Artifact-free and high-temporal-resolution in vivo opto-electrophysiology with microLED optoelectrodes.

The combination of in vivo extracellular recording and genetic-engineering-assisted optical stimulation is a powerful tool for the study of neuronal circuits. Precise analysis of complex neural circuits requires high-density integration of multiple cellular-size light sources and recording electrodes. However, high-density integration inevitably introduces stimulation artifact. We present minimal-stimulation-artifact (miniSTAR) μLED optoelectrodes that enable effective elimination of stimulation artifact. A multi-metal-layer structure with a shielding layer effectively suppresses capacitive coupling of stimulation signals. A heavily boron-doped silicon substrate silences the photovoltaic effect induced from LED illumination. With transient stimulation pulse shaping, we reduced stimulation artifact on miniSTAR μLED optoelectrodes to below 50 μVpp, much smaller than a typical spike detection threshold, at optical stimulation of >50 mW mm–2 irradiance. We demonstrated high-temporal resolution (<1 ms) opto-electrophysiology without any artifact-induced signal quality degradation during in vivo experiments. MiniSTAR μLED optoelectrodes will facilitate functional mapping of local circuits and discoveries in the brain.

be 5.37 × 10 -19 F m -1 , and that between the n-GaN layer and one recording electrode interconnect was 2.3 × 10 -16 F m -1 . When all the interconnects were assumed to be floating at both ends, capacitive voltage coupling between the LED interconnects and the recording electrode interconnect was -48.96 dB (3.57 mV coupling for 1 V LED voltage), and the coupling between the n-GaN layer and the recording electrode interconnects was -0.06 dB (0.99 mV coupling for 1 mV n-GaN voltage). Voltage distribution inside the one-metal-layer optoelectrode shank (and the air surrounding the optoelectrode) is shown in Supplementary Figure 2c (bottom).
Electrostatic simulation of shielded LED optoelectrode structure predicted that the coupling would be greatly reduced (Supplementary Figure 2d,bottom). The reduction of coupling between LED interconnects and the recording interconnects was greater than 46 orders of magnitude (to approximately -975 dB). Coupling between the n-GaN layer and the recording interconnects was reduced by 8 orders of magnitude (to approximately -60 dB). It should be noted that the simulation expected ideal groundconnected shielding layer and floating electrodes and therefore the values seem greatly exaggerated.

Supplementary Note 3. Simulation of PV-induced voltage around LED optoelectrode shanks during optical stimulation
We built a 3D model of a LED optoelectrode shank and simulated the effect of illumination on the silicon substrate (Supplementary Figure 5a). The doping density of boron, an acceptor dopant, inside the silicon substrate and the intensity of the optical illumination were varied. We observed a series of phenomena that result in the buildup of the electrostatic potential at the substrate-electrolyte interface and in turn generation of a voltage pulse with negative polarity in the recorded signal (Supplementary Figure   5b). First, optical illumination induced electron-hole pair generation inside the silicon substrate, and optically generated carriers redistributed inside the substrate separately depending on their types. The difference between electron and hole distribution patterns gave rise to the electric field inside the substrate, and, in turn, the electrostatic potential of the substrate-electrolyte interface changed. Because the electrolyte is connected to the common reference pin of the amplifier chip which is then connected to the inverting inputs of the low-noise amplifiers in the IC, the resulting output waveform would have a negative polarity.
Supplementary Figure 5c shows calculated substrate-electrolyte interface electrostatic potential (voltage) for substrates with different doping densities under illumination with different intensities. It is worth noting that, while higher doping density resulted in lower voltage at lower irradiance, with higher irradiance, the voltage on lightly doped (typically referred to as p -) substrates became higher than that on substrates that are almost intrinsic (not doped, typically referred to as HR or FZ, especially if the silicon substrate was float-zone grown for high-purity and low doping density). As can be seen in Supplementary   Figure 5d, it was calculated that the interface voltage from a substrate with boron doping density of 5 × 10 16 cm -3 (psubstrate) can be as high as that from the substrate with boron doping density of 4 × 10 12 cm -3 (FZ substrate) under illumination with irradiance as high as 50 mW mm -2 . On the other hand, the interface voltage from the substrate with a boron doping density of 1 × 10 20 cm -3 (p + substrate) was kept relatively low even with high-intensity illumination.

Si substrates with different doping densities
Simple μLED test structures were built using a process similar to the fabrication process for μLED optoelectrodes. After formation of μLED structures, whose LED mesa dimensions are identical to those on the μLED optoelectrodes (23 × 10 m, only 15 × 10 m of which is exposed on the front side), GaN-on-Si GaN/InGaN MQW LED wafers with μLED test structures were diced into small (4 × 10 mm) pieces each of which contains nine μLEDs. The pieces were then mounted on the PCBs and connected in the same way as the actual μLED optoelectrodes are connected to the PCBs.
The electrical and optical characteristics of each LED on the μLED test structures were characterized using the setup and the procedure identical to those for characterization of the actual LED optoelectrodes, outlined in Methods.

Electrostatics simulation for calculation of mutual capacitances between interconnects
COMSOL Multiphysics (COMSOL Inc., Burlington, MA) was used for finite element analysis of mutual capacitance distribution among the metal traces (interconnects) on the shanks of the LED optoelectrode. A 2D model of the shank was built by drawing the cross-section of the optoelectronic shank, and the electrostatics physics interface was imported to calculate the mutual capacitance values of 100-m long segments of the shanks of LED optoelectrodes with and without shielding layers. Built-in material properties (dielectric constants) of air and silicon dioxide were used. Each interconnect plus the n-GaN layer was assigned either terminal (V = 0) or floating potential (Q = 0) boundary condition. The automatic terminal sweep was used for the calculation of the Maxwell capacitance matrix. Mutual capacitance values were then extracted from the matrix. For calculation of capacitive voltage coupling magnitude, all the boundaries that correspond to recording electrode interconnects were defined as terminals with floating potential, all the boundaries that correspond to LED ground (cathode) interconnects and shielding layer were defined as grounds (V = 0), and then, assuming the LED and n-GaN voltages of 1 V, the voltage values (in dB) were reported.

In vitro characterization of photovoltaic voltage induction on LED optoelectrodes and electrode arrays on non-Si substrates
Identical to the characterization of LED-drive-induced stimulation artifact, characterization of photovoltaic voltage induction was conducted in 1 × PBS solution (prepared using 10 × PSB purchased from MP Biomedicals, Solon, OH). A fiber-optic cannula (CFMXD10, Thorlabs, Newton, NJ) was attached to a clear plastic container (Container Store, Coppell, TX) through a hole drilled on a side of the container using a 3D printed frame and glue. PBS was poured into the container until the exposed optical fiber tip of the cannula is submerged approximately 2.5 mm under the surface of PBS. The optoelectrode, attached to a 3-axis micromanipulator on a stereotaxic frame (Model 962, David Kopf Instruments, Tujunga, CA), was lowered into the container until its shanks were sufficiently submerged into the PBS.
The position of the optoelectrode was precisely adjusted using the micromanipulator so that the tips of the optoelectrode are exactly 1.5 mm away from the tip of the optical fiber, while the top side of the optoelectrode which has the electrodes and the LED are facing the optical fiber.
Optical stimulation was provided using a fiber-coupled LED light source (M470F3, Thorlabs), whose spectrum ( peak = 470 nm) is similar to those of the LEDs on the LED optoelectrodes. The optical power at the end of the fiber optic cannula was measured beforehand using the combination of the integrating sphere and the spectrometer. 5-Hz, 50-ms long (25 % duty ratio) rectangular voltage pulses with varying on-voltage levels were used as the LED drive signal. Pulses with 0 V low-level (off-time) voltage and high-level (on-time) voltage that would generate the irradiance the same as the precharacterized irradiance was used.
Intan RHD2132 neural signal amplifier headstage PCB (RHD2132, Intan Technologies, Los Angeles, CA) recorded the induced voltage signal. Data collection and processing followed the procedure identical to that has been previously outlined in Methods for LED-drive-induced stimulation artifact characterization.

Device physics simulation for calculation of photovoltaic-effect-induced electrostatic potential buildup inside silicon substrate
Sentaurus TCAD suite (Synopsis, Mountain View, CA) was used for finite element analysis of carrier generation and electrostatic potential buildup inside LED optoelectrode's silicon substrate during LED illumination. A 3D model of a shank was built using Sentaurus Device Editor and the shank's silicon substrate was given variable boron doping density. Two contacts, each of which indicates GaN-AlN interface and the silicon-PBS interface, were created and appropriately assigned. Using Sentaurus Device, carrier distribution and electrostatic potential buildup during irradiation of specified intensity were calculated. Ground (V(t) = 0) and floating (Q(t) = Q 0, stationary ) boundary conditions were applied to the ground and the electrolyte contacts, respectively, before simulation. First, steady-state conditions were For each iteration, different irradiance value was used, while the wavelength of the light was kept at 470 nm. The voltage of the silicon-PBS interface before, during, and after specified irradiation and optical generation of electrons and holes were recorded and reported for each combination of boron doping density and irradiance.

Measurement of stimulation artifact during current-based LED driving
Setup for in vitro characterization of stimulation artifact, described in Methods, was used. Two miniSTAR LED optoelectrodes, identical to those used for characterization of stimulation artifact resulting from voltage driving, were used. Instead of a Keysight 33220A function generator, a custom FPGA-controlled current source (DAC8750, Texas Instruments, Dallas, TX) was used as the LED driver.
Current pulses with different rise times and rising and falling edge shapes were generated using SPI commands created with a custom script written in MATLAB (MathWorks, Natick, MA). Low-and highlevel values of the current pulses were kept as 0 A and 75 µA regardless of the type of waveform. First, LED wafers were cleaned in acetone, isopropyl alcohol (IPA), and then deionized wafer (DI H 2 O) for removal of organic residue. Next, the wafers were exposed to hydrochloric acid (HCl) for surface cleaning. Following wafer surface cleaning, chlorine-based reactive ion etching (RIE) was used for LED mesa definition (using LAM 9400, Lam Research Corporation, Fremont, CA). Plasma-enhanced chemical vapor deposition (PECVD) of silicon dioxide (SiO2) was used for surface passivation of LED structure (using GSI ULTRADEP 2000, GSI Lumonics, Novanta Inc., Bedford, MA, USA). After SiO 2 PECVD, SiO2 on top of the p-and n-GaN contact sites were etched using C4F8/SF6-based RIE (using LAM 9400). Lift-off patterned, electron-beam (e-beam) evaporated nickel/gold (Ni/Au, deposited using SJ-20, Denton Vacuum, Moorestown, NJ) and titanium/aluminum/titanium/gold (Ti/Al/Ti/Au, deposited using Enerjet Evaporator, K. J. Lesker, Jefferson Hills, PA and Denton Vacuum) metal stacks were used for p-GaN contact and n-GaN contact, respectively. Following deposition, the p-GaN contact metal stack was annealed at 500 °C in a N2/O2 environment (using JetFirst 150, SEMCO Inc., Irving, TX).
Bilayer stack of atomic-layer-deposited (ALD) aluminum oxide (Al2O3, deposited using Oxford OpAL, Oxford Instruments, Abingdon, UK) and PECVD SiO 2 (deposited using P5000 PECVD, Applied Materials, Santa Clara, CA) composed the passivation layers, and e-beam evaporated Ti/Au (deposited using Enerjet Evaporator) formed the metal layers. After deposition of ALD Al 2 O 3 and PECVD SiO 2 bilayer above the top metal layer, the top passivation was partially etched etching using a combination of RIE (using LAM 9400) and wet etching (using dilute buffered hydrofluoric acid) to expose recording electrode contact sites. Recording electrodes were defined using lift-off patterning of sputter-deposited titanium/platinum/iridium (Ti/Pt/Ir, deposited using LAB 18, K. J. Lesker) stack.
Double-sided plasma dicing process, consisting of patterned front-side deep reactive ion etching (DRIE) followed by backside plasma thinning, was conducted (using STS Pegasus 4, SPTS Technologies, Orbotech Ltd., Yavne, Israel). Released optoelectrodes were cleaned in xylenes, acetone, and then IPA before pick up and assembly.

Test used
Two-sided Mann-Whitney U test with Bonferroni correction

Fig. 2d
Test used Two-sided Mann-Whitney U test with Bonferroni correction

Test used Two-sided Mann-Whitney U test with Bonferroni correction Samples and categories
Peak-to-peak magnitudes of stimulation artifact, recorded from all sites on shielded LED optoelectrodes fabricated using LED wafer with moderately boron-doped silicon substrate, measured during optical stimulation using LEDs resulting in LED surface irradiance of 1.5, 3, 7.5, 15, 30, 45, 60

Fig. 4b, left
Test used Two-sided Mann-Whitney U test with Bonferroni correction Samples and categories Peak-to-peak magnitudes of stimulation artifact, recorded from all sites on shielded LED optoelectrodes fabricated using LED wafer with FZ-silicon substrate (FZ-Si), shielded LED optoelectrodes fabricated using LED wafer with moderately borondoped silicon substrate (p --Si), and shielded LED optoelectrodes fabricated using LED wafer with heavily boron-doped silicon substrate (p + -Si)

Statistics provided in figure
Box plots with whiskers and outliers (denoting median, IQR, EVs and outliers)

Test used
Two-sided Mann-Whitney U test with Bonferroni correction

Samples and categories
Peak-to-peak magnitudes of stimulation artifact, recorded from two sites at the bottom of each shank (sites 1 & 2) on shielded LED optoelectrodes fabricated using LED wafer with FZ-silicon substrate (FZ-Si), shielded LED optoelectrodes fabricated using LED wafer with moderately boron-doped silicon substrate (p --Si), and shielded LED optoelectrodes fabricated using LED wafer with heavily boron-doped silicon substrate (p + -Si)

Statistics provided in figure
Box plots with whiskers and outliers (denoting median, IQR, EVs and outliers)

Test used
Two-sided Mann-Whitney U test with Bonferroni correction

Samples and categories
Peak-to-peak magnitudes of stimulation artifact, recorded from two sites at the top of each shank (sites 7 & 8) on shielded LED optoelectrodes fabricated using LED wafer with FZ-silicon substrate (FZ-Si), shielded LED optoelectrodes fabricated using LED wafer with moderately boron-doped silicon substrate (p --Si), and shielded LED optoelectrodes fabricated using LED wafer with heavily boron-doped silicon substrate (p + -Si)

Statistics provided in figure
Box plots with whiskers and outliers (denoting median, IQR, EVs and outliers)

Test used
Two-sided Mann-Whitney U test

Samples and categories
Peak-to-peak magnitudes of stimulation artifact, recorded from different sites on each shank (sites 1 -8) on shielded LED optoelectrodes fabricated using LED wafer with heavily boron-doped silicon substrate; comparing two sites with same LED-tointerconnect distances

Test used Two-sided Mann-Whitney U test with Bonferroni correction Samples and categories
Peak-to-peak magnitudes of stimulation artifact, recorded from different sites on each shank (sites 1 -8) on shielded LED optoelectrodes fabricated using LED wafer with heavily boron-doped silicon substrate; comparing pairs of two sites with same LEDto-interconnect distances with each other pair

Test used Two-sided Mann-Whitney U test with Bonferroni correction Samples and categories
Peak-to-peak magnitudes of stimulation artifact, recorded from two sites at the bottom of each shank (sites 1 & 2) on MiniSTAR LED optoelectrodes during LED driving with voltage pulses with 0 V low-level voltage and 5 ns rise time (0V-5ns), with pulses with 0 V low-level voltage and 1 ms rise time (0V-1ms), with pulses with 2.8 V low-level voltage and 5 ns rise time (2P8V-4ns), and with pulses with 2.8V lowlevel voltage and 1 ms rise time (2P8V-1ms)

Test used
Kruskal-Wallis test

Samples and categories
Current through LEDs fabricated on LED wafer with FZ-silicon substrate (FZ-Si), LEDs fabricated on LED wafer with moderately boron-doped silicon substrate (p --Si), and LEDs fabricated on LED wafer with heavily boron-doped silicon substrate (p + -Si), at 4 V of forward bias voltage, measured from five different locations on each wafer (B, C, T, L, and R).

Test used
Kruskal-Wallis test

Samples and categories
Radiant flux generated from LEDs fabricated on LED wafer with FZ-silicon substrate (FZ-Si), LED wafer with moderately boron-doped silicon substrate (p --Si), and LED wafer with heavily boron-doped silicon substrate (p + -Si), at 4 V of forward bias voltage, measured from five different locations on each wafer (B, C, T, L, and R).

Test used Kruskal-Wallis test Samples and categories
Maximum plug efficiency of LEDs fabricated on LED wafer with FZ-silicon substrate (FZ-Si), LEDs fabricated on LED wafer with moderately boron-doped silicon substrate (p --Si), and LEDs fabricated on LED wafer with heavily borondoped silicon substrate (p + -Si), measured from five different locations on each wafer (B, C, T, L, and R).

Statistics provided in figure
Box plots with whiskers and outliers (denoting median, IQR, EVs and outliers) n 8, 9, 9, 9, & 8  Test used Two-sided Mann-Whitney U test with Bonferroni correction

Samples and categories
Peak-to-peak magnitudes of PV-induced voltage signal, recorded from all sites on shielded LED optoelectrodes fabricated using LED wafer with FZ-silicon substrate (FZ-Si), shielded LED optoelectrodes fabricated using LED wafer with moderately boron-doped silicon substrate (p --Si), and shielded LED optoelectrodes fabricated using LED wafer with heavily boron-doped silicon substrate (p + -Si)

Statistics provided in figure
Box plots with whiskers (denoting median, IQR, and EVs)

Test used
Two-sided Mann-Whitney U test with Bonferroni correction

Samples and categories
Peak-to-peak magnitudes of PV-induced voltage signal, recorded from all sites on shielded LED optoelectrodes fabricated using LED wafer with heavily boron-doped silicon substrate (p + -Si), electrode arrays fabricated using soda-lime glass substrate (G), and electrode arrays fabricated using LED-on-sapphire substrate (S)

Statistics provided in figure
Box plots with whiskers (denoting median, IQR, and EVs)

Test used
Two-sided Mann-Whitney U test with Bonferroni correction Samples and categories Peak-to-peak magnitudes of stimulation artifact, recorded from all sites on control LED optoelectrodes (i.e. non-shielded optoelectrodes with moderate boron doping of the silicon substrate, 1ML), LED optoelectrodes with shielding layer and moderately boron-doped the silicon substrate (SO), and miniSTAR optoelectrodes.

Test used
Two-sided Mann-Whitney U test

Samples and categories
Peak-to-peak magnitudes of stimulation artifact, recorded from different sites on each shank (sites 1 -8) on miniSTAR LED optoelectrodes during optical stimulation using LED 1; comparing two sites with same LED-to-interconnect distances

Test used
Two-sided Mann-Whitney U test

Samples and categories
Peak-to-peak magnitudes of stimulation artifact, recorded from different sites on each shank (sites 1 -8) on miniSTAR LED optoelectrodes during optical stimulation using LED 2; comparing two sites with same LED-to-interconnect distances

Test used
Two-sided Mann-Whitney U test

Samples and categories
Peak-to-peak magnitudes of stimulation artifact, recorded from different sites on each shank (sites 1 -8) on miniSTAR LED optoelectrodes during optical stimulation using LED 3; comparing two sites with same LED-to-interconnect distances

Test used
Two-sided Mann-Whitney U test

Samples and categories
Peak-to-peak magnitudes of stimulation artifact, recorded from a few selected sites (only among sites 1 and 2) on a miniSTAR optoelectrodes during optical stimulation; comparing the magnitude recorded before and after electroplating

Test used
Kruskal-Wallis test

Samples and categories
Peak-to-peak magnitudes of stimulation artifact, recorded from two sites at the bottom of each shank (sites 1 & 2) on miniSTAR optoelectrodes during LED driving with current pulses different shapes -trapezoidal, sinusoidal, and sigmoidal -with 10 -90 % rise times of approximately 1 ms.

Statistics provided in figure
Box plots with whiskers (denoting median, IQR, and EVs) and values evaluated at a few selected harmonic frequencies (in thin solid lines. Only the prime numbered harmonics (f = 5 × (2, 3, 5, 7, etc.)) are shown for better visibility). The frequency range of 250 Hz < f < 10 kHz is highlighted with a shade of grey.