Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2025 May 22;17(22):32764–32772. doi: 10.1021/acsami.4c22829

Dielectrophoresis-Enhanced Graphene Field-Effect Transistors for Nano-Analyte Sensing

Nezhueyotl Izquierdo 1, Ruixue Li 1, Peter R Christenson 1, Sang-Hyun Oh 1, Steven J Koester 1,*
PMCID: PMC12147070  PMID: 40401923

Abstract

Dielectrophoretic (DEP) sensing is an extremely important sensing modality that enables the rapid capture and detection of polarizable particles of nanoscale size. This makes it a versatile tool for applications in medical diagnostics, environmental monitoring, and materials science. Because DEP relies upon the creation of sharp electrode edges, its sensitivity is fundamentally limited by the electrode thickness. Graphene, with its monolayer thickness, enables scaling of the DEP force, allowing trapping of particles at graphene edges at ultralow voltages. However, to date, this enhanced trapping efficiency of graphene has not been translated into an effective sensing geometry. Here, we demonstrate the expansion of graphene DEP trapping capability into a graphene field effect transistor (GFET) geometry that allows the trapped particles to be electrically detected. This four-terminal multifunctional hybrid device structure operates in three distinct modes: DEP, GFET, and DEP-GFET. By segmenting the channel of the GFET into multiple parallel channels, greatly increased density of particle trapping is demonstrated using fluorescence microscopy analysis. We show further enhancement of the trapping efficiency using engineered “nanosites,” which are holes in the graphene with size on the order of 200–300 nm. Scanning electron microscope analysis of immobilized gold nanoparticles (AuNPs) shows trapping efficiency >90% for properly engineered nanosites. We also demonstrate real-time, rapid electrical sensing of AuNPs, with >2% current change occurring in 4.1 s, as well as rapid sensing of a variety of biomolecule-coated nanoparticles. This work shows that graphene DEP is an effective platform for nanoparticle and biomolecule sensing that overcomes diffusion-limited and Brownian motion-based interactions.

Keywords: graphene, GFET, dielectrophoresis, sensor, trapping, DNA, polystyrene

graphic file with name am4c22829_0008.jpg

graphic file with name am4c22829_0007.jpg

1. Introduction

As a two-dimensional (2D) material, graphene possesses favorable properties as a transducer channel material for biosensors, including high mobility, chemical and mechanical stability, biocompatibility, and ease of surface functionalization. In a liquid medium, graphene field-effect transistor (GFET) sensors have the potential for fast response due to their strong surface sensitivity. However, traditional GFET sensors often rely upon diffusion and Brownian motion to produce a recognition event. Therefore, the detection of target particles, particularly for solutions approaching the limit of detection (LOD), can still require a long incubation time in order to observe a measurable response. − The integration of dielectrophoresis (DEP) within the device structure could offer a promising method to speed up the response time of GFET sensors by overcoming the Brownian motion to concentrate target particles locally through an attractive force. DEP is a well-established technique and the immobilization of various biomolecules and particles, including DNA, enzymes, polystyrene (PS) beads, , and gold nanoparticles (AuNPs) have been demonstrated. Furthermore, the use of DEP to enhance sensitivity and reduce response times is a common practice in many sensor platforms, as reviewed by Henriksson et al., who described how DEP improves mass transfer, a key bottleneck in traditional biosensor geometries.

Fast response time is a critical requirement across a broad range of biosensing platforms. Electrochemical sensors, optical sensors such as surface plasmon resonance (SPR), and impedance-based biosensors all rely on the ability to detect binding events or chemical changes in real time. These technologies highlight the general importance of rapid signal transduction for applications in clinical diagnostics, environmental monitoring, and point-of-care testing. While this work focuses on GFETs, the emphasis on achieving fast, sensitive detection is relevant across many biosensing approaches. As an example of using DEP to enhanced response time, Sharma et al., integrated DEP into a photonic biosensor, improving the response time from 60 to 1 min, with improved sensitivity. In another example, Szymborski et al., developed a DEP-based surface-enhanced Raman scattering (SERS) sensor for detecting cancer cells in microfluidic chips. DEP was shown to improve the sensor LOD down to 20 cells/mL, with a response time of only 7 min.

Despite the benefit of DEP for improving sensor operation, its use in conjunction with GFET sensors is very limited in the literature. Most notably, Kumar et al. used DEP to enhance GFET sensor operation and detect antibiotic-resistant bacteria at the single-cell level. Their DEP-assisted GFETs achieved a response time of approximately 5 min, a 9-fold reduction in response time, demonstrating the effectiveness of DEP in accelerating sensor performance. However, the approach used by Kumar et al. was limited in that the DEP and sensing were conducted sequentially, rather than simultaneously, by applying a series of DC biases to the liquid gate itself, and then measuring the FET. While straightforward, this sequential approach limited the response time improvement that could be achieved.

To overcome the limitations of previous DEP-GFET sensors, the inherent properties of graphene can be utilized. In particular, Barik et al. showed that the DEP force is especially strong at the atomically thin edges of graphene, and is capable of efficiently trapping both PS beads and biologically relevant species such as DNA at ultralow voltages. However, while that report demonstrated the potential of graphene for use in DEP trapping, the device structure was not well-suited for use in sensors, since the trapping only occurred only to the periphery of the device. Furthermore, a two-terminal “varactor” structure , was used, which is not ideal for biosensing using DEP, since the AC bias would have to be applied to the same buried electrode needed to interrogate the sensor, making real-time readout difficult.

In this work, we overcome the limits of this prior work by proposing and demonstrating a novel four-terminal DEP-GFET geometry. This device separates the buried and liquid gate electrodes, allowing the sensor readout and DEP force to be applied independently and potentially, simultaneously. It also uses a GFET geometry for sensor readout, where the graphene can be patterned into smaller segments that provides high sensitivity by providing a large area of the sensor that interacts with the trapped particles. In this way, it meets the key metrics needed for an effective DEP-enhanced sensor: high trapping efficiency, high sensitivity, and fast response time.

2. Experimental Section

Device Structure and Fabrication

Our novel sensor design is shown in Figure a, where the complete fabrication process flow is provided in the Supporting Information (Figure S1). The device structure consists of four terminals: a buried local back gate (LBG) is used for controlling DEP trapping, and a liquid gate (LG) electrode modulates the GFET sensor. The source and drain electrodes are located on opposite sides of the LBG. A top view of a completed fabricated device is shown in Figure b. A critical aspect of our design is that the graphene channel between source and drain consists of several parallel strips, and the LBG is broken into three separate fingers. In this design, each graphene strip was 25-μm wide. These dimensions were chosen to increase the number of edge trapping sides, compared a conventional design with a single wide sheet, while still ensuring high fabrication yield. While these dimensions can be further optimized, the design trapping density was found to be sufficient for sensor operation, as will be shown later in the manuscript.

1.

1

Open in a new tab

(a) Schematic diagram of DEP-GFET device, and (b) color-enhanced top-down optical micrograph of completed DEP-GFET device. The width of each graphene channel (green) is 25 μm while the designed width of each local back gate electrode (pink) is 7.5 μm. The total designed source-to-drain spacing is 70 μm.

Modes of Operation

The four-terminal geometry allows the sensor to be operated in one of three modes: GFET only, DEP only, and DEP-GFET, as illustrated in Figure . In this section, we briefly describe these different modes of operation, where more details are provided in the Supporting Information. In the “GFET” sensing mode, DEP is not used and the device operates as a conventional GFET sensor. In this configuration (Figure a), the sensor tracks the Dirac point, V Dirac, and/or the drain current, I D, response to the analyte concentration. In the second “DEP” mode of operation (Figure b), DEP is used to trap particles at the graphene edges, and fluorescence microscopy is used to identify trapping at the graphene edges. A final sensing mode is the “DEP-GFET” configuration, depicted in Figure c, where the two sensing configurations are combined. This mode has the advantage of separating DEP and sensing sweeps, potentially allowing the trapping to DEP bias to be applied during sensing. Additional details on the specific sensing modes enabled by this configuration are provided below. The measurement setup is also described in Figure S2.

2.

2

Open in a new tab

Diagram of different operating regimes for sensors. (a) GFET mode, (b) DEP mode, and (c) DEP-GFET mode.

3. Results and Discussion

Fluorescence Measurements of PS Beads in DEP Mode

As an initial step to characterize the DEP trapping efficiency, we analyzed the trapping of PS beads to test the device operation with fluorescence (FL) microscopy (Figure ). PS beads were used to establish the baseline DEP trapping behavior since the strong fluorescence signal facilitates optical observation of trapping at graphene edges, and pulsed PS bead trapping can be performed with minimal permanent trapping. As depicted schematically in Figure a and b, a localized increase in the concentration of fluorophore-decorated particles resulting from the DEP force should produce FL intensity bands at the trap sites located at the edges of the graphene. Here, fluorophore-decorated PS particles (42 nm, Bangs Laboratories, Inc.) were selected to evaluate the trapping capabilities for the device structure. In all measurements, to reduce any influence of the GFET bias on DEP, we kept the liquid gate at 0 V during trapping. In this way, particle movement was driven purely by DEP and not by any DC electrostatic attraction to the gate. The experiments are performed in 0.01X phosphate buffered saline (PBS) solution to minimize Joule heating and ionic shielding effects. For the DEP-off state (1 mVPP, 800 kHz) trap sites are vacant, evidenced by the FL image and the FL 3D surface plot map (Figure c). In comparison, the DEP-on state (1.3 VPP, 800 kHz) results in essentially full trap site occupancy (∼99%) (Figure d). Thus, the multichannel graphene array geometry is effective, with the trap site length increased by 5-fold compared to the situation that would have occurred without the segmented graphene channel. An FL intensity line profile along the length of the LBG electrode is provided in Figure e. A comparison of the DEP-off and DEP-on states shows an increase in the FL intensity at all ten graphene channel edge positions. In Figure f, three sequential DEP (on–off) pulses are measured by monitoring the mean FL intensity at all trap sites (SI Video 1). After baseline correction, the maximum FL intensity value during DEP-on is 55× the calculated baseline (DEP-off) standard deviation, and the FL intensity increases to 50% of the maximum fluorescence in only 4.5 s, and no permanent trapping of the PS beads is observed. We also evaluated the effect of increasing V PP on the FL intensity. A series of on–off DEP pulses with sequential 100 mVPP increases was performed and the results are presented in Figure g. The normalized FL intensity increases in response to increasing the trapping bias, as expected. Additional FL trapping results are shown in SI Video 2.

3.

3

Open in a new tab

(a) Schematic depiction of DEP-off and DEP-on states. (b) Graphic depiction of intended nanoparticle trapping mechanism at the graphene edges. (c–d) Fluorescent images and 3D intensity maps of polystyrene (PS) bead DEP trapping in both the (c) DEP-off and (d) DEP-on state. (e) Plot of fluorescence intensity vs distance across the graphene sensor strip with DEP off (black) and DEP on (green). The data is extracted from results shown in Figure c and d. Trapping is isolated to engineered trap sites located at the graphene-edge and local back gate intersection. (f) Tracking of fluorescence intensity (black) at the trap sites shows a reproducible baseline and trapping response for three sequential DEP pulses (red). (g) Filtered (Savitzky–Golay) data for trapping intensity (black line) at the trap sites shows a response to increasing DEP voltage (V PP) (blue).

The fluorescence data clearly show that diffusion alone results in minimal analyte accumulation at the graphene surface, evidenced by a very weak fluorescence response. This indicates that, under natural diffusion, target analytes do not reach the graphene surface in sufficient concentration to produce a strong fluorescence signal. However, upon applying DEP, a significant fluorescence signal increase is observed, demonstrating a rapid and targeted concentration of analytes at the sensor surface. This sharp increase confirms that DEP effectively improves the limitations of passive diffusion, actively driving analytes to the graphene and enabling a much stronger fluorescence response.

Fluorescence and SEM of AuNPs in DEP Mode

To precisely count the number of trapped particles and determine the precise location of the graphene participating in trapping along the channel edge, we performed additional experiments using gold nanoparticles (AuNPs). AuNPs can be imaged in a scanning electron microscope (SEM), yet are also biologically relevant with applications in biomedicine and sensing. The AuNPs had 150 nm diameter and were trapped using DEP at a frequency of 1.5 MHz and a LBG bias of 2 VPP. In this regime, permanent trapping occurred, meaning that after removing the DEP excitation, the particles remained attached to the graphene edges through van der Waals forces, which allowed for subsequent SEM imaging. In Figure a, a comparison of DEP trapping (FL) and post-DEP immobilization (BF) is shown. SEM characterization (Figure b and c) of immobilized AuNPs provides a straightforward method to evaluate trap sites with high spatial resolution and allow particle counting and trapping morphology analysis. The AuNP chains remained attached to graphene without any external bias, implying a nonelectrostatic adhesion mechanism such as van der Waals forces. However, the precise attachment mechanism is not particularly relevant, since in practical applications, lower DEP bias would be used so that permanent attachment does not occur.

4.

4

Open in a new tab

Designing nanosite trapping. (a) Fluorescence (FL) and bright field (BF) images of DEP trapping of AuNPs showing post-DEP particle immobilization above a threshold value of V PP. (b–c) False color SEM images of trapping locations of (b) pristine and (c) defective graphene channels. (d) FL images before and after DEP showing AuNP trapping at electron-beam-lithography-defined nanosite. (e) Schematic diagram of nanosite design spelling out the letters U-M-N. (f) False color SEM images of AuNP trapping at broad nanosite edges. (g) COMSOL simulation of three neighboring nanosites confirming the presence of strong electric field gradients. (h) FL line profile analysis of nanosite (red) trapping along the LBG shows a substantial increase in FL across the graphene channel width compared to edge trapping (black). (i) Comparison of FL immobilization for different nanosite dimensions H L = 100, 300, and 500 nm. The immobilization of particles per hole increases as H L increases, thus improving the immobilization efficiency.

From the SEM images, it is clear that AuNPs form a linear chain at the engineered trap site locations along the exposed graphene edges. In Figure b, approximately 60 AuNPs are seen to be trapped along the ∼ 7.5 μm trap site length, and the strong clustering just at the graphene edge is a good indication of the locality of the DEP force. However, while the segmentation of graphene enhances the DEP interaction area of the graphene sensor, further improvement is needed. Therefore, we evaluated the use of electron-beam lithography to create high-density nanoscale trap sites or “nanosites” within the graphene channel, with the intent to have a similar trapping effect as metallic nanopores used in previous work, , except with greatly simplified fabrication.

The nanosites consisted of etched “holes” in the graphene created using a short oxygen plasma, where the fabrication details are described in the Supporting Information. As shown from the FL microscopy images in Figure d and SI Video 3, trapping using nanosites spelling out the letters U-M-N were used, was clearly effective. An example schematic of the complex nanosite pattern design is presented in Figure e, and the specific location of the trapping is shown in Figure f. The regions between the nanosites are less favorable for trapping compared to the edges, which provide direct access to the wider graphene regions. This can be traced to series resistance effects from the graphene, as the AuNPs prefer to trap at sites where the resistance relative to the contact electrode is minimized. COMSOL simulations (Figure g) further confirm that the nanosites also have a strong electric field gradient which accounts for their excellent trapping behavior.

We also investigated the effect of the nanosite dimensions. Here, the hole spacing, H S, was fixed while the hole length, H L, was varied. We observed that increasing H L was found to increase the number of trapped AuNPs, resulting in an overall increase in the FL intensity. A line profile comparison of edge trapping and edge plus nanosite trapping shows a significant boost in normalized FL signal within the graphene channel (Figure h). The permanent trapping occupancy (PTO), defined in the Supporting Information, was measured for nanosites with H L = 100, 300, and 500 nm, and it was found that PTO ≥ 90% for H L > 300 nm, showing that the nanosites must be larger than the AuNPs in order to trap effectively. Figure i further shows that more particles are trapped per nanosite as H L is increased. Additional analysis of the trapping statistics is shown in Figure S3. We note that the nanosite size, shape, and density will likely need to be optimized depending upon the target particle size and trapping properties. Furthermore, depending on the application, the DEP conditions could be adjusted if reversible trapping is required, but also that the permanent capture could still be quite useful for creating functionalized surfaces with AuNPs.

DEP-GFET Sensing of AuNPs

In the next aspect of our work, we performed measurements in the DEP-GFET sensing mode, where the GFET and DEP measurements are combined. While the nanosite patterning (Figure ) was explored to gauge the maximum achievable trapping efficiency, the DEP-GFET sensing measurements were performed on devices without nanosite holes, using the original multistrip design depicted in Figure b. To perform these measurements in a controllable manner, care must be taken to avoid measurement artifacts. To highlight this challenge, we measured I D-V G curves in buffer solution with no particles while applying the DEP signal, to understand the effect on the device characteristics. Since the DEP bias is an AC signal, its main effect is to smear the I D-V G characteristics near the Dirac point, an effect that increases with increasing V PP. These results are shown in Figures a and b and indicate that while the current modulation is reduced, a Dirac point can still be observed even at V PP = 2.5 V. The smearing effect is also found to be weaker at higher frequencies, primarily due to RC delay effects, which weaken the effect DEP signal that reaches the gate electrode (Figure c and d). However, to minimize interference effects between gate electrodes, during sensing the DEP signal was never applied to the LBG at the same time as the V G sweep to the LG.

5.

5

Open in a new tab

(a) I D–V G plot for a typical DEP-GFET device showing the effect of increasing V pp on the GFET characteristics. The results show that a smearing effect occurs, but that the Dirac point remains observable, even up to V pp = 2.5 V. (b) Plot showing I D vs time for same data in (a). (c) I D–V G plot showing effect of DEP frequency on GFET characteristics. Here V pp = 0.5 V. The results show that lower frequencies have a strong effect on the characteristics due to RC time delays which reduce the actual voltage that reaches the active device. (d) Plot showing I D vs time for same data in (c).

Instead, we established a three-step DEP-GFET sensing protocol, and this is depicted in Figure a and described in detail in the Supporting Information. In brief, steps 1 and 3 are used to measure the V Dirac shift before and after applying DEP as an indirect sensing method. However, this method is only responsive to immobilized particles. Alternatively, in step 2, the response to the DEP excitation can be measured by monitoring I D, allowing for real-time sensing.

6.

6

Open in a new tab

Sensing measurement protocol for DEP-GFETs. (a) Measurement sequence for Dirac voltage shift protocol, (b–c) results of Dirac voltage shift protocol, showing measurement after three sequential 2 min DEP pulses for (b) a control sample, and (c) a sample of 100 pM AuNPs produced a −40 mV shift. (d) Dirac voltage shift after 2, 4, and 6 min of applied DEP for control and 100 pM AuNP sample. (e) Results for the same sample in (d), except measured using the continuous I D-sensing mode, showing the current changes by 2% in only 4.1 s. (f) Results showing current shift for PBS control, ds-DNA (1 nM), streptavidin-coated spheres (1 pM), and SARS COV2 virus-like particles (VLPs) (5 fM).

Our initial measurements in DEP-GFET mode used fluorophore-decorated-AuNPs (100 pM, AuNP in 0.01X PBS) to evaluate the sensing protocol, and the results were compared with 0.01X PBS controls. The first sensing method utilized V Dirac tracking. A comparison of the relative V Dirac shift, after three sequential 2 min DEP pulses, shows that a response only occurs in the presence of the active sample. Specifically, after an aggregate exposure of 6 min, the control sample produced only a −2 mV shift in V Dirac (Figure b), while the sample of 100 pM AuNPs produced a −40 mV shift (Figure c). This shift relative to the control is consistent over multiple trials on the same sample and the statistical results are shown in Figure d.

A second superior sensing method was developed to monitor I D during application of the DEP force. Measuring the sensor by monitoring I D allows the response to be observed nearly instantaneously, and much faster than the V Dirac mode, which requires a baseline and response transfer curve measurement, reducing the granularity of the time response, adding several minutes to the total sensor response time beyond that needed for incubation. Using this measurement configuration, the DEP-I D response was measured for AuNPs and the results are shown in Figure e. In this mode, I D was kept constant during DEP application in order to minimize crosstalk between the two gate electrodes. Here, DEP-I D sensing of AuNPs is shown using both the electrical (red) and FL (black) response. The current response is almost immediate and changes by 2% in only 4.1 s, while the FL response takes much more time to build up, highlighting the benefit of the DEP-I D sensing mode. Although Figure f shows a continuously rising fluorescence signal due to ongoing particle accumulation during short DEP pulses, the key electrical response in the graphene occurs very rapidly (within tens of seconds) and quickly reaches a new steady state, as evidenced in Figures e and f. These results show the power of the real-time DEP sensing method. While traditional transfer curve measurements (e.g., Dirac point shifts) can capture the steady-state electrical characteristics of a GFET sensor, they are less effective at revealing the rapid dynamics of dielectrophoretic (DEP) trapping. Our real-time current monitoring during the DEP trapping protocol provides a more sensitive and immediate measure of particle capture.

DEP-GFET Sensing of Biomolecules

In the final aspect of our work, we used the DEP-GFET sensing scheme (in I D mode) to detect various biologically relevant targets, including double-stranded DNA (ds-DNA), PS latex-beads coated with streptavidin protein, and a virus-like-particle (VLP). In all measurements described below and shown in Figure f, I D was measured for 10 min and normalized to the starting DEP-I D value. We also performed several independent trials for each analyte to ensure the observed responses are consistent. For the control measurements, a baseline in 0.01X PBS buffer was established. Here, the current was virtually unchanged (I D increased by 2 ± 5% over 7 trials over 10 min). Details of the particle preparation and measurement conditions are provided in the Supporting Information. As shown in Figure f, ds-DNA produced an I D change of −24 ± 4% (over 7 trials), while the streptavidin-coated PS beads produced an average I D change of −44% ± 6% (over 4 trials). Finally, the VLPs produced the largest change of −51 ± 12% (over 6 trials). FL imaging of VLP trapping is also shown in Figure S4, where it can be seen that VLPs are isolated to the DEP trapping regions, as expected. The fact that the control (buffer-only) experiments showed a very small drift underscores that these changes are a result of the analyte exposure, and not associated with noise or device drift.

For all the measurements above, the reproducibility of the response was tested before and after DEP, to ensure that the change in current was not due to sample degradation or other nonsensing-related effect. Multiple DEP pulses in succession were analyzed to assess the reproducibility of the response, as well as potential graphene channel degradation. The maximum DEP-I D does not decrease after sequential DEP pulses, implying that graphene is not damaged during the application of DEP. Furthermore, SEM images of graphene-edge trap sites (Figure b) and/or nanosites (Figure f) do not show graphene damage provided that the DEP excitation amplitude remained in the range of 0–2.5 VPP.

The results demonstrate that the observed GFET electrical response is due to DEP-enhanced analyte accumulation at the graphene surface. The DEP-GFETs are already incubated in the target analyte solution when the DEP force is applied. Therefore, this method prevents false responses arising from adding the target solution which can change the ionic strength. The measurements before and after DEP-force are applied were taken in a consistent solution. Furthermore, for each sample conditionStreptavidin, VLP, and DNAwe fitted an exponential decay trend to the I D data over the first 30 s to model early response behavior under DEP influence. Similar to the AuNP results described above, from these fitted trends, we identified the earliest time at which each condition’s current reached the 2% threshold. This 2% threshold was chosen because it exceeds the noise level and represents a clearly detectable change. This provides a metric for early detection capability. While the overall signal continues to grow beyond 2%, it typically stabilizes to a steady value after roughly 60 s, which can be considered the full settling time of the sensor. Streptavidin (10 pM), VLP (5 fM), and ds-DNA (1 nM) each demonstrated rapid response, reaching the 2% threshold within 3.9, 9.4, and 14.5 s, respectively, indicating that our DEP-GFET concept is effective for a wide range of targets in attracting analytes to the sensor surface. Complete data showing the individual response curves are provided in Figure S5. These DEP-assisted measurements compare favorably with the even the best response times observed in traditional GFET biosensors upon analyte addition. −

We believe the different response behavior between AuNPs and PS-coated particles arises from the stronger surface binding of AuNP due to their stronger polarization compared to PS particles. This leads to a more rapid saturation behavior in the electrical response, as the particles occupy the sites and then do not readily debind. However, PS particles have greater surface mobility where attachment can occur through a steady state balance between the surface and liquid concentration, where an individual particle may attach and detach many times throughout the sensing process. This is reflected in the longer time frame for saturation of the signal in PS-based sensing.

In addition to the response time, the different frequency-dependent trapping behavior is a result of the different dielectric properties of PS beads and AuNPs. PS beads, being dielectric particles, experience DEP forces primarily based on their permittivity contrast with the surrounding medium, typically resulting in positive DEP at lower frequencies where the particle permittivity exceeds that of the medium. In contrast, AuNPs, with their high conductivity and metallic nature, exhibit much stronger polarization and can display positive DEP at significantly higher frequencies due to their free electron response and high effective permittivity. These intrinsic material differences lead to distinct trapping behaviors: PS beads tend to show optimal trapping at lower frequencies and moderate Vpp levels, whereas AuNPs require higher frequency operation and show stronger DEP forces at comparable voltages. These differences were experimentally observed on our device, with optimized frequency windows and voltage thresholds differing for each particle type.

Limitations and Methods for Improvement

While we have demonstrated the DEP-GFET operation for a variety of nanoparticles, a current limitation of our work is that the interaction is largely nonspecific. It is well-established that GFET-based biosensors can provide selectivity by incorporating surface functionalization. − Therefore, a next step would be to determine the degree to which surface functionalization can be combined with DEP to enable selectivity. Graphene functionalization achieved by strong π-π interactions are unlikely to be disrupted by the DEP force.

DEP could also be used in other ways to impart selectivity. For instance, while we focused on sensing target analytes trapped by positive DEP, negative DEP could provide a mechanism to reduce nonspecific adsorption, a persistent challenge for all sensor technologies. By removing unwanted nonspecific interactions at the surface of the graphene channel, the engineered probe-target response is isolated, improving the specificity and selectivity. Creating an array of multiplexed DEP-GFET devices working in unison could also provide pseudoselectivity simply by operating individual sensors at different DEP trapping frequencies. Another interesting application of our device would be to use DEP to perform surface functionalization. Several groups have published results where AuNPs on graphene were used as functionalization elements. However, such techniques can damage the graphene, produce nonuniform coverage, and require long incubation times (e.g., 30 min). , Therefore, our DEP technique could be used as a simpler means of AuNP functionalization, where the particle position can be controlled precisely on a rapid time-scale.

Finally, in this proof-of-concept work, a detailed analysis of sensor response vs analyte concentration was not performed for each target. We demonstrated detection at one concentration for each analyte to illustrate versatility and rapid response. Future studies will focus on calibration of the DEP-GFET sensor, determining its sensitivity and dynamic range for specific targets. We do note that the large response at pM levels suggests we have headroom to detect even lower concentrations if needed.

4. Conclusion

In conclusion, we have demonstrated a multimodal four-terminal DEP-GFET device design that can effectively trap and sense a variety of nanoscale particles. Using FL microscopy, we provided evidence of improved DEP efficiency at engineered trap sites, located at the graphene-edge intersections with a LBG electrode. Immobilization of AuNPs at the trap sites further provided a technique to determine trapping location with precision, number of particles, and trapping morphology. Exposed graphene edges can be significantly increased by nanosite patterning to form complex and dense nanosite arrays of different dimensions. Lastly, a current sensing scheme was developed to provide real-time DEP-I D sensor data. A collection of samples was investigated which demonstrate the device’s usefulness in sensing biological analytes, such as n-gene dsDNA, streptavidin-coated PS beads, and SARS-CoV2 VLPs. The response demonstrated rapid kinetics and significantly greater magnitude compared to the control drift signal. The demonstrated DEP-GFET technology provides a simple yet powerful way to concentrate target species at the surface of a GFET sensor without relying upon diffusion-limited transport, which requires a long sensor response time.

Supplementary Material

am4c22829_si_001.pdf (704.9KB, pdf)
am4c22829_si_002.zip (24.6MB, zip)

Acknowledgments

This work was supported primarily by Grip Molecular Technologies. Device fabrication was performed at the Minnesota Nano Center at the University of Minnesota, which receives partial support from the National Science Foundation (NSF) through the National Nanotechnology Coordinated Infrastructure (NNCI) under Award No. ECCS-2025124. Portions of this work were also carried out in the University of Minnesota Characterization Facility, which receives capital equipment funding from the University of Minnesota MRSEC under NSF Award No. DMR-2011401.

The data used in this paper are available from the corresponding authors upon reasonable request.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsami.4c22829.

  • Details of the graphene transfer, device fabrication, measurement setup and DEP measurement parameter determination, COMSOL simulations, additional sensing details and data, and SEMs (PDF)

  • Videos of particle trapping (ZIP)

‡.

N.I. and R.L. contributed equally.

The authors declare the following competing financial interest(s): Two of the authors, S.J.K. and S.H.O., serve on the Scientific Advisory Board of Grip Molecular Technologies Inc., and have several patents relating to the graphene sensor technology. S.J.K. also holds an equity interest in Grip Molecular Technologies Inc.. The University of Minnesota (UMN) and the inventors are entitled to standard royalties should licensing revenue be generated from these inventions. These interests have been reviewed and managed by UMN in accordance with its Conflict of Interest policies.

References

  1. Shahdeo D., Roberts A., Abbineni N., Gandhi S.. Graphene Based Sensors. Compr. Anal. Chem. 2020;91:175–199. doi: 10.1016/bs.coac.2020.08.007. [DOI] [Google Scholar]
  2. Hwang E. H., Adam S., Das Sarma S.. Carrier Transport in Two-Dimensional Graphene Layers. Phys. Rev. Lett. 2007;98:2–5. doi: 10.1103/PhysRevLett.98.186806. [DOI] [PubMed] [Google Scholar]
  3. Fernandes E., Cabral P. D., Campos R., Machado G. Jr., Cerqueira M. F., Sousa C., Freitas P. P., Borme J., Petrovykh D. Y., Alpuim P.. Functionalization of Single-Layer Graphene for Immunoassays. Appl. Surf. Sci. 2019;480:709–716. doi: 10.1016/j.apsusc.2019.03.004. [DOI] [Google Scholar]
  4. Fu W., Jiang L., van Geest E. P., Lima L. M. C., Schneider G. F.. Sensing at the Surface of Graphene Field-Effect Transistors. Adv. Mater. 2017;29:1603610. doi: 10.1002/adma.201603610. [DOI] [PubMed] [Google Scholar]
  5. Vishnubhotla R., Sriram A., Dickens O. O., Mandyam S. V., Ping J., Adu-Beng E., Johnson A. T. C.. Attomolar Detection of ssDNA Without Amplification and Capture of Long Target Sequences With Graphene Biosensors. IEEE Sensors J. 2020;20:5720–5724. doi: 10.1109/JSEN.2020.2973949. [DOI] [Google Scholar]
  6. Xu L., Ramadan S., Rosa B. G., Zhang Y., Yin T., Torres E., Shaforost O., Panagiotopoulos A., Li B., Kerherve G., Kim D. K., Mattevi C., Jiao L. R., Petrov P. K., Klein N.. On-Chip Integrated Graphene Aptasensor with Portable Readout for Fast and Label-Free COVID-19 Detection in Virus Transport Medium. Sens Diagn. 2022;1:719–730. doi: 10.1039/D2SD00076H. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Yin T., Xu L., Gil B., Merali N., Sokolikova M. S., Gaboriau D. C. A., Liu D. S. K., Mustafa A. N. M., Alodan S., Chen M., Txoperena O.. et al. Graphene Sensor Arrays for Rapid and Accurate Detection of Pancreatic Cancer Exosomes in Patients’ Blood Plasma. ACS Nano. 2023;17:14619–14631. doi: 10.1021/acsnano.3c01812. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Lerner M. B., Matsunaga F., Han G. H., Hong S. J., Xi J., Crook A., Perez-Aguilar J. M., Park Y. W., Saven J. G., Liu R., Johnson A. T. C.. Scalable Production of Highly Sensitive Nanosensors Based on Graphene Functionalized with a Designed G Protein-Coupled Receptor. Nano Lett. 2014;14:2709–2714. doi: 10.1021/nl5006349. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Pethig, R. Dielectrophoresis: Theory, Methodology and Biological Applications; Wiley, 2017. 10.1002/9781118671443. [DOI] [Google Scholar]
  10. Pohl, H. A. Dielectrophoresis: The Behavior of Neutral Matter in Nonuniform Electric Fields; Cambridge University Press: 1978. [Google Scholar]
  11. Tuukkanen S., Toppari J. J., Kuzyk A., Hirviniemi L., Hytönen V. P., Ihalainen T., Törmä P.. Carbon Nanotubes as Electrodes for Dielectrophoresis of DNA. Nano Lett. 2006;6:1339–1343. doi: 10.1021/nl060771m. [DOI] [PubMed] [Google Scholar]
  12. Laux E. M., Kaletta U. C., Bier F. F., Wenger C., Hölzel R.. Functionality of Dielectrophoretically Immobilized Enzyme Molecules. Electrophoresis. 2014;35:459–466. doi: 10.1002/elps.201300447. [DOI] [PubMed] [Google Scholar]
  13. Barik A., Chen X., Oh S.-H.. Ultralow-Power Electronic Trapping of Nanoparticles with Sub-10 nm Gold Nanogap Electrodes. Nano Lett. 2016;16:6317–6324. doi: 10.1021/acs.nanolett.6b02690. [DOI] [PubMed] [Google Scholar]
  14. Chen Q., Yuan Y. J.. A Review of Polystyrene Bead Manipulation by Dielectrophoresis. RSC Adv. 2019;9:4963–4981. doi: 10.1039/C8RA09017C. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Fu K., Chen S., Zhao J., Willis B. G.. Dielectrophoretic Assembly of Gold Nanoparticles in Nanoscale Junctions for Rapid. Miniature Chemiresistor Vapor Sensors. ACS Sensors. 2016;1:444–450. doi: 10.1021/acssensors.6b00041. [DOI] [Google Scholar]
  16. Henriksson A., Neubauer P., Birkholz M.. Dielectrophoresis: An Approach to Increase Sensitivity, Reduce Response Time and to Suppress Nonspecific Binding in Biosensors? Biosensors. 2022;12:784. doi: 10.3390/bios12100784. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Monošík R., Stred’anský M., Šturdík E.. Application of Electrochemical Biosensors in Clinical Diagnosis. J. Clin. Lab Analysis. 2012;26:22–34. doi: 10.1002/jcla.20500. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Špringer T., Bocková M., Slabý J., Sohrabi F., Čapková M., Homola J.. Surface Plasmon Resonance Biosensors and Their Medical Applications. Biosens. Bioelectron. 2025;278:117308. doi: 10.1016/j.bios.2025.117308. [DOI] [PubMed] [Google Scholar]
  19. Chen Y.-S., Huang C.-H., Pai P.-C., Seo J., Lei K. F.. A Review on Microfluidics-Based Impedance Biosensors. Biosensors. 2023;13:83. doi: 10.3390/bios13010083. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Sharma A., Han C.-H., Jang J.. Rapid Electrical Immunoassay of the Cardiac Biomarker Troponin I Through Dielectrophoretic Concentration Using Imbedded Electrodes. Biosens. Bioelectron. 2016;82:78–84. doi: 10.1016/j.bios.2016.03.056. [DOI] [PubMed] [Google Scholar]
  21. Szymborski T. R., Czaplicka M., Nowicka A. B., Trzcinska-Danielewicz J., Girstun A., Kaminska A.. Dielectrophoresis-Based SERS Sensors for the Detection of Cancer Cells in Microfluidic Chips. Biosensors. 2022;12:681. doi: 10.3390/bios12090681. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kumar N., Wang W., Ortiz-Marquez J. C., Catalano M., Gray M., Biglari N., Hikari K., Ling X., Gao J., van Opijnen T., Burch K. S.. Dielectrophoresis Assisted Rapid, Selective and Single Cell Detection of Antibiotic Resistant Bacteria with G-FETs. Biosens Bioelectron. 2020;156:112123. doi: 10.1016/j.bios.2020.112123. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Barik A., Zhang Y., Grassi R., Nadappuram B. P., Edel J. B., Low T., Koester S. J., Oh S.-H.. Graphene-Edge Dielectrophoretic Tweezers for Trapping of Biomolecules. Nat. Commun. 2017;8:1867. doi: 10.1038/s41467-017-01635-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Koester S. J.. High Quality Factor Graphene Varactors for Wireless Sensing Applications. Appl. Phys. Lett. 2011;99:163105. doi: 10.1063/1.3651334. [DOI] [Google Scholar]
  25. Ebrish M. A., Shao H., Koester S. J.. Operation of Multi-Finger Graphene Quantum Capacitance Varactors using Planarized Local Bottom Gate Electrodes. Appl. Phys. Lett. 2012;100:143102. doi: 10.1063/1.3698394. [DOI] [Google Scholar]
  26. Gallo-Villanueva R. C., Sano M. B., Lapizco-Encinas B. H., Davalos R. V.. Joule Heating Effects on Particle Immobilization in Insulator-Based Dielectrophoretic Devices. Electrophoresis. 2014;35:352–361. doi: 10.1002/elps.201300171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Freedman K. J., Otto L. M., Ivanov A. P., Barik A., Oh S.-H., Edel J. B.. Nanopore Sensing at Ultra-Low Concentrations Using Single-Molecule Dielectrophoretic Trapping. Nat. Commun. 2016;7:10217. doi: 10.1038/ncomms10217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Barik A., Otto L. M., Yoo D., Jose J., Johnson T. W., Oh S.-H.. Dielectrophoresis-Enhanced Plasmonic Sensing with Gold Nanohole Arrays. Nano Lett. 2014;14:2006–2012. doi: 10.1021/nl500149h. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Fathi-Hafshejani P., Azam N., Wang L., Kuroda M. A., Hamilton M. C., Hasim S., Mahjouri-Samani M.. Two-Dimensional-Material-Based Field-Effect Transistor Biosensor for Detecting COVID-19 Virus (SARS-CoV-2) ACS Nano. 2021;15:11461–11469. doi: 10.1021/acsnano.1c01188. [DOI] [PubMed] [Google Scholar]
  30. Kumar N., Towers D., Myers S., Galvin C., Kireev D., Ellington A. D., Akinwande D.. Graphene Field Effect Biosensor for Concurrent and Specific Detection of SARSCoV-2 and Influenza. ACS Nano. 2023;17:18629–18640. doi: 10.1021/acsnano.3c07707. [DOI] [PubMed] [Google Scholar]
  31. Fenoy G. E., Marmisollé W. A., Azzaroni O., Knoll W.. Acetylcholine Biosensor Based on the Electrochemical Functionalization of Graphene Field-Effect Transistors. Biosensors & Bioelectron. 2020;148:111796. doi: 10.1016/j.bios.2019.111796. [DOI] [PubMed] [Google Scholar]
  32. Kang H., Wang X., Guo M., Dai C., Chen R., Yang L., Wu Y., Ying T., Zhu Z., Wei D., Liu Y., Wei D.. Ultrasensitive Detection of SARS-CoV-2 Antibody by Graphene Field-Effect Transistors. Nano Lett. 2021;21:7897–7904. doi: 10.1021/acs.nanolett.1c00837. [DOI] [PubMed] [Google Scholar]
  33. Wang G., Zhang M., Zhu M., Zhang T., Qian X., Liu Y., Ma X., Dai C., Wei D., Zhu Z.. et al. Ultraprecise Detection of Influenza Virus by Antibody-Modified Graphene Transistors. Sensors. 2025;25:959. doi: 10.3390/s25030959. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Béraud A., Sauvage M., Bazán C. M., Tie M., Bencherif A., Bouilly D.. Graphene Field-Effect Transistors as Bioanalytical Sensors: Design, Operation and Performance. Analyst. 2021;146:403–428. doi: 10.1039/D0AN01661F. [DOI] [PubMed] [Google Scholar]
  35. Fernandes E., Cabral P. D., Campos R., Machado G. Jr., Cerqueira M. F., Sousa C., Freitas P. P., Borme J., Petrovykh D. Y., Alpuim P.. Functionalization of Single-Layer Graphene for Immunoassays. Appl. Surf. Sci. 2019;480:709–716. doi: 10.1016/j.apsusc.2019.03.004. [DOI] [Google Scholar]
  36. Zhen X. V., Swanson E. G., Nelson J. T., Zhang Y., Su Q., Koester S. J., Bühlmann P.. Non-Covalent Monolayer Modification of Graphene Using Pyrene and Cyclodextrin Receptors for Chemical Sensing. ACS Appl. Nano Mater. 2018;1:2718–2726. doi: 10.1021/acsanm.8b00420. [DOI] [Google Scholar]
  37. Shin D., Kim H. R., Hong B. H.. Gold Nanoparticle-Mediated Non-Covalent Functionalization of Graphene for Field-Effect Transistors. Nanoscale Adv. 2021;3:1404–1412. doi: 10.1039/D0NA00603C. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Muszynski R., Seger B., Kamat P. V.. Decorating Graphene Sheets with Gold Nanoparticles. J. Phys. Chem. C. 2008;112:5263–5266. doi: 10.1021/jp800977b. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

am4c22829_si_001.pdf (704.9KB, pdf)
am4c22829_si_002.zip (24.6MB, zip)

Data Availability Statement

The data used in this paper are available from the corresponding authors upon reasonable request.

Articles from ACS Applied Materials & Interfaces are provided here courtesy of American Chemical Society

RESOURCES