The authors present a study on the evolution behaviors of the transfer characteristics of MoS2 and WSe2 field-effect transistor biosensors when they are subjected to tumor necrosis factor-alpha and streptavidin solutions with varying analyte concentrations. Both MoS2 and WSe2 sensors exhibit very low detection limits (∼60 fM for tumor necrosis factor-alpha detection; ∼70 fM for streptavidin detection). However, WSe2 sensors exhibit the higher linear-regime sensitivities in comparison with MoS2 sensors. In particular, WSe2 sensors exhibit high linear-regime sensitivities up to ∼1.54%/fM for detecting streptavidin at a concentration of ∼70 fM. Such relatively higher sensitivities obtained from WSe2 sensors are attributed to their intrinsic ambipolar transfer characteristics, which make their ON-state carrier concentrations significantly lower than those of MoS2 sensors, and therefore, the target-molecule-induced doping effect results in more prominent channel conductance modulation in WSe2 transistor sensors than in MoS2 sensors. Furthermore, this work strongly implies that the target-molecule-induced surface scattering also plays an important role in determining the response behaviors of the sensors made from atomically layered materials. Especially, the competition between target-molecule-induced p-doping and surface-scattering effects is responsible for the sensor behavior variation observed in the p-type conduction branch of WSe2 sensors. This work advances the critical device physics highly relevant with the fabrication and implementation of reliable nanoelectronic biosensors based on emerging atomically layered semiconductors.

Atomically layered transition metal dichalcogenides (TMDCs), such as molybdenum disulfide (MoS2) and tungsten diselenide (WSe2), have been widely studied as attractive materials for enabling new nanoelectronic and optoelectronic device applications.1–13 Especially, semiconducting TMDCs can be potentially used for making new transistor-based nanoelectronic biosensors with unprecedented biosensing capability. Wang et al. and Sarkar et al. recently demonstrated that MoS2 transistor biosensors can enable 100–400 femtomolar (fM)-level detection limits for detecting cancer-related biomarkers.14,15 Our recent work demonstrated multiple MoS2 transistor biosensors with highly consistent sensor response behaviors.16 Such sensors can work synergistically to enable a standard curve for quantifying tumor necrosis factor-alpha (TNF-α) molecules at fM-level detection limits.16 Lee et al. realized the direct antibody functionalization on bare MoS2 surfaces with no need of the dielectric coating, which is based on the hydrophobicity of MoS2 layers.17 This work has significantly simplified the fabrication process for making MoS2 biosensors.17 Currently, the potential application of other semiconducting TMDCs in making electronic biosensors remains experimentally unexplored. Especially, other TMDCs such as WSe2 and WS2, which have very different transport properties as compared to MoS2, could enable new device mechanisms for improving the biodetection capability. The transistor sensor structures based on these TMDCs are worth being created and compared with MoS2-based sensors in terms of transport characteristics.

In this article, we report a study on the sensor response behaviors of MoS2 and WSe2 field-effect transistor (FET) biosensors. Such sensors were tested for TNF-α and streptavidin detections. The transfer characteristics of these FET sensors measured at varying analyte concentrations were systematically investigated and compared. We found that although both MoS2 and WSe2 sensors exhibit very low detection limits, WSe2 sensors exhibit the higher linear-regime (or ON-state) sensitivities in comparison with MoS2 sensors. This is attributed to the ambipolar transport property of WSe2 FETs. Such a property makes the ON-state free carrier concentration in a WSe2 FET channel significantly lower than that in a MoS2 channel, and therefore, the target-molecule-induced p-doping effect results in a more prominent conductance modulation in the WSe2 sensor than in the MoS2 sensor. Furthermore, our work also implied that in addition to the target-molecule-induced doping effect, the target-molecule-induced surface scattering of free carriers also plays an important role in determining the response behaviors of the sensors made from atomically layered materials. Especially, we observed a variation of the sensor response behavior in the p-type conduction branch of WSe2 sensors, which is attributed to the competition between target-molecule-induced p-doping and surface-scattering effects.

Figure 1(a) schematically illustrates the FET biosensor with a semiconducting few-layer TMDC channel. In this work, pristine MoS2 and WSe2 FETs were fabricated using our previously reported printing methods.18–20 Specifically, few-layer MoS2 and WSe2 flakes (thickness, 5–7 nm) were exfoliated onto p+-Si substrates coated with 300 nm thick SiO2 layers. The drain/source (D/S) contacts were fabricated on MoS2 and WSe2 flakes using photolithography followed with deposition of Ti/Au (5/50 nm) and lift-off in solvents. In an as-fabricated FET, the p+-doped Si substrate was used as the back gate with a 300 nm thick SiO2 back-gate dielectric. For the biosensing test, the D/S contacts were passivated by 100 nm thick SiO2 films, which were deposited using a sputtering tool. Figures 1(b) and 1(c) show the top-view optical micrographs of a representative pristine MoS2 FET sensor and a representative WSe2 FET sensor, respectively. For all sensors, the TMDC channel thickness is 5–7 nm, the channel length is ∼10 μm, and the channel width exhibits a device-to-device variation in the range of 7–30 μm. The FET characterization was performed using a HP4145 semiconductor parameter analyzer connected with a CPX-HF LakeShore probe station.

Fig. 1.

(Color online) (a) Schematic illustration of the FET biosensor with a few-layer TMDC sensing channel; (b) and (c) the optical micrographs of a MoS2 FET sensor and a WSe2 FET sensor, respectively. For both sensors, the channel thickness is ∼5–7 nm; the SiO2 back-gate dielectric thickness is 300 nm; the channel length is 10 μm; the channel width exhibits a device-to-device variation in the range of 7–30 μm; and the D/S contacts are 5 nm Ti/50 nm Au, which are passivated with 100 nm thick SiO2 films.

Fig. 1.

(Color online) (a) Schematic illustration of the FET biosensor with a few-layer TMDC sensing channel; (b) and (c) the optical micrographs of a MoS2 FET sensor and a WSe2 FET sensor, respectively. For both sensors, the channel thickness is ∼5–7 nm; the SiO2 back-gate dielectric thickness is 300 nm; the channel length is 10 μm; the channel width exhibits a device-to-device variation in the range of 7–30 μm; and the D/S contacts are 5 nm Ti/50 nm Au, which are passivated with 100 nm thick SiO2 films.

Close modal

We functionalized MoS2 and WSe2 FET channels with antihuman TNF-α antibody receptors for detecting TNF-α molecules. The functionalization process is described as follows. First, the devices are immersed in 5% (3-aminopropyl) triethoxysilane (APTES) in ethanol for 1 h. After silanization, the devices are rinsed with phosphate buffered saline (PBS) followed with N2 blow dry. The sensors are subsequently dipped into 5% glutaraldehyde in PBS for 2 h followed with rinsing with PBS. Finally, 50 μg/ml antihuman TNF-α antibody solution droplets are dropped on the sensor surfaces, and all sensors are incubated for 45 min. After the antibody functionalization, the sensors are ready for the TNF-α detection test, which involves the following steps: (1) incubating an as-functionalized FET sensor in a target TNF-α solution (solvent: PBS) for 20–30 min to assure that the binding reaction between target molecules and receptors reaches to the equilibrium state; (2) cleaning up unreacted TNF-α molecules with deionized (DI) water; and (3) blowing dry the sensor and measuring its transfer characteristics using a semiconductor analyzer. To detect streptavidin molecules, biotin receptors need to be functionalized on MoS2 and WSe2 FET channels. The functionalization process includes (1) soaking the sensors in 5% APTES in ethanol for 1 h followed with rinsing with PBS and N2 blow dry; (2) coating the sensors with 0.1 mg/ml (+)-Biotin N-hydroxysuccinimide ester (NHS-biotin) in PBS and incubating the sensors for 1 h. After biotin functionalization, streptavidin detection is performed using the process same as that used for detecting TNF-α molecules.

Figure 2 shows the transfer characteristics [i.e., drain/source current (IDS)–gate voltage (VG) curves] of a representative pristine MoS2 FET [Figs. 2(a) and 2(b)] and a representative pristine WSe2 FET [Figs. 2(c) and 2(d)] measured under a fixed drain/source voltage (VDS = 1 V). Figures 2(a) and 2(c) display the IDS-VG curves plotted in the linear scale, and Figs. 2(b) and 2(d) display the respective curves plotted in the semilogarithmic scale. The transfer characteristics of the MoS2 FET exhibit a typical n-type conduction behavior (i.e., electron-dominated conduction behavior) within the VG range of ±100 V (this is an applicable VG range for 300 nm thick SiO2 gate dielectrics). For our MoS2 FETs, the field-effect electron mobility values were measured to be in the range of 30–55 cm2/Vs. The transfer characteristics of the WSe2 FET exhibit a typical ambipolar conduction behavior [i.e., both p-type (or hole-dominated) and n-type (or electron-dominated) conduction branches appear within the VG range of ±100 V]. For our WSe2 FETs, the n-branch (or electron) mobility values were measured to be in the range of 1–3 × 10−3 cm2/Vs, and the p-branch (or hole) mobility values were measured to be in the range of 16–28 cm2/Vs. For all WSe2 FETs, their IDS-VG characteristic curves have an electrically neutral point (ENP, VGENP ∼ −20−10 V). The areal carrier density (na) in a 2D-layer-based FET biased at the ON-state can be evaluated using na = LIDS(ON)/eμWVDS, in which e is the elementary charge; μ is the carrier mobility determined in the transistor characterization; L and W are the FET channel length and width, respectively; IDS(ON) is the ON-state drain–source current measured from the FET; and VDS is the drain–source voltage applied across the FET channel (here, VDS = 1 V). The na values for MoS2 FETs, n-type conduction branches of WSe2 FETs, and p-type conduction branches of WSe2 FETs are estimated in the ranges of 3–6 × 1012, 5–8 × 1010, and 1–2 × 1011 cm−2, respectively. The difference between MoS2 and WSe2 FETs in their transfer characteristics is attributed to the intrinsically different electronic band structures of MoS2 and WSe2 layers.21 In addition, as shown in Fig. 2(d), there is a slight current fluctuation appearing in the proximity of the electrically neutral point of the WSe2 FET. Because the OFF current level measured at this electrically neutral point is about 1–2 pA and the precision of our semiconductor analyzer is about 1–5 pA, such an observed current fluctuation is mainly attributed to the electrical noise from the measurement tool.

Fig. 2.

(Color online) Transfer characteristics of (a) or (b) a representative pristine MoS2 FET that exhibits a typical n-type conduction behavior and (c) or (d) a representative pristine WSe2 FET that exhibits a typical ambipolar conduction behavior. (a) and (c) The IDS-VG curves plotted in the linear scale; (b) and (d) the corresponding curves plotted in the semilogarithmic scale. For most MoS2 FETs, the field-effect mobility values are in the range of 30–55 cm2/Vs. For most WSe2 FETs, the n-branch (or electron) mobility values are in the range of 1–3 × 10−3 cm2/Vs, and the p-branch (or hole) mobility values are in the range of 16–28 cm2/Vs.

Fig. 2.

(Color online) Transfer characteristics of (a) or (b) a representative pristine MoS2 FET that exhibits a typical n-type conduction behavior and (c) or (d) a representative pristine WSe2 FET that exhibits a typical ambipolar conduction behavior. (a) and (c) The IDS-VG curves plotted in the linear scale; (b) and (d) the corresponding curves plotted in the semilogarithmic scale. For most MoS2 FETs, the field-effect mobility values are in the range of 30–55 cm2/Vs. For most WSe2 FETs, the n-branch (or electron) mobility values are in the range of 1–3 × 10−3 cm2/Vs, and the p-branch (or hole) mobility values are in the range of 16–28 cm2/Vs.

Close modal

Figure 3 displays the sensor response behaviors of two representative MoS2 FET biosensors, which were used for TNF-α and streptavidin detections, respectively. In particular, Fig. 3(a) shows the transfer characteristics of an antibody-functionalized MoS2 FET, which were measured at a set of incremental TNF-α concentrations (i.e., n = 0, 60 fM, 600 fM, 6 pM, and 60 pM). Figures 3(b) and 3(c) show the transfer characteristics of a biotin-functionalized MoS2 FET, which were measured at a set of incremental streptavidin concentrations (i.e., n = 0, 70 fM, 700 fM, 7 pM, and 70 pM). In either of these two detection tests, the IDS measured under a given VG in the linear regime (or the ON state) of the MoS2 FET decreases with increasing the analyte concentration (n), as denoted by the downward arrows in Fig. 3. Such a sensor response behavior is referred to as “IDS-reduced response behavior” for the following discussion. This IDS-reduced response behavior was reliably observed in all MoS2 FET biosensors for both TNF-α and streptavidin detections. This response behavior is attributed to two possible mechanisms: (1) the target-molecule-induced surface-charge-transfer doping into MoS2 layers;22 (2) the target-molecule-induced surface scattering of the free carriers transporting in MoS2 layers. Here, we tentatively speculate that the adsorption of TNF-α or streptavidin molecules on a MoS2 FET sensor results in a p-doping effect in MoS2 layers, which could reduce the effective free electron concentration in the MoS2 FET channel, resulting in an IDS-reduced response behavior. Furthermore, the linear-regime (or ON-state) channel conductance of a FET with a few-layer TMDC channel is anticipated to be extremely sensitive to the surface scattering effect because of the atomically thin thickness of the channel (5–7 nm). The increase of the analyte concentration (n) results in the higher areal density of target molecules adsorbed on the sensor area at the equilibrium state of the binding reaction, which is expected to result in more prominent surface scattering of free carriers and degrade the ON-state channel conductance, therefore leading to an IDS-reduced response behavior. Based on this speculation, although currently we cannot quantitatively evaluate the relative contribution magnitudes of target-molecule-induced p-doping and surface-scattering effects, it is expected that MoS2 FET sensors always exhibit the IDS-reduced response behavior in TNF-α and streptavidin detections because both of these two effects are expected to result in conductance reduction in n-type (or electron-dominated) MoS2 FET channels. The selectivity (or specificity) of such MoS2 FET sensors was previously studied by us.16 Specifically, the sensors functionalized with antihuman TNF-α antibody receptors were employed for detecting interleukin-6 (IL-6) cytokine molecules that are not specific to TNF-α antibody receptors.16 Under such a detection condition, the sensors generated negligible electrical responses to IL-6 molecules and therefore exhibit a high detection specificity.16 

Fig. 3.

(Color online) Sensor response behavior of MoS2 FET biosensors: (a) Transfer characteristics of an antibody-functionalized MoS2 FET measured at a set of incremental TNF-α concentrations (n = 0, 60 fM, 600 fM, 6 pM, and 60 pM); (b) transfer characteristics of a biotin-functionalized MoS2 FET measured at a set of incremental streptavidin concentrations (n = 0, 70 fM, 700 fM, 7 pM, and 70 pM); (c) zoomed view of the transfer characteristic curves displayed within the dashed box in (b). For both detections, the IDS measured under a given VG in the linear regime of the MoS2 FET decreases with increasing n (i.e., an IDS-reduced response behavior, as denoted by the downward arrows).

Fig. 3.

(Color online) Sensor response behavior of MoS2 FET biosensors: (a) Transfer characteristics of an antibody-functionalized MoS2 FET measured at a set of incremental TNF-α concentrations (n = 0, 60 fM, 600 fM, 6 pM, and 60 pM); (b) transfer characteristics of a biotin-functionalized MoS2 FET measured at a set of incremental streptavidin concentrations (n = 0, 70 fM, 700 fM, 7 pM, and 70 pM); (c) zoomed view of the transfer characteristic curves displayed within the dashed box in (b). For both detections, the IDS measured under a given VG in the linear regime of the MoS2 FET decreases with increasing n (i.e., an IDS-reduced response behavior, as denoted by the downward arrows).

Close modal

Figure 4 shows the sensor response behaviors of two representative WSe2 FET biosensors, which were used for TNF-α and streptavidin detections, respectively. Their response behaviors are noticeably different from those of MoS2 FET sensors. In particular, Fig. 4(a) displays the transfer characteristics of an antibody-functionalized WSe2 FET, which were measured at a set of incremental TNF-α concentrations (n = 0, 60 fM, 300 fM, 600 fM, 3 pM, and 6 pM). Figure 4(b) is the zoomed view of the transfer characteristic curves displayed within the dashed box denoted in Fig. 4(a). Figure 4(c) shows the transfer characteristics of a biotin-functionalized WSe2 FET, which were measured at a set of incremental streptavidin concentrations (n = 0, 70 fM, 700 fM, 7 pM, and 70 pM). Figure 4(d) shows the zoomed view of the transfer characteristic curves displayed within the dashed box denoted in Fig. 4(c). In the TNF-α detection test, after the WSe2 FET channel is functionalized with antihuman TNF-α antibody receptors, the sensor response in its p-type conduction branch is significantly suppressed, but its n-type conduction branch still exhibits a prominent IDS-reduced response behavior, as denoted by the downward arrow in Fig. 4(b). We have not identified the reason responsible for the suppression of the p-branch response caused by the TNF-α antibody functionalization. In the streptavidin detection test, both n- and p-type conduction branches of the WSe2 FET exhibit detectable responses, as shown in Fig. 4(d). The n-type conduction branch exhibits an IDS-reduced response behavior [denoted by the downward arrow in Fig. 4(d)], which is similar to the response behavior of the MoS2 sensor in streptavidin detection [Figs. 3(b) and 3(c)]. However, within the p-type conduction branch of the WSe2 FET sensor, the IDS measured under a given VG in the linear regime (or ON state) increases with increasing the streptavidin concentration (n) [i.e., an IDS-enhanced response behavior is observed, as denoted by the upward arrow in Fig. 4(d)]. This IDS-enhanced response behavior further implies that the adsorption of streptavidin molecules on the WSe2 FET channel induces a p-doping effect in WSe2 layers and increases the hole-dominated channel conductance.

Fig. 4.

(Color online) Sensor response behavior of WSe2 FET biosensors: (a) or (b) Transfer characteristics of an antibody-functionalized WSe2 FET measured at a set of incremental TNF-α concentrations (n = 0, 60 fM, 300 fM, 600 fM, 3 pM, and 6 pM) [here, (b) is the zoomed view of the transfer characteristic curves displayed within the dashed box in (a)]; (c) or (d) transfer characteristics of a biotin-functionalized WSe2 FET measured at a set of incremental streptavidin concentrations (n = 0, 70 fM, 700 fM, 7 pM, and 70 pM) [here, (d) is the zoomed view of the transfer characteristic curves displayed within the dashed box in (c)]. For the TNF-α detection, the p-branch response of the WSe2 FET is significantly suppressed, but its n-type conduction branch still exhibits a prominent IDS-reduced response behavior, as denoted by the downward arrow. For the streptavidin detection, both n- and p-type conduction branches of the WSe2 FET exhibit detectable responses. The n-type branch exhibits an IDS-reduced response behavior (denoted by the downward arrow), whereas in the p-type branch, the IDS measured under a given VG in the linear regime increases with increasing n (i.e., an IDS-enhanced response behavior, as denoted by the upward arrow).

Fig. 4.

(Color online) Sensor response behavior of WSe2 FET biosensors: (a) or (b) Transfer characteristics of an antibody-functionalized WSe2 FET measured at a set of incremental TNF-α concentrations (n = 0, 60 fM, 300 fM, 600 fM, 3 pM, and 6 pM) [here, (b) is the zoomed view of the transfer characteristic curves displayed within the dashed box in (a)]; (c) or (d) transfer characteristics of a biotin-functionalized WSe2 FET measured at a set of incremental streptavidin concentrations (n = 0, 70 fM, 700 fM, 7 pM, and 70 pM) [here, (d) is the zoomed view of the transfer characteristic curves displayed within the dashed box in (c)]. For the TNF-α detection, the p-branch response of the WSe2 FET is significantly suppressed, but its n-type conduction branch still exhibits a prominent IDS-reduced response behavior, as denoted by the downward arrow. For the streptavidin detection, both n- and p-type conduction branches of the WSe2 FET exhibit detectable responses. The n-type branch exhibits an IDS-reduced response behavior (denoted by the downward arrow), whereas in the p-type branch, the IDS measured under a given VG in the linear regime increases with increasing n (i.e., an IDS-enhanced response behavior, as denoted by the upward arrow).

Close modal

Figures 3 and 4 also show that both MoS2 and WSe2 sensors exhibit very low detection limits (∼60 fM for TNF-α detection; ∼70 fM for streptavidin detection). To quantitatively compare the detection sensitivities of MoS2 and WSe2 FET sensors used for TNF-α and streptavidin quantifications, Fig. 5 plots the standard curves [i.e., the sensor response quantity (R) plotted as a function of the analyte concentration (n)] for Fig. 5(a) TNF-α quantification results acquired using a MoS2 FET sensor and a WSe2 FET sensor (only the n-type branch response is available for both sensors), and Fig. 5(b) streptavidin quantification results measured using a MoS2 sensor and a WSe2 sensor (both n- and p-branch responses of the WSe2 FET sensor are plotted). Here, the sensor response quantity (R) is defined as the relative change in IDS caused by the introduction of the analyte solution with a specific concentration (n); i.e., R = 100% × (IDS − IDS n=0)/IDS n=0, in which IDS n=0 is the IDS measured at n = 0 under a specific VG (VDS is always set to 1 V in this work). In Fig. 5, the R value at each n for a specific FET sensor is the mean value of (IDS − IDS n=0)/IDS n=0 data measured under different VG values within the linear regime (or ON-state regime, i.e., VG > threshold voltage Vth for the n-branch responses; VG < Vth for the p-branch responses) of the FET. Specifically, to evaluate the R values from the n-branch of a MoS2 sensor for TNF-α detection, the n-branch of a MoS2 sensor for streptavidin detection, the n-branch of a WSe2 sensor for TNF-α detection, the n-branch of a WSe2 sensor for streptavidin detection, and the p-branch of a WSe2 sensor for streptavidin detection, the VG ranges used for statistically calculating R values are 0–100, −75–100, 80–100, −15–100, and −100 to −65 V, respectively. The standard deviation of each statistically averaged R value is also plotted in Fig. 5. Figure 5 shows that for either TNF-α or streptavidin detection, the WSe2 FET sensor exhibits the higher sensitivity as compared to the MoS2 FET. In particular, the detection sensitivity at a specific analyte concentration (n) is the slope of the standard curve (or the R-n response curve) at n. For the TNF-α detection at n ∼ 60 fM, the WSe2 FET sensor exhibits a sensitivity of −0.47%/fM, and the MoS2 sensor exhibits a sensitivity of −0.18%/fM. For the streptavidin detection at n ∼ 70 fM, the detection sensitivities achieved from the n-/p-type conduction branches of the WSe2 FET sensor and the MoS2 sensor are −0.29%/fM, +1.54%/fM, and −0.08%/fM, respectively. The relatively higher sensitivity obtained from the WSe2 sensor is attributed to its intrinsic ambipolar transfer characteristic within the applicable VG range of ±100 V. Because of this ambipolar IDS-VG characteristic, given an applicable VG range (e.g., ±100 V for 300 nm SiO2 gate dielectrics), the ON-state free carrier concentration (typically 1010–1011 cm−2) in a WSe2 FET sensor is significantly lower than that in a MoS2 FET sensor (typically larger than 1012 cm−2). Therefore, the target-molecule-induced p-doping effect results in much more prominent conductance modulation in the WSe2 sensor than in the MoS2 sensor.

Fig. 5.

(Color online) Standard curves [i.e., the sensor response quantity (R) as a function of the analyte concentration (n)] for (a) TNF-α quantification measured by a MoS2 FET sensor and a WSe2 FET sensor (only the n-type branch response is available); (b) streptavidin quantification measured by a MoS2 FET sensor and a WSe2 sensor (both n- and p-type branch responses are plotted). For either of these two detections, the WSe2 FET sensor exhibits the higher sensitivity as compared to the MoS2 FET. For the TNF-α detection at n ∼ 60 fM, the WSe2 FET sensor exhibits a sensitivity of −0.47%/fM; the MoS2 sensor exhibits a sensitivity of −0.18%/fM. For the streptavidin detection at n ∼ 70 fM, the sensitivities achieved from the n-/p-type conduction branches of the WSe2 FET sensor and the MoS2 sensor are measured to be −0.29%/fM, +1.54%/fM, and −0.08%/fM, respectively. The relatively higher sensitivity of the WSe2 sensor is attributed to its ambipolar transport character.

Fig. 5.

(Color online) Standard curves [i.e., the sensor response quantity (R) as a function of the analyte concentration (n)] for (a) TNF-α quantification measured by a MoS2 FET sensor and a WSe2 FET sensor (only the n-type branch response is available); (b) streptavidin quantification measured by a MoS2 FET sensor and a WSe2 sensor (both n- and p-type branch responses are plotted). For either of these two detections, the WSe2 FET sensor exhibits the higher sensitivity as compared to the MoS2 FET. For the TNF-α detection at n ∼ 60 fM, the WSe2 FET sensor exhibits a sensitivity of −0.47%/fM; the MoS2 sensor exhibits a sensitivity of −0.18%/fM. For the streptavidin detection at n ∼ 70 fM, the sensitivities achieved from the n-/p-type conduction branches of the WSe2 FET sensor and the MoS2 sensor are measured to be −0.29%/fM, +1.54%/fM, and −0.08%/fM, respectively. The relatively higher sensitivity of the WSe2 sensor is attributed to its ambipolar transport character.

Close modal

In this work, we further found that the WSe2 FETs treated with O2 plasma exhibit a more prominent p-type conduction behavior in comparison with pristine ones. This is attributed to the O2-induced p-doping in TMDC layers.19,20,23 Figure 6 shows the sensor response behavior of a representative O2-doped WSe2 FET. The IDS-VG characteristic curves of such O2-doped WSe2 FETs exhibit a more positive electrically neutral point (VGENP = 87–100 V) and therefore a broader applicable VG range (−100–87 V) for showing the p-type conduction branch, as compared to those of pristine WSe2 sensors. The FET sensor shown in Fig. 6 was specifically functionalized with biotin receptors for detecting streptavidin molecules. Multiple transfer characteristic curves of this WSe2 FET sensor were measured at a set of incremental streptavidin concentrations (n = 0, 70 fM, 350 fM, 700 fM, and 3.5 pM). As shown in Fig. 6, in the VG range relatively close to the electrically neutral point (i.e., VG = 62–87 V for this representative device, which is referred to as the “slightly p-type regime”), the sensor exhibits an IDS-enhanced response behavior, as denoted by the upward arrow. This behavior is similar to that observed in the p-type branch of an undoped WSe2 sensor [Fig. 4(d)]. However, in the VG range significantly away from the electrically neutral point (i.e., VG = −100−62 V for this device, which is referred to as the “deeply p-type regime”), the sensor exhibits an IDS-reduced response behavior, as denoted by the downward arrow in Fig. 6. The transition between IDS-reduced and IDS-enhanced response behaviors occurs at VG ∼ 62 V for this device (all other O2-doped WSe2 FETs exhibit very similar transition behaviors). Such a transition behavior is attributed to the competition between the target-molecule-induced p-doping effect and the target-molecule-induced surface scattering effect. In particular, within the slightly p-type regime, the free carrier concentration is relatively low, and the target-molecule-induced p-doping effect plays a more dominant role in determining the sensor response behavior as compared to the target-molecule-induced surface scattering effect, therefore resulting in an IDS-enhanced response behavior similar to that observed in undoped WSe2 sensors. Within the deeply p-type regime, free holes are greatly populated by more negative back-gate voltages. In this case, the carrier concentration modulation caused by the target-molecule-induced p-doping becomes less dominant, but the target-molecule-induced surface scattering effect is expected to be more critical in determining the sensor response behavior because the higher carrier concentration leads to the higher probability for free carriers to be scattered by the potential disturbance induced by adsorbed target molecules. Because the surface scattering effect always degrades the carrier mobility, it results in an IDS-reduced sensor response behavior. The transition of the response behaviors observed in the p-type branch of O2-doped WSe2 FET sensors can serve as an important evidence to support our speculation discussed above; i.e., target-molecule-induced doping and surface-scattering effects coexist, and both effects contribute to the sensor response behaviors of the FET sensors made from emerging atomically layered semiconductors.

Fig. 6.

(Color online) Transfer characteristics of a biotin-functionalized WSe2 FET sensor exhibiting a more prominent p-type conduction behavior, measured at a set of incremental streptavidin concentrations (n = 0, 70 fM, 350 fM, 700 fM, and 3.5 pM). In the VG range relatively close to the electrically neutral point of the WSe2 sensor (i.e., VG = 62–87 V for this device), the sensor exhibits an IDS-enhanced response behavior, as denoted by the upward arrow. However, in the VG range significantly away from the electrically neutral point (i.e., VG = −100–62 V for this device), the sensor exhibits an IDS-enhanced response behavior, as denoted by the downward arrow. The transition between IDS-reduced and IDS-enhanced response behaviors occurs at VG ∼ 62 V for this device. Such a transition behavior is attributed to the competition between the target-molecule-induced p-doping effect and the target-molecule-induced surface scattering effect.

Fig. 6.

(Color online) Transfer characteristics of a biotin-functionalized WSe2 FET sensor exhibiting a more prominent p-type conduction behavior, measured at a set of incremental streptavidin concentrations (n = 0, 70 fM, 350 fM, 700 fM, and 3.5 pM). In the VG range relatively close to the electrically neutral point of the WSe2 sensor (i.e., VG = 62–87 V for this device), the sensor exhibits an IDS-enhanced response behavior, as denoted by the upward arrow. However, in the VG range significantly away from the electrically neutral point (i.e., VG = −100–62 V for this device), the sensor exhibits an IDS-enhanced response behavior, as denoted by the downward arrow. The transition between IDS-reduced and IDS-enhanced response behaviors occurs at VG ∼ 62 V for this device. Such a transition behavior is attributed to the competition between the target-molecule-induced p-doping effect and the target-molecule-induced surface scattering effect.

Close modal

Finally, to evaluate the stability of our FET sensors for streptavidin quantification, a biotin-functionalized WSe2 FET sensor was characterized at a set of incremental streptavidin concentrations (n = 0, 70, and 700 fM), and at each streptavidin concentration, the IDS-VG characteristic of this FET sensor was consecutively measured for five times for a stability check of the sensor response behavior. Figure 7 displays the p-branch transfer characteristics measured from this sensor. These characterization results show that the transfer characteristic curves measured at a given streptavidin concentration are consistent with each other, and the typical IDS signal variation measured at a given streptavidin concentration is much smaller than the typical IDS difference between two streptavidin concentrations. Therefore, our FET sensors exhibit a good stability of the sensor response behavior.

Fig. 7.

(Color online) P-branch transfer characteristics of a biotin-functionalized WSe2 FET sensor measured at a set of incremental streptavidin concentrations (n = 0, 70, and 700 fM). At each streptavidin concentration, the IDS-VG characteristic of this FET sensor was consecutively measured for five times for a stability check of the measurement.

Fig. 7.

(Color online) P-branch transfer characteristics of a biotin-functionalized WSe2 FET sensor measured at a set of incremental streptavidin concentrations (n = 0, 70, and 700 fM). At each streptavidin concentration, the IDS-VG characteristic of this FET sensor was consecutively measured for five times for a stability check of the measurement.

Close modal

We present the fabrication of MoS2 and WSe2-based FET biosensors as well as the systematical comparison of their transfer characteristics measured at various TNF-α and streptavidin concentrations. Both MoS2 and WSe2 sensors exhibit very low detection limits (∼60 fM for TNF-α detection; ∼70 fM for streptavidin detection). Our WSe2 sensors exhibit the higher linear-regime sensitivities as compared to our MoS2 sensors. Specifically, our WSe2 sensors exhibit very high linear-regime sensitivities up to 1.54%/fM for detecting streptavidin at concentration of ∼70 fM. The relatively higher sensitivities of WSe2 sensors are attributed to their ambipolar IDS-VG characteristic, which is expected to make the ON-state carrier concentration in a WSe2 sensor significantly lower than that in a MoS2 sensor given an applicable VG range. Therefore, the target-molecule-induced doping effect results in more prominent conductance modulation in the WSe2 sensor than in the MoS2 sensor. Furthermore, our work strongly implies that in addition to the target-molecule-induced p-doping effect, the target-molecule-induced surface scattering of free carriers may also play a critical role in determining the ultimate response behaviors of the sensors made from atomically layered semiconductors. Especially, a sensor behavior transition appears in the p-type branch of O2-doped WSe2 sensors. This phenomenon is attributed to the competition between target-molecule-induced p-doping and surface-scattering effects. This work advances the critical device physics for designing and producing next-generation nanoelectronic biosensors based on emerging atomically layered semiconductors.

This work was supported by NSF Grant No. ECCS-1452916. The authors would like to thank staff of the University of Michigan's Lurie Nanofabrication Facility for providing the support of AFM imaging and device fabrication.

1.
S.
Das
and
J.
Appenzeller
,
Appl. Phys. Lett.
103
,
103501
(
2013
).
2.
T.
Korn
,
S.
Heydrich
,
M.
Hirmer
,
J.
Schmutzler
, and
C.
Schüller
,
Appl. Phys. Lett.
99
,
102109
(
2011
).
3.
K. F.
Mak
,
C.
Lee
,
J.
Hone
,
J.
Shan
, and
T. F.
Heinz
,
Phys. Rev. Lett.
105
,
136805
(
2010
).
4.
V.
Podzorov
,
M. E.
Gershenson
,
C.
Kloc
,
R.
Zeis
, and
E.
Bucher
,
Appl. Phys. Lett.
84
,
3301
(
2004
).
5.
J.
Pu
,
Y.
Yomogida
,
K. K.
Liu
,
L. J.
Li
,
Y.
Iwasa
, and
T.
Takenobu
,
Nano Lett.
12
,
4013
(
2012
).
6.
B.
Radisavljevic
,
A.
Radenovic
,
J.
Brivio
,
V.
Giacometti
, and
A.
Kis
,
Nat. Nanotechnol.
6
,
147
(
2011
).
7.
Y. J.
Zhang
,
T.
Oka
,
R.
Suzuki
,
J. T.
Ye
, and
Y.
Iwasa
,
Science
344
,
725
(
2014
).
8.
Q. Y.
He
,
Z. Y.
Zeng
,
Z. Y.
Yin
,
H.
Li
,
S. X.
Wu
,
X.
Huang
, and
H.
Zhang
,
Small
8
,
2994
(
2012
).
9.
C. L.
Tan
and
H.
Zhang
,
Chem. Soc. Rev.
44
,
2713
(
2015
).
10.
H.
Li
,
J. M. T.
Wu
,
Z. Y.
Yin
, and
H.
Zhang
,
Acc. Chem. Res.
47
,
1067
(
2014
).
11.
X.
Huang
,
Z. Y.
Zeng
, and
H.
Zhang
,
Chem. Soc. Rev.
42
,
1934
(
2013
).
12.
Z. Y.
Yin
 et al,
ACS Nano
6
,
74
(
2012
).
13.
Z. Y.
Yin
,
Z. Y.
Zeng
,
J. Q.
Liu
,
Q. Y.
He
,
P.
Chen
, and
H.
Zhang
,
Small
9
,
727
(
2013
).
14.
D.
Sarkar
,
W.
Liu
,
X.
Xie
,
A. C.
Anselmo
,
S.
Mitragotri
, and
K.
Banerjee
,
ACS Nano
8
,
3992
(
2014
).
15.
L.
Wang
,
Y.
Wang
,
J. I.
Wong
,
T.
Palacios
,
J.
Kong
, and
H. Y.
Yang
,
Small
10
,
1101
(
2014
).
16.
H.
Nam
,
B. R.
Oh
,
P.
Chen
,
M.
Chen
,
S.
Wi
,
W.
Wan
,
K.
Kurabayashi
, and
X.
Liang
,
Sci. Rep.
5
,
10546
(
2015
).
17.
J.
Lee
,
P.
Dak
,
Y.
Lee
,
H.
Park
,
W.
Choi
,
M. A.
Alam
, and
S.
Kim
,
Sci. Rep.
4
,
7352
(
2014
).
18.
H.
Nam
,
S.
Wi
,
H.
Rokni
,
M. K.
Chen
,
G.
Priessnitz
,
W.
Lu
, and
X. G.
Liang
,
ACS Nano
7
,
5870
(
2013
).
19.
S.
Wi
,
H.
Kim
,
M.
Chen
,
H.
Nam
,
L. J.
Guo
,
E.
Meyhofer
, and
X.
Liang
,
ACS Nano
8
,
5270
(
2014
).
20.
M. K.
Chen
,
S. J.
Wi
,
H.
Nam
,
G.
Priessnitz
, and
X. G.
Liang
,
J. Vac. Sci. Technol., B
32
,
06FF02
(
2014
).
21.
C.
Gong
,
H. J.
Zhang
,
W. H.
Wang
,
L.
Colombo
,
R. M.
Wallace
, and
K. J.
Cho
,
Appl. Phys. Lett.
103
,
053513
(
2013
).
22.
W.
Chen
,
D. C.
Qi
,
X. Y.
Gao
, and
A. T. S.
Wee
,
Prog. Surf. Sci.
84
,
279
(
2009
).
23.
M.
Chen
,
H.
Nam
,
S.
Wi
,
J.
Lian
,
X.
Ren
,
L. F.
Bian
,
S.
Lu
, and
X. G.
Liang
,
Appl. Phys. Lett.
103
,
142110
(
2013
).