3D-Printed Carbon Nanoneedle Electrodes for Dopamine Detection in Drosophila

Abstract

In vivo electrochemistry in small brain regions or synapses requires nanoelectrodes with long straight tips for submicron scale measurements. Nanoelectrodes can be fabricated using a Nanoscribe two-photon printer, but annealed tips curl if they are long and thin. We propose a new pulling-force strategy to fabricate a straight carbon nanoneedle structure. A micron-width bridge is printed between two blocks. The annealed structure shrinks during pyrolysis, and the blocks create a pulling force to form a long, thin, and straight carbon bridge. Parameterization study and COMSOL modeling indicate changes in the block size, bridge size and length affect the pulling force and bridge shrinkage. Electrodes were printed on niobium wires, insulated with aluminum oxide, and the bridge cut with focused ion beam (FIB) to expose the nanoneedle tip. Annealed needle diameters ranged from 400 nm to 5.25 μm and length varied from 50.5 μm to 146 μm. The electrochemical properties are similar to glassy carbon, with good performance for dopamine detection with fast-scan cyclic voltammetry. Nanoelectrodes enable biological applications, such as dopamine detection in a specific Drosophila brain region. Long and thin nanoneedles are generally useful for other applications such as cellular sensing, drug delivery, or gas sensing.

Graphical Abstract

University of Virginia and researcher Twitter username: Jill Venton (@jventon)

Introduction

The development of implantable microelectrodes has significantly benefited neuroscience by allowing real-time neurotransmitter detection in specific brain regions. Typical electrodes are made from carbon fibers with a diameter of 7 μm;^[1–3] however, new designs are needed for use in small brain regions or synapses to understand localized signaling. First, electrode size should be on the sub-micron scale for sensing of single vesicles or in small brain regions, such as Drosophila, which has brain regions that are less than 10 micrometers.^[2,4–6] Second, the electrode needs to have a long and thin tip for easy implantation with little damage. Third, the electrode material needs to have good electrochemical performance to ensure rapid sensing. Carbon is the traditional material for neurotransmitter sensing because carbon has good biocompatibility, excellent electrochemical performance, and a wide potential window.^[7–9] However, new methods are needed to reproducibly fabricate long, thin submicron carbon probes.

Recently, many groups have developed different approaches for making carbon nanoelectrodes for neurotransmitter detection and cellular recording. The Huang group and the Sombers group successfully etched carbon fibers into nanoelectrodes and applied them for neurotransmitter detection in synapses and single cells.^[10,11] Hegarty et al. designed carbon loaded microneedles for pH measurements with reduced damage during insertion.^[12,13] Nanopipettes with a nano-size cavity have been used in many studies, such as electrochemical sensing, imaging, and cell counting.^[14–17] The Robinson group designed a special microfluidic device to actuate flexible carbon nanotube fiber for electrophysiology.^[18] Xu et al. fabricated a scalable three-dimensional (3D) field-effect transistor (FET) array to characterize intracellular and intercellular signal conduction.^[19] The Lieber group developed U-shaped nanowire field-effect transistor arrays (U-NWFETs) to achieve intracellular recording.^[20] Our group fabricated carbon nanospike-coated nanoelectrodes by depositing carbon nanomaterials on metal wires with a nanotip.^[21] We also developed 3D-printed carbon electrodes made with photolithography by polymerizing and then carbonizing photoresist.^[22–24] The printed carbon structure has a nanotip, which is then cut by a focused ion beam to form a disk. Although these electrodes are nano-sized, they are too blunt for implantation in small brain regions. A long and thin needle-shaped probe would reduce the tissue damage.^[25–27]

3D-printing with laser lithography enables customized carbon micro- and nanoelectrode fabrication with a higher reproducibility. A structure is printed in a thin layer of photoresist by a focused laser using two-photon adsorption and the photoresist is then pyrolyzed to carbon with a glassy-carbon-like surface.^[25,28] Shrinkage is observed after pyrolysis because the decomposition of the polymer generates byproducts, including H₂O, CO, CH₂O, and CO₂.^[29,30] but the shape of the final structure is well retained.^[29,31] To build a needle shape, we printed a micron-diameter cylinder. However, curling of the ultra-small tip in nano-sized needles with a high aspect ratio occurred due to the asymmetric force in the needle structure during annealing.^[23] Similarly, del Campo and Greiner observed a curling of the printed cylinder structure when they fabricated a microcylinder array using SU-8 photoresist.^[32] Curled or bent sensors are difficult to implant into tissue, and could cause tissue damage and probe tip breakage. Thus, new methods are needed to fabricate long, straight nanostructures.

Here, we designed a novel pulling method to generate a straight, carbon nanoneedle structure without any curling. Using a Nanoscribe 2-photon lithography printer, a nanoneedle bridge was printed between two blocks. During pyrolysis, a pulling force is generated along the bridge when shrinking and the bridge is stretched to a long, thin, and straight nanoneedle structure. The end is cut off using focused ion beam (FIB) milling to make a carbon nanoneedle electrode that has good properties for dopamine detection with fast-scan cyclic voltammetry (FSCV). With the high aspect ratio, the carbon nanoneedle enables electrochemical sensing in Drosophila larva. Long nanoneedles could also be used potentially in drug delivery by piercing the cell membrane^[33,34] or be made in arrays for applications in microelectronic development, such as super capacitor for energy storage and gas sensors devices.^[35,36]

Results and Discussion

The goal of this study is to make long and thin nanoneedle electrodes. With the Nanoscribe 3D printer, polymer structures can be printed down to about 1 μm. A customized structure is printed in the photoresist by using the laser to trigger chemical cross-linking of the photoresist polymer, and then developing it to rinse away non-cross-linked photoresist. During annealing at a high temperature, pyrolysis occurs which carbonizes the structure and shrinks it.^[37] Shrinkage occurs during annealing and carbonization, which reduces final structure size by 30% to 50%.^[29,30] Thus, one advantage of this technique is that the final structure is smaller than the printer resolution. However, as Figure 1 shows, when the tip is printed long and thin with a high aspect ratio, the tip curls during pyrolysis due to gravity and slight asymmetries in printing. Curled tips cannot be implanted into tissue.

Figure 1.

A) The tip curls during annealing when it is thin and long. B) New design of a nanoelectrode. A bridge is printed between two blocks, which will pull it into a long, straight nanoneedle when annealed. C) SEM image of curled tip D) SEM of straight tip after FIB cutting

Thus, a new method was designed to keep the tip straight during pyrolysis using a pulling force between two blocks, as illustrated in Figure 1B. The bridge is printed between a block on a metal wire and an additional block, and it shrinks to a smaller width during annealing. The bridge will remain straight due to the pulling force of the block. This design solves the curling problem and enables the construction of a long, thin tip. Figure 1C and D present SEM images of curled tip and pulled, straight tip after FIB cutting.

Optimization of nanoneedles with different bridge diameters, bridge lengths and block sizes

Ultra-small structure construction is challenging because the printed polymer needs to be dense enough to avoid deformation of the final structure.^[38] Thus, we optimized the printing parameters (Fig. S1 and Table S1) and found for 1 μm bridges, the optimal slicing and hatching distance was 100 nm (or 0.1 μm), which is the practical minimum value for Nanoscribe. For larger structures, the density can be decreased to shorten the printing time; for example, slicing and hatching distances of 0.5 μm are sufficient for bridges wider than 4 μm.

Next, we performed a parameterization study to understand the effects of different parameters on the resulting 3D printed structures. Figure 2 and Table 1 show different parameters, comparing the effects of bridge width, bridge length, and block size on the final structure. Figure 2A presents printed designs for bridge diameters of 5, 10, 15 and 20 μm while Fig. 2B shows designs for bridge lengths of 30, 60, 90, and 120 μm and Fig. 2C block sizes of 25, 50, 75, and 100 μm. Figure 2D, E, and F show the actual structures after annealing, with bridge diameters indicated. Compared to the square shadow around the block structure, which is the initial printing trace, obvious shrinkage occurred after annealing. The needle-like bridge between the two blocks is stretched and pulled long and straight due to mechanical stress and mass loss during pyrolysis. For the bridge diameter (Fig. 2D), the smaller designed bridges result in smaller diameters, and the smallest printed bridge shrunk from about 5 μm to 1.5 μm. For the bridge lengths (Fig. 2E), diameter was kept constant at 10 μm and the shortest length led to a smaller diameter, as pulling force was greater for the shorter bridge. For block size (Fig. 2F), smaller blocks induced greater shrinkage and pulling force (i.e. increased length) and smaller diameter. For the larger blocks, there is also some non-ideality in the block shrinkage behavior, as the block appears to have liquefied and bulged out in the center during the annealing process.

Figure 2.

Block and bridge structures printed on a silicon wafer with varying bridge diameters, bridge lengths, and block sizes. A) Designs with different bridge diameters. B) Desgins with multiple bridge lengths. C) Designs with various block diameters. A to C are on the same scale. D) SEM image of annealed samples from the designed structures in A, varying bridge diameter. Bridge length, diameter, and block size are given. E) SEM image of annealed samples with varied bridge length (design in B). F) SEM image of annealed samples with different sizes (design in C). D to F are all on the same scale. G) Simulation of shrinkage after annealing for the various bridge widths (design In A) using external strain, 0.55. Bridge diameters and lengths are noted. H) Simulation of annealed structures with different bridge lengths (design in B) using external strain, 0.55. I) Simulation of annealed structures with different block sizes (design in C) by using external strains, 0.45, 0.50, 0.53, and 0.55 for 25, 50, 75, and 100 μm blocks, respectively. G to I are on the same scale.

Table 1.

Dimensions and shrinkage of the 3D printed structures

Designed Block Sizes (μm)	Designed Bridge Diameter (μm)	Designed Bridge Length (μm)	Experimental bridge diameter (μm)	Simulation bridge diameter (μm)	Experimental bridge diameter shrinkage ratio	Simulation bridge diameter shrinkage ratio
100	5.0	100	1.45	1.35	0.290	0.270
100	10.0	100	2.62	2.84	0.262	0.284
100	15.0	100	3.93	4.60	0.262	0.307
100	20.0	100	5.24	5.90	0.262	0.295
100	10.0	30	2.48	2.37	0.248	0.237
100	10.0	60	2.55	3.22	0.255	0.322
100	10.0	90	2.55	3.56	0.255	0.356
100	10.0	120	2.76	3.73	0.276	0.373
25	10.0	100	2.27	2.43	0.227	0.243
50	10.0	100	2.45	2.61	0.245	0.261
75	10.0	100	2.55	2.70	0.255	0.270
100	10.0	100	2.62	2.84	0.262	0.284

Table 1 lists printed bridge diameters, as well as linear shrinkage of the bridge. Bridge shrinakge is larger than for the smaller block because of the pulling force induced by the block shrinkage. We printed test structures four times and calculated relative standard deviations (RSDs) for the bridge diameter for each structure ranged from 5.25–7.5%. Thus, the printing and annealing methods show good reproducibility and produce similar structures with similar bridge diameters every time.

Simulation of the shrinkage behavior

To predict and visualize effects of the mass loss and concomitant volumetric shrinkage of the 3D-printed structures on narrowing of the bridge, we simulated the 3D model using COMSOL Multiphysics with the external strain subnode of the Solid Mechanics module. The external strain values in our models were determined by experimentally observed pyrolysis-induced decreases in the linear sizes of the printed blocks.

The designed structures used for modeling are shown in Figure 2A, B, and C and the COMSOL modeled structures after shrinking are shown in Figures 2G, H, I (detailed drawing of structure also in Fig. S2). In Fig. 2, the bottom planes of the block structure have no shrinkage because they were a fixed constraint, but the top of the block reduces in size due to the volumetric shrinkage. The lighter blue and green colors of the bridges indicate negative stress, so a pulling force was generated along the cylinder. The model simulates deformations in solid materials associated with their mass loss and concomitant volumetric shrinkage (negative strain). In the model, we assume the polymer is an elastic solid material, while its mass loss during pyrolysis is accounted for by isotropic strain. Fig. S3 shows a flow chart and Fig. S4 shows images of how we determined the external strain and Poisson’s ratio for the material from the block shrinkage. Fig. S5 and S6 show modeled effects of varying Poisson’s ratio and bridge length on the relative shrinkage and Table S2 lists the values used for the model.

Bridge diameter and shrinkage ratio calculations for Fig. 2 are listed in Table 1. The material is compressible, with a Poisson’s ratio smaller than 0.5 (the parameter for simulation models is 0.3).^[39] For the first two simulations in Fig. 2G and H, a constant external strain of 0.55 was an average value determined from blocks that did not suffer from the liquefaction. For the simulation in Fig. 2I, the external strain differs with the block size, because the smaller blocks shrink more. Thus, external strain varied from 0.45 to 0.55 (note that lower values are higher shrinkage). In general, the trends for the predicted bridge widths match the trends for the experimental results. The smallest final bridges result from smaller printed bridges, shorter bridges, and smaller blocks and induce the lowest background charging and dopamine Faradaic currents. Nanoelectrodes are a disk electrode, and the active surface area depends on bridge diameter that is exposed after FIB cutting. Thus, for nanoneedle development, we used smaller blocks and bridges. If longer bridges are required, the length was not as big a factor in controlling the bridge diameter, so bridges of 100 μm are feasible.

Nanoneedle construction by focused ion beam cutting

To design new nanoneedle electrodes, we used the knowledge from the parameterization study to optimize parameters to make nano-sized tips. After pyrolysis, a focused ion beam (FIB) cut the bridge to generate a long nanoneedle tip. Figure 3 shows a series of pyrolyzed structures with various widths. Figure 3A has the thinnest bridge, which is 330 nm wide (printed structure was 1 μm diameter, 50 μm length, and the block size was 25 μm). Thus, a nano-sized bridge is possible with printing a 1 μm diameter bridge. Figure 3B has a width of 1.1 μm (printed structure was 3 μm diameter, 50 μm length, and the block size was 25 μm), whereas the bridge in Figure 3C is 2.2 μm wide (printed structure was 5 μm, 50 μm length, and the block size was 25 μm). As shown in Figure 3, the bridges are still straight after they were cut, with smooth edges. There is a slight bending for the smallest diameter, but the tip is still straight enough to allow tissue insertion. The results indicate that FIB cutting is neat and precisely controlled and allows the length to be customized.

Figure 3.

The pyrolyzed carbon structure of the bridges with different widths after FIB cutting: A) 330 nm, B) 1.1 μm, C) 2.2 μm. The accelerating voltage of the FIB was 30 kV.

Construction of nanoneedle on metal wires

For electrode fabrication, the structure must be printed on a wire inert to dopamine and that could be etched to a small diameter to preventing major brain damage. Nb wires (with a 50 μm diameter) were chosen and then etched to a sharp 1 μm tip in diameter.^[21,40] The metal wire must be mounted to meet several requirements: First, the metal wire must be above the Si wafer to allow a suspended bridge that shrinks. Second, the anchors of the structure must be mounted to the Si substrate to generate a pulling force and survive the high temperature of annealing. Third, the adhesion between the printed structure and the metal wire must be strong.

The wire assembly and fabrication process are illustrated in Figure 4A. The metal wire was placed on a Si wafer and several blocks were 3D printed as anchors to mount and stabilize the wire. Pyrolyzed blocks for mounting Nb wire were squeezed by the tweezer and no carbon remained on the wire to influence the implantation of the nanoelectrode. The complex structure was printed in a bottom-up direction from the Si wafer, where the left part covers the metal tip, and the right block adheres to the wafer tightly. The left part serves as a base of the nanoneedle on the metal wire and the right part is removed to expose the needle tip after pyrolysis. Figure 4B and 4C show the final structure with a glassy carbon like surface after pyrolysis. Compared with the original cubic design, the pyrolyzed bulk structure has a curved surface due to liquefaction around the wire during annealing and surface tension during shrinkage. The bridge was pulled long and straight with a length of approximately 110 μm and diameter of approximately 600 nm. To ensure that blocks could stay on Nb wires and bridges are thin and long enough after annealing to be applied in the fly experiments, we chose the optimized parameters to print the structures (1 μm bridge diameter, 100 μm bridge length, and 50 μm block size).

Figure 4.

The printing procedure of the nanoneedle on a Nb wire. A) The illustration of the design. The whole structure was printed separately, including a base block printed on a silicon substrate and a complex structure printed on the Nb wire. B, C) The printed complex structure was pyrolyzed to carbon. Two bulk structures were connected by a thin and straight bridge in between. D) After depositing with Al₂O₃, the bridge was cut to a needle by FIB with 30 kV accelerating voltage. The needle has a diameter around 600 nm with a straight shape. The zoomed-in image was obtained at the cutting position on the tip.

After pyrolysis, the sample was covered with a thin film of Al₂O₃ (100 nm thickness) by atomic layer deposition to insulate the surface. EDS data (Figure S7, Table S3) indicates the aluminum oxide was uniformly coated on the pyrolyzed carbon. When the surface was fully coated with Al₂O₃, there was no characteristic dopamine oxidation current (Figure S11), which means that all active sites were blocked, and no dopamine could be adsorbed to the electrode surface. After FIB cutting of the bridge, Figure 4D shows the nanoneedle electrode with a straight and long tip. The enlarged image of the cross section indicates a nano-sized exposed carbon area surrounded by a layer of the non-conductive Al₂O₃.

The Nb wire with a carbon nanoneedle structure on the tip was then fabricated into a nanoelectrode for dopamine detection with fast-scan cyclic voltammetry (FSCV). Figure 5A shows a small background current (approximately 20 nA) due to the small sensing area. Although background currents are always larger than Faradaic currents, background currents are stable during the experiment, and background subtraction is performed to obtain the final CVs with FSCV. The nanoneedle electrode produces a typical dopamine CV shape (Figure 5B) with oxidation peak around 0.8 V and a reduction peak at −0.3 V. The current for 10 μM dopamine was 0.60 nA. To investigate dopamine adsorption on the electrode surface, we varied the scan rate and plotted the log peak current (i) vs. log scan rate (v) for the electrodes. In Figure 5C, electrochemical signals were plotted against scan rates and current is linear with scan rate. In Figure 5D, the fitted line also shows a linear correlation between the scan rate and the current and the slope of the log-log plot is close to 1. They both indicate an adsorption-controlled process on the nanoneedle electrode, because the current is proportional to the scan rate when the electrochemistry is governed by adsorption.^[41]

Figure 5.

The electrochemical performance of the carbon nanoneedle electrode for dopamine detection with fast-scan cyclic voltammetry. A dopamine waveform was used to obtain the signal (scan rate of 400 V/s, sweep from −0.4 V to 1.3 V at 10 Hz). A) Background current of the nanoelectrode. B) The background-subtracted CV for 10 μM dopamine. C) Current response with scan rate ranging from 100 V/s to 1000 V/s. D) log-log plot of scan rate test result. E) Dopamine concentration tests (1–500 μM). F) Dopamine linear range (1–50 μM). Error bar = SEM, n=4 electrodes.

In Figure 5E and F, Faradaic currents were plotted respectively against varying concentrations. For Figure 5E, as the concentration exceeds 50 μM, the electrode surface becomes saturated and the current plateaus due to all the dopamine adsorption sites being occupied, and the kinetics are then diffusion controlled. ^[5] To quantify the sensitivity of the dopamine detection on nanoneedle electrodes, we determined the slope of the line linear portion of the concentration range, from 1 to 50 μM. The sensitivity is 1.5*10⁻² nA/μM, with a LOD of approximately 0.56 μM. Figure S8 plots stability of the current for 4 hours, showing the current is stable with repeated measurements over the time frame of a typical biological experiment. The results show that the 3D printed nanoneedle electrodes have a good electrochemical performance for dopamine detection. With this novel fabrication method of the carbon nanoneedles, we produced disk electrodes but, in the future, a cylinder electrode could be fabricated by removing some of the aluminum oxide coating from the length of the tip.

To show that our carbon nanoneedle electrodes could be used in biological applications, we inserted them into Drosophila larvae ventral nerve cord (VNC) tissue to measure dopamine with FSCV. ^[42–44] Figure 6A shows a dissected-out VNC and GFP expression shows dopamine neuron clusters are found in the optic lobes and the lateral periphery near the neuromuscular junction. In Figure 6B, the zoomed in picture also shows dopamine neurons near the mid-line that will be targeted for electrode placement. ^[42] Figure 6C and 6D shows the nanoneedle electrode inserted near the mid-line dopamine neurons. The nanoneedle was inserted into the tissue without breaking. Only the exposed disk of nanoelectrodes, after FIB cutting, provided active sites for detection adsorbed dopamine molecules. To measure dopamine in tissue, a picospritzing capillary with 10 mmol dopamine was inserted into the tissue. ^[42–44] Dopamine is injected into the tissue, which is measured by the nanoelectrode. The oxidation peak is slightly shifted more to the right because of tissue adsorption to the electrode, which slows electron transfer. ^[45] Figure 6E shows a FSCV dopamine signal detected with the nanoneedle in tissue. The nanoneedle electrode will sample from fewer neurons or a single cluster in the larva VNC, compared to the CFME, because of its smaller size. As the nanoelectrode size is still bigger than synapses, dopamine electrochemical detection was performed in extracellular fluid. Dopamine currents increased at first and then decreased because of the reuptake by neurons. Fig. S9 shows the nanoneedle before and after insertion in the larval VNC and it does not break after tissue insertion. Fig. S10 shows additional data for replicates. CV shapes have small differences, which are caused by the ionic and pH change at the electrode surface during in the Drosophila brain. However, we could continuously obtain stable dopamine oxidation currents in the larval VNC. This nanoneedle electrode can withstand being inserted into biological tissue and will be useful for future electrochemical experiments.

Figure 6.

Carbon nanoneedle electrodes can be used for biological experiments. A. A th-Gal4/Cyo; UAS-mCD8-GFP cross shows GFP expression of dopamine (DA) neuron clusters in a Drosophila larva (fruit fly) ventral nerve cord (VNC). B. Zoomed in image shows dopamine neurons surround the mid-line of the VNC and will be targeted for electrode placement C. Nanoneedle electrode is inserted into the VNC next to a capillary filled with 10 mmol dopamine. Dopamine was injected into tissue are measured with the nanoneedle and FSCV. D. The zoomed in image of the nanoneedle inserted in the tissue. E. FSCV Dopamine signal using the nanoneedle electrode. A current vs time trace shows dopamine increasing and decreasing and the cyclic voltammogram confirms dopamine is detected.

While we used the nanoneedles for neurotransmitter sensing, the ability to make long, straight carbon nanoneedles could be useful for other fields. For example, long nanoneedles also could be used potentially in single cell analysis by piercing the cell membrane and doing sensing inside cells or even specific organelles. ^[33,34] In addition, a more hollow structure could be designed to allow specific drug delivery. 3D printing could also be used to make arrays of long nanoneedles for applications in microelectronics, such as supercapacitors for energy storage and gas sensors devices.^[35,36] Thus, this work presents a novel way to fabricate long, straight carbon nanoneedles that could be useful in a variety of applications and devices.

Conclusions

In this paper, we designed a new method to make long, straight carbon nanoelectrodes with 3D printing. By printing a bridge structure between two blocks, a long, thin, and straight carbon structure was generated by pulling force during pyrolysis. The simulation model shows that the pulling induces more strain on the nanoneedle, causing more shrinkage. To produce electrodes, structures were printed on metal wires and insulated with a thin film of Al₂O₃. After cutting with FIB, a straight, nanosized needle was fabricated with customizable width and length. The carbon nanoneedle electrodes show a reliable electrochemical performance for dopamine detection, even in Drosophila tissue. Considering the ultra-small and sharp tip of the needle, nanoelectrodes can be easily implanted in small organisms in future studies, including fruit fly brains or in a synapse between neurons.