Performance Analysis of MOS-HEMT as a Biosensor: A Dielectric Modulation Approach

A dielectric modulated MOS-HEMT is investigated for different neutral and charged biomolecules with cavities embedded on the source side. To analyze the performance of the device, neutral and charged biomolecules are immobilized in the cavity independently. The neutral biomolecules are emulated by dielectric constant in the cavity and the charged biomolecules are emulated by varying interface charge at oxide/AlGaN interface. The proposed structure is simulated using the ATLAS Silvaco device simulation tool. The sensitivity of the device is optimized by various parameters such as barrier mole fraction and cavity length. The performance of the device is evaluated by the change in drain current and shift in threshold voltage. It has been observed that the drain current and threshold voltage varies with dielectric modulation of the cavity and change in biomolecular charge. The effect of neutral and charged biomolecules is also analyzed for channel potential and channel conductance. As observed from the results, the variation in gd\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {g}_{\mathrm {d}}$$\end{document} due to neutral biomolecules is very high as compared to charged biomolecules. It shows very less variation in gd\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {g}_{\mathrm {d}}$$\end{document} for charged ranges from 3×1012/cm2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$3\times 10^{12}/\mathrm {cm}^2$$\end{document} to -5×1011/cm2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$-5\times 10^{11}/\mathrm {cm}^2$$\end{document}.


Introduction
In recent years HEMT based biosensors have attracted researchers due to their quick response, simple detection, and biocompatibility [1]. High scalability, less area, higher sensitivity, low cost, and user-friendly make field-effect transistors prominent for bio-sensing applications. GaN-based materials have unique features which make them attractive for high RF applications. The dielectric modulated approach enables sensing of biomolecules in a Nano-gap/Nano-cavity region. Silicon-based devices are not chemically stable. They showed very low sensitivity towards biomolecules and are not compatible with high-temperature conditions due to their small band-gap compared to GaN. The polarization of AlGaN/AlN/GaN HEMT generates the electric field that pulls the surface state electrons to the empty conduction band (CB) states at the heterojunction, resulting in 2DEG formation [2]. The presence of high sheet charge density and mobility increases the sensitivity [3]. Various research has been done on HEMT based biosensor for pH detection [4][5][6][7], Immuno-detection using floating gate AlGaN HEMT [8], deoxyribonucleic acid.
(DNA) [9], Urea testing [10], Prostate-specific antigen testing [11,12], Monokine induced by interferon gamma [13], CerbB2 detection [14], CO and H 2 detection [3,15], C-reactive protein detection [16], troponin I testing [17,18] and carcinoembryonic antigen (CEA) detection respectively. To enhance the sensitivity of the proposed HEMTs device, various structures are also optimized like Circular gate HEMT, Double gate, Gateless devices, Nano-cavity under the gate region towards the drain side, and Nano-cavity at the source and drain both sides, respectively. The effect of half and fully filled cavity on sensitivity was also reported. In the previous reports, the effect of mole fraction on biomolecule sensitivity is not reported [19]. In this work, Gated MOS-HEMT is proposed to sense biological agents. The cavity on the source side is embedded for the first time using HEMT. The performance metrics of the device depends on threshold voltage sensitivity, drain current variation, and channel conductance variation. The device is capable of detecting both neutral and charged biomolecules. This paper presents a HEMT with a Nano-cavity that can be used for various bio sensing applications. The rest of the paper is organized as Section 2 presenting the HEMT device architecture and simulation parameters procedure. The results and discussions are reported in Section 3, which is later summarized as the conclusion.

Device Architecture and Simulation Parameters
The proposed device structure and enlarged cavity view of MOS-HEMT are shown in Fig. 1(a) and (b), respectively. Figure 2 presents the tentative fabrication steps of the device. The proposed structure may be fabricated on an Al 2 O 3 substrate, as shown in Fig. 2. The epitaxial growth can be done by MOCVD. Ti/Al/Ni/Au will use to make S/D ohmic contact, and Ni/Au will use as a gate material [19]. S/D contact can be form using an electron beam evaporator and annealed using a rapid thermal annealing(RTA) system. A 10nm insulating layer of SiO 2 is deposited as gate dielectric using plasma enhanced chemical vapor deposition between source and drain. A 2nm Al 2 O 3 layer is deposited using ALD. For gate contact, an 8nm SiO 2 layer is deposited. Gate is formed by electron beam evaporation. Now the unwanted oxide over Al 2 O 3 is removed by dry oxidation etching to make a cavity. The device consists gate length = 1 m, Oxide thickness ( t ox ) = 10nm, source to gate length ( L SG ) and gate to drain length ( L GD ) is 1 m and 2 m respectively. The space between gate to drain ( L GD ) is kept larger than the space between source to gate ( L SG ) to increases the breakdown voltage of the device [20]. Al x Ga 1−x N thickness is kept as 25nm, the Buffer layer thickness is 3 m, AlN spacer layer thickness is 3nm and AlN nucleation layer thickness is 100nm. The AlN spacer reduces the Coulomb scattering of charge carriers at the interface and enhances the channel's carrier concentration. The nucleation layer is used to reduce lattice mismatch of GaN and the Substrate layer [21]. According to the proposed design cavity is considered here as a sensing area for biomolecules. The cavity length and height of the proposed device are 500nm and 7nm, respectively. The mole fraction of the barrier layer considered is x=0.30. When we fill the cavity with biomolecules, a stern layer will form, represented by SiO 2 over the gate oxide. Afterward, SiO 2 oxide is then etched to create the cavity of 500nm length and 7nm height on the source side below the gate region. Si 3 N 4 will be used as a passivation layer over the exposed area of the device to prevent the device from other environmental effects. Sensing of neutral and charged biomolecules is achieved with the proposed architecture. Only permittivity is associated with neutral biomolecules, whereas charged biomolecules are associated with both permittivity and charge. The AlGaN/GaN HEMT has high electron mobility and is then etched to create the cavity of 500nm length high saturation velocity due to high electron confinement at the heterojunction. The basic working principle of the GaN/AlGaN MOS-HEMT device is the modulation of polarization charge density at the heterojunction, as given in Eq. (1) [22].
where oxide oxide is the dielectric constant of oxide, AlGaN is the dielectric constant and d AlGaN is width of the barrier layer, respectively, q b is the gate barrier height. The where is the charge density at the interface, E C is the conduction band offset, f Fermi level separation from conduction band of buffer layer and n oxide is the average charge density of oxide. Q Polar is the polarization charge density at hetero junction. Models used in the simulation analysis are Field-dependent drift velocity (FLDMOB), concentrationdependent mobility (CONMOB), Albrct.n, Shockley-Read-Hall (SRH), Polarization, and Calc. strain. The recombination effect is accounted for SRH, while the low field mobility effects accounted for albrct. n. The simulation models and methods depict a good agreement with the experimentally measured response of [23], as shown in Fig. 3. The dimensions used for simulation are the same as experimental.

Results and Discussion
This section presents the analysis of dielectric modulation based MOS-HEMT biosensor for the neutral, negatively charged, and positively charged bio-molecules. The neutral biomolecules analysed for the study have different dielectric constant values given in Table 1. The device is analysed for different charges range from 3 × 10 12 ∕cm 2 to −5 × 10 12 ∕cm 2 , which includes the DNA charge ranges from −1 × 10 12 ∕cm 2 to (2) . The dielectric of air (K=1) is considered as reference to differentiate the change in device performance parameters after exposure to different biomolecules. To obtain the sensitivity of the devices for various biomolecule drain currents and the threshold voltage has been considered. The AlGaN/AlN/GaN MOS-HEMT shows a sheet charge density of 1.39 × 10 13 ∕cm 2 .
As depicted in Fig. 4, the channel potential drops as on the increase of dielectric of the cavity. The highest obtained channel potential is for urease. In the case of the charged biomolecules cavity is filled with dielectric constant K=2. The total capacitance of the device is the sum of the capacitance of four regions under the gate area. Capacitance for oxide and cavity gave by Eqs. (1) and (2) (3) 1.64 Streptavidin [10] 2.1 Biotin [25] 2.63 Glucose-Oxidase (Gox) [26] 3.46 Zein [27] 5 kertain [27][28][29] 8

Fig. 4 Channel potenial of different neutral biomolecules in the gate region
The capacitance of the cavity is defined by, The total effective capacitance is defined by, Now the device total capacitance is defined as follows: Where C AlGaN is the capacitance of barrier layer that is defined as follows: Where the permittivity of AlGaN is calculated using the following equation; Where m (0.22,0.25,0.30) is the mole fraction of the barrier layer. t AlGaN is the thickness of the AlGaN layer. To obtain the electrical characteristics in the presence of neutral biomolecules, we considered the dielectric constant of the cavity. The effect on the electrical characteristics of charged biomolecules is also analyzed. Figure 5(a) presents the transfer characteristics of the device for different neutral biomolecules. Figure 5(b-d) illustrates the variation in threshold voltage for Streptavidin, Zein and Keratin molecules w.r.t K=1. The maximum thresold voltage shift obtained is -4.36V for Keratin. The thresold voltage reduces as on increase of dielectric of the cavity. The transfer charactersticts in the presence of charged molecules i.e. 3 × 10 12 ∕cm 2 , −5 × 10 12 ∕cm 2 , −1 × 10 12 ∕cm 2 and −1 × 10 11 ∕cm 2 with constant dielectric value i.e =2 is plotted in Fig. 6. The figure depicts that with the increase of the concentration of the negatively charged biomolecules the thresold voltage reduces due to the reduction of charge carriers in the channel and increases in the presence of positively charged molecules, respectively. The device output characteristics are shown in Figs. 7, 8 for neutral and Fig. 9 for charged species, respectively. The device is biased at V GS = 0V and V GS = −1V . The device is biased at V GS = −1V because of maximum trans-conductance at this gate voltage. As depicted in Figs. 7 and 8 the drain current will start decreasing on increasing of the dielectric. The maximum current observed for neutral molecules is 656mA/ mm for urease. Figure 9(a) and (b) presents the charged biomolecules output characteristics. The drain current increases when the biomolecule carries a positive charge. Applying positive charge on the interface will attract more electrons to the channel, whereas the increment of charge carriers will increase the current in the device.
In the case of a negative charge, the current will decrease due to the charge carriers' depletion. Figures 10 and 11 shows the drain current variation induced by different biomolecules. It is observed that the variation in drain current for neutral biomolecules increases by increasing of dielectric of the nanocavity. In the case of neutral biomolecules, the maximum drain current variation obtained for keratin is 86.2mA/mm, whereas the minimum current variation is 12.4mA/mm for urease at V GS = −1V . The maximum current obtained at V GS = 0V is 50.5mA/mm for keratin, whereas the minimum is 8.6mA/mm for urease. In the case of charged biomolecules, the more negative the charge more the relative change in current w.r.t = 0 . The device is sensible for charge ranges from 3 × 10 12 ∕cm 2 to −5 × 10 12 ∕cm 2 . The maximum current measured is 710(mA/ mm) at 3 × 10 12 ∕cm 2 due to an increase in positive charge will increase the charge carrier concentration in the channel. The device is biased at V GS = 0V to observe the pure effect of neutral and charged species on the device. When a molecule consists of charges and dielectric, the device parameters are affected by both dielectric and charge density. To analyze the effect of DNA charge, varying its charge from −1 × 10 12 ∕cm 2 [9] to −1 × 10 11 ∕cm 2 [9].
As observed that increases of the dielectric constant, the channel conductance decrease due to the reduction of drain current. It decreases with an increase in drain voltage. Using the following formula the channel conductance can be calculated; The channel conductance variation can be calculated by using the following formula; g d Should be higher to achieve high biomolecule sensitivity. Figures 12 and 13 show the channel conductance for neutral and charged biomolecules. The channel conductance decreases with the increase in negative charge associated with a biomolecule. Reduction in g d indicate the enhancement of device resistance due to capacitive coupling of higher negative charge in the cavity. The relative change in the channel conductance Δg d is shown in Figs. 14 and 15 for neutral and charged biomolecules, respectively. The channel conductance variation increases on increasing dielectric of   Figure 16 depicts the effect of neutral and charged biomolecules on the threshold voltage. Figure 16(a) depicts the effect of neutral bio-molecules on the threshold voltage. The threshold voltage is extracted using constant current method with 10 −7 mA reference current. The threshold voltage shows a positive shift with the increase of dielectric of neutral biomolecules, as shown in Fig. 16(a).

Threshold Voltage Variation
The variation in V th is depicted in Fig. 16(b). The variation is higher for high k biomolecules. The threshold voltage variation can be calculated by using the following equations Where, V th(Ref .) is the threshold voltage when the cavity is empty, V th(Bio) is when the cavity is filled with a biomolecule, V th(Neutral) is when no interface charge is applied and V th(charged_Bio) is when the cavity is filled with charged biomolecules. Figure 17(a) indicates that the threshold voltage decreases as the negative charge of biomolecules in the cavity region of the device increases, whereas threshold voltage increases when the positive charge of biomolecules in the cavity region of the device increases.  Figure 17(b) indicates that the threshold variation will have a high positive value of variation when increases the positive charge of the biomolecule whereas a high negative value of threshold variation with an increase in the negative charge of the biomolecule. The dielectric of neutral and charged biomolecules influences the change in threshold voltage. The sensitivity was calculated using the following formula; The V TH sensitivity was calculated using the following formula; In the case of neutral biomolecules, increasing the dielectric would increase sensitivity due to an increase in the threshold voltage variation, as seen in Fig. 18(a). Figure 18(b) depicts the threshold voltage sensitivity due to the charge of biomolecules. Higher sensitivity is correlated with a higher negative charge, as shown in the figure. The highest sensitivity observed is for −5 × 10 12 ∕cm 2 in case of the charged biomolecule.

Impact of Cavity Length on Device Sensitivity
The effect of different cavity lengths on sensing is plotted and explained in this section. The cavity lengths considered i.e. 100nm, 300nm, and 500nm. To measure the sensitivity, the shift in threshold voltage and drain current variation are used. The change in threshold voltage for different cavity lengths is plotted for charged and neutral biomolecules. Figure 19(a) and (b) demonstrate the effect of cavity lengths on the drain current of neutral biomolecules. The maximum current variation was for K=8 obtained at 300nm cavity length for K=8 at VGS=0V is 22.9mA/mm. To measure the drain current variation, the following formula is used;  where I DS (Ref .) is the current at K=1 when cavity is not occupied with neutral/charged biomolecule. I DS(Bio) is the current when cavity is filled with neutral/charged biomolecules. The impact of cavity length on sensitivity for charged biomolecules is plotted in Fig. 20(a) and (b). The sensitivity is high for 500nm cavity length due to availabilty of large area to intreact with the sensing surface.

Impact of Cavity Length on Drain Current Sensitivity
For various cavity lengths, the drain current sensitivity is calculated at V DS =1V. The plots of sensitivity for The plots of sensitivity for 100nm, 300nm, and 500nm cavity lengths are shown in Fig. 21(a) and (b), respectively. The drain current sensitivity for neutral and charged biomolecules is calculated using the following equation; Where I DS(Ref .) is the current value when the cavity is empty, I DS(Bio) is when the cavity has some biomolecule. As on increasing dielectric and cavity length of neutral and charged biomolecules, the sensitivity also increases. By increasing cavity length, the sensitivity will increase due to the increase in surface area for biomolecule. Higher value of negative charged biomolecule indicate more variation in drain current that will lead to higher sensitivity. These plots show the pure effect of biomolecules on device sensitivity at V GS = 0V.

Impact of Cavity Length on Threshold Voltage Sensitivity
The device's sensitivity for neutral and charged biomolecules is calculated due to threshold voltage variation using Eq. 16. The change in sensitivity is due to the shift in threshold voltage for different neutral/charged biomolecules. Figure 22(a) and (b) show the sensitivity for neutral and charged biomolecules, respectively. For neutral biomolecules, the maximum sensitivity for K=8 is 26%, whereas, for charged biomolecules, 51% for -5 × 10 12 ∕cm 2 . The sensitivity will increase as the cavity length increases. The sensitivity for high k biomolecule will be higher. In the case of charged Fig. 19 The drain current variation at V GS =0V for (a) 100nm and (b) 300nm cavity lengths for neutral biomolecules Fig. 20 The drain current variation at V GS = 0V for (a) 100nm and (b) 300nm cavity lengths for charged biomolecules biomolecules, the sensitivity is affected by both charge and dielectric. Figure 22(b) shows the sensitivity plot for charged biomolecule at the constant value of dielectric (K=2). The graph depicts that the sensitivity of a particular biomolecule increase as the negative charge associated with it increases.

Impact of Mole Fraction of Barrier Layer
The polarization field in the device increases as the mole fraction increases. The polarization is caused by a lattice mismatch between the GaN and the barrier layer. With an increase in Al composition of the barrier layer, the polarization fields increase. Due to the control on the conduction band discontinuity at heterojunction increase of mole fraction of barrier increases the drain current, which will allow high electron confinement in the channel. Different mole fraction values are considered here (x=0.22,0.25,0.30) to evaluate the mole fraction effect.
As an increase of mole fraction, the band-gap also increases. The higher the band gap exhibits higher stability of the device at high-temperature conditions. The highest band-gap was obtained at x=0. 30    respectively. The maximum threshold sensitivity obtained is 68% for −5 × 10 12 ∕cm 2 at x=0.22 and minimum is 50% at x=0.30. Table 2 presents the comparison between the previously reported work and the present study. The drain current variation obtained in the present work is higher than the previous for neutral bio-molecules, indicating higher sensitivity. The present study reports that the drain current sensitivity for Keratin is 99.9, which is 57.8% greater than reported [27]. The threshold voltage sensitivity is reported in this work is 26% which is 4% higher than reported [27].

Comparative Analysis
The maximum threshold voltage variation observed for charged biomolecules is -2.62V, more elevated than the previously reported work (0.1-0.5) [27]. Table 3 presents the comparative analysis of threshold voltage variation. The DNA

Conclusions
A dielectric modulated MOS-HEMT is proposed for the detection of neutral and charged biomolecules. It exhibits good sensitivity for different biomolecules. The threshold voltage and drain current sensitivity obtained for neutral bio-molecules in the range of 0.06-2v and 8.9-50.7mA/mm. The analyzed voltage shift for DNA charge is up to 1.32V, and the current variation is 43.02mA/mm. The maximum threshold sensitivity obtained is 26% for keratin and 50.10% for charged biomolecules, respectively. The maximum sensitivity obtained is 74.04% for charged biomolecules. As analysed from the results that the device with higher cavity length(500nm) and lower mole fraction(x=0.22) will be suitable for high sensing performance. The study indicates that cavity based MOS-HEMT is suitable for bio-sensing applications.
Acknowledgements The authors would like to the Department of Electronics and Communication Engineering, Malaviya National Institute of Technology for providing necessary support for carrying out the simulation work.
Author Contributions This work was proposed and done by Ritu Poonia (Author 1). Aasif Mohammad Bhat provided the necessary support regarding simulations and data interpretation. C. Periasamy and Chitrakant Sahu supervised the work and made important discussions and modifications to the final manuscript.

Funding
The authors have not received any funding for this work.

Declarations
Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Consent to Publication NA
Informed consent Informed consent was obtained from all individual participants included in the study.

Conflicts of interest
The authors declare that they have no conflict of interest.