide was of interaction between physiologically established oscillatory chosen as a substrate due to its thermoxidative stability, high mechanical strength, insulating properties and chemical resistance63,85, as well as its expected bio- dynamics, theta, spindle and gamma oscillations, and ECoG- compatibility and previously reported stability for chronic implants4,5. A first metal derived ISA in freely moving rodent. Consistent with published layer (10 nm Ti/100 nm Au) was deposited by electron-beam vapor on a previously intracranial work73, volume conducted hippocampal theta mea- photodefined-negative AZ 5412E (Clariant, Germany) and then structured by a sured at the cortical surface also modulated cortical gamma lift-off process. Afterwards, the graphene grown by chemical-vapor deposition on Cu was transferred (process done by Graphenea s.a.). In platinum devices for power, thus demonstrating that the developed technology is biocompatibility studies, another photolithography, metal evaporation and lift-off sufficiently sensitive to characterize known gamma dynamics. followed the first one. Graphene was then patterned by oxygen plasma (50 sccm, While power of ISA modulating sleep spindle oscillations 300 W for 1 min) in a reactive ion etching (RIE). The photodefinable resist used to power during SWS was much lower than that during REM, a protect the graphene in the channel region was HIPR 6512, chosen to minimize the level of contamination. After the graphene etching, a second metal layer was related hypersynchronous thalamo-cortical rhythm, HVS, was patterned on the contacts following the same procedure as for the first layer. The associated with much larger DC transients. Long-term wireless lift-off step was followed by an annealing in ultra-high vacuum to improve the recording allowed us to evaluate the characteristics of HVS in a contact resistance and to eliminate resist residues from the graphene channel. 12 NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w ARTICLE Subsequently, the transistors were insulated with a 3-µm-thick photodefinable SU- skin removed to expose the skull but not the temporal muscle. The periosteum was 8 epoxy photoresist (SU-8 2005 Microchem), keeping uncovered the active area of removed using a bone scraper. The skin around the perimeter of the removed tissue the transistors channel. The SU-8 photoresist was chosen as insulating material was glued to the bone using Vetbond tissue adhesive (3 M, UK). A craniotomy (~4 because it is photodefinable and because its use in chronic implants has been mm x 6 mm) was performed using a high speed surgical micro drill. Lambda was previously reported86,87. The use of a photodefinable passivation polymer is used as a posterior reference for the craniotomy, which was positioned at least 1 required in the current graphene technology because etching of the passivation mm lateral to the midline, to avoid the sagittal sinus. The drilling region was rinsed layer would also etch the underlying graphene channel. The polyimide substrate regularly with saline to prevent heat damage. Once the bone around the border of was structured in a reactive ion etching process using a thick AZ9260-positive the craniotomy was sufficiently thin, all bone shavings and other debris were photoresist (Clariant) layer as an etching mask. The neural probes were then peeled removed using compressed air and the bone flap was gently removed. The cortical off from the wafer and placed in a zero insertion force connector to be interfaced surface was kept moist using Ringer’s solution. A fine needle with the tip bent at a with our custom electronic instrumentation. Finally, the devices were rinsed in 90° angle was used to gently lift the dura away from the cortical surface and ethanol to eliminate remaining resist residues on the graphene channel. another needle was used to create a slit in the dura, carefully positioned to avoid blood vessels. A pocket was created by lifting the dura next to the opening using fine forceps and the device was carefully placed on the cortical surface. The dura Phase-amplitude coupling evaluation. The signal inversion observed between was then repositioned to hold the device in place. A glass window (UQG Optics, channels in the infra-slow frequency band (0.005–0.05 Hz) was quantitatively UK) of the appropriate size was positioned to fill the craniotomy and was fixed in evaluated by calculating the probability density of a signal amplitude as a function place using dental cement (Superbond C&B, Prestige Dental). The animals received of its phase with respect to a second signal. In order to estimate the phase between a subcutaneous injection of 0.9% saline (1 ml) and were placed in a recovery cage the two signals the Hilbert transform of each of them was computed using the until the anesthetic had worn off. python library scipy and the difference between their phase calculated. A two- dimensional histogram was then used to express the probability density of the signal amplitude in the amplitude-phase space (Fig. 3c). Behavioral testing for biocompatibility evaluation. All animals had pre-surgical behavioral baselines taken at the age of 5 weeks. One week later, all animals were assigned to one of five groups; graphene electrode, platinum electrode, blank Ethical approval and animal handling. The experiments in-vivo were in accor- electrode, sham surgery (no electrode implanted), or naive (no surgery). Animals dance with the European Union guidelines on protection of vertebrates used for were then tested at one or two timepoints—2 weeks, 2 and 6 weeks, or 6 and experimentation (Directive 2010/63/EU of the European Parliament and of the 12 weeks post-surgery. Timepoints were chosen in line with ISO 10993 definitions; Council of 22 September 2010). Electrophysiological experiments with Long Evans where prolonged exposure is classified >24 h but <30 days, and permanent expo- rats were carried out under the German Law for Protection of Animals (TierSchG) sure is defined as >30 days (ISO 10993-6:2007). and were approved by the local authorities (ROB-55.2-2532.Vet_02-16-170). Prior to the first exposure to NOR rats were placed into the empty arenas the Experimental procedures using Sprague Dawley rats for biocompatibility assess- day before testing for 20 min with their cage mates for acclimatization purposes. ment were carried out according to the United Kingdom Animals (Scientific The square Plexiglas boxes (measuring 52 cm by 52 cm at the base with a height of Procedures) Act, 1986 and approved by the Home Office and the local Animal 30 cm) had a white floor and black walls. Animals were acclimatized to the NOR Ethical Review Group, University of Manchester. Rats were kept under standard arena before experiments began by placing them into the arena for 3 min while conditions (room temperature 22 ± 2 °C, 12:12 h light–dark cycle, lights on at there were no objects within the arena. The NOR test comprises a training and 10:00), with food and water available ad libitum. testing trial, separated by an inter-trial period. In the training session, two identical objects were placed within the arena, such as two bottles of the same shape and Implantation of the graphene-sensor arrays for electrophysiological mea- size. In the testing session, two new objects were placed in the arena, one object surements. As described previously in Garcia-Cortadella et. al.34, an adult Long identical to that during the training session, and one completely new object, such as Evans rat, weighing 580 g, was anaesthetized with MMF (Midazolam 2 mg/kg, a can. For training sessions, animals were placed in the arena and allowed to Medetomidin 0.15 mg/kg, Fentanyl 0.005 mg/kg). 1 h after, MMF induction Iso- explore for 3 min, before returning to the home cage. Animals were left in the flurane was supplemented at 1% to maintain the rat anaesthetized and Metamizol home cage for 30 min, before being placed into the arena for the testing session, was given at 110 mg/kg. The posterior-dorsal area of the head was shaved, the skin again remaining in the arena to freely explore for 3 min. The time the animal spent locally disinfected with Povidone-iodine and subcutaneously infiltrated with local interacting with the objects was measured in both the training and testing trials. In anesthetic Bupivacaine. The skin was then incised and the dorsal skull cleaned a healthy animal, the animal should spend more time interacting with the novel carefully by blunt dissection. The dried skull was covered with UV-curing adhesive object during the testing session. Ideally these tests should not be used >3 times for Optibond (Kerr) and a 3D-printed base ring was anchored to skull with screws and any animal, and therefore animals in the 12 week implantation group were tested Metabond cement (Parkell). for NOR at baseline and then at 6 and 12 weeks post implantation. All other Symmetric craniotomies with a maximum width of 5 mm were performed animals were tested at every selected timepoint available before sacrifice. bilaterally, extending between +2 mm and −8 mm with respect to Bregma in the Videos of NOR test trials were manually scored by blinded researchers using an anterior-posterior axis. The dura mater was incised and removed within these online stopwatch (http://jackrrivers.com/program/). Animals were classed as craniotomies. A further craniotomy of 1 × 1 mm was performed over the interacting with an object if their nose or paws touched the object. The amount of cerebellum. All craniotomies were covered with prepolymerized time spent interacting with the two objects was analyzed, and a discrimination ratio polydimethylsiloxane (PDMS) (Sylgard 184, Dow Corning, USA) with mixing ratio was determined by dividing time spent interacting with the novel object with time 1:10 and sealed with Vetbond (Animal Care Products, USA). The skin margins spent interacting with the known object. A discrimination ratio >0.5 indicated an around the implant were sutured and the implant closed with a protective cap. animal had a preference for the novel object, a sign of normal cognition. After 1 week of recovery the g-SGFET array was implanted under Isofluran Anesthesia (5% induction 1% maintenance). After partial opening and sidewards flapping of the polymer covering the right hemisphere the array was placed onto Tissue collection and processing. At 2 weeks, 6 weeks or 12 weeks post the pial surface positioned such as to cover the posterior aspect of the right implantation of devices, animals were culled using an appropriate method for the hemisphere (ca −7 to −2 mm from bregma). In addition, two Pt-Ir wires were type of subsequent analysis. For histology, animals underwent perfusion fixation implanted at either side of the g-SGFET array. One proximal to the array, the other using heparinised saline, followed by 4% paraformaldehyde (PFA; Sigma-Aldrich, distally on the opposing hemisphere. The polymer cover was flapped back into UK; 441244) in phosphate-buffered saline (Sigma-Aldrich, UK; D8537). Tissue was position with the flexible cable of the g-SGFET array leaving through the remaining stored a minimum of 24 h in PFA, transferred to a sucrose solution for 48 h, and slit. A second PDMS cover was used to cover both the incised polymer and array, frozen before cryosectioning 40 sections at 25 µm per animal. Cryosections were anchored to the skull with Vetbond and Evoflow (Ivoclar Vivadent, Liechtenstein) stained for one of three markers: (i) ionized calcium binding adaptor molecule 1 and sealed with silicon gel 3-4680 (Dow Corning, USA). Finally, an Ag/AgCl (Iba1) to quantify microglial population, (ii) terminal deoxynucleotidyl transferase electrode was placed in contact with the cerebellum as a reference for the recording dUTP nick-end labeling (TUNEL) staining to assess apoptosis, or (iii) haemotixylin of neural activity. and eosin (H&E) to assess gross morphology of the brain tissue. Tissue sections were blocked in 5% goat serum in PBS with 0.1% triton-X, before incubation with Iba1 primary antibody (1:200, 019-19741, Wako) overnight. A goat anti-rabbit Implantation of the graphene, platinum, and PI devices for biocompatibility Alexa 594 secondary antibody (1:1000, A11012, Invitrogen) was used for visuali- evaluation. Sprague Dawley rats (200–280 g) were anesthetized using isoflurane zation. For immunofluorescently stained tissue, a DAPI counterstain was per- inhalation (typically at 3.5% for induction and between 1.5 and 2.5% for main- formed before slides were mounted using ProLong Gold mountant (P10144, tenance) in 100% oxygen. The top of the animal’s head was shaved and the animal ThermoFisher). was positioned within a stereotactic frame with tooth and ear bar fixation. Animals For TUNEL staining, the manufacturer’s instructions were followed using a were placed on a heated blanket, with a pulse oximeter attached to the foot and a DeadEnd™ Colorimetric TUNEL System (G7360, Promega). Following rectal probe inserted to monitor body temperature. Viscotears liquid gel (Bausch & diaminobenzidine (DAB) visualization of the TUNEL staining, slides were Lomb, UK) was applied to the eyes for protection during the procedure. Depth of counterstained using methyl green (0.1% w/v aqueous solution, Alfa Aesar). H&E anesthesia was confirmed and maintained throughout the surgery via absence of a staining was performed as standard, using a 1-min haemotoxylin staining time, pedal reflex. All experimental animals received a subcutaneous injection of followed by an acetic acid rinse, and a 30-s eosin staining time. Slides for both TUNEL buprenorphine (0.03 mg/kg). The head was swabbed with iodine and a large flap of and H&E staining were mounted using DPX mountant (06522, Sigma-Aldrich). NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications 13 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w Slides were imaged using the 3D Histec Pannoramic250 slide scanner, and images was calculated and subsequently z-scored. Periods with z-score values exceeding 0.7 analyzed using CaseViewer (Version 2.2, 3DHistech Ltd). TUNEL-positive cells were were labeled HVS. Candidate HVS periods shorter than 1 s were skipped to counted and averaged across the cortical surface in forty 25 µm sections per minimize false positives from occasional sharp single wave transients of hemisphere. Microglial cells were individually classified into one of four undetermined physiological nature. Spectral profiles of the individual HVS events morphologies; Grade 0 (resting/ramified), Grade 1 (de-ramifying/re-ramifying), were post-hoc classified to remove artifact contamination. Onset and offset of peak Grade 2 (activated/ameboid) or Grade 3 (clustered & activated) as previously power, instantaneous frequency and power of the first spectral peak were extracted described88. Activation was determined as a percentage of total microglial cells, which from each event. were either Grade 3 or 4. Theta states were defined based on the ratio between power in the theta (5–9.5 For ELISA, animals were culled by rising concentration of CO2, before cardiac Hz) and delta (2–4 Hz) ranges on the theta reference channel. Gaps shorter than puncture was performed to extract blood. Brain tissue was extracted, snap frozen in 10 s were concatenated with neighboring Theta periods. Theta coinciding with an liquid nitrogen, and stored at −80 °C until further use. The extracted blood was Inactive motor state was labeled Inactive Theta. All Inactive Theta periods that collected in a blood collection tube (Vacutainer, Becton Dickson, UK) and allowed were preceded by a SW state within 1 s and which were longer than 5 seconds were to clot at room temperature for 15–30 min. The tube was centrifuged at 5000RPM considered REM sleep. All remaining Theta periods were considered Awake Theta. for 10 min at 4 °C, and the resulting serum supernatant was collected and stored at Finally, all periods that were neither SWS nor REM nor Awake Theta were defined −80 °C until further use. There were insufficient serum samples to be run from the as Awake Nontheta. It should be noted that the existence of low amplitude naive control group, so this group was excluded from analysis. Brain tissue was microstates during NREM sleep has been described previously89. It remains to be lysed by addition of liquid nitrogen and grinding the tissue to create a powder, to determined to which degree a subset of periods with low amplitude in the SW which NP-40 lysis buffer (150 mM NaCl, 50 mM Tric-Cl, 1% Nonidet range assigned to Awake-Nontheta in this study maps onto states defined as low P40 substitute, Fluka, pH adjusted to 7.4) containing protease and phosphatase activity sleep microstates by Miyawaki and colleagues89. inhibitor (Halt™ Protease and Phosphatase Inhibitor Cocktail, ThermoFisher All states were considered to be mutually exclusive with the exception of HVS, Scientific) was added to the tissue. Samples were centrifuged at 5000RPM for which was considered an event occurring during but not interrupting ongoing 10 min, and the supernatant stored at −80 °C until further use. ELISA kits for four background states. Therefore, for each HVS episode the brain states immediately cytokines were used, IL-17a (437904, Biolegend), IFN-γ (439007, Biolegend), TNF- preceding and following were merged if they belonged to the same state. All spectral α (438204, Biolegend), and IL-6 (437107, Biolegend). Manufacturer instructions analysis was performed using custom-developed Matlab implementation of were followed for all four kits. multitaper estimate90. Analysis of the g-SGFET performance was done in Python. ISA brain dynamics analysis was limited to immobility and REM sleep, for which potential influence of motion artifacts on the infra-slow fluctuations could be discarded. Motion capture and behavioral states classification. The rat was recorded for up to 24 h during spontaneous behavior in a recording arena of 100 × 100 cm to which it was prehabituated. During the recording session it had ad libitum access to food and Statistical analysis. For the evaluation of the yield and homogeneity of the g- water. The battery of the wireless system was exchanged once every 6 h. A motion SGFETs performance, 9 neural probes with 64 g-SGFETs each were characterized capture (Mocap) system (Optitrack) using passive reflective markers and 8 cameras in vitro. This data is plotted in the boxplots in Fig. 2, the boxes extend from the was used to track the motion of the animal head in three-dimensional space. Four lower to the upper quartiles, with a line at the median. The whiskers extend 1.5 reflective markers were anchored to the protective cap and their position averaged to times the inter-quartile range and the data points beyond the whiskers are indi- infer the position in 3D and orientation of the head. Instantaneous headspeed in 3D cated by a dot. The longitudinal evaluation of g-SGFETs stability in-vivo was was computed as time derivative of the modulus of the spatial coordinates. Motive performed with one 64-channel array implanted on the cortex of a rat. The box- 2.2 software was used for the analysis of Mocap data. plots shown in Fig. 3 are defined as those in Fig. 2. In panels a and e of Fig. 3, the For the classification of the behavior as active or inactive, periods where statistical sample are all 8 g-SGFET connected to the DC-coupled channels of the headspeed exceeded 100 mm/s were labeled as active. In a second step active headstage. Panels d and f of Fig. 3 correspond to all 64 g-SGFETs on the array. periods shorter than 5 seconds were skipped while gaps in active periods shorter Finally, panel g shows the normalized response of the 10 g-SGFETs on the posi- than 5 s were concatenated with the neighboring active state. Timepoints, that did tions under the highest induced electric field during bipolar stimulation. not fall under this definition of active, were labeled as inactive. The ratio of active For biocompatibility assessment, three device types; platinum, graphene, and and inactive periods varied substantially across total recording period (between polyimide (blank) were fabricated. The rats in each of these groups were implanted 6.6% and 88.0% active per hour, mean 40.2% ± 4.7% active per hour). with one of the three device types on the parietal cortex of the brain. A fourth Rear events were defined as short elevations of the head to heights that group of animals had the full surgery without the implantation of any device (sham necessitate the animal to stand on its hindlimbs. After visual inspection of z control) and a fifth group (naive) had no surgery. For NOR testing, the number of elevations throughout the recordings a threshold of 200 mm elevation from ground rats used was n = 7 for all groups at all timepoints, except 12 weeks, which had n = was determined to effectively separate punctuate rearing onsets/offsets from 3–7 depending on the group. For cytokine detection in the brain tissue, the number ongoing height variations at lower z positions (see supplementary of rats was n = 4 at 2 and 6 weeks, and n = 3 at 12 weeks post implantation. For information S10). While active and inactive states were defined as mutually microglial activation, the number of rats was n = 3 at 2 and 12 weeks and n = 2 at exclusive, rear events were considered to be a substrate of the active motor state. 6 weeks (or 3 for the contralateral hemisphere). In all cases, the contralateral hemisphere was also evaluated as a control. Data where n = 3 or higher were Neural signals processing and analysis. The wireless headstage was controlled analyzed using a two-way ANOVA to compare all timepoints and interventions. using Multi Channel Experimenter 2.12.1 software and the recorded data was Dunnett’s multiple comparisons were then performed at each timepoint comparing converted to HDF5 format using Multi Channel Data Manager 1.13.1. Neural data each surgical intervention to the control. *p < 0.05, **p < 0.005, ***p < 0.001, was calibrated according to the transconductance of g-SGFETs using Python ****p < 0.0001 are indicated for each surgical intervention in Fig. 4. 2.7 scripts and exported to Neuroscope software for data exploration (see data and The measurement of gamma activity modulation during rearing was evaluated code availability statement). Analysis of neural signals was carried out using Matlab for 163 rearing events measured in the course of a 24 h recording in one rat. HVS 2016b scripts (see code availability statement). were also measured during the same 24 h recording. Five-hundred sixty-six events were detected during immobility and 92 events during REM sleep. Differences in spectral content of HVS during distinct states (IMM and REM) was evaluated by a Brain states classification. Two channels of the epicortical array were chosen for Wilcoxon ranksum test (the p-values are indicated in the main text). The the separation of brain states. One channel was selected from the posterior area of electrophysiological data shown corresponds to one rat. The large statistical sample the array, putatively overlying the hippocampal formation and exhibiting promi- of events allows illustrating the capabilities of the graphene-based technology, nent theta oscillations. A second channel was selected from the frontal area, however, the interpretation of these results should be subject to variability across recording from the region and neighborhood of somatosensory cortex, where high- measurements and animals. voltage spindles are expressed most prominently. Power spectra were computed on The modulation of ISA power by REM state vs. SWS was evaluated in two ways: whitened LFP signals in the range from 1 to 200 Hz using multitaper methods first, for the period directly around the state transition (−30 to 30 s around the using 4-s windows in sliding 0.5 s steps. First, slow wave (SW) states were identified REM onset). Second, we evaluated the ISA power in both states over their entire as periods were the z-score of the summed power of delta (1–4 Hz), alpha-beta duration, not only in the SWS-REM transition. ISA power comparison between ranges (10–25 Hz) exceeded −0.1. Gaps shorter than 5 s were concatenated with REM vs. SWS states was restricted to the 44 REM episodes lasting more than 40 s neighboring periods. (see Supplementary Information S12B). To test for frequency-specific changes SW states coinciding with inactive motor state and >10 s temporal distance across the state transition we compared the distributions of median spectral power from the last preceding high-voltage spindle (HVS) event were defined as slow across trials for the 30 s pre vs. post SWS-REM transition for each frequency bin. wave sleep (SWS). Following previous literature we assumed incompatibility Significance was assessed by permutation test for each frequency bin (n = 1000 between HVS and deep slow wave sleep in unanaesthetized Long Evans rats84. HVS permutations, see Fig. 5d example channel and Supplementary Information S12C are close in frequency to theta oscillations, but differ markedly in their expression for all working DC channels). Increase of ISA and concurrent decrease of 1–4 Hz of multiple higher harmonics due to their strongly non-sinusoidal waveshape. power during SWS-REM transition is significant after permutation test (n = 1000 Owing to this fact spectral power in the frequency band from 20 to 50 Hz on the permutations) on all DC channels except one excluded channel, same as for HVS reference channel was used to selectively detect HVS episodes and distinguish longitudinal evaluation in Fig. 3a, e, due to poor signal to noise (Supplementary them from theta activity. The mean of multitaper power spectra from 20 to 50 Hz Information S12C). Second, we tested for statistical differences between 14 NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w ARTICLE distributions of the integrated power in the ISA (0.01–0.1 Hz) band in SWS and 15. Hébert, C. et al. Flexible graphene solution-gated field-effect transistors: REM states using Wilcoxon rank sum test (see Fig. 5e, test results for each channel efficient transducers for micro-electrocorticography. Adv. Funct. Mater. in Supplementary Information S12D). 1703976. https://doi.org/10.1002/adfm.201703976 (2017). Modulation of the LFP power by the DC signal-derived ISA and LFP phase was 16. Jastrzebska-Perfect, P. et al. Translational neuroelectronics. Adv. Funct. Mater. quantified using instantaneous fast frequency power-weighted resultant length of https://doi.org/10.1002/adfm.201909165 (2020) the instantaneous slow frequency phase vectors normalized by mean LFP power in 17. Cea, C. et al. Enhancement-mode ion-based transistor as a comprehensive the respective band72 for which magnitude reflects the strength of LFP power interface and real-time processing unit for in vivo electrophysiology. Nat. modulation to a preferred phase of the ISA or LFP. ISA and LFP phase and LFP Mater. 19, 679–686 (2020). power were computed as angle and absolute value of the analytical signal of the 18. Khodagholy, D. et al. In vivo recordings of brain activity using organic respective AC and DC channel signals that were band-pass filtered, with 0.04 and transistors. Nat. Commun. 4, 1575 (2013). 0.4 Hz bandwidth, respectively. Significance of the modulation was tested based on 19. Rivnay, J. et al. Organic electrochemical transistors. Nat. Rev. Mater. 3, 1–14 1000 surrogate phase-power pairs randomly shifted with respect to each other by (2018). up to 100 s. Resulting empirical p-value was corrected following false-discovery- 20. Dankerl, M. et al. Diamond transistor array for extracellular recording from rate control procedure at the error rate of 0.001. For constructing topographic electrogenic cells. Adv. Funct. Mater. 19, 2915–2923 (2009). maps of theta and spindle power band modulation we used mean modulation 21. Fromherz, P., Offenhäusser, A., Vetter, T. & Weis, J. A neuron-silicon strength for the LFP power band 8–9 Hz (theta) and 10–14 Hz (spindle band), and junction: a Retzius cell of the leech on an insulated-gate field-effect transistor. the ISA phase frequency of 0.05-0.1 Hz computed for every AC channel (LFP Science (80-.). 252, 1290–1293 (1991). power) and one fronto-medial DC channel (ISA phase). 22. Spyropoulos, G. D., Gelinas, J. N. & Khodagholy, D. Internal ion-gated organic electrochemical transistor: a building block for integrated Reporting summary. Further information on research design is available in the Nature bioelectronics. Sci. Adv. 5, eaau7378 (2020). Research Reporting Summary linked to this article. 23. Lee, C., Wei, X., Kysar, J. W. & Hone, J. Measurement of the elastic properties and intrinsic strength of monolayer graphene. Science (80-.). 321, 385–388 (2008). Data availability 24. Zandiatashbar, A. et al. Effect of defects on the intrinsic strength and stiffness Device characterization and raw electrophysiological data examples are available in GIN of graphene. Nat. Commun. 5, 1–9 (2014). repository with the identifier (https://doi.org/10.12751/g-node.4zw2lt). The complete 25. Banhart, F., Kotakoski, J. & Krasheninnikov, A. V. Structural defects in electrophysiological dataset is available from the corresponding author upon reasonable graphene. ACS Nano 5, 26–41 (2011). request. Source data are provided with this paper. 26. Bendali, A. et al. Purified neurons can survive on peptide-free graphene layers. Adv. Healthc. Mater. 2, 929–933 (2013). Code availability 27. Kostarelos, K., Vincent, M., Hebert, C. & Garrido, J.A. Graphene in the Design Custom code for the analysis of devices characterization is provided in GIN repository and Engineering of Next-Generation Neural Interfaces. Adv. Mater. 29, 42 with the identifier (https://doi.org/10.12751/g-node.4zw2lt). The complete code for (2017). analysis of electrophysiological data is available from the corresponding author upon 28. Balasubramanian, K. et al. Reversible defect engineering in graphene grain request. boundaries. Nat. Commun. 10, 1090 (2019). 29. Banszerus, L. et al. Ultrahigh-mobility graphene devices from chemical vapor deposition on reusable copper. Sci. Adv. 1, e1500222 (2015). Received: 22 April 2020; Accepted: 8 December 2020; 30. Mitra, A. et al. Spontaneous infra-slow brain activity has unique spatiotemporal dynamics and laminar structure. Neuron 98, 297–305.e6 (2018). 31. Pan, W.-J., Thompson, G. J., Magnuson, M. E., Jaeger, D. & Keilholz, S. Infraslow LFP correlates to resting-state fMRI BOLD signals. Neuroimage 74, 288 (2013). References 32. Krishnan, G. P., González, O. C. & Bazhenov, M. Origin of slow spontaneous 1. Khodagholy, D. et al. NeuroGrid: Recording action potentials from the surface resting-state neuronal fluctuations in brain networks. Proc. Natl Acad. Sci. of the brain. Nat. Neurosci. 18, 310–315 (2015). USA. https://doi.org/10.1073/pnas.1715841115 (2018) 2. Lee, W. et al. Integration of organic electrochemical and field-effect transistors 33. Masvidal-Codina, E. et al. High-resolution mapping of infraslow cortical brain for ultraflexible, high temporal resolution electrophysiology arrays. Adv. activity enabled by graphene microtransistors. Nat. Mater. 18, 280–288 (2019). Mater. 28, 9722–9728 (2016). 34. Garcia-Cortadella, R. et al. Distortion-free sensing of neural activity using 3. Khodagholy, D. et al. Organic Electronics for High-resolution graphene transistors. Small 1906640. https://doi.org/10.1002/smll.201906640 Electrocorticography of the Human Brain. Sci. Adv. 2, 11 (2016). (2020) 4. Chiang, C.-H. et al. Development of a neural interface for high-definition, 35. Hiltunen, T. et al. Infra-Slow EEG fluctuations are correlated with resting- long-term recording in rodents and nonhuman primates. Sci. Transl. Med. 12, state network dynamics in fMRI. J. Neurosci. 34, 356–362 (2014). eaay4682 (2020). 36. Mitra, A., Snyder, A. Z., Tagliazucchi, E., Laufs, H. & Raichle, M. E. 5. Chung, J. E. et al. High-Density, long-lasting, and multi-region Propagated infra-slow intrinsic brain activity reorganizes across wake and electrophysiological recordings using polymer electrode arrays. Neuron 101 slow wave sleep. Elife 4, e10781 (2015). (21-31), e5 (2019). 37. Grooms, J. K. et al. Infraslow electroencephalographic and dynamic resting 6. Garcia-Cortadella, R. et al. Switchless multiplexing of graphene active sensor state network activity. Brain Connect. 7, 265–280 (2017). arrays for brain mapping. Nano Lett. acs.nanolett.0c00467. https://doi.org/ 38. Kelly, A. M. C., Uddin, L. Q., Biswal, B. B., Castellanos, F. X. & Milham, M. P. 10.1021/acs.nanolett.0c00467 (2020) Competition between functional brain networks mediates behavioral 7. Schaefer, N. et al. Multiplexed neural sensor array of graphene solution-gated variability. Neuroimage 39, 527–537 (2008). field-effect transistors. 2D Mater. 7, 025046 (2020). 39. Vanhatalo, S. et al. Infraslow oscillations modulate excitability and interictal 8. Viventi, J. et al. Flexible, foldable, actively multiplexed, high-density electrode epileptic activity in the human cortex during sleep. Proc. Natl Acad. Sci. 101, array for mapping brain activity in vivo. Nat. Neurosci. 14, 1599–1605 (2011). 5053–5057 (2004). 9. Fang, H. et al. Capacitively coupled arrays of multiplexed flexible silicon 40. Watson, B. O., Hengen, K. B., Gonzalez Andino, S. L. & Thompson, G. J. transistors for long-term cardiac electrophysiology. Nat. Biomed. Eng. 1, 0038 Cognitive and physiologic impacts of the infraslow oscillation. Front. Syst. (2017). Neurosci. 12, 44 (2018). 10. Chang, E. F. Towards large-scale, human-based, mesoscopic 41. Vanhatalo, S., Voipio, J. & Kaila, K. Full-band EEG (FbEEG): an emerging neurotechnologies. Neuron 86, 68–78 (2015). standard in electroencephalography. Clin. Neurophysiol. 116, 1–8 (2005). 11. Lacour, S. P., Courtine, G. & Guck, J. Materials and technologies for soft 42. Kovac, S., Speckmann, E.-J. & Gorji, A. Uncensored EEG: the role of DC implantable neuroprostheses. Nat. Rev. Mater. 1, 1–14 (2016). potentials in neurobiology of the brain. Prog. Neurobiol. 165–167, 51–65 12. Nguyen, J. K. et al. Mechanically-compliant intracortical implants reduce the (2018). neuroinflammatory response. J. Neural Eng. 11, 056014 (2014). 43. Nelson, M. J., Pouget, P., Nilsen, E. A., Patten, C. D. & Schall, J. D. Review of 13. Moshayedi, P. et al. The relationship between glial cell mechanosensitivity and signal distortion through metal microelectrode recording circuits and filters. J. foreign body reactions in the central nervous system. Biomaterials 35, Neurosci. Methods 169, 141–157 (2008). 3919–3925 (2014). 44. Chan, A. W., Mohajerani, M. H., LeDue, J. M., Wang, Y. T. & Murphy, T. H. 14. Song, E., Li, J., Won, S. M., Bai, W. & Rogers, J. A. Materials for flexible Mesoscale infraslow spontaneous membrane potential fluctuations bioelectronic systems as chronic neural interfaces. Nat. Mater. 19, 590–603 recapitulate high-frequency activity cortical motifs. Nat. Commun. 6, 7738 (2020). (2015). NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications 15 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w 45. Leopold, D. A. Very slow activity fluctuations in monkey visual cortex: 77. Sirota, A. et al. Entrainment of neocortical neurons and gamma oscillations by implications for functional brain imaging. Cereb. Cortex 13, 422–433 (2003). the hippocampal theta rhythm. Neuron 60, 683–697 (2008). 46. Mackin, C. et al. A current–voltage model for graphene electrolyte-gated field- 78. Kandel, A. & Buzséki, G. Cellular-synaptic generation of sleep spindles, spike- effect transistors. IEEE Trans. Electron Devices 61, 3971–3977 (2014). and-wave discharges, and evoked thalamocortical responses in the neocortex 47. Schaefer, N. et al. Improved metal-graphene contacts for low-noise, high- of the rat. J. Neurosci. 17, 6783–6797 (1997). density microtransistor arrays for neural sensing. Carbon N. Y. 161, 647–655 79. Kandel, A. & Buzsáki, G. Cerebellar neuronal activity correlates with spike and (2020). wave EEG patterns in the rat. Epilepsy Res. 16, 1–9 (1993). 48. Rivnay, J. et al. High-performance transistors for bioelectronics through 80. Voipio, J., Tallgren, P., Heinonen, E., Vanhatalo, S. & Kaila, K. Millivolt-scale tuning of channel thickness. Sci. Adv. 1, e1400251 (2015). DC shifts in the human scalp EEG: evidence for a nonneuronal generator. J. 49. Masvidal-Codina, E. et al. High-resolution mapping of infraslow cortical brain Neurophysiol. 89, 2208–2214 (2003). activity enabled by graphene microtransistors. Nat. Mater. 18, 280–288 (2019). 81. Pan, W. J., Thompson, G. J., Magnuson, M. E., Jaeger, D. & Keilholz, S. 50. Zurutuza, A. & Marinelli, C. Challenges and opportunities in graphene Infraslow LFP correlates to resting-state fMRI BOLD signals. Neuroimage 74, commercialization. Nat. Nanotechnol. 9, 730–734 (2014). 288–297 (2013). 51. Deng, B., Liu, Z. & Peng, H. Toward mass production of CVD graphene films. 82. Grooms, J. K. et al. Infraslow electroencephalographic and dynamic resting Adv. Mater. 31, 1800996 (2019). state network activity. Brain Connect. 7, 265–280 (2017). 52. Khodagholy, D. et al. High transconductance organic electrochemical 83. Lazic, K., Ciric, J. & Saponjic, J. Sleep spindle dynamics during NREM and transistors. Nat. Commun. 4, 1–6 (2013). REM sleep following distinct general anaesthesia in control rats and in a rat 53. Mavredakis, N., Garcia Cortadella, R., Bonaccini Calia, A., Garrido, J. A. & model of Parkinson’s disease cholinopathy. J. Sleep Res. 28, e12758 (2019). Jiménez, D. Understanding the bias dependence of low frequency noise in 84. Kozák, G., Földi, T. & Berényi, A. Spike-and-wave discharges are not single layer graphene FETs. Nanoscale 10, 14947–14956 (2018). pathological sleep spindles, network-level aspects of age-dependent absence 54. West, B. J. & Shlesinger, M. The noise in natural phenomena. Am. Sci. 78, seizure development in rats. eNeuro 7, ENEURO.0253-19.2019 (2020). 40–45 (1990). 85. Rubehn, B. & Stieglitz, T. In vitro evaluation of the long-term stability of 55. Neto, A. H. C., Guinea, F., Peres, N. M. R., Novoselov, K. S. & Geim, A. K. The polyimide as a material for neural implants. Biomaterials 31, 3449–3458 (2010). electronic properties of graphene. Rev. Mod. Phys. https://doi.org/10.1103/ 86. Lu, H. et al. SU8-based micro neural probe for enhanced chronic in-vivo RevModPhys.81.109 (2007) recording of spike signals from regenerated axons. In Sensors, 2006 IEEE, 56. Pinto, H. & Markevich, A. Electronic and electrochemical doping of graphene 66–69 (IEEE, 2006). by surface adsorbates. Beilstein J. Nanotechnol. 5, 1842–1848 (2014). 87. Márton, G. et al. A multimodal, SU-8-platinum-polyimide microelectrode 57. Fu, W. et al. Biosensing near the neutrality point of graphene. Sci. Adv. 3, array for chronic in vivo neurophysiology. PLoS ONE 10, e0145307 (2015). e1701247 (2017). 88. Al-Ahmady, Z. S. et al. Selective liposomal transport through blood brain 58. Prodanov, D. & Delbeke, J. Mechanical and biological interactions of implants barrier disruption in ischemic stroke reveals two distinct therapeutic with the brain and their impact on implant design. Front. Neurosci. 10, 11 opportunities. ACS Nano 13, 12470–12486 (2019). (2016). 89. Miyawaki, H., Billeh, Y. N. & Diba, K. Low activity microstates during sleep. 59. Prasad, A. & Sanchez, J. C. Quantifying long-term microelectrode array Sleep 40, zsx066 (2017). functionality using chronic in vivo impedance testing. J. Neural Eng. 9, 026028 90. Mitra, P. P. & Pesaran, B. Analysis of dynamic brain imaging data. Biophys. J. (2012). 76, 691–708 (1999). 60. Degenhart, A. D. et al. Histological evaluation of a chronically-implanted electrocorticographic electrode grid in a non-human primate. J. Neural Eng. 13, 046019 (2016). Acknowledgements 61. Sun, F. T., Arcot Desai, S., Tcheng, T. K. & Morrell, M. J. Changes in the This work has been funded by the European Union’s Horizon 2020 research and electrocorticogram after implantation of intracranial electrodes in humans: innovation program under Grant Agreement No. 732032 (BrainCom) and Grant The implant effect. Clin. Neurophysiol. 129, 676–686 (2018). Agreement No. 696656 and 785219 (Graphene Flagship). The ICN2 is supported by the 62. Salatino, J. W., Ludwig, K. A., Kozai, T. D. Y. & Purcell, E. K. Glial responses Severo Ochoa Centres of Excellence program, funded by the Spanish Research Agency to implanted electrodes in the brain. Nat. Biomed. Eng. 1, 862–877 (2017). (AEI, grant no. SEV-2017-0706), and by the CERCA Program/Generalitat de Catalunya. 63. Hassler, C., Boretius, T. & Stieglitz, T. Polymers for neural implants. J. Polym. R.G.C. is supported by the International Ph.D Program La Caixa-Severo Ochoa (Pro- Sci. Part B Polym. Phys. 49, 18–33 (2011). grama Internacional de Becas “la Caixa”-Severo Ochoa). This work has made use of the 64. Mendes, G. C. C., Brandão, T. R. S. & Silva, C. L. M. Ethylene oxide Spanish ICTS Network MICRONANOFABS partially supported by MICINN and the sterilization of medical devices: A review. Am. J. Infect. Control 35, 574–581 ICTS “NANBIOSIS”, more specifically by the Micro-NanoTechnology Unit of the CIBER (2007). in Bioengineering, Biomaterials, and Nanomedicine (CIBER-BBN) at the IMB-CNM. 65. Huang, C., Irwin, M. G., Wong, G. T. C. & Chang, R. C. C. Evidence of the This work is within the project FIS2017-85787-R funded by the “Ministerio de Ciencia, impact of systemic inflammation on neuroinflammation from a non-bacterial Innovación y Universidades” of Spain, the “Agencia Estatal de Investigación (AEI)”, and endotoxin animal model. J. Neuroinflammation 15, 147 (2018). the “Fondo Europeo de Desarrollo Regional (FEDER/UE)”. A.S. and G.S. were also 66. Stephenson, R., Lim, J., Famina, S., Caron, A. M. & Dowse, H. B. Sleep-wake supported by Bundesministerium fü r Bildung und Forschung [grant number behavior in the rat: Ultradian rhythms in a light-dark cycle and continuous 01GQ0440]. R.G.C. acknowledges that this work has been done in the framework of the bright light. J. Biol. Rhythms 27, 490–501 (2012). Ph.D in Electrical and Telecommunication Engineering at the Universitat Autònoma de 67. Vyazovskiy, V. V., Achermann, P. & Tobler, I. Sleep homeostasis in the rat in Barcelona. We thank Eduardo Blanco Hernández for assistance with the preprocessing of the light and dark period. Brain Res. Bull. 74, 37–44 (2007). the motion tracking data. 68. Trachsel, L., Tobler, I. & Borbely, A. A. Sleep regulation in rats: effects of sleep deprivation, light, and circadian phase. Am. J. Physiol. -Regul. Integr. Comp. Author contributions Physiol. 251, R1037–R1044 (1986). R.G.C. fabricated the graphene sensor arrays and performed their characterization 69. Lecci, S. et al. Coordinated infraslow neural and cardiac oscillations mark in vitro, he characterized the headstage and contributed to the recording and analysis of fragility and offline periods in mammalian sleep. Sci. Adv. 3, e1602026 (2017). electrophysiological activity. G.S. developed the implantation methodology and did the 70. Thompson, G. J. et al. Phase-amplitude coupling and infraslow (<1 Hz) surgery. He also contributed to electrophysiological signals recording and analysis. C.J. frequencies in the rat brain: relationship to resting state fMRI. Front. Integr. designed the wireless headstage and software for the operation of active sensor arrays. X. Neurosci. 8, 41 (2014). I. contributed to the design and fabrication of the neural probes. A.G., S.S., I.S., and E.S. 71. Sirota, A. et al. Entrainment of neocortical neurons and gamma oscillations by performed the experiments for the biocompatibility study. E.M. fabricated and tested the the hippocampal theta rhythm. Neuron 60, 683–697 (2008). devices for the biocompatibility study. K.K., S.S., and A.G., designed and interpreted the 72. Pesaran, B. et al. Investigating large-scale brain dynamics using field potential biocompatibility data. K.K. and S.S. overall coordinated the team for the biocompatibility recordings: analysis and interpretation. Nat. Neurosci. 21, 903–919 (2018). study. A.G.B. contributed to the design of the recording system. A.S. coordinated the 73. Sirota, A. & Buzsáki, G. Interaction between neocortical and hippocampal experiments for electrophysiological data-acquisition and contributed to the analysis of networks via slow oscillations. Thalamus Relat. Syst. 3, 245–259 (2007). electrophysiological signals. J.A.G. coordinated the team in the ICN2 for the fabrication 74. Sturman, O., Germain, P. L. & Bohacek, J. Exploratory rearing: a context- and of the sensor arrays. R.G.C., G.S., S.S., A.S., and J.A.G. wrote the manuscript. stress-sensitive behavior recorded in the open-field test. Stress 21, 443–452 (2018). 75. Lever, C., Burton, S. & O’Keefe, J. Rearing on hind legs, environmental Competing interests novelty, and the hippocampal formation. Rev. Neurosci. 17, 111–133 (2006). Patent application (no. P201831068) filled by CSIC, ICREA, CIBER, ICN2, and IDI- 76. Barth, A. M., Domonkos, A., Fernandez-Ruiz, A., Freund, T. F. & Varga, V. BAPS; concerning a graphene transistor system for measuring electrophysiological sig- Hippocampal network dynamics during rearing episodes. Cell Rep. 23, nals (pending); inventors who are co-authors in the present article are A.G.B., X.I., E.M., 1706–1715 (2018). and J.A.G. 16 NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20546-w ARTICLE Additional information Open Access This article is licensed under a Creative Commons Supplementary information is available for this paper at https://doi.org/10.1038/s41467- Attribution 4.0 International License, which permits use, sharing, 020-20546-w. adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Correspondence and requests for materials should be addressed to A.S. or J.A.G. Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless Peer review information Nature Communications thanks Jonathan Viventi and other, indicated otherwise in a credit line to the material. If material is not included in the anonymous, reviewers for their contributions to the peer review of this work. Peer review article’s Creative Commons license and your intended use is not permitted by statutory reports are available. regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Reprints and permission information is available at http://www.nature.com/reprints licenses/by/4.0/. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. © The Author(s) 2021 NATURE COMMUNICATIONS | (2021)12:211 | https://doi.org/10.1038/s41467-020-20546-w | www.nature.com/naturecommunications 17