Chapter 4: Integration of Graphene Field-Effect-Transistors on a Perforated Microfluidic Platform
4.5 SPCM Measurements of Whole Mice Retina
To probe electrical activities of live tissues with gFETs, THY1.2-YFP transgenic mice retina expressing green fluorescent protein for RGC somas and axons, was studied on the microfluidic gFET platform through photocurrent measurements. Figure 4.6 shows an optical image of the bare device, and a fluorescence image after retina was loaded and negative pressure was started at the suction channels. Note that the integrity of the tissue is not compromised by the two-suction channel device during the application of negative pressure.
Figure 4.6. a. Optical image of the gFET platform prior to retina loading. b. Fluorescent image of THY1.2-YFP mice retina during optoelectronic measurements. Scale bar is 300 μm.
In our experiments, photocurrent mapping of the bare graphene transistors was performed prior to placing the retina onto the graphene transistors. Figure 4.7a,b shows an optical image and the corresponding photocurrent map of a 100 x 60 μm2 gFET during optoelectronic measurements.
Activity is strongest at the electrode/graphene junctions due to the Schottky-like barriers (Figure 4.7b, Figure 4.4a), while small to no photocurrent signals can be detected at the flat band region between the electrodes (Figure 4.7b, Figure 4.4a). The opposite polarity at the drain and source terminals indicates the two directions in which graphene band bends.
95
In contrast, obvious photocurrent responses were detected across the entire graphene region when retina was sitting on top of the transistors’ substrate (Figure 4.7c,d). Interestingly, the polarity at the graphene-electrode interface changed, indicating that the retina membrane potential can turn the bending direction of the graphene electronic band and alter the doping from p-type to n-type. This observation was corroborated across measurements of multiple retina tissues and invites further study to see if the electrical activities in retina not only acts as local gates, but also as a global gate through its membrane potential. While abundant photocurrent signals were detected in the graphene channel, the data is complex to analyze and does not directly correspond to the neuronal structures obtained by confocal imaging of THY1.2-YFP retina (Figure 4.7e).
Figure 4.7. a,b. Optical and scanning photocurrent images of bare graphene transistor, respectively. c,d. Fluorescence and scanning photocurrent images of the same graphene transistor underneath THY1.2-YFP mice retina, respectively e. Overlapped fluorescence (c) and photocurrent (d) images. f. Zoomed-in view of region of interest in photocurrent image. Scale bars are 40 μm for (a – c), and 10 μm for (f).
96
It is hypothesized that the reason for the vast amount of photocurrent data is that many different cellular structures in the retina can alter the local electrochemical potential of the graphene. The first three retina layers are the inner limiting membrane, the nerve fiber layer (RNFL), and the ganglion cell layer (GCL), respectively. In our experiments, the inner limiting membrane is removed by enzyme digestion to leave the RGCs exposed to the electronics. Initially, we envisioned that RGCs and axons were the predominant cellular processes capable of dynamically changing the extracellular membrane potential in the RNFL/GCL, and modulating the local electrochemical environment of graphene. However, others have demonstrated that RGCs compromise between 50 – 60% of the total neuronal population in the GCL of rodents.243 In addition to RGCs, small glia and cholinergic amacrine cells have also been found in the RNFL/GCL.243,244 The mean thickness of the RNFL + GCL in young and adult mice were reported to be approximately 85 μm and 89 μm, respectively. Figure 3.34 in chapter 3 indicates that action potentials can be detected by graphene electrodes positioned ~14 μm away from the activity source. Further characterization is needed to identify the magnitude and maximum distance from the action potential source to the gFETs necessary to separate the photo-excited EHPs and generate photocurrent signals.
As shown in Figure 4.8, gFETs can also be used as standard electrodes. Action potentials, most likely from RGCs, with similar waveform and spike duration as those measured with MEA were detected in multiple electrodes. Using gFETs as conventional electrodes during the initial steps of retina measurements could drastically reduce the time to identify which transistors detect obvious neural activity, and later focus on those transistors for photocurrent measurements.
97
Figure 4.8. Measurements of graphene field effect transistor as a traditional MEA.
To better comprehend the capabilities and limitations of scanning photocurrent microscopy for electrophysiological studies, we are currently designing a signal source to generate highly localized artificial signals. By using a tungsten microprobe with a 0.5 μm point radius connected to a LabView signal generator through a data acquisition (DAQ) system, we should be able to create focal voltage pulses at different locations of the graphene channels. The microprobe will have a co-axial configuration with the tungsten probe in the middle, which is then coated with a layer of parylene, 50 nm thick Au layer, and then another layer of parylene, which should allow for confining the signal to the end of the tungsten tip. The end of the probe will be cut open using the focused ion beam to expose a small open window at the end (0.5 μm point radius) to maximize the spatial resolution of the experiments. The probe will be mounted on a x-y-z micromanipulator to allow for the independent control of the probe. Furthermore, a wide range of voltages (10 μV to 10 V) can be generated in LabView and transmitted to the probe through the DAQ. The amplitude of action potentials detected in the MEA studies range from 30 to 300 μV. Therefore, this system will allow to mimic pulses within these limits and beyond. We hope that by performing a thorough characterization on the effect that location and signal amplitude have on generating photocurrent data by SPCM, we will gain a better understanding of the capabilities and limitations of this unique biosensing technique.
98