Highly Efficient Microscale Gallium Arsenide Solar Cell Arrays as Optogenetic Power Options

Optogenetics is one of the most powerful investigation tools in neuroscience research. A major engineering challenge is the wireless power supply needed to operate the light-emitting diodes (LEDs) and generate over 1 mW/mm² optical power density required to activate opsins. Here, we describe design strategies to construct gallium arsenide microscale solar cells and approaches to integrate them into array structures as efficient optogenetic power options. The photovoltaic (PV) system outputs an electric power of 2.30 mW with an open-circuit voltage (V_oc) of 4.97 V and a short-circuit current (I_sc) of 0.59 mA under direct infrared illumination. We show that this power level is enough to operate both blue and yellow LEDs and provide optical power densities of 3.5 and 2.3 mW/mm², respectively. This paper provides a guideline to design efficient PV systems as power supplies for optogenetics and other biomedical implants.

SECTION 1. Introduction

Optogenetics has emerged as a core technology to study brain function, and helps to provide new therapeutic options for neurological disorders such as epilepsy and Parkinson's disease [1], [2]. It uses genetic engineering methods to insert opsins into specific cell types and then controls the cell behavior with light [3]. Although the last decade has witnessed an exponential growth of genetically encoded optogenetic actuators [4], the corresponding optoelectronic systems necessary for practical and effective light delivery to activate opsins have developed much more slowly. Early stage light delivery has been achieved by optical fibers coupled to external laser diodes. Recently, LEDs have been extensively used as illumination sources due to their advantages in illumination stability, versatility, and ease of light switching [5]–​[8]. Several studies have suggested that using external fiber optics to deliver optical stimuli from LEDs to targeted cells or neurons would introduce undesired physical tethers that restrain naturalistic behavior of animals [9], [10]. As a result, optoelectronic tools beyond optical fiber-based systems are highly desirable to enable sophisticated optogenetic manipulation with limited disruption of animal behavior to match the advances in genetics and biochemistry. An ideal technical platform would require wireless on-board powering of the LEDs with miniaturized circuitry and this remains as a major technological challenge for optogenetics.

Over the last few years, several impressive wireless powering methods have been developed for optogenetics, including rechargeable batteries [11], far-field [10], [12], mid-field [8], and near-field [13] energy harvesters. However, batteries suffer from weight, space, and limited capacity concerns while wireless energy harvesters have disadvantages such as device location/orientation limitations, interferences from obstacles, and antennas with large dimensions. Solar cells that convert light into electricity represent another promising power option for biomedical applications. Both organic and inorganic types of solar cells have been recently demonstrated in various bio-integrated electronic devices, such as to power cardiac pacemakers, organic electrochemical transistors, and cortical electrical stimulators [14]–​[18]. However, in terms of optogenetics, the overall electrical power outputs of these solar cell systems are still not enough to operate LEDs and generate optical power densities above the threshold (1 mW/mm²) for optogenetic activation of a neuronal circuit [13]. To maximize the potential of solar cells as a power option for bioelectronics, it is highly desirable to design high-performance PV systems as local optogenetic power generators. The emerging transfer printing-based assembly techniques have enabled the heterogeneous integration of inorganic materials or devices from their donor wafers into flexible, unconventional polymeric substrates [19], [20]. Single junction GaAs photodiodes have been utilized for efficient photodetection in implantable biosensors by taking advantage of the transfer printing techniques [9]. On the other hand, despite the fact that GaAs thin-film solar cell is a top candidate in the thin-film solar cell market due to its high power conversion efficiency, long-term stability, and reasonable cost, their applications as power options for bio-integrated electronics are yet to be demonstrated.

Here, we present micro-fabrication schemes for single junction GaAs μ-SCs and approaches to integrate them into an array structure (3 × 5) onto flexible, and transparent substrates as potential practical wireless power generators for optogenetics. The individual solar cell shows a $V_{oc}$ of 0.92 V while the integrated PV system exhibits a $V_{oc}$ of 4.97 V and generates approximately 2.30 mW of electrical power. This output power is among the highest reported values for wireless biomedical power supplies [15], [21]–​[23]. A proof of concept demonstration shows that the integrated PV system is able to power both blue and yellow microscale LEDs and yield optical power densities above the threshold for optogenetic research. Due to the use of the transparent substrate, the integrated system is compatible with optical imaging setups and provides a potential alternative for wireless optogenetics with simultaneous optical imaging.

SECTION 2. Experimental Details

2.1 Fabrication of the μ-SCs

The μ-SCs were built on epitaxially grown layers of Te-doped n-type GaAs top contact layer (100 nm, 1 × 10¹⁹ cm⁻³), Si-doped n-type GaAs top contact layer (100 nm, 2 × 10¹⁸ cm⁻³), Si-doped n-type In_0.50Ga_0.50P window layer (25 nm, 2 × 10¹⁸ cm⁻³), Si-doped n-type GaAs emitter layer (100 nm 2 × 10¹⁸ cm⁻³), Zn-doped p-type GaAs base layer (2500 nm, 1 × 10¹⁷ cm⁻³), Zn-doped p-type Al_0.30Ga_0.70As BSF layer (100 nm, 5 × 10¹⁸ cm⁻³) and Zn-doped p-type GaAs bottom contact layer (2000 nm, 3-5 × 10¹⁹ cm⁻³), respectively. Fig. 1 shows a schematic fabrication flow of the μ-SCs, which followed a sequential wet-etching process, as we previously reported [9]. The first step was to define the n-GaAs contact by photolithography (AZ nLOF 2070, Integrated Micro Materials, spin coated at 3000 rpm, developed with AZ 917MIF) and electron-beam deposition of Cr/Au bilayers (5 nm/150 nm). The Te and Si doped GaAs layers were removed by a mixture of H₃PO₄/H₂O₂/H₂O (3:1:25 volume ratio), with the Cr/Au layers as protection masks. The active regions of the μ-SCs were defined by photolithography with AZ 5214-E (MicroChem Corp, spin coated at 3000 rpm; developed with AZ 300MIF), followed by wet etching with HCl/H₃PO₄ (1:1 volume ratio) and H₃PO₄/H₂O₂/H₂O (3:1:25 volume ratio) to remove the In_0.50Ga_0.50P window layer, GaAs emitter layer and base layer, respectively. Creating the p-GaAs contacts (Cr/Au, 5 nm/150 nm) for the μ-SCs followed procedures similar to those described for the n-GaAs contacts. The μ-SC dimensions (400 μm × 400 μm) were then defined by lithographically patterned AZ 5214-E protection layer and wet-etching process in H₃PO₄/H₂O₂/H₂O (3:1:25 volume ratio). Finally, the Al_0.95Ga_0.05As sacrificial layer was selectively undercut by immersion in an ethanol/HF mixture (1.5:1 volume ratio). The μ-SCs were temporarily held to the GaAs substrate via 4 patterned breakable photoresist anchors (2 μm, AZ 5214-E, MicroChem Corp). Notably, the dimensions of the μ-SCs were easily programmable by adjusting the sizes of isolation masks, leading to controllable light harvesting areas for different applications.

Fig. 1.

Fabrication scheme of GaAs μ-SCs.

2.2 Fabrication of the μ-SC Arrays

Fig. 2 shows a schematic fabrication scheme for the integrated 3 × 5 μ-SC array. A polyethylene terephthalate (PET) film (75 μm thick) was laminated on a glass slide with a thin layer of polydimethylsiloxane (PDMS, Dow Corning Sylgard 184, spin coated at 3000 rpm) as adhesive. Photolithography (AZ P4620, MicroChem Corp, spin coated at 3000 rpm, developed with AZ 400K) and wet etching defined the 3 × 5 Cr markers (10 nm) for the array. A 2 μm thick SU-8 epoxy (SU-8 2002, MicroChem Corp) was spin coated on the PET film as an adhesive layer. The μ-SCs were then picked up from the GaAs source wafer and transfer printed to SU-8 layer directly on top of the Cr markers by PDMS elastomer stamps (400 μm × 400 μm × 50 μm). The device was then cured by baking at 110°C for 10 mins. The numbers of μ-SCs connected in series and parallel were controlled through the arrangement of Cr markers. Photolithography defined SU-8 layer (SU-8 2002, MicroChem Corp, spin coated at 3000 rpm, developed with SU-8 developer) encapsulated the μ-SCs and selectively opened the contact areas for p- and n-GaAs contacts. Photolithography (AZ nLOF 2070, Integrated Micro Materials, spin coated at 3000 rpm, developed with AZ 917MIF) patterned and sputtered Cr/Au metals (10 nm/500 nm) served as interconnections for the μ-SC array and electrical contacts for the LEDs. A blue LED (270 μm × 220 μm × 50 μm, C460TR2227-0216, Cree Inc.) or a yellow LED (300 μm × 300 μm × 100 μm, TCE12-589, Three Five Materials Inc.) was placed on the electrical contacts by transfer printing and soldering (Indalloy 290; Indium Corporation) at 150°C. Spin coated SU-8 (7 μm) and PDMS elastomer (10 μm) encapsulated the device. It is important to note that PDMS is a well-known biocompatible material and this makes our design fully compatible with future in vitro or in vivo experiments.

Fig. 2.

Schematic illustration of key fabrication steps for a GaAs μ-SC solar cell array.

2.3 Measurement

The current versus voltage (I-V) characteristics were measured with a Keithley 2400 source meter under 100 mW/cm² standard AM 1.5G irradiation from a 1000W Oriel 91192 solar simulator. A reference silicon solar cell calibrated the light intensity of the solar simulator. A 774 nm LED (Thorlabs M780L3) served as the IR (near-infrared) light source. The EQE was collected using a halogen lamp coupled to a monochromator. Optical power densities of the IR, blue, and yellow LEDs were measured with an AvaSpec-ULS2048L StarLine Versatile Fiber-optic Spectrometer.

SECTION 3. Results

Fig. 3(a) reveals the multilayer stacked vertical structure of the GaAs μ-SCs with Au contacts on both p and n sides. Fig. 3(b) presents an optical image of an encapsulated solar cell array, demonstrating its excellent flexibility and transparency. The array design contains 15 μ-SCs, 5 columns interconnected in series and 3 rows interconnected in parallel. A single μ-SC has a dimension of 400 μm × 400 μm × 5 μm in Fig. 3(b). We would like to point out that this design allows easily adjusting the μ-SC numbers or changing the interconnections to increase the $I_{sc}$, $V_{oc}$, and overall power output to satisfy the energy requirements of different biomedical implants. In addition, the entire system only weighs 23 mg. This light weight property makes it advantageous for bio-applications.

Fig. 3.

(a) Schematic image of the solar cell stack structure. (b) Left, optical image of an encapsulated flexible GaAs solar cell array. Scale bar, 5 mm. Right, optical image of a single GaAs solar cell. Scale bar, 200 μm.

Fig. 4(a) summarizes the I-V curves of an individual GaAs μ-SC under different light intensities. At 1 sun illumination (AM 1.5G spectrum), the μ-SC exhibits an $I_{sc}$ of 20.1 μA, a $V_{oc}$ of 0.92 V, and a fill factor (FF) of 79.3%, generating an overall output power of 14.6 μW. Table 1 lists the average photovoltaic performance of GaAs μ-SCs and arrays under either 1 sun illumination or IR illumination. The average $I_{sc}$, $V_{oc}$, and FF of a single cell under AM 1.5G 1 sun illumination are 19.8 ± 0.2 μA, 0.92 ± 0.01 V, and 78.5 ± 0.4%, respectively. The little difference between the average and best photovoltaic results suggests the good reproducibility of the solar cell performance. For GaAs solar cells, characterizing the light intensity dependence of $I_{sc}$ is important to determine the quality of the devices. In Fig. 4(b), $I_{sc}$ shows a power law dependence on light intensity from 1000 W/m² to 10 W/m², indicating the absence of trap states and the good quality of the solar cells. Fig. 4(c) depicts the external quantum efficiency (EQE) of the solar cell, which defines the efficiency at different spectral wavelength. The solar cell exhibits a wide EQE response from 400 nm to 860 nm, with maximum EQEs reaching over 85% from 650 nm to 850 nm, respectively. The I-V curve of the array is plotted in Fig. 4(d). The 3 × 5 array design increases the $I_{sc}$, $V_{oc}$ from 20.1 μA and 0.92 V to 57.3 μA and 4.53 V, with a total output power of 202 μW, respectively. The slight difference between measured output power and expected one based on the single μ-SC performance (219 μW) is due to variations in cell performance and contact resistances in the array structure [23], [24]. The average output power under AM 1.5G 1 sun illumination is 196 ± 3 μW (Table 1).

Fig. 4.

Photovoltaic performance of GaAs solar cells. (a) Light intensity–dependent I-V characteristics of the solar cell. Red, blue, dark yellow, olive, violet, and black represent the light intensity of 1000, 500, 100, 39, 10 W/m² and the dark condition, respectively. (b) The dependence of $I_{sc}$ on light intensity of a single GaAs solar cell. (c) EQE spectrum of a GaAs solar cell. (d) I-V characteristics of a single solar cell (red) and a 3 × 5 solar cell array (black) under 1 sun illumination.

Table 1 Average photovoltaic parameters for single cells and arrays based on 5 devices

For optogenetic research, opsins sensitive to blue and yellow lights, such as channelrhodopsin-2 (ChR2) and halorhodopsin are widely used to stimulate and inhibit cell activity [25], [26]. Thus, an IR (near-infrared) LED would serve as a good portable light source to prevent undesired crosstalk between illumination of the solar cells and activation of the opsins. In addition, the μ-SC exhibits high and stable EQEs around 88% from 715 nm to 800 nm, as shown in Fig. 4(c), assuring good PV performance under IR illumination. Importantly, our previous study has demonstrated that IR light exhibits over 60% transmission through the skin and is biocompatible under low intensities, making it a good option for in vivo bio-related applications [23]. The emission maximum of the IR LED used in this study is at 774 nm with a full width at half maximum of 28 nm, as shown in Fig. 5(a). Fig. 5(b) presents the I-V curves of a single cell and a solar cell array under 200 mW/cm² IR illumination. The intensity of the light source should be carefully controlled to reduce the possibility of introducing a burn to the skin. The power level used in this paper is similar to those in low level NIR light therapy for treating wound healing and pain relief [27]. The single cell generates an output power of 153.5 μW with $I_{sc}$, $V_{oc}$, and FF of 202.7 μA, 1.00 V, and 75.4%, respectively. The average power output over 5 devices is 150 ± 4 μW (Table 1). The dramatically improved $I_{sc}$ results from both the higher light intensity and EQEs in the IR region. The increase in $V_{oc}$ is due to the fact that $V_{oc}$ shows a weak logarithmic scaling with optical generation rate, which is affected by light intensity [28]. The μ-SC array generates 2.30 mW electrical power, a $V_{oc}$ of 4.97 V, and an $I_{sc}$ of 0.59 mA under direct IR illumination, with an average power output of 2.23 mW (Table 1). This value is among the highest reported output powers in wireless bioelectronic power supplies.

Fig. 5.

(a) Normalized emission spectrum of the IR LED light source. (b) I-V characteristics of a single solar cell (red) and a 3 × 5 solar cell array (black) under 774 nm IR illumination.

The output power from the solar cell array is enough to operate blue and yellow LEDs, as shown in Fig. 6(a). The I-V characteristics of the two LEDs are presented in Fig. 6(b). The blue and yellow LEDs have turn-on voltages of 2.7 V and 1.8 V, respectively, which are significantly lower than the $V_{oc}$ (4.97 V) of the μ-SC array. The emission maximums of the two LEDs are 468 nm and 594 nm, which is presented in Fig. 6(c). When the μ-SC array is under 200 mW/cm² IR illumination, the measured optical output light densities at the surface of the blue and yellow LEDs are ∼3.5 mW/mm² and 2.3 mW/mm², which match well with the optical excitation requirements of ChR2 and halorhodopsin in optogenetic experiments [13]. Thus this PV design is sufficient to serve as wireless power supply for optogenetics.

Fig. 6.

GaAs solar arrary integrated with different LEDs. (a) Optical images of the blue and yellow LEDs turned on by the solar cell arrays during exposure to IR illumination. (b) I-V characteristics of the blue and yellow LEDs. (c) Emission spectra of the blue and yellow LEDs with peaks at 468 nm and 594 nm, respectively.

SECTION 4. Conclusion

In summary, we successfully demonstrated approaches to construct GaAs μ-SC arrays on flexible and transparent PET substrates. The array architecture allows fine tuning of the output voltages, currents, and powers over a wide range to satisfy diverse requirements in biomedical devices. The current design is able to generate 2.30 mW of practical electrical power. Detailed studies show that this integrated system could power different LEDs and generate optical output power densities relevant to activate opsins for optogenetic research (>1 mW/mm²), indicating its great potential as an attractive option for wireless optogenetics. This work indicates that photovoltaic system could be used for various biomedical devices after proper design and optimization of the performance. Future developments that can integrate different solar cells (such as GaAs, silicon, and CIGS solar cells) with microscale chip batteries will further solve the power reliability issues due to the movement of animals regarding to light sources and make this technology more attractive for neuroscience research.

ACKNOWLEDGMENT

This work was performed in part at The George Washington University Nanofabrication and Imaging Center.