Skip to main content
NCBI home page
ACS Omega logoLink to ACS Omega
. 2021 Nov 3;6(45):30281–30291. doi: 10.1021/acsomega.1c02222

Ion-Selective Membrane-Coated Graphene–Hexagonal Boron Nitride Heterostructures for Field-Effect Ion Sensing

Nowzesh Hasan †,‡, Urna Kansakar †,‡, Eric Sherer §, Mark A DeCoster †,‡, Adarsh D Radadia †,‡,§,*
PMCID: PMC8600519  PMID: 34805660

Abstract

graphic file with name ao1c02222_0006.jpg

An intrinsic ion sensitivity exceeding the Nernst–Boltzmann limit and an sp2-hybridized carbon structure make graphene a promising channel material for realizing ion-sensitive field-effect transistors with a stable solid–liquid interface under biased conditions in buffered salt solutions. Here, we examine the performance of graphene field-effect transistors coated with ion-selective membranes as a tool to selectively detect changes in concentrations of Ca2+, K+, and Na+ in individual salt solutions as well as in buffered Locke’s solution. Both the shift in the Dirac point and transconductance could be measured as a function of ion concentration with repeatability exceeding 99.5% and reproducibility exceeding 98% over 60 days. However, an enhancement of selectivity, by about an order magnitude or more, was observed using transconductance as the indicator when compared to Dirac voltage, which is the only factor reported to date. Fabricating a hexagonal boron nitride multilayer between graphene and oxide further increased the ion sensitivity and selectivity of transconductance. These findings incite investigating ion sensitivity of transconductance in alternative architectures as well as urge the exploration of graphene transistor arrays for biomedical applications.

1. Introduction

Ion-sensitive field-effect transistors (ISFETs) have revolutionized the field of ion sensing by reducing the sensor size and response times, enabling mass production, and tighter integration with electronics for drift compensation and data processing.1−3 Ion sensors play an integral role in biomedical diagnostics,4 environmental monitoring of water resources,5,6 and quality control of food and water products.7 The need for ISFETs with outstanding sensitivity, selectivity, repeatability, response time, and stability in biological fluids remains unaddressed to electronically interface with cells and tissues during the in vitro and in vivo experiments essential to understand the disordered physiological processes associated with diseases or injury and its rapid diagnosis on the bedside.8 Work to date has explored tailoring the semiconductor–oxide–electrolyte interface using proteins and nucleic acid sequences to selectively detect a wide range of biomolecules,9,10 increasing the intrinsic sensitivity using various gate dielectric,11−13 nanoscale channel materials such as silicon nanowires,14,15 carbon nanotubes,16,17 organic semiconductors,18−20 and graphene21,22 and in situ amplification of the intrinsic sensitivity using strategies involving dual (solution/bottom) gating23 and parallel channels of different areas.24 Graphene and organic semiconductor ISFETs allow overcoming two limiting aspects of silicon analogues. (a) The intrinsic sensitivity for silicon ISFETs due to the oxide–liquid interface has been restricted to the Nernst–Boltzmann limit (2.3 kBT/q ∼59 mV/decade for monovalent ions) set by the classic Boltzmann distribution of ions at the semiconductor–liquid interface and the site-binding mechanism proposed by Bergveld and co-workers.25,26 Although intrinsic sensitivity beyond the Nernst–Boltzmann limit has been achieved with silicon ISFETs by reducing ion-binding sites,27 or controlling the counter-ion size,28 this requires scaling the transistor channel down to ∼25 nm (costly complex lithography) and significant sample preparation. (b) The oxide–liquid interface is sensitive to ion migration especially when solution-gated (the most sensitive way to operate) for prolonged periods in biological electrolytes.29 The resulting drift in silicon ISFETs is currently addressed in commercial devices (Honeywell DuraFET) through the use of an internal reference, which increases device dimensions.30 Floating gate design commercialized by Ion Torrent for DNA sequencing also lowers the drift but at the cost of sensitivity.31

Intrinsic sensitivity exceeding the Nernst–Boltzmann limit has been demonstrated by our group as well as others with solution-gated ISFETs based on graphene, which due to its sp2-hybridized carbon network is also expected to be chemically stable when solution-gated in biological electrolytes.32−34 In prior work, we have shown a sensitivity as high as −164 mV/log [K+] and −57 mV/log [Ca2+] with graphene ISFETs.33 While the exact transduction mechanism remains unproven, a change in the double-layer composition (both at gate and channel) is believed to be primarily responsible for the observed change in charge carrier concentration and mobility in graphene.33,35 Enhanced sensitivities up to −198 mV/log [K+] and −110 mV/log [Ca2+] can be obtained by placing graphene on a dangling bond-free, chemically inert substrate such as hexagonal boron nitride (hBN), which is believed to increase the flatness of graphene, improve heat spreading in FETs, and open band gap in graphene leading to enhanced charge carrier mobility, reduced carrier inhomogeneity, and improved high-bias performance.36−44 This translates to an ion sensitivity normalized with respect to the drain-source voltage of 1980 mV/V/decade between 0.1 and 1000 mM KCl, which is larger than that claimed using organic electrochemical transistors in ref (45). However, to make graphene ISFETs ideal for in vitro and in vivo temporal recording applications such as a neural probe,46 it is important to induce ion selectivity. Ion-selective solvent-polymeric membrane,47 similar to those in ion-selective electrodes, holds the potential to induce ion selectivity.48−51 However, its impact on the inherent ion sensitivity of graphene, the resulting selectivity, and stability in biological electrolytes is unknown.

Here, we characterize the sensitivity, selectivity, and stability of graphene ISFETs coated with ion-selective membranes that are specific to either Na+, K+, or Ca2+. Characterization is performed with respect to shift in Dirac voltage (VDirac) and transconductance (gm), which is the slope of the linear region on either side of the transfer curve. A shift in VDirac indicates a change in the Fermi level of the graphene channel, while a change in gm indicates a change in charge carrier mobility and/or gate capacitance. Both VDirac and gm follow a Nernst-like (logarithmic) response to changes in ion concentrations. The paper is organized as follows. First, we show the proof-of-principle using ion-selective membranes and characterize its impact on the inherent ion sensitivity of graphene on SiO2 or hBN. Then, we characterize the sensitivity and selectivity of the Na+, K+, and Ca2+ ISFETs using individual solutions of NaCl, KCl, and CaCl2, followed by testing in buffered Locke’s solution with varying concentrations of either NaCl, KCl, or CaCl2. Finally, we present results on repeatability and stability up to 60 days and discuss the implications of these findings and directions of future research work.

2. Results and Discussion

2.1. Proof-of-Principle and Impact on Inherent Ion Sensitivity

ISFET devices (shown in Figure 1a–c) were fabricated with chemical vapor deposition (CVD) graphene directly on a 280 nm thick thermal SiO2 or with a 13 nm thick hBN spacer layer, here onwards referred to as SiO2 and hBN devices, respectively. The 200 nm Au/25 nm Cr electrodes were used as the source and the drain, and a poly(dimethylsiloxane) (PDMS) well was used to hold liquid samples over the device. Graphene and hBN films were characterized via confocal Raman microscopy (Figure 1d,e). Raman spectrum of multilayer hBN exhibited the characteristic E2g phonon mode at 1367 cm–1.52−54 Raman spectrum of graphene showed a G peak for graphene around 1580 cm–1 arising from in-plane vibrations of the sp2-hybridized carbons,55−57 a D peak around 1350 cm–1 arising from the out-of-plane vibration that for the sp3-hybridized carbons (defects and residues), and a 2D peak at 2690 cm–1 indicating the layer breathing vibration of graphene. The ratio of peak intensities, I2D/IG, could be strongly affected by the p-doped sample and it decreases with the increase of the doping level as reported in a previous study.55 The ratio of peak intensities, I2D/IG (>1), implies that the graphene was monolayer thick. A D* band appearing as a shoulder at ∼1615 cm–1 on the G band, typically referred to result from defect and edge effects, was not seen in the Raman spectra. A 2500-points Raman mapping was performed over a 50 μm × 50 μm area to examine the quality of transferred graphene, as shown in Figure S1. It shows that the G and 2D peak intensities did not show significant spatial variation, and a separate D* peak was absent. The D peak intensity showed spatial variation, and the defect ratio ID/IG was found to be 0.25 ± 0.20 with a distribution shown in Figure S1.

Figure 1.

Figure 1

Open in a new tab

Selective ion sensing demonstrated with calcium ionophore II on an hBN device. (a) Schematic cross-section of the hBN device. (b) Optical image of a chip (1.4 cm × 1.4 cm) that was fabricated for this research, each containing 12 sensors. (c) Optical microscope image of the 70 μm × 10 μm graphene strip spanning across four metal electrodes, all covered with an electrically insulating layer made of SU-8 epoxy. (d) Raman spectrum of multilayer hBN. (e) Raman spectrum of graphene. (f–h) Transfer curves recorded as the membrane-coated graphene on hBN were gated through solutions of (f) CaCl2, (g) KCl, and (h) NaCl of varying concentrations (0.1–1000 mM), in each case. The drain-to-source voltage (VDS) was held constant at 0.1 V. The color intensity of each transfer curve is plot darker for increasing salt concentration. (i) Shift in VDirac recorded on changing CaCl2 concentration on an hBN device before (blue filled triangles) and after (orange filled circles) casting the Ca2+ ion-selective membrane. Shift recorded relative to value at 0.1 mM. (j) Corresponding shift in gm recorded on the p-carrier dominant side of transfer curves. The dashed lines indicate curves (y = m × log10[x] + c) fit to the data points of respective color, and the text in the respective color indicates its slope (m).

A 2 μm thick SU-8 layer was used to electrically isolate the source and drain electrodes, and a 20 μm × 10 μm window in the SU-8 layer was patterned to expose a 10 μm × 10 μm graphene to the salt solutions during testing. The ion-selective membranes for Na+, K+, or Ca2+ were formed by drop-casting 1 μL of the ionophore cocktail over the sensor; this resulted in a 1.39 μm thick Na+ selective membrane, a 1.03 μm thick K+ selective membrane, or a 32.88 μm thick Ca2+ selective membrane as shown in Figure S2. The solutions to cast ion-selective membranes were made as per cocktail solutions sold by a vendor. The Ca2+ selective membrane ended up being thicker; this may be intended to reduce the permeation of smaller ions Na+ and K+ and provide higher selectivity. The quantum capacitance of graphene, which is a few μF/cm2, is high compared to a few hundred nF/cm2 for a 100 nm thick SiO2. This, in series with the double-layer capacitance (a few μF/cm2), allows graphene to be directly gated through the sample solution without an intermediate insulator layer.58−60 Detection with graphene ISFET involves recording transfer curves, which are a plot of the drain-to-source current (IDS) as a function of the gate voltage (VG) typically applied to an Ag/AgCl wire dipped in the sample solution. An ambipolar transfer curve such as those shown in Figure 1f–h is obtained with p-carriers being majority on the right side of the curve, n-carriers being majority on the left side, and both sides meeting at the lowest conductive point, VDirac. The bidirectional Id–Vg sweeps in graphene ISFET did show hysteresis in transfer curves, which can be attributed to charge trapping. Reduced hysteresis was found in devices with graphene on hBN. However, to compare the performance of multiple devices with and without hBN, the Id–Vg sweep was primarily made from positive to negative Vg.

The results reported in this study were collected with five ISFETs with graphene on SiO2 and three ISFETs with graphene on hBN, here onwards referred to as SiO2 devices and hBN devices. Three SiO2 devices and one hBN device were coated with a Ca2+ selective membrane, one SiO2 device and one hBN device were coated with a K+ selective membrane, and one SiO2 device and one hBN device were coated with a Na+ selective membrane. Three SiO2 devices were characterized in the solutions of varying Ca2+ concentrations before and after being coated with a Ca2+ selective membrane; here onwards, these results are referred to as without (w/o) ion-selective membrane. This way each device was tested in solutions containing the target cation prior to being coated with a target cation-selective membrane.

Figure 1f–h shows that a Ca2+ selective membrane imparts ion selectivity to an hBN device; a relatively larger left shift in VDirac and a drop in gm with an increase in Ca2+ concentration compared to when Na+ or K+ concentrations were varied. The left shift of VDirac can be explained by the rise in surface potential of graphene, which is expected to increase the doped charge carrier concentration in graphene.33 This, in turn, is expected to decrease the overall mobility of charge carriers in graphene and thus reduce gm according to the Drude model. The rise in gate capacitance with ion concentration has little effect on gm. Figure 1i,j shows a comparison of VDirac and gm before and after casting a Ca2+ selective membrane on the same hBN device. An overall lower magnitude of VDirac was obtained for a given CaCl2 concentration, and the ion sensitivity of VDirac was found to decrease by 36%, from −110 to −70 mV/decade. The magnitude of gm was found to increase, and the ion sensitivity of gm on the hBN device reduced from −1.6 × 10–3 to −1.4 × 10–3 mS/decade but stayed within the same order of magnitude. Similarly, casting a Ca2+ ion-selective membrane on a SiO2 device lowered the overall magnitude of VDirac and lowered the ion sensitivity of VDirac by 29%, from −57 to −40 mV/decade (Figure S3). The overall magnitude of gm was found to increase, and the ion sensitivity of gm reduced from −2.3 × 10–4 to −1.8 × 10–4 mS/decade. These performance changes can be explained by the ion screening resulting from the membrane, which decreases the total ions reaching graphene and results in relatively lower doping-induced charge carriers in graphene than when without a membrane. According to the Drude model, the reduced charge carrier concentration leads to an increased charge carrier mobility, and thereby a higher magnitude of gm and its lower sensitivity to changes in ion concentration.

The reduction of graphene’s inherent ion sensitivity by casting a Ca2+ selective membrane was confirmed on three SiO2 devices, as shown in Figure S4. Further, the VDirac and gm on the hBN device (Figure 1) demonstrated a higher ion sensitivity than SiO2 devices (Figure S4); the gm was almost an order magnitude more sensitive to changes in ion concentrations, while the VDirac was 1.8 times more sensitive. The variation of VDirac sensitivity among SiO2 devices was found to range from 2.6 to 12.1%, while the variation of gm sensitivity was found to range from 9 to 14.5% among SiO2 devices; this is in agreement to our prior work.33 Similar experiments were conducted to examine the effect of the K+ selective and Na+ selective membranes on the magnitude and ion sensitivity of the VDirac and gm on SiO2 and hBN devices. The results are presented in Figures S5 and S6. For the K+ ionophore membrane, the ion sensitivity of VDirac dropped by 26 and 32% upon membrane coating for the SiO2 and hBN devices, respectively. For the Na+ ionophore membrane, the ion sensitivity of VDirac dropped by 22 and 24% upon membrane coating for the SiO2 and hBN devices, respectively. The ion sensitivity of gm reduced in each case but stayed within the same order of magnitude.

2.2. Testing in Individual Salt Solutions

The shift in VDirac and gm for the SiO2 and hBN devices coated with Ca2+, K+, or Na+ selective membranes were recorded using 0.1–1000 mM of either Ca2+, Na+, or K+ (Figure 2). The shift in VDirac was calculated with respect to the lowest ion concentration measured. The slope of the graph plotting the VDirac or gm against log10 (ion concentration) was inferred as sensitivity. The selectivity of the ISFET was calculated by taking a ratio of the sensitivity toward the target ion to that toward the interfering ion. For example, when the ion concentration was measured in terms of shift in VDirac, the Ca2+ ionophore membrane on a SiO2 device in Figure 2a exhibits a selectivity of 1.33 against Na+ and 1.48 against K+, respectively. When the sensitivity was measured in terms of gm, a higher selectivity of ∼200 was obtained against both Na+ and K+. This was confirmed on three different SiO2 devices coated with the Ca2+ ionophore coating, as shown in Figure S7. Similarly, the sensitivity values from Figure 2 were used to calculate the selectivity values shown in Table 1.

Figure 2.

Figure 2

Open in a new tab

Sensitivity and selectivity evaluated in individual salt solutions. (a–f) Results obtained using SiO2 devices. (g–l) Results obtained using hBN devices. The left column shows results from devices coated with a Ca2+ selective membrane, the middle column shows results from devices coated with a K+ selective membrane, and the right column shows results from devices coated with a Na+ selective membrane. (a–c, g–i) The response recorded in terms of the shift in VDirac. (d–f, j–l) The response recorded in terms of gm. In each graph, orange circles, black asterisks, and blue triangles represent data recorded in CaCl2, NaCl, and KCl, respectively. The dashed lines indicate curves (y = m × log10[x] + c) fit to the data with respective color, and the text in the respective color indicates its slope.

Table 1. Selectivity of ISFETs Measured in the Individual Salt Solutiona.

graphene support  
  No Membrane
  SVDiracCa2+/Na+ SgmCa2+/Na+ SVDiracCa2+/K+ SgmCa2+/K+
SiO2 0.33 0.13 0.35 0.21
hBN 0.53 0.23 0.56 0.35
  Ca2+ Selective Membrane
  SVDiracCa2+/Na+ SgmCa2+/Na+ SVDiracCa2+/K+ SgmCa2+/K+
SiO2 1.33 98.9 1.48 196
hBN 2.15 665 2.37 2070
  K+ Selective Membrane
  SVDiracK+/Ca2+ SgmK+/Ca2+ SVDiracK+/Na+ SgmK+/Na+
SiO2 11.1 100 2.65 9.83
hBN 11.3 144 2.33 3.80
  Na+ Selective Membrane
  SVDiracNa+/Ca2+ SgmNa+/Ca2+ SVDiracNa+/K+ SgmNa+/K+
SiO2 15 80.8 2.87 9.21
hBN 14.5 137 2.62 7.66
Open in a new tab
a

SVDiracCa2+/Na+ denotes selectivity of detecting Ca2+ over Na+ with sensitivity measured in terms of the shift in VDirac. Values rounded to first three accurate digits.

The following observations can be made from the data in Figure 2 and Table 1. (1) Ion selectivity of the devices was enhanced using the ionophoretic membranes as shown by the calculated selectivity of detecting Ca2+ in Table 1. For example, the selectivity of Ca2+ over Na+ for the SiO2 device increased from 0.33 without the ion-selective membrane to 1.33 with the ion-selective membrane. (2) In each of the cases shown in Table 1, the selectivity calculated using gm exhibits a higher value compared to that calculated using the shift in VDirac. The values for selectivity measured in terms of gm exhibited a larger spread, ranging from 3.8 to 2070, while the values for selectivity measured in terms of VDirac ranged from 1.33 to 15. (3) Compared to SiO2 devices, hBN devices recorded a larger sensitivity of VDirac or gm for the target ion as shown in Figure 2; the gain being the highest for the Ca2+ selective membrane and the least for the K+ selective membrane. In some cases, hBN devices also recorded a larger sensitivity of VDirac or gm for the interfering ion. By examining the selectivity values in Table 1, it can be seen that hBN devices yielded an order magnitude higher selectivity for the Ca2+ membrane when measured in terms of gm but a deterioration in selectivity against K+ on the Na+ membrane and against Na+ on the K+ selective membrane.

2.3. Testing in HEPES-Buffered Locke’s Solution

The ISFETs were then used to detect changes in concentrations of Ca2+, K+, and Na+ in a HEPES-buffered Locke’s solution (pH 7.4). The standard Locke’s solution formulation used incubate excised brain and for calcium imaging in our lab contains 154 mM NaCl, 5.6 mM KCl, 3.6 mM NaHCO3, 2.3 mM CaCl2, 1.2 mM MgCl2, 5.6 mM glucose, and 5 mM HEPES. Three sets of solutions were prepared by modification of the standard formulation; each set either contained a variation in concentration of Ca2+ from 0 to 2.3 mM, K+ from 0 to 5.6 mM, or Na+ from 0 to 157.6 mM. The compositions of the nine solutions used in this test are listed in Table S1. The osmolarity was balanced using choline chloride. The changes in VDirac or gm of ion-selective membrane-coated ISFETs were recorded in each of these modified Locke’s solutions as well as the standard Locke’s solution. The results for SiO2 and hBN devices are shown in Figure 3. The sensitivity values from Figure 3 were used to calculate the selectivity values shown in Table 2.

Figure 3.

Figure 3

Open in a new tab

Sensitivity and selectivity evaluated in a HEPES-buffered Locke’s solution. (a–f) Results obtained using SiO2 devices. (g–i) Results obtained using hBN devices. The left column shows results from devices coated with a Ca2+ selective membrane, the middle column shows results from devices coated with a K+ selective membrane, and the right column shows results from devices coated with a Na+ selective membrane. (a–c, g–i) Response recorded in terms of the shift in VDirac. (d–f, j–l) Response recorded in terms of gm. In each graph, orange circles, black asterisks, and blue triangles represent data recorded in CaCl2, NaCl, and KCl, respectively. The dashed lines indicate curves (y = m × log10[x] + c) fit to the data with respective color, and the text in the respective color indicates its slope.

Table 2. Selectivity of ISFETs Measured in a HEPES-Buffered Locke’s Solutiona.

graphene support  
  Ca2+ Selective Membrane
  SVDiracCa2+/Na+ SgmCa2+/Na+ SVDiracCa2+/K+ SgmCa2+/K+
SiO2 1.57 45.4 1.76 228
hBN 2.09 1610 2.28 3710
  K+ Selective Membrane
  SVDiracK+/Ca2+ SgmK+/Ca2+ SVDiracK+/Na+ SgmK+/Na+
SiO2 13 72.7 2.65 5.96
hBN 12.8 279 2.67 8.69
  Na+ Selective Membrane
  SVDiracNa+/Ca2+ SgmNa+/Ca2+ SVDiracNa+/K+ SgmNa+/K+
SiO2 12.6 60.5 2.65 8.95
hBN 13.8 111 2.55 7.85
Open in a new tab
a

SVDiracCa2+/Na+ denotes selectivity of detecting Ca2+ over Na+ with sensitivity measured in terms of the shift in VDirac. Values rounded to first three accurate digits.

Comparing the sensitivities measured in terms of shift in VDirac obtained for the target and interfering ions in Locke’s solution (Figure 3) to that obtained in individual salt solutions (Figure 2), it was remarkable that the sensitivities were within 8 mV/decade of each other, except the hBN device coated with K+ selective membrane, which recorded an increase in sensitivity by 31 mV/decade. Further, the sensitivities measured in terms of gm using Locke’s solution (Figure 3) were found to be within the same order of magnitude with a slight improvement over those recorded in the individual salt solutions (Figure 2); an increase of 2.54 times was recorded in the case of K+ on the Ca2+ selective membrane. Also, it is remarkable that the selectivity values obtained in individual salt solutions (Table 1) are comparable or better than those obtained in Locke’s solution (Table 2). The observations (1) and (2) made with testing in individual salt solutions were also noted true when testing with Locke’s solution. For each case in Table 2, the selectivity calculated from the shift in gm exhibits a higher value compared to that calculated using the shift in VDirac. Also, hBN devices recorded a larger ion sensitivity of VDirac or gm for target ions compared to SiO2 devices, as shown in Figure 3.

Testing sensitivity in Locke’s solution also provided an insight into the underlying transduction mechanism of graphene ISFETs. Sensing selectivity in ISFET literature, akin to ion-selective electrodes, has been described using the Nikolsky–Eisenman equation as follows for concentration C of target species i with valency zi in the presence of other species j with valency zj.

2.3. 1

where A is a constant and L is the detection limit. This model assumes that the presence of multiple ionic species induces a competitive binding behavior in the ion-selective membrane and the surface of the ISFET gate at equilibrium. The fitting of data from testing with Locke’s solutions (Figure 3) to the Nikolski–Eisenmann equation as per the Bayesian model analysis did not converge as described in the “Bayesian Analysis” section in the Supporting Information. An alternative is to consider the additive binding behavior in the presence of multiple ionic species, which can be explained by the below equation that has been recently described to explain the behavior of 0D silicon ISFET with reduced density of charged sites.27

2.3. 2

The data from testing with Locke’s solutions (Figure 3) was found to fit eq 2 as shown in the “Bayesian Analysis” section in the Supporting Information. Such additive binding behavior in the ion-selective membrane is highly unlikely due to the vast amount of experimental data showing otherwise. However, such additive binding behavior on the graphene surface potentially would explain why the data fit to eq 2.

2.4. Repeatability and Stability

Repeatability, also called the variability of measurement (instrumental), for a SiO2 device coated with the Ca2+ selective membrane was measured by consecutively recording transfer curves thrice while varying CaCl2 concentration in the gating solution from 0.1 to 1000 mM. From these transfer curves, the values for VDirac and gm were used to calculate the repeatability, (%)R, as follows

2.4. 3

where SD represents the standard deviation among recorded values for VDirac or gm and span is the maximum value among three trials. The obtained values for VDirac and gm are plotted in Figure S8. The calculated values of standard deviation, span, and 1 – (%)R are tabulated in Table S2. The plot of normalized standard deviation (SD/span) in VDirac values as a function of CaCl2 concentrations did not show a discernable trend as evident from Figure 4a; an average value of 0.28% and a maximum value of 0.44% were obtained. The average (%)R in the measurement of VDirac was calculated to be 99.56%. Likewise, the normalized standard deviation in gm was also found not to be a discernable function of CaCl2 concentration, as shown in Figure 4b; an average normalized standard deviation of 0.03% and a maximum value of 0.07% were obtained. The average (%)R in the measurement of gm was calculated to be 99.93%, which is superior to 99.5% obtained with current commercial ISFETs (Microsens MSFET-3330). The lack of such data for coated commercial ISFETs made it difficult to make a direct comparison.

Figure 4.

Figure 4

Open in a new tab

Repeatability and stability measurements for ISFETs coated with the Ca2+ selective membrane cast on a SiO2 device. (a, b) Normalized standard deviation in the measurement of VDirac (a) and (b) gm as a function of CaCl2 concentration during three separate experiments on the same device. The black hollow circles are data points corresponding to CaCl2 concentration tested, and the dashed black lines show the average value. (c) VDirac and (d) gm measured when the same device was gated using fresh 10 mM CaCl2 at different time points. The filled red and blue circles are data points indicating VDirac and gm, respectively, and the dashed gray lines indicate the extremes recorded.

The stability of a SiO2 device coated with a Ca2+ selective membrane was gauged by conducting an experiment at regular intervals (15 days) for a total period of 60 days. In each experiment, a transfer curve was recorded while gating through CaCl2 concentrations ranging from 0.1 to 1000 mM. The VDirac and gm values were seen to increase with time as shown in the plot for 10 mM CaCl2 in Figure 4c,d. The data for the rest of the concentrations is provided in Figures S9 and S10. Reproducibility (%)Rp was calculated using eq 1 with SD now calculated as the standard deviation in VDirac or gm values over the 60 days and span as the maximum value obtained over the 60 days. The detailed calculation of (%)Rp for VDirac and gm is provided in Tables S3 and S4, respectively. The value for (%)Rp in the measurement of VDirac ranged between 98.45 and 99.12%. Similarly, the value for (%)Rp in the measurement of gm ranged between 98.34 and 98.92%. The average ion sensitivity of VDirac and gm over the 60 days was found to be −39.2 ± 0.6 mV/decade and −2 × 10–4 ± 3.6 × 10–6 mS/decade, respectively, with a maximum deviation of 3.8 and 4.3%, respectively (see Figure S11). A higher value of (%)Rp translates to less frequent calibrations for continuous monitoring applications. The values of (%)Rp obtained here are far superior compared to most nonencapsulated silicon-based ISFETs, which have a characteristic drift of ∼1 mV/h, which further translates to a (%)Rp value of 90%. Ruggedized encapsulation of silicon ISFETs such as in the case of Honeywell DuraFET has allowed achieving ((%)Rp) values of 99.5%.

2.5. Practical Implications

Here, we showed that monitoring gm on hBN devices coated with ion-selective membranes allowed sensing changes in concentrations of Ca2+, K+, and Na+ with good sensitivity, selectivity, repeatability, and stability. For example, with the Ca2+ membrane-coated hBN device, the fluctuation of K+ concentration from 5.6 to 2.8 mM and Na+ concentration from 157.6 to 78.8 mM in Locke’s solution resulted in a change in gm equivalent to that would have resulted from the standard Ca2+ concentration of 2.3 mM to change by 0.4 and 1 μM, respectively. Similarly, with the K+ membrane-coated hBN device, the fluctuation of Ca2+ concentration from 2.3 to 1.15 mM and Na+ concentration from 157.6 to 78.8 mM in Locke’s solution resulted in a change in gm equivalent to that would have resulted from the standard K+ concentration of 5.6 mM to change by 13 μM and 0.46 mM, respectively. Likewise, with the Na+ membrane-coated hBN device, the fluctuation of Ca2+ concentration from 2.3 to 1.15 mM and K+ concentration from 5.6 to 2.8 mM in Locke’s solution resulted in a change in gm equivalent to that would have resulted from the standard Na+ concentration of 157.6 mM to change by 1 and 14.5 mM, respectively.

A parallelization could conceivably be achieved for the synchronized detection of Na+, K+, and Ca2+ using a sensor array of ISFETs coated alternately with solvent-polymeric membranes specific to Na+, K+, and Ca2+. Such sensing arrays would be of immense value on a Petri dish or a neural probe for the detection of extracellular ion concentrations. Body fluid concentrations of the three most important cations, Na+ (135–145 mM), K+ (3.5–5.0 mM), and Ca2+ (1.1–1.3 mM) are tightly regulated for normal functions. While Na+ levels maintain an osmotic balance and a control over fluid movement between compartments, the Ca2+ and K+ levels help establish membrane potentials essential for firing or resting in neurons and muscle fibers. Studying the dynamics of these cations concurrently could conceptually help improve our understanding of how abnormal functions result in a complex organ like the brain.61,62 However, in our experiments, we did not optimize membrane formulations, they were adapted from the literature,63 and low volumes of it were cast to achieve the thinnest films. To realize highly sophisticated arrays of ISFETs in a Petri dish or a neural probe, the following should be considered. The concentrations of ionophore, plasticizer, and additives as well as the membrane thickness could be optimized to achieve an application-specific balance between sensitivity, selectivity, and response time. In addition, this optimization has to be done at the target operating temperature, which would be 37 °C for most biological experiments. Further, it is difficult to infer the concentration of a specific ion with high confidence using gm recorded on a single ISFET. This confidence could be improved using duplicate ISFETs, reference ISFETs, and using membranes doped with other target-specific ionophores; however, this would increase the device size or decrease spatial resolution. Alternatively, results from ISFETs for detecting other expected ions, like that reported here, can be correlated to increase the confidence in reporting ion concentration and also identifying a change in the concentration of more than one ion. The results of such a combination of sensors can be bolstered with the use of nonlinear signal processing methods such as the partial least squares regression, artificial neural networks, and Bayesian blind source separation to confidently reconstruct ion concentrations.

3. Conclusions

In summary, we evaluated different aspects of ISFETs coated with ion-selective membranes to selectively detect changes in concentrations of Ca2+, K+, and Na+ in solution. The membrane coating lowers the inherent sensitivity of the VDirac and gm in an ISFET but imparts noticeable selectivity toward the target ion. The graphene–hBN heterostructure results in devices with higher ion sensitivity of VDirac and gm compared to devices with graphene on SiO2. Results from testing with individual salt solutions and buffered Locke’s solutions show that in comparison to VDirac, monitoring gm provides a higher selectivity in sensing targeted ions. Both VDirac and gm were measured with greater than 99.5% repeatability. Using experiments over 60 days, we show that measurements of VDirac and gm are more than 98% reproducible and the ion sensitivity of VDirac and gm stays within 3.8 and 4.3% of that recorded at day 0, respectively. This demonstrates the stability of the membrane–graphene structure in a biological electrolyte. These results warrant the use of graphene ISFET-based tools for biological studies sensing ion concentration changes outside the cell.

4. Methodology and Materials

4.1. Materials

Copper foils (20 μm thick) with monolayer graphene or multilayer hBN films grown by the chemical vapor deposition (CVD) method were obtained from Graphene Labs Inc. These CVD graphene films were 1–10 μm in grain size and mostly monolayer with 10–30% bilayer islands. Poly(methyl methacrylate) (996 kDa) was obtained from ALDRICH, and it was dissolved in anisole from Fluka to prepare a 5 (w/v)% PMMA solution. Copper etchant type CE-100 was obtained from TRANSENE Company, Inc. A general method to transfer hBN and graphene films from copper to silicon substrates as required to build the needed devices is described in ref (33).

4.2. Raman Spectroscopy

Raman spectroscopy was carried using a Horiba Scientific XploRA Plus confocal Raman microscope equipped with a 532 nm laser, 100 × 0.95 NA objective 1200 lines/mm diffraction grating, a 300 μm slit, and a 200 μm hole. The setup achieved a beam spot diameter of 1 μm. Raman spectrum for mono/bilayer CVD graphene and multilayer hBN exhibits characteristic peaks, as shown in Figure 1c,d.

4.3. Ion-Selective Cocktail Preparation and Film Deposition

The ion-selective membrane solution was prepared by dissolving the ion-selective cocktail with the corresponding solvent. The sodium ion-selective membrane cocktail comprised of sodium ionophore X (1% w/w; Selectophore, Sigma-Aldrich), sodium tetrakis [3,5-bis(trifluoromethyl)phenyl] borate (0.55% w/w; Selectophore, Sigma-Aldrich), poly(vinyl chloride) (33% w/w; Sigma-Aldrich), and bis(2-ethylhexyl) sebacate (65.45% w/w; Sigma-Aldrich). The sodium ion-selective membrane solution was prepared by dissolving 100 mg of the ion-selective cocktail in 660 μL of tetrahydrofuran (anhydrous, ≥99.9%; Sigma-Aldrich).63,64 The potassium ion-selective membrane cocktail comprised of valinomycin (2% w/w; Sigma-Aldrich), sodium tetraphenyl borate (0.6% w/w; Sigma-Aldrich), poly(vinyl chloride) (32.7% w/w; Sigma-Aldrich), and bis(2-ethylhexyl) sebacate (64.7% w/w; Sigma-Aldrich). The potassium ion-selective membrane solution was prepared by dissolving 100 mg of the ion-selective cocktail in 350 μL of cyclohexanone (≥99.5%; Selectophore, Sigma-Aldrich). The calcium ion-selective membrane cocktail comprised of calcium ionophore II (9.2% w/w; Selectophore, Sigma-Aldrich), sodium tetrakis[3,5-bis(trifluoromethyl)phenyl]borate (4.5% w/w; Selectophore, Sigma-Aldrich), bis(2-ethylhexyl) sebacate (56.3% w/w; Sigma-Aldrich), and poly(vinyl chloride) (30% w/w; Sigma-Aldrich). The calcium ion-selective membrane solution was prepared by dissolving 111 mg of the ion-selective cocktail in 333 mg of tetrahydrofuran (anhydrous, ≥99.9%; Sigma-Aldrich). The ionophore membrane was coated on the exposed graphene surface by drop-casting 1 μL of the ion-selective solution over a 20 μm × 10 μm sensing window using the solvent evaporation process at room temperature (25 °C).

4.4. Ion-Selective HEPES-Buffered Locke’s Solution

Within 248.75 mL of purified deionized water, 2250 mg of NaCl (154 mM), 104.4 mg of KCl (5.6 mM), 75.6 mg of NaHCO3 (3.6 mM), 84.5 mg of CaCl2·2H2O (2.3 mM), 61 mg of MgCl2·6H2O (1.2 mM), and 252.3 mg of C6H12O6 (5.6 mM) were dissolved via vortexing. Then, 1.25 mL of 1 M stock C8H18N2O4S (5 mM) with pH 7.4 was added into the dissolved solution. Within a vacuum chamber, the dissolved liquid was passed through the filter using vacuum filtration. The HEPES-buffered fluid was stored inside the refrigerator under 4 °C.

4.5. Electrical Measurements

All electrical measurements were accomplished with a probe station inside a Faraday cage. Resistance and transfer curves were measured using a dual source-measurement unit (Keithley 2636A). A poly(dimethylsiloxane) (PDMS) well was punched out to hold the electrolyte over the graphene strip. Electrochemical top-gate circuit setup was completed by immersing an Ag/AgCl electrode into the electrolyte. The resistance of the graphene strip was measured in air by sweeping the drain-to-source voltage (VDS) from 0 to 100 mV in pulsed mode (1 ms pulse width, 50 ms time period). The pulsed mode was used to avoid significant Joule heating effects during measurements. Over ion-selective membranes (Ca2+, Na+, and K+), electrolytes (NaCl, KCl, and CaCl2) of varying concentration (0.1, 1, 3, 5, 7, 9, 10, 100, and 1000 mM) were used as the gate electrolyte in measuring transfer curves on graphene ISFETs with or without hBN as an underlying dielectric layer. To generate transfer curves, VDS was held constant at 100 mV DC across the drain and the source while sweeping the gate voltage, VG. Using the Ca2+ ionophore membrane on graphene ISFET over SiO2, three trials of transfer curve measurement data are recorded using CaCl2 with varying concentrations (0.1–1000 mM) for testing repeatability. For sensor stability testing, five sets of transfer curve measurement data are recorded with the Ca2+ ionophore membrane on graphene ISFET fabricated on SiO2 over 15 days apart. In both cases for repeatability and reproducibility data recording, VDS was held constant at 100 mV DC across the drain and the source while sweeping the gate voltage. For the Ca2+ ionophore membrane, using varying Ca2+ concentrations (0, 0.58, 0.77, 1.15, 1.53, 1.73, and 2.3 mM) in Locke’s solutions, transfer curves are recorded from graphene ISFETs with and without hBN. To observe the selectivity, transfer curves are recorded with varying Na+ concentrations (0, 78.8, and 157.6 mM) and K+ concentrations (0, 2.8, and 5.6 mM) in Locke’s solutions. Likewise, for the Na+ ionophore membrane, using varying Na+ concentrations (0, 39.4, 52.53, 78.8, 105.07, 118.2, and 157.6 mM) in Locke’s solutions, transfer curves are recorded from graphene ISFETs with and without hBN. To observe the selectivity, transfer curves are recorded with varying Ca2+ concentrations (0, 1.15, and 2.3 mM) and K+ concentrations (0, 2.8, and 5.6 mM) in Locke’s solutions. For the K+ ionophore membrane, using varying K+ concentrations (0, 1.4, 1.87, 2.8, 3.7, 4.2, and 5.6 mM) in Locke’s solutions, transfer curves are recorded from graphene ISFETs with and without hBN. To observe the selectivity, transfer curves are recorded with varying Ca2+ concentrations (0, 1.15, and 2.3 mM) and Na+ concentrations (0, 78.8, and 157.6 mM) in Locke’s solutions.

Acknowledgments

This material is based upon work partly supported by the Research Competitiveness Subprogram from the Louisiana Board of Regents through the Board of Regents Support Fund under the Contract Number LEQSF(2013-2016)-RD-A-09; an Institutional Development Award (IDeA) from the National Institute of General Medical Sciences of the National Institutes of Health under grant number P20GM103424; the Research Enhancement Award (Subcontract 75537) by the Louisiana Board of Regents Support Fund [LEQSF(2010-2015)-LaSPACE]; and the support of NASA [Grant Number NNX10AI40H]. The authors are thankful to the staff at the Institute for Micromanufacturing and the Center for Biomedical Engineering and Rehabilitation Science at Louisiana Tech University.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.1c02222.

  • Figure S1: Raman mapping of transferred graphene; Figure S2: change in ion-selective membrane film thickness and diameter with varying amount of membrane cocktail solution; Figure S3: selective ion sensing demonstrated with calcium ionophore II on a SiO2 device; Figure S4: impact of casting the Ca2+ selective membrane on the sensitivity of gm and VDirac to changes in CaCl2 concentrations; Figure S5: impact of casting the K+ selective membrane on the sensitivity of VDirac and gm to changes in KCl concentrations; Figure S6: impact of casting the Na+ selective membrane on the sensitivity of gm and VDirac to changes in NaCl concentrations; Figure S7: response of Ca2+ selective membrane-coated SiO2 devices recorded in single-salt solutions (NaCl, KCl, or CaCl2); Figure S8: repeatability measurements for ISFETs coated with the Ca2+ selective membrane on the SiO2 device; Figure S9: reproducibility of VDirac for a SiO2 device coated with the Ca2+ selective membrane; Figure S10: reproducibility of gm for a SiO2 device coated with the Ca2+ selective membrane; Figure S11: ion sensitivity of VDirac and gm over 60 day period for a SiO2 device coated with the Ca2+ selective membrane; Table S1: repeatability calculation using normalized standard deviations in (a) Dirac voltage (VDirac) and (b) transconductance (gm) based on three consecutive trials; Table S2: reproducibility calculation using VDirac recorded over a period of 60 days when measured with different CaCl2 solutions; Table S3: reproducibility calculation using gm recorded over a period of 60 days when measured with different CaCl2 solutions (0.1–1000 mM); Table S4: composition of solutions that were used to obtain results in Figure 3; and A section on Bayesian analysis of data in Figure 3 (PDF)

Author Present Address

∥ Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States

Author Present Address

⊥ Albert Einstein College of Medicine, Bronx, New York 10461, United States.

Author Present Address

# Corteva Agriscience, Indianapolis, Indiana 46268, United States.

Author Contributions

N.H. and A.D.R. conceived and designed the experiments; N.H. fabricated the chips and N.H. performed the experiments; and M.A.D. and U.K. prepared the various buffered Locke’s solution. N.H. and A.D.R. analyzed the data; E.S. performed the Bayesian analysis; N.H. and A.D.R. wrote the paper; and M.A.D. provided essential edits to the paper.

The authors declare no competing financial interest.

Supplementary Material

ao1c02222_si_001.pdf (2.1MB, pdf)

References

  1. Bergveld P. Development of an ion-sensitive solid-state device for neurophysiological measurements. IEEE Trans. Biomed. Eng. 1970, BME-17, 70–71. 10.1109/TBME.1970.4502688. [DOI] [PubMed] [Google Scholar]
  2. Moser N.; Lande T. S.; Toumazou C.; Georgiou P. ISFETs in CMOS and emergent trends in instrumentation: A review. IEEE Sens. J. 2016, 16, 6496–6514. 10.1109/JSEN.2016.2585920. [DOI] [Google Scholar]
  3. Martinoia S.; Massobrio G.; Lorenzelli L. Modeling ISFET microsensor and ISFET-based microsystems: a review. Sens. Actuators, B 2005, 105, 14–27. 10.1016/S0925-4005(04)00107-8. [DOI] [Google Scholar]
  4. Duroux P.; Emde C.; Bauerfeind P.; Francis C.; Grisel A.; Thybaud L.; Arstrong D.; Depeursinge C.; Blum A. The ion sensitive field effect transistor (ISFET) pH electrode: a new sensor for long term ambulatory pH monitoring. Gut 1991, 32, 240–245. 10.1136/gut.32.3.240. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. McLaughlin K.; Dickson A.; Weisberg S. B.; Coale K.; Elrod V.; Hunter C.; Johnson K. S.; Kram S.; Kudela R.; Martz T.; et al. An evaluation of ISFET sensors for coastal pH monitoring applications. Reg. Stud. Mar. Sci. 2017, 12, 11–18. 10.1016/j.rsma.2017.02.008. [DOI] [Google Scholar]
  6. Jimenez-Jorquera C.; Orozco J.; Baldi A. ISFET based microsensors for environmental monitoring. Sensors 2010, 10, 61–83. 10.3390/s100100061. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Sandifer J. R.; Voycheck J. J. A review of biosensor and industrial applications of ph-isfets and an evaluation of honeywell’s “durafet”. Microchim. Acta 1999, 131, 91–98. 10.1007/PL00021393. [DOI] [Google Scholar]
  8. Dvir T.; Timko B. P.; Kohane D. S.; Langer R. Nanotechnological strategies for engineering complex tissues. Nat. Nanotechnol. 2011, 6, 13. 10.1038/nnano.2010.246. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Toumazou C.; Georgiou P. Piet Bergveld-40 years of ISFET technology: From neuronal sensing to DNA sequencing. Electron. Lett. 2011, 47, S7–S12. 10.1049/el.2011.3231. [DOI] [Google Scholar]
  10. Syu Y.-C.; Hsu W.-E.; Lin C.-T. Field-Effect Transistor Biosensing: Devices and Clinical Applications. ECS J. Solid State Sci. Technol. 2018, 7, Q3196–Q3207. 10.1149/2.0291807jss. [DOI] [Google Scholar]
  11. Chen S.; Bomer J. G.; Carlen E. T.; van den Berg A. Al2O3/silicon nanoISFET with near ideal Nernstian response. Nano Lett. 2011, 11, 2334–2341. 10.1021/nl200623n. [DOI] [PubMed] [Google Scholar]
  12. Schöning M. J.; Brinkmann D.; Rolka D.; Demuth C.; Poghossian A. CIP (cleaning-in-place) suitable “non-glass” pH sensor based on a Ta2O5-gate EIS structure. Sens. Actuators, B 2005, 111–112, 423–429. 10.1016/j.snb.2005.03.053. [DOI] [Google Scholar]
  13. Bousse L.; Mostarshed S.; van der Schoot B.; De Rooij N. Comparison of the hysteresis of Ta2O5 and Si3N4 pH-sensing insulators. Sens. Actuators, B 1994, 17, 157–164. 10.1016/0925-4005(94)87044-6. [DOI] [Google Scholar]
  14. Cui Y.; Wei Q.; Park H.; Lieber C. M. Nanowire nanosensors for highly sensitive and selective detection of biological and chemical species. Science 2001, 293, 1289–1292. 10.1126/science.1062711. [DOI] [PubMed] [Google Scholar]
  15. Stern E.; Klemic J. F.; Routenberg D. A.; Wyrembak P. N.; Turner-Evans D. B.; Hamilton A. D.; LaVan D. A.; Fahmy T. M.; Reed M. A. Label-free immunodetection with CMOS-compatible semiconducting nanowires. Nature 2007, 445, 519. 10.1038/nature05498. [DOI] [PubMed] [Google Scholar]
  16. Allen B. L.; Kichambare P. D.; Star A. Carbon nanotube field-effect-transistor-based biosensors. Adv. Mater. 2007, 19, 1439–1451. 10.1002/adma.200602043. [DOI] [Google Scholar]
  17. Sorgenfrei S.; Chiu C.-y.; Gonzalez R. L. Jr; Yu Y.-J.; Kim P.; Nuckolls C.; Shepard K. L. Label-free single-molecule detection of DNA-hybridization kinetics with a carbon nanotube field-effect transistor. Nat. Nanotechnol. 2011, 6, 126. 10.1038/nnano.2010.275. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Knopfmacher O.; Hammock M. L.; Appleton A. L.; Schwartz G.; Mei J.; Lei T.; Pei J.; Bao Z. Highly stable organic polymer field-effect transistor sensor for selective detection in the marine environment. Nat. Commun. 2014, 5, 2954 10.1038/ncomms3954. [DOI] [PubMed] [Google Scholar]
  19. Schmoltner K.; Kofler J.; Klug A.; List-Kratochvil E. J. Electrolyte-Gated Organic Field-Effect Transistor for Selective Reversible Ion Detection. Adv. Mater. 2013, 25, 6895–6899. 10.1002/adma.201303281. [DOI] [PubMed] [Google Scholar]
  20. Benfenati V.; Toffanin S.; Bonetti S.; Turatti G.; Pistone A.; Chiappalone M.; Sagnella A.; Stefani A.; Generali G.; Ruani G.; et al. A transparent organic transistor structure for bidirectional stimulation and recording of primary neurons. Nat. Mater. 2013, 12, 672. 10.1038/nmat3630. [DOI] [PubMed] [Google Scholar]
  21. Ohno Y.; Maehashi K.; Yamashiro Y.; Matsumoto K. Electrolyte-gated graphene field-effect transistors for detecting pH and protein adsorption. Nano Lett. 2009, 9, 3318–3322. 10.1021/nl901596m. [DOI] [PubMed] [Google Scholar]
  22. Dong X.; Shi Y.; Huang W.; Chen P.; Li L. J. Electrical detection of DNA hybridization with single-base specificity using transistors based on CVD-grown graphene sheets. Adv. Mater. 2010, 22, 1649–1653. 10.1002/adma.200903645. [DOI] [PubMed] [Google Scholar]
  23. Spijkman M. J.; Myny K.; Smits E. C.; Heremans P.; Blom P. W.; De Leeuw D. M. Dual-gate thin-film transistors, integrated circuits and sensors. Adv. Mater. 2011, 23, 3231–3242. 10.1002/adma.201101493. [DOI] [PubMed] [Google Scholar]
  24. Go J.; Nair P. R.; Reddy B. Jr; Dorvel B.; Bashir R.; Alam M. A. Coupled heterogeneous nanowire–nanoplate planar transistor sensors for giant (> 10 V/pH) Nernst response. ACS Nano 2012, 6, 5972–5979. 10.1021/nn300874w. [DOI] [PubMed] [Google Scholar]
  25. Bergveld P.; DeRooij N.; Zemel J. Physical mechanisms for chemically sensitive semiconductor devices. Nature 1978, 273, 438. 10.1038/273438a0. [DOI] [Google Scholar]
  26. Van Hal R.; Eijkel J. C.; Bergveld P. A novel description of ISFET sensitivity with the buffer capacity and double-layer capacitance as key parameters. Sens. Actuators, B 1995, 24, 201–205. 10.1016/0925-4005(95)85043-0. [DOI] [Google Scholar]
  27. Sivakumarasamy R.; Hartkamp R.; Siboulet B.; Dufrêche J.-F.; Nishiguchi K.; Fujiwara A.; Clément N. Selective layer-free blood serum ionogram based on ion-specific interactions with a nanotransistor. Nat. Mater. 2018, 17, 464. 10.1038/s41563-017-0016-y. [DOI] [PubMed] [Google Scholar]
  28. Parizi K. B.; Xu X.; Pal A.; Hu X.; Wong H. P. ISFET pH sensitivity: Counter-Ions play a key role. Sci. Rep. 2017, 7, 41305 10.1038/srep41305. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Jamasb S.; Collins S.; Smith R. L. A physical model for drift in pH ISFETs. Sens. Actuators, B 1998, 49, 146–155. 10.1016/S0925-4005(98)00040-9. [DOI] [Google Scholar]
  30. Martz T. R.; Connery J. G.; Johnson K. S. Testing the Honeywell Durafet for seawater pH applications. Limnol. Oceanogr.: Methods 2010, 8, 172–184. 10.4319/lom.2010.8.172. [DOI] [Google Scholar]
  31. Merriman B.; Torrent I.; Rothberg J. M. Progress in ion torrent semiconductor chip based sequencing. Electrophoresis 2012, 33, 3397–3417. 10.1002/elps.201200424. [DOI] [PubMed] [Google Scholar]
  32. Ang P. K.; Chen W.; Wee A. T. S.; Loh K. P. Solution-gated epitaxial graphene as pH sensor. J. Am. Chem. Soc. 2008, 130, 14392–14393. 10.1021/ja805090z. [DOI] [PubMed] [Google Scholar]
  33. Hasan N.; Hou B.; Moore A. L.; Radadia A. D. Enhanced ionic sensitivity in solution-gated graphene-hexagonal boron nitride heterostructure field-effect transistors. Adv. Mater. Technol. 2018, 3, 1800133 10.1002/admt.201800133. [DOI] [Google Scholar]
  34. Yan F.; Zhang M.; Li J. Solution-Gated Graphene Transistors for Chemical and Biological Sensors. Adv. Healthcare Mater. 2014, 3, 313–331. 10.1002/adhm.201300221. [DOI] [PubMed] [Google Scholar]
  35. Loan P. T. K.; Wu D.; Ye C.; Li X.; Tra V. T.; Wei Q.; Fu L.; Yu A.; Li L.-J.; Lin C.-T. Hall effect biosensors with ultraclean graphene film for improved sensitivity of label-free DNA detection. Biosens. Bioelectron. 2018, 99, 85–91. 10.1016/j.bios.2017.07.045. [DOI] [PubMed] [Google Scholar]
  36. Dean C. R.; Young A. F.; Meric I.; Lee C.; Wang L.; Sorgenfrei S.; Watanabe K.; Taniguchi T.; Kim P.; Shepard K. L.; et al. Boron nitride substrates for high-quality graphene electronics. Nat. Nanotechnol. 2010, 5, 722–726. 10.1038/nnano.2010.172. [DOI] [PubMed] [Google Scholar]
  37. Geim A. K.; Grigorieva I. V. Van der Waals heterostructures. Nature 2013, 499, 419. 10.1038/nature12385. [DOI] [PubMed] [Google Scholar]
  38. Kim K.; Choi J.-Y.; Kim T.; Cho S.-H.; Chung H.-J. A role for graphene in silicon-based semiconductor devices. Nature 2011, 479, 338. 10.1038/nature10680. [DOI] [PubMed] [Google Scholar]
  39. Meric I.; Dean C. R.; Petrone N.; Wang L.; Hone J.; Kim P.; Shepard K. L. Graphene field-effect transistors based on boron–nitride dielectrics. Proc. IEEE 2013, 101, 1609–1619. 10.1109/JPROC.2013.2257634. [DOI] [Google Scholar]
  40. Moon P.; Koshino M. Electronic properties of graphene/hexagonal-boron-nitride moiré superlattice. Phys. Rev. B 2014, 90, 155406 10.1103/PhysRevB.90.155406. [DOI] [Google Scholar]
  41. Yankowitz M.; Xue J.; LeRoy B. J. Graphene on hexagonal boron nitride. J. Phys.: Condens. Matter 2014, 26, 303201 10.1088/0953-8984/26/30/303201. [DOI] [PubMed] [Google Scholar]
  42. Kovtyukhova N. I.; Perea-López N.; Terrones M.; Mallouk T. E. Atomically Thin Layers of Graphene and Hexagonal Boron Nitride Made by Solvent Exfoliation of Their Phosphoric Acid Intercalation Compounds. ACS Nano 2017, 11, 6746–6754. 10.1021/acsnano.7b01311. [DOI] [PubMed] [Google Scholar]
  43. Pratik K. C.; Rai A.; Ashton T. S.; Moore A. L. APCVD hexagonal boron nitride thin films for passive near-junction thermal management of electronics. Nanotechnology 2017, 28, 505705 10.1088/1361-6528/aa97ae. [DOI] [PubMed] [Google Scholar]
  44. Choi D. S.; Poudel N.; Park S.; Akinwande D.; Cronin S. B.; Watanabe K.; Taniguchi T.; Yao Z.; Shi L. Large Reduction of Hot Spot Temperature in Graphene Electronic Devices with Heat-Spreading Hexagonal Boron Nitride. ACS Appl. Mater. Interfaces 2018, 10, 11101–11107. 10.1021/acsami.7b16634. [DOI] [PubMed] [Google Scholar]
  45. Ghittorelli M.; Lingstedt L.; Romele P.; Crăciun N. I.; Kovács-Vajna Z. M.; Blom P. W.; Torricelli F. High-sensitivity ion detection at low voltages with current-driven organic electrochemical transistors. Nat. Commun. 2018, 9, 1441 10.1038/s41467-018-03932-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Masvidal-Codina E.; Illa X.; Dasilva M.; Calia A. B.; Dragojević T.; Vidal-Rosas E. E.; Prats-Alfonso E.; Martínez-Aguilar J.; Jose M.; Garcia-Cortadella R.; et al. High-resolution mapping of infraslow cortical brain activity enabled by graphene microtransistors. Nat. Mater. 2019, 18, 280–288. 10.1038/s41563-018-0249-4. [DOI] [PubMed] [Google Scholar]
  47. Morf W. E.The Principles of Ion-Selective Electrodes and of Membrane Transport.; Elsevier, 2012; Vol. 2. [Google Scholar]
  48. Li H.; Zhu Y.; Islam M. S.; Rahman M. A.; Walsh K. B.; Koley G. Graphene field effect transistors for highly sensitive and selective detection of K+ ions. Sens. Actuators, B 2017, 253, 759–765. 10.1016/j.snb.2017.06.129. [DOI] [Google Scholar]
  49. Maehashi K.; Sofue Y.; Okamoto S.; Ohno Y.; Inoue K.; Matsumoto K. Selective ion sensors based on ionophore-modified graphene field-effect transistors. Sens. Actuators, B 2013, 187, 45–49. 10.1016/j.snb.2012.09.033. [DOI] [Google Scholar]
  50. Kim J.; Wang L.; Bourouina T.; Cui T. Ion sensitive field effect transistor based on graphene and ionophore hybrid membrane for phosphate detection. Microsyst. Technol. 2019, 25, 3357–3364. 10.1007/s00542-018-4200-z. [DOI] [Google Scholar]
  51. Li P.; Liu B.; Zhang D.; Sun Ye.; Liu J. Graphene field-effect transistors with tunable sensitivity for high performance Hg (II) sensing. Appl. Phys. Lett. 2016, 109, 153101 10.1063/1.4964347. [DOI] [Google Scholar]
  52. Jin C.; Kim J.; Suh J.; Shi Z.; Chen B.; Fan X.; Kam M.; Watanabe K.; Taniguchi T.; Tongay S. Interlayer electron–phonon coupling in WSe2/hBN heterostructures. Nat. Phys. 2017, 13, 127. 10.1038/nphys3928. [DOI] [Google Scholar]
  53. Geick R.; Perry C.; Rupprecht G. Normal modes in hexagonal boron nitride. Phys. Rev. 1966, 146, 543 10.1103/PhysRev.146.543. [DOI] [Google Scholar]
  54. Serrano J.; Bosak A.; Arenal R.; Krisch M.; Watanabe K.; Taniguchi T.; Kanda H.; Rubio A.; Wirtz L. Vibrational properties of hexagonal boron nitride: inelastic X-ray scattering and ab initio calculations. Phys. Rev. Lett. 2007, 98, 095503 10.1103/PhysRevLett.98.095503. [DOI] [PubMed] [Google Scholar]
  55. Casiraghi C.Raman Spectroscopy of Graphene. In Spectroscopic Properties of Inorganic and Organometallic Compounds: Techniques, Materials and Applications; the Royal Society of Chemistry: London, U.K: 2012; pp 29–56. [Google Scholar]
  56. Ferrari A. C. Raman spectroscopy of graphene and graphite: disorder, electron–phonon coupling, doping and nonadiabatic effects. Solid State Commun. 2007, 143, 47–57. 10.1016/j.ssc.2007.03.052. [DOI] [Google Scholar]
  57. Malard L.; Pimenta M.; Dresselhaus G.; Dresselhaus M. Raman spectroscopy in graphene. Phys. Rep. 2009, 473, 51–87. 10.1016/j.physrep.2009.02.003. [DOI] [Google Scholar]
  58. Xia J.; Chen F.; Li J.; Tao N. Measurement of the quantum capacitance of graphene. Nat. Nanotechnol. 2009, 4, 505. 10.1038/nnano.2009.177. [DOI] [PubMed] [Google Scholar]
  59. Fang T.; Konar A.; Xing H.; Jena D. Carrier statistics and quantum capacitance of graphene sheets and ribbons. Appl. Phys. Lett. 2007, 91, 092109 10.1063/1.2776887. [DOI] [Google Scholar]
  60. Chen F.; Qing Q.; Xia J.; Li J.; Tao N. Electrochemical gate-controlled charge transport in graphene in ionic liquid and aqueous solution. J. Am. Chem. Soc. 2009, 131, 9908–9909. 10.1021/ja9041862. [DOI] [PubMed] [Google Scholar]
  61. Raimondo J. V.; Burman R. J.; Katz A. A.; Akerman C. J. Ion dynamics during seizures. Front. Cell. Neurosci. 2015, 9, 419 10.3389/fncel.2015.00419. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Schwartzkroin P. A.; Baraban S. C.; Hochman D. W. Osmolarity, ionic flux, and changes in brain excitability. Epilepsy Res. 1998, 32, 275–285. 10.1016/S0920-1211(98)00058-8. [DOI] [PubMed] [Google Scholar]
  63. Gao W.; Emaminejad S.; Nyein H. Y. Y.; Challa S.; Chen K.; Peck A.; Fahad H. M.; Ota H.; Shiraki H.; Kiriya D.; et al. Fully integrated wearable sensor arrays for multiplexed in situ perspiration analysis. Nature 2016, 529, 509. 10.1038/nature16521. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Bandodkar A. J.; Molinnus D.; Mirza O.; Guinovart T.; Windmiller J. R.; Valdés-Ramírez G.; Andrade F. J.; Schöning M. J.; Wang J. Epidermal tattoo potentiometric sodium sensors with wireless signal transduction for continuous non-invasive sweat monitoring. Biosens. Bioelectron. 2014, 54, 603–609. 10.1016/j.bios.2013.11.039. [DOI] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

ao1c02222_si_001.pdf (2.1MB, pdf)

Articles from ACS Omega are provided here courtesy of American Chemical Society

RESOURCES