Vertical-junction photodiodes for smaller pixels in retinal prostheses

Abstract

Objective.To restore central vision in patients with atrophic age-related macular degeneration, we replace the lost photoreceptors with photovoltaic pixels, which convert light into current and stimulate the secondary retinal neurons. Clinical trials demonstrated prosthetic acuity closely matching the sampling limit of the 100μm pixels, and hence smaller pixels are required for improving visual acuity. However, with smaller flat bipolar pixels, the electric field penetration depth and the photodiode responsivity significantly decrease, making the device inefficient. Smaller pixels may be enabled by (a) increasing the diode responsivity using vertical p-n junctions and (b) directing the electric field in tissue vertically. Here, we demonstrate such novel photodiodes and test the retinal stimulation in a vertical electric field.Approach.Arrays of silicon photodiodes of 55, 40, 30, and 20μm in width, with vertical p-n junctions, were fabricated. The electric field in the retina was directed vertically using a common return electrode at the edge of the device. Optical and electronic performance of the diodes was characterizedin-vitro, and retinal stimulation threshold measured by recording the visually evoked potentials in rats with retinal degeneration.Main results.The photodiodes exhibited sufficiently low dark current (<10 pA) and responsivity at 880 nm wavelength as high as 0.51 A W^-1, with 85% internal quantum efficiency, independent of pixel size. Field mapping in saline demonstrated uniformity of the pixel performance in the array. The full-field stimulation threshold was as low as 0.057±0.029mW mm^-2with 10 ms pulses, independent of pixel size.Significance.Photodiodes with vertical p-n junctions demonstrated excellent charge collection efficiency independent of pixel size, down to 20μm. Vertically oriented electric field provides a stimulation threshold that is independent of pixel size. These results are the first steps in validation of scaling down the photovoltaic pixels for subretinal stimulation.

Keywords: neural stimulation; photodiode; photovoltaics; retinal prosthesis; vertical junction; visually evoked potential.

Figure 1.

Diagram of the photovoltaic pixel with (a) planar and (b) vertical junctions. The dark blue region represents the highly doped area of the same polarity as the background doping for making ohmic contact, while red has the opposite doping polarity. The p⁺ region is connected to the active electrode, while the n⁺—to the return electrode.

Figure 2.

(a) Layout of the photovoltaic arrays with 40 μm pixels. ‘Bipolar honeycomb’ pixels (left column) have a common return electrode on top of the honeycomb walls, whereas ‘monopolar flat’ pixels (right column) have a return electrode in the periphery of the device. The size of the pixel is denoted with ‘s’. (b) Electric potential in the medium relative to infinity produced by full-field illumination, with electric current of 100 nA per pixel. Note different scale in vertical and horizontal directions, introduced for better visibility. (c) Zoom into the single pixel area outlined in (b) (green box). In this case, the horizontal and vertical scales are the same, and potential is plotted relative to the center of the active electrode right above the device, showing similar voltage drop across the bipolar cell for these two electrode geometries. Histology of the degenerate retina (RCS) with a diagram of the bipolar and ganglion cells (BC and RGC) is shown to scale for comparison. Confocal microscopy of the retinal cells migrating into the honeycombs is shown in supplemental figure 1.

Figure 3.

(a) Normalized photoreceptor sensitivity [36] and ocular transmittance to the retina [37] as a function of wavelength, showing high transmittance and low photoreceptor sensitivity near 880 nm wavelength. (b) Silicon absorption depth, and typical silicon photodiode responsivity (A/W) [35] as a function of wavelength, showing absorption depth around 25 μm and high responsivity near 880 nm wavelength.

Figure 4.

Fabrication process flow for photodiodes with vertical p–n junctions. Layers shown not to scale for better visibility.

Figure 5.

(a) Trenches of 1.7 μm on top and 1.2 μm at the bottom, etched for pixel isolation and p–n junction formation. Magnified top and bottom sections are shown on the right. (b) Cross-section of the device with trenches filled by polycrystalline silicon (pointed by the red arrows). The area adjacent to the trench is n-doped, and the adjacent dark stripes are the depletion zones (*). Dash line shows the top of the SiO₂ layer not visible due to the white background. (c) SIMS profile of dopant concentration across the vertical junction overlying the scanning electron microscope (SEM) image, confirming consistent and uniform doping across the trench.

Figure 6.

(a) Diagram of the stack used to design the anti-reflection coating. The thicknesses of the silicon, buried silicon oxide, and titanium are 30, 0.5, and 0.2 μm, respectively. The silicon carbide and top silicon oxide thicknesses are varied. (b) Reflectance of the front surface, as a function of SiC and SiO₂ thickness, calculated by the transfer matrix method. White dash line indicates a safe SiO₂ thickness for avoiding pinholes. The ‘×’ marks our attained layers of 57 nm SiO₂ and 208 nm SiC, achieving 2.4% reflectance. (c) Reflectivity as a function of the incidence angle.

Figure 7.

(a), (b) Optical images showing 1.5 mm implants with pixel size 55, 40, 30, and 20 μm (cropped from four different arrays). (c) SEM image of the released device with 20 μm pixels, placed on a porcine retinal pigment epithelium for scale, and (d) SEM of a single 20 μm pixel, showing a central active electrode, surrounded by photosensitive area, and the hexagonal trench covered by metal buried under SiC to connect to the outer return electrode ring.

Figure 8.

I–V characteristic of diodes with pixel sizes of 55, 40, and 30 μm, shown here on (a) linear and (b) logarithmic scales. (c) Diffusion capacitance as a function of the forward bias voltage. Below 0.6 V, these capacitances are far below the capacitances of the active electrodes, shown by the dash lines. (d) Responsivity of the photodiodes with four different pixel sizes, showing consistently high values independent of size.

Figure 9.

(a)–(c) FIB cross-section of the active electrode. The electrode is composed of Pt, Ti, and SIROF layers, with their thicknesses decreasing toward the edge of the electrode. (d) Average capacitance of the active electrodes decreases with decreasing pixel size due to SIROF thickness non-uniformity. (e) Peak electric potential in the medium 20 μm above the device with 40 μm pixels, generated by full-field illumination at 5 mW mm⁻² irradiance. (f) Waveform of the electric potential measured 20 μm above the centers of the devices is nearly the same with 20 and 40 μm pixels. The y-axis label is shared with (e).

Figure 10.

(a) Fundus of the rat eye with a subretinal implant to the right of the optic nerve. (b) OCT image of the implant under the retina and (c) higher magnification demonstrating close apposition of the INL to the implant. (d) Typical VEP in response to the full-field stimulation with 10 ms pulses at various irradiances, repeated at 2 Hz (averaged over 500 trials), with * indicating the stimulation artifact, and arrows pointing at the negative and positive peaks. (e) Average VEP amplitude as a function of the incident irradiance on the retina, with implants having 20 and 40 μm pixels (n = 4 per group). The amplitude is normalized to the RMS of noise in each animal. Error bars represent standard deviation. Dash line depicts the 6.5× (RMS noise) level, which corresponds to the 95% confidence interval.