Abstract
This work reports an emerging structure of gateallaround ferroelectric area tunneling fieldeffect transistor (FATFET) by considering ferroelectric and a nepitaxial layer enveloped around the overlapped region of the source and channel to succeed with complete area of tunneling probability. To accomplish this, ferroelectric (\(\hbox{Hf}_{0.5}\hbox{Zr}_{0.5}\hbox{O}_2\)) is exploited and modeled to boost the FATFET performance through internalvoltage (\(V_{{\rm int}}\)) amplification. The corresponding modeling approach to estimate the ferroelectric parameters along with \(V_{{\rm int}}\) calculations of the metalferroelectricinsulator (MFIS) option through capacitance equivalent method is addressed. Using these options the proposed device outperforms effectively in delivering superior DC and RF performance among possible options of the \(\hbox{Si}_{1x}\hbox{Ge}_x\) ferroelectric TFETs. The significance of proposed design is examined with recently reported ferroelectric TFETs. Our results show 10time advancement on the \(I_{{\rm on}}\), reduced steep or average subthreshold swing (< 25 mV/dec), frequencies higher than 150 GHz, and insignificant to linearity deviations at low bias points. Furthermore, 2order reduction in energy efficiency is succeeded with the proposed design environment.
Background
The inevitability of new principal devices such as tunneling fieldeffect transistors (TFETs) is under exploration to meet the aspects of low power consumption. TFETs are the devices that deliver low offstate current (\(I_{{\rm off}}\)) and subthreshold swing (SS) lower than 60 mV/dec [1]. However, based on the experimental and computational works in TFETs it is identified that achieving (1) low average or steeper SS (\(\hbox{SS}_{{\rm avg}}\)) instead of impressive point or minimum SS (\(\hbox{SS}_{{\rm min}}\)), (2) high onstate current (\(I_{{\rm on}}\)), and (3) low energy efficiency or switching energy (SE) are the key challenges.
Line TFETs (LTFETs) through strong verticalgate fields have been aided with substantial tunneling probability as associated to the point TFETs [2,3,4]. Additionally, it is identified that the ferroelectric materials can introduce excess electric field (E) and polarization (P) than the conventional dielectrics [5, 6]. Therefore, the ferroelectric materials have become beneficial to help in amplification of tunneling probability through reduction in tunneling length (\(\lambda\)) by the virtue of internal voltage (\(V_{{\rm int}}\)). Therefore, several demonstrations (theoretical and practical) have shown that the application of ferroelectricity can advance the performance of TFETs [7, 8]. Among existing ferroelectric options, the HZO as \(\hbox{Hf}_{0.5}\hbox{Zr}_{0.5}\hbox{O}_2\) has been proven to be a most appropriate option in fulfillment with the experimental approaches [9]. However, the recent demonstrations still lag behind utilization of ferroelectricity due to unfitted geometrical options of TFETs (pointtunneling) that cannot harvest a complete area of tunneling.
To address the aforementioned key challenges in TFETs, the modeled device is equipped with improved area of tunneling using scaled nepitaxial layer, material options with source as \(\hbox{Si}_{1x}\hbox{Ge}_x\) (with x = 0.4 as the Ge fraction) and gatedielectric as \(\hbox{Hf}_{0.5}\hbox{Zr}_{0.5}\hbox{O}_2\), and the gateallaround (GAA) nanosheet geometry. The significance of scaling line tunneling (by nepitaxy and overlapped source (\(L_{{\rm sov}}\))) and the ferroelectric options have been demonstrated in our recent articles [10, 11]. Here, the work is extended to detailed investigation of ferroelectric effect dependency, the device reliability by analyzing the linearity behavior, and finally the cumulative comparison with recently explored structures of ferroelectric line TFETs. Here, the ferroelectric dependency elaborates modeling of ferroelectric parameters such as \(V_{{\rm int}}\) calculations; influence of remnant polarization (\(P_{{\rm r}}\)) and coercive fields (\(E_{{\rm c}}\)) on the proposed geometry.
Methods
Device Design and Methodology
The proposed design is processed by using TCAD simulations and in the view of emerging technology nodes (e.g., sub3nm) [14, 15]. A 3D device simulation platform has been quantified and validated for respective models of electron bandtoband tunneling (BTBT) and trapassistedtunneling (TAT) models [16, 17], along with the parameter calibrations of \(\hbox{Si}_{0.6}\hbox{Ge}_{0.4}\) to achieve faithful results as similar to our earlier work [18, 19]. The initial parameters to model the several physical models are brief as follows. For example, the calculation of BTBT can be evaluated based on the tunneling mass of electron and hole, energy bandgap, and so on which are the initial parameters based on materials to evaluate additional parameters. The detailed procedure and calculations that undergo were calculated in our previous article [18]. Current article highlights the procedure to model the ferroelectric parameters, where \(E_{{\rm c}}\) and \(P_{{\rm r}}\) are the initial guess for analyzing the parameters of ferroelectric. The detailed procedure is illustrated in subsequent sections. The design is equipped with \(\hbox{Si}_{0.6}\hbox{Ge}_{0.4}\) as source to achieve a good figure of merit in terms of \(I_{{\rm on}}\) and \(I_{{\rm off}}\). This is due to the fact that an added Ge content reduces the effective bandgap and tunneling mass that are beneficial for tunneling rate or \(I_{{\rm on}}\). In contrast, a greater reduction of bandgap or tunneling mass is proportionate for large current during offstate regime (i.e., \(I_{{\rm off}}\)). Therefore, an optimum addition of Ge content is most suitable for a good figure of \(I_{{\rm on}}/I_{{\rm off}}\) [19].
Modeling of Ferroelectric Parameters
Here, ferroelectric parameters are calibrated and plugged into simulations using Landau–Khalatnikov (L–K) approach for further strengthening of device simulations [12]. The MFIS geometry with its capacitance equivalent circuit as depicted in Fig. 1a is considered as a ferroelectric option for proposed TFETs. It is worth mentioning that several discussions have been made on quasistatic (QS) and nonquasistatic (NQS) behavior of ferroelectric capacitance (\(C_{{\rm fe}}\)) [20,21,22,23]. The findings are, 1) stabilization of negative capacitance (NC) at microscopic level, 2) slow switching dynamics of ferroelectricity, and 3) ambiguity between L–K (assuming \(C_{{\rm fe}}\) intrinsically negative) and Miller (domainwall propagation delay (\(\tau\)) as \(R_{{\rm fe}}C_{{\rm fe}}\), where \(R_{{\rm fe}}\) is the ferroelectric resistance) models [20]. From these findings, it has been concluded (experimentally) that the NC with QS is still valid without NQS (will be discussed further) due to slow switching dynamics of ferroelectricity at microscopic level. Hence, the L–K approach with QS based ferroelectric is modeled here.
Polarization charge in ferroelectric (\(Q_{{\rm fe}})\) is [21],
where \(V_{{\rm fe}}\), \(t_{{\rm fe}}\), and \(A_{{\rm fe}}\) are voltage, thickness, and crosssectional area of the ferroelectric. The rate of change of \(Q_{{\rm fe}}\) is expressed as [24, 25],
where \(\rho\) defines the frictional inertia and lefthand side of (2) represents the NQS behavior of the ferroelectric as NC. The free energy of ferroelectric (U) is expressed as,
Here, \(\alpha\), \(\beta\), and \(\gamma\) are the anisotropic constants; \(V_{{\rm fe}}\) is the voltage across ferroelectric region. For simplicity, up to fourthorder polynomials (secondorder phase transition) is considered. Evaluating the (2) using (3),
Under steadystate condition of ferroelectric (\(\hbox{d}Q_{{\rm fe}}/\hbox{d}t=0\)),
In an isolated ferroelectric region \(V_{{\rm fe}} = 0\) and \(Q_{{\rm fe}}=\pm Q_0\), where \(Q_0\) is the initial polarization charge in the ferroelectric, implies
and
In general, the charge in ferroelectric fluctuates from two magnitudes as \(+Q_{{\rm c}}\) to \(Q_{{\rm c}}\) (Fig. 1b) at \(V_{{\rm fe}} = V_{{\rm c}}\), which implies \(\hbox{d}V_{{\rm fe}}/\hbox{d}Q_{{\rm fe}}=0\). Thus, \(Q_{{\rm c}}\) and \(V_{{\rm c}}\) are evaluated as,
and
By employing (6) in (11), the \(\alpha\) and \(\beta\) can be derived as,
and
External electric field (\(E_{{\rm ext}}\)) in the ferroelectric depends on the applied voltage (V), i.e.,
Hence, \(V_{{\rm c}}\) and \(Q_0\) are functions of the coercive field (\(E_{{\rm c}}\approx V_{{\rm c}}/t_{{\rm fe}}\)) and P (see (1)), respectively. In addition, the factor of remnant polarization (\(P_{{\rm r}}\)) is derived from P, i.e.,
where \(\chi\) represents the electric susceptibility of the ferroelectric and \(E_{{\rm c}}\le E_{{\rm ext}}\le E_{{\rm c}}\). The \(E_{{\rm c}}\) and \(P_{{\rm r}}\) are measured experimentally as \(\approx\) 1 MV/cm and 1–20 μC/cm^{2} by properly controlling the ferroelectricity in \(\hbox{Hf}_{1x}\hbox{Zr}_{{x}}\hbox{O}_2\) [9]. Therefore, using expressions (12)–(15), the evaluated \(\alpha\) and \(\beta\) at \(P_{{\rm r}}=10\) μC/cm^{2} and \(E_{{\rm c}} = 1\) MV/cm are of \(1.299\times 10^{11}\) cm/F and \(6.4952\times 10^{20}\) cm\(^5\)/FC^{2}, respectively. Recalling (5), the \(V_{{\rm fe}}\) can be identified to be a beneficial factor for the enhancement of \(V_{{\rm int}}\) to achieve high tunneling probability in TFETs. The numerically solved data (above expressions) and the extracted data from TCAD simulations are depicted in Fig. 1c. This signifies the simulations are properly tuned according to the standard expressions. In addition, simulations are also calibrated with measurement (or experimental [6]) data and are observed as consistent, which is depicted in Fig. 1d [13].
Modeling of \(V_{{\rm int}}\)
The formation of \(V_{{\rm int}}\) for the MFIS (Fig. 1a) can be evaluated through a voltagedivider rule for the series connected lumped elements of capacitance (since distributed charge on the oxide and ferroelectric is the product of capacitance and potential). Here, \(V_{{\rm int}}\) for both the QS and NQS is evaluated as follows.
Based on Fig. 1a (top) i.e., QS, the \(V_{{\rm int}}\) is expressed as,
where \(C_{{\rm IS}}\) and \(C_{{\rm fe}}\) are the capacitance across insulatorsemiconductor and the ferroelectric regions, respectively.
The \(V_{{\rm int}}\) of NQS is evaluated by applying the current rule in Fig. 1a (bottom),
Note that the applied and internal voltages (\(V_\mathrm{G}\) and \(V_{{\rm int}}\)) are assumed to be variation with time. Further (18) simplifies as,
Switching the factors of \(\hbox{d}V_{{\rm int}}/\hbox{d}t\), \(\hbox{d}V_{{\rm G}}/\hbox{d}t\) as \(\Delta V_{{\rm int}}\) and \(\Delta V_\mathrm{G}\), respectively, then
The expressions (17) and (21) signify the \(V_{{\rm int}}\) or \(\Delta V_{{\rm int}}\) calculations for QS and NQS; the \(V_{{\rm int}}\) of QS can be visualized in (21) at \(\partial Q_{{\rm fe}}/\partial t\rightarrow 0\). As a result, higher \(V_{{\rm int}}\) implies lower \(V_{{\rm G}}\) requirements, meaning that operating at low voltage is beneficial to the device.
According to NCFET studies, the gateoxide (\(t_{{\rm ox}}\)) has to be thinner to utilize the benefits of ferroelectricity and hysteresis free operation [27]. The thinner \(t_{{\rm ox}}\) plays the role of \(\lambda\) reduction through its resultant low effective oxide thickness (\(\hbox{EOT}\,\rightarrow \,t_{{\rm ox}}\)) [28], which is beneficial for improved TFETs performance. In addition, the use of low nanosheet geometrical thickness (\(t_{{\rm ns}}\)) can further shrink the \(\lambda\), based on [28]
Here, \(\epsilon _{{\rm ns}}\) and \(\epsilon _{{\rm ox}}\) are the permittivity of nanosheet and gateoxide. Hence, the low \(t_{{\rm ox}}\) or EOT in association with low \(t_{{\rm ns}}\) can afford enough source to have reasonably high electron BTBT at the tunneling junction by dropping \(\lambda\) and amplifying \(V_{{\rm int}}\).
Design of ATFET and FATFET
The demonstration of the GAA nanosheet structure of FATFET is depicted in Fig. 2. Figure 2a illustrates a novel design by stacking nepitaxial layer over the channel and source overlapped regions (\(L_{{\rm sov}}=5\) nm) to improve the factor of vertical tunneling. The complete (area of) tunneling along the source (\(p^{++}\))n and \(p^{++}\)channel (p) can be recognized (arrows) in Fig. 2b. In addition, the drain region expansion is ignored here compared to our previously reported structure [28] because we identified that the drain expansion has the least significance on the tunneling rate and this ignorance will be beneficial for reduction in device complexity at fabrication environment.
A simplified version of Fig. 2c depicts the processing steps necessary to build the proposed device architecture. The process begins with the epitaxy of SiGe/Si on top of a passivation layer over a substrate. This passivation layer can be utilized to generate multiple nanosheets and also for GAA employment at the end. In continuation, the source and drain epitaxy can be grown on top of the substrate. The process is continued for source–channel revealing through elimination of the passivation layer, subsequently p and ntype implantations for source(S), channel, and drain (D) regions. The nanosheet must then be trimmed to replace with epitaxy (n) growth. Recent demonstrations have shown growth of 4 nm thick nanosheets with GAA nanosheet Si/SiGe epitaxy using 2 nm trimming [29]. Further, the trimmed nanosheet is substituted with epitaxy layer growth and trailed by selective doping of ntype implantation aimed at epitaxy region. The advanced processing techniques like metalorganic vaporphase epitaxy (MOCVD) [30], neutral beam etching (NBE) can be employed for growth and etching mechanisms [31]. At last, the gateoxide and ferroelectric layer can be deposited by atomic layer deposition (ALD) and or pulsed layer deposition (PLD) [32], followed by TiN deposition for metal formation. The device specifications and materials used are listed in Table 1.
The principle mechanism of ATFET (without ferroelectric) and FATFET works on both vertical or line and horizontal or lateral tunneling approaches (see Fig. 2b with vertical and horizontal arrows). Point tunneling is the other approach that leads to lateral tunneling with gateedge field (hence, point tunneling) [4]. Notably, the concept of line tunneling is a leading mechanism than the point tunneling [33]. However, the line tunneling requires higher biasing voltages to exhibit its significance than point tunneling. Therefore, the ferroelectric material can be able to solve this issue by inducing an additional \(V_{{\rm int}}\) to reduce the bias level of line tunneling, i.e., by FATFET. The total external voltage as the applied gatevoltage (\(V_\mathrm{G}\)) and \(V_{{\rm int}}\) at reasonable drainbias (\(V_\mathrm{D}\)) will make enough band bending, resulting in sharp reduction of \(\lambda\). Besides, the scaled nepitaxial region and \(L_{{\rm sov}}\) provides enough room to accommodate a large area of BTBT. The relevant information is depicted in Fig. 3, showing the energy band profile during off (\(V_\mathrm{G}=0\) V, \(V_\mathrm{D}=0.5\) V) and onstates (\(V_\mathrm{G}=V_\mathrm{D}=0.5\) V) along source–channel–drain regions of with and without NC (by ferroelectric) devices. The effect of source, channel, and drain Fermi levels (\(E_{{\rm fs}}, E_{{\rm fc}}\), and \(E_{{\rm fd}}\)) along with their reductions in energy levels: \(\Delta E_\mathrm{s}, \Delta E_{{\rm ch}}\) and \(\Delta E_\mathrm{d}\), are evaluated and depicted. Here, the slope of current density (Fig. 3(right)) can be reduced through large \(V_{{\rm int}}\); ultimately, it helps to accomplish low \(\hbox{SS}_{{\rm avg}}\). Thus, the grouping of ferroelectric film (\(\hbox{Hf}_{0.5}\hbox{Zr}_{0.5}\hbox{O}_2\)), \(\hbox{Si}_{0.6}\hbox{Ge}_{0.4}\) as source, and selectivity of nepitaxial layer with nanosheet geometry options would consequently improve characteristics of FATFET.
Results and Discussion
The significance of FATFET with respect to ATFET in terms of line tunneling rate, DCRF, and linearity analyses is explored here. The discussions are further extended to cumulative comparison of the DC characteristics and SE evaluations of proposed and the explored structures.
Energy Band Calculations
The energy band structure of the proposed FATFET is extracted and analyzed (Fig. 3) based on the energy and builtin potential calculations, as explained below. The terms represented in the energy band diagram (Fig. 3), i.e., \(\Delta E_\mathrm{s}\), \(\Delta E_{{\rm ch}}\) and \(\Delta E_{{\rm d}}\), can be defined with \(\Delta E_{{\rm s}}=E_{{\rm VBmax}}E_\mathrm{f}\) of the source, \(\Delta E_{{\rm ch}}=E_\mathrm{f}E_{{\rm CBmin}}\) of the channel, and \(\Delta E_{{\rm d}}=E_\mathrm{f}E_{{\rm CBmin}}\) of the drain, respectively. Here, \(E_{{\rm CBmin}}\) and \(E_{{\rm VBmax}}\) are the conduction band minimum and valence band maximum. The calculated \(\Delta E_{{\rm s}}\), \(\Delta E_{{\rm ch}}\), and \(\Delta E_{{\rm d}}\) (difference in Fermi level) are \(0.096\), 0.195, and \(0.0304\) eV for FATFET. Since the source is strained with that of \(\hbox{Si}_{1x}\hbox{Ge}_x\) with \(x = 0.4\), the effective mass of density of states (\(m_\mathrm{V}\)) in valence band is calculated via linear approximation, as [34]
where \(m_0\) is the electron mass in free space. Consequently, the source–channel and source–drain builtin potentials \(\phi _{{\rm sc}}\) and \(\phi _{{\rm sd}}\) can be expressed as
and
The calculated builtin potentials in the proposed structure (FATFET) are of \(\phi _{{\rm sc}}=0.719\) eV and \(\phi _{{\rm sd}}=1.1364\) eV. Based on these calibrations it is clear that the higher \(V_{{\rm int}}\) makes provision for deeper band bending and thus higher electron BTBT. Therefore, the current density increases with respect to the sourcedrain Fermi level difference (\(F_{{\rm S}}(\mathrm{E})F_{{\rm D}}(\mathrm{E})\)) as depicted in Fig. 3 (right).
Role of Ferroelectricity in FATFET Compared with ATFET
Due to strong verticalfields with the benefits of high P and E through ferroelectric, the FATFET is able to attain higher electron BTBT than the ATFET, as depicted in Fig. 4. More precisely, excess BTBT rate at position \(p^{++}n\) (vertical tunneling) and \(p^{++}p\) (lateral tunneling) are observed from Fig. 4b, c through the cutlines \(\hbox{C}_2\) and \(\hbox{C}_3\). It is clear that the FATFET has higher vertical (1.5 times) and lateral tunneling (1order) rates.
The comparative \(I_{{\rm D}}\)–\(V_{{\rm G}}\) characteristics of ATFET and FATFETs are depicted in Fig. 5a at multiple bias points (\(V_{{\rm D}} = 0.01\) and 0.5 V). With the aforementioned benefits of amplification in \(V_{{\rm int}}\), reduction in \(\lambda\), and tunneling rate enhancements; the FATFET achieves with high \(I_{{\rm on}}\) having benefited \(\approx\) 0.2 V of \(V_{{\rm int}}\). The hysteresis operation is also shown by varying the \(t_{{\rm fe}}\) from 3 to 20 nm. It is evident that the reduction in \(t_{{\rm fe}}\) below 5nm suppresses the hysteresis [13, 27]. Whereas, the \(I_{{\rm on}}\) affects marginally in TFETs due to its dependency on EOT. Though the \(V_{{\rm fe}}\) increases proportionally with \(t_{{\rm fe}}\), this will not be beneficial for TFETs to deliver high\(I_{{\rm on}}\). Since \(\lambda\) enlarges for an increased \(t_{{\rm fe}}\) due to high EOT (see (22)) [28]. Hence it is affordable to use low \(t_{{\rm fe}}\) that will be beneficial not only the high\(I_{{\rm on}}\) (especially in TFETs) but also nonhysteresis operation.
Significance of \(P_{{\rm r}}\) and \(E_{{\rm c}}\) in FATFET
The ferroelectric parameters of \(P_{{\rm r}}\) and \(E_{{\rm c}}\) are crucial in estimating the device behavior for the optimized ferroelectric option. Hence, the significance of \(P_{{\rm r}}\) and \(E_{{\rm c}}\) is analyzed in terms of \(I_{{\rm D}}\)–\(V_{{\rm G}}\) and SS, which are depicted in Fig. 6. The results are consistent to recent demonstrations [7], i.e., the variation in \(E_{{\rm c}}\) replicates stronger influence than \(P_{{\rm r}}\). Due to the fact that the tunneling is more pronounced to \(E_{{\rm c}}\) and which is sensitive in vertical TFET geometries. The same can be viewed in Fig. 6a that the \(>0.1\) MV/cm of \(E_{{\rm c}}\) is sensitive to \(I_{{\rm off}}\). Figure 6b reveals the marginal variation in performance (\(I_{{\rm on}}\)) at fixed \(E_{{\rm c}}\) but eventually \(I_{{\rm off}}\) becomes sensitive at very high \(P_{{\rm r}}\). It is because \(Q_{{\rm fe}}\) increases at higher \(P_{{\rm r}}\) and is thus responsible for decrease in tunneling window (Fig. 3). Nevertheless, the significance of \(P_{{\rm r}}\) is minor as compared to \(E_{{\rm c}}\) by the view of less variation in energy level across the tunneling window. Hence, from the perspective of \(\hbox{SS}_{{\rm avg}}\) and \(I_{{\rm off}}\), it can be concluded that the acceptable range of \(P_{{\rm r}}\) and \(E_{{\rm c}}\) for an optimum ferroelectric utilization is below 30 μC/cm^{2} and 0.25 MV/cm (Fig. 6c).
Significance of Interface Defects in FATFET
The interface and/or extended defects which can cause carrier scattering, as well as traps that are degradation factors of the transistors are explored here. It means that phonons experience scattering at interfaces and contribute to net recombination rate or TAT by which degradation of TFETs performance is observed. The influence of net recombination rate or TAT is captured by dynamic nonlocal path TAT (by Hurkx) and discussed here [16]. The model not only captures the position dependent electron and hole TAT but also the defect level TAT through phonon assisted tunneling processes. This model dynamically creates a tunneling path based on the energy profile for both aforementioned TAT’s. Here, the defect level can be identified along the tunneling path direction (\(0\le X \le L\)), where ’X’ being the tunneling path direction (similar to electron tunneling direction shown in Fig. 3), ’0’ and ’L’ are the starting and ending positions of tunneling path lengths. The electron and hole occupation probabilities at the defect level can be determined by balancing the net electron and hole capture rate. The simulated profile of electron TAT (i.e., position dependent) and the defect level TAT of FATFET are depicted in Fig. 7 at multiple bias points. The distribution of electron TAT can be observed along the \(p^{++}\)n and pn junctions, whereas defect level TAT is seen at the Si/\(\hbox{Si}_{0.6}\hbox{Ge}_{0.4}\) interfaces from Fig. 7a, b. In addition, the low bias or offstate (\(V_{{\rm D}}=0.5\) V, \(V_{{\rm G}}=0\) V) is less effective than at high bias point (\(V_{{\rm D}}=0.5\,\mathrm{V}=V_{{\rm G}}=0.5\) V). The electron TAT is more significant than defect level TAT even though the magnitude of defect level TAT is slightly higher. Nevertheless, the overall contribution of TAT on the proposed device is minor, therefore the effect of \(I_{{\rm off}}\) current is less than 1order higher as shown in Fig. 7c.
RF and Linearity Metrics of FATFETs
The frequency response of FATFET and ATFET is analyzed in terms of unitygain cutoff frequency (\(f_{{\rm t}}\)) and maximum oscillation frequency [35], respectively. It is to be noticed from Fig. 8 that the frequency terms are higher in FATFET than the ATFET because of its high drive current capability (i.e., \(I_{{\rm on}}\)) as discussed before. A substantial benefit in terms of \(V_{{\rm G}}\) requirement can be noticed in FATFET compared to ATFET targeted with maximum frequency achievement. The linearity test is crucial to signify the amount of power wastage by the device due to nonlinearity behavior. A nonlinear device could switch the biasing point and cause reduction in gain and frequency terms. Hence, the transconductance should be high enough and low distortion oriented at the operating point (active or depletion region) of TFETs. The transconductance coefficients can be evaluated as,
where n = 1, 2, 3, .... For the linearity test, higherorder transconductance coefficients such as \(g_{m_1}\) and \(g_{m_2}\) and transconductance generation factor (TGF) are considered [36, 37]. Figure 9a depicts the \(g_{m1}\) of FATFET and ATFET, stating the transconductance continuum with \(V_{{\rm G}}\) without harmonic distortion in FATFET than ATFET. In addition, the FATFET is achieved with \(\ge\) 10time improvement in \(g_{m_1}\) than the ATFET, whereas, higherorder \(g_{m_2}\) as depicted in Fig. 9b has two or more harmonic distortions both in FATFET and ATFET, respectively. This signifies that the device is nonlinear and should be taken care of. Nevertheless, the device still operates linearly during the active region (0.2–0.3 V) or below the gateoverdrive voltage (\(V_{{\rm G}}V_{{\rm t}}\)) point, where \(V_{{\rm t}}\) is the threshold voltage. TGF as shown in Fig. 9c shows that the FATFET has better linearity compared with ATFET. The nonlinear behavior of TGF in ATFET is due to nonsteep swing of \(I_{{\rm D}}\)–\(V_{{\rm G}}\) behavior as per Fig. 5a compared with FATFET. From these analyses, it is understood that the FATFET has better linearity improvement compared with ATFET; however, more analyses are needed to address other linearity factors that will be considered in the future work.
Significance of Proposed FATFET
The significance of proposed geometry (FATFET) is addressed through cumulative comparison of recently reported NCTFETs [7, 26]. Though the reported structures are with planar and dissimilar materials options, here we have reimplemented the structures ( [7, 26]) with nanosheet and \(\hbox{Si}_{0.6}\hbox{Ge}_{0.4}\) as source. The geometrical comparisons are listed in Fig. 10 with merits and demerits of each (right of Fig. 10). That is, the proposed geometry (Fig. 10a) has the flexibility in scaling the tunneling junction (\(p^{++}\)n) by simply varying the \(L_{{\rm sov}}\). This resembles the advantage of area tunneling feature can easily adjusted even for future technology nodes, whereas Fig. 10a and b (the reported structures) is restricted with weak \(p^{++}\)–p tunneling. Furthermore, the explored structure in Fig. 10b is not flexible in scaling \(p^{++}\)n junction for future technology nodes. Though Fig. 10c can be easily extendable for future technology nodes, the tunneling rate is insufficient due to nonavailability of vertical tunneling mechanism. With the aforementioned features in device reliability, the study is extended to DCperformance comparison in terms of \(I_{{\rm D}}\)–\(V_{{\rm G}}\) and \(\hbox{SS}_{{\rm avg}}\), which is depicted in Fig. 11. The results signify that the proposed structure has 10time improvement in \(I_{{\rm on}}\) and 2order advancement in \(I_{{\rm on}}/I_{{\rm off}}\), and benefited with \(\hbox{SS}_{{\rm avg}}\) (evaluated from \(I_{{\rm off}}\) to \(10^{7}\) orders of current [38]) than the explored structure [7]. In summary, the proposed design has shown with improvements in DC than the recently reported devices and has further scope to improve through novel material options.
Energy Efficient FATFETs
It is meaningful to perform energy efficiency tests on the proposed FATFETs being the TFETs as the energy efficient devices. The energy efficiency or switching energy (SE) is evaluated by [7]
and depicted in Fig. 12. Here, \(V_{{\rm DD}}\) is the supply voltage, \(\xi\) is the active time ratio with 1/800 and 1/10,000 for logic and memory applications. It is observed that the proposed (FATFET) has impressive performance in both logic and memory environments at supply voltages < 0.15 V. Specifically, 2order reduction in SE is achieved in both logic and memory environments because of the impressive \(I_{{\rm on}}{/}I_{{\rm off}}\) (Fig. 11).
Conclusions
A promised TFET design called FATFET has been presented in this work. The modeling of ferroelectric parameters along with \(V_{{\rm int}}\) calculations of MFIS geometry having QS and NQS has been addressed for the promised design. The benefits of ferroelectric with the design of FATFET has been shown in comparison to ATFET (without ferroelectric). Furthermore, the significance of \(P_{{\rm r}}\) and \(E_{{\rm c}}\) on FATFET for an optimized design is addressed. The results signifies that the high \(E_{{\rm c}}\) and \(P_{{\rm r}}\) are undesirable for stable performance of TFETs. With the optimized option of \(P_{{\rm r}}\) and \(E_{{\rm c}}\); FATFET has shown in delivering high \(I_{{\rm on}}\), lower \(I_{{\rm off}}\), low \(\hbox{SS}_{{\rm avg}}\), and higher frequency terms (\(f_{{\rm t}}\) and \(f_{{\rm max}}\)), respectively. Furthermore, the linearity analysis signifies that FATFET has the stabilized linear behavior (within the operating regime) compared to ATFET. Compared to recently reported structures, the proposed FATFET design proves to be impressive in terms of DC and SE.
Availability of data and materials
All data generated or analyzed during this study are included in this published article.
Abbreviations
 TFET:

Tunneling fieldeffect transistor
 LTFET:

Line TFET
 ATFET:

Area TFET
 FATFET:

Ferroelectric ATFET
 MFIS:

Metalferroelectricinsulator
 SS:

Subthreshold swing
 SE:

Switching energy
 HZO:

Hafniumzirconiumoxide
 GAA:

Gateallaround
 TCAD:

Technology computer aided design
 BTBT:

Bandtoband tunneling
 TAT:

Trapassistedtunneling
 L–K:

Landau–Khalatnikov
 QS:

Quasistatic
 NQS:

Nonquasistatic
 NC:

Negative capacitance
 Si:

Silicon
 SiGe:

Silicon–germanium
 NBE:

Neutral beam etching
 MOCVD:

Metalorganic vapourphase epitaxy
 ALD:

Atomic layer deposition
 PLD:

Pulsed layer deposition
 EOT:

Effective oxide thickness
 TGF:

Transconductance generation factor
References
Seabaugh BAC, Zhang Q (2010) Lowvoltage tunnel transistors for beyond CMOS logic. Proc IEEE 98(12):2095–2110. https://doi.org/10.1109/JPROC.2010.2070470
Cheng W, Liang R, Xu G, Yu G, Zhang S, Yin H, Zhao C, Ren TL, Xu J (2020) Fabrication and characterization of a novel Si line tunneling TFET with high drive current. IEEE J Electron Devices Soc 8:336–340. https://doi.org/10.1109/JEDS.2020.2981974
Thoti N, Lakshmi B (2017) RF performance enhancement in multifin TFETs by scaling inter fin separation. Mater Sci Semicond Process 71:304–309. https://doi.org/10.1016/j.mssp.2017.08.014
Zhang M, Guo Y, Zhang J, Yao J, Chen J (2020) Simulation study of the doublegate tunnel fieldeffect transistor with step channel thickness. Nanoscale Res Lett. https://doi.org/10.1186/s11671020033607
Verhulst AS, Saeidi A, Stolichnov I, Alian A, Iwai H, Collaert N, Ionescu AM (2020) Experimental details of a steepslope ferroelectric InGaAs tunnelFET with highquality PZT and modeling insights in the transient polarization. IEEE Trans Electron Devices 67(1):377–382. https://doi.org/10.1109/TED.2019.2954585
Saeidi A, Rosca T, Memisevic E, Stolichnov I, Cavalieri M, Wernersson LE, Ionescu AM (2020) Nanowire tunnel FET with simultaneously reduced subthermionic subthreshold swing and offcurrent due to negative capacitance and voltage pinning effects. Nano Lett 20(5):3255–3262. https://doi.org/10.1021/acs.nanolett.9b05356
Kobayashi M, Jang K, Ueyama N, Hiramoto T (2017) Negative capacitance for boosting tunnel FET performance. IEEE Trans Nanotechnol 16(2):253–258. https://doi.org/10.1109/TNANO.2017.2658688
Lee MH, Lin JC, Wei YT, Chen CW, Tu WH, Zhuang HK, Tang M (2013) Ferroelectric negative capacitance heterotunnel fieldeffecttransistors with internal voltage amplification. In: Technical Digest—international electron devices meeting, IEDM, pp 104–107. https://doi.org/10.1109/IEDM.2013.6724561
Kim SJ, Mohan J, Summerfelt SR, Kim J (2019) Ferroelectric Hf_{0.5}Zr_{0.5}O_{2} thin films: a review of recent advances. Minerals Met Mater Soc 71(1):246–255. https://doi.org/10.1007/s1183701831405
Thoti N, Li Y, Kola SR, Samukawa S (2020) Highperformance metalferroeletricsemiconductor nanosheet line tunneling field effect transistors with strained SiGe. In: International conference on simulation of semiconductor processes and devices (SISPAD), pp 375–378. https://doi.org/10.23919/SISPAD49475.2020.9241591
Thoti N, Li Y, Kola SR, Samukawa S (2020) New proficient ferroelectric nanosheet line tunneling FETs with strained SiGe through scaled nepitaxial layer. In: IEEE 20th international conference on nanotechnology (IEEENANO), pp. 319–322. https://doi.org/10.1109/NANO47656.2020.9183460
Khan AI, Radhakrishna U, Chatterjee K, Salahuddin S, Antoniadis DA, Member S (2016) Negative capacitance behavior in a leaky ferroelectric. IEEE Trans Electron Devices 63(11):4416–4422. https://doi.org/10.1109/TED.2016.2612656
Thoti N, Li Y (2021) Promised design of energyefficient negativecapacitance vertical tunneling FET. ECS J Solid State Sci Technol 10(7):075002. https://doi.org/10.1149/21628777/ac10b4
Synopsys S (2018) Sentaurus TCAD. Synopsys, Inc
International Roadmap for Devices and Systems—More than Moore—IRDS (2020). https://irds.ieee.org/editions/2020
Hurkx GAM, Klaassen DBM, Knuvers MPG (1992) A new recombination model for device simulation including tunneling. IEEE Trans Electron Devices 39(2):331–338. https://doi.org/10.1109/16.121690
Jena D, Fang T, Zhang Q, Xing H (2008) Zener tunneling in semiconducting nanotube and graphene nanoribbon pn junctions. Appl Phys Lett 93(11):112106–13. https://doi.org/10.1063/1.2983744.0806.0139
Thoti N, Li Y, Kola SR, Samukawa S (2020) Optimal intergate separation and overlapped source of multichannel line tunnel FETs. IEEE Open J Nanotechnol 1:38–46. https://doi.org/10.1109/ojnano.2020.2998939
Thoti N, Li Y (2020) Influence of fringingfield on DC/AC characteristics of Si_{1−x}Ge_{x} based multichannel tunnel FETs. IEEE Access 8:208658–208668. https://doi.org/10.1109/ACCESS.2020.3037929
Saha AK, Datta S, Gupta SK (2018) “Negative capacitance’’ in resistorferroelectric and ferroelectricdielectric networks: apparent or intrinsic? J Appl Phys 123(10):105102–118. https://doi.org/10.1063/1.5016152
Obradovic B, Rakshit T, Hatcher R, Kittl JA, Rodder MS (2018) Modeling transient negative capacitance in steepslope FeFETs. IEEE Trans Electron Devices 65(11):5157–5164. https://doi.org/10.1109/TED.2018.2868479.1809.02053
Liu Z, Bhuiyan MA, Ma TP (218) A critical examination of  quasistatic negative. In: Technical Digest—international electron devices meeting, pp 711–714. https://doi.org/10.1109/IEDM.2018.8614614
Cao W, Banerjee K (2020) Is negative capacitance FET a steepslope logic switch? Nat Commun 11(1):1–8. https://doi.org/10.1038/s41467019137979
Rabe K, Ahn C, Triscone J (2007) Physics of ferroelectrics: a modern perspective. Springer, New York, pp 1–388
Nathaniel N, Ahluwalia R, Srolovitz DJ (2012) Depletionlayerinduced size effects in ferroelectric thin films: a Ginzburg–Landau model study. Phys Rev B Condens Matter Mater Phys 86(9):1–7. https://doi.org/10.1103/PhysRevB.86.094104
Hu VPH, Lin HH, Lin YK, Hu C (2020) Optimization of negativecapacitance verticaltunnel FET (NCVTFET). IEEE Trans Electron Devices 67(6):2593–2599. https://doi.org/10.1109/TED.2020.2986793
Lin CI, Khan AI, Salahuddin S, Member S, Hu C (2016) Effects of the variation of ferroelectric properties on negative capacitance FET characteristics. IEEE Trans Electron Devices 63(5):2197–2199. https://doi.org/10.1109/TED.2016.2514783
Thoti N, Li Y (2021) pSiGe nanosheet line tunnel fieldeffect transistors with ample exploitation of ferroelectric. Jpn J Appl Phys 60(5):054001–15. https://doi.org/10.35848/13474065/abf13e
Mochizuki S, Bhuiyan M, Zhou H, Zhang J, Stuckert E, Li J, Zhao K, Wang M, Basker V, Loubet N, Guo D, Haran B, Bu H (2020) Stacked gateallaround nanosheet pFET with highly compressive strained Si_{1−x}Ge_{x} channel, pp 231234. https://doi.org/10.1109/IEDM13553.2020.9372041
Detchprohm T, Ryou JH, Li X, Dupuis RD (2020) Future aspects of MOCVD technology for epitaxial growth of semiconductors. In: Metalorganic vapor phase epitaxy (MOVPE): growth, materials properties, and applications, pp 507–548. Wiley, New York. https://doi.org/10.1002/9781119313021.ch14
Samukawa S (2019) Neutralbeam technologies for novel nanomaterials and nanodevices: suppressing the formation of defects at the atomic layer level. IEEE Nanatechnol Mag 13(6):21–33. https://doi.org/10.1109/MNANO.2019.2941034
Nukala P, AntojaLleonart J, Wei Y, Yedra L, Dkhil B, Noheda B (2019) Direct epitaxial growth of polar (1–x)HfO2(x)ZrO2 ultrathin films on silicon. ACS Appl Electron Mater 1:2585–2593. https://doi.org/10.1021/acsaelm.9b00585
Verhulst AS, Leonelli D, Rooyackers R, Groeseneken G (2011) Drain voltage dependent analytical model of tunnel fieldeffect transistors. J Appl Phys 110(2):024510–110. https://doi.org/10.1063/1.3609064
Levinshtein ME, Rumyantsev SL, Shur MS (2001) Properties of advanced semiconductor materials : GaN, AlN, InN, BN, SiC, SiGe. Wiley, New York, pp 1–194
Thoti N, Li Y (2022) Gateallaround nanowire vertical tunneling FETs by ferroelectric internal voltage amplification. Nanotechnology 33(5):055201. https://doi.org/10.1088/13616528/ac2e26
Bansal AK, Kumar M, Gupta C, Hook TB, Dixit A (2018) Series resistance reduction with linearity assessment for vertically stacked junctionless accumulation mode nanowire FET. IEEE Trans Electron Devices 65(8):3548–3554. https://doi.org/10.1109/TED.2018.2845541
Garg N, Pratap Y, Gupta M, Kabra S (2020) Reliability assessment of GaAs/AlO junctionless FinFET in the presence of interfacial layer defects and radiations. IEEE Trans Device Mater Reliab 20(2):452–458. https://doi.org/10.1109/TDMR.2020.2991662
Thoti N, Li Y, Sung Wl (2022) Significance of work function fluctuations in SiGe/Si heteronanosheet tunnelFET. IEEE Trans Electron Devices 69(1):434–438
Acknowledgements
This work was supported in part by the Ministry of Science and Technology, Taiwan, under Grant MOST 1102221EA49139, Grant MOST 1102218E492003MBK, Grant MOST 1092221E009033, and Grant MOST 1092634F009030, and in part by the ”Center for mmWave Smart Radar Systems and Technologies” under the Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education in Taiwan.
Funding
Grant MOST 1102221EA49139, Grant MOST 1102218E492003MBK, Grant MOST 1092221E009033, and Grant MOST 1092634F009030.
Author information
Authors and Affiliations
Contributions
YL supervised the study. NT and YL performed the simulation and preparation, reviewed, and editied theoriginal draft. Both authors read and approved the final manuscript.
Authors’ information
Narasimhulu Thoti received the B.Tech. and M.Tech. degrees from Jawaharlal Nehru Technological University, Hyderabad, India, in 2009 and 2011, respectively. He is currently pursuing the Ph.D. degree with the Parallel and Scientific Computing Laboratory, National Yang Ming Chiao Tung University, Hsinchu, Taiwan. His research interests include the modeling and fabrication of emerging semiconductor devices. He was a recipient of the Best Poster Paper Award from IWPSD 2019, Kolkata, India, and IEDMS 2021, Tainan City, Taiwan. Yiming Li is currently a Full Professor of Electrical and Computer Engineering with National Yang Ming Chiao Tung University, Hsinchu, Taiwan. He has authored or coauthored more than 200 research papers appearing in international book chapters, journals, and conferences. His current research interests include computational electronics, device physics, semiconductor nanostructures, modeling and parameter extraction, biomedical and energy harvesting devices, and optimization techniques. He has been on the Program Committee of IEDM since 2011.
Corresponding author
Ethics declarations
Competing interests
The authors declare that they have no competing interests.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Thoti, N., Li, Y. Design of GAA Nanosheet Ferroelectric Area Tunneling FET and Its Significance with DC/RF Characteristics Including Linearity Analyses. Nanoscale Res Lett 17, 53 (2022). https://doi.org/10.1186/s11671022036908
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s11671022036908
Keywords
 Ferroelectric
 Gateallaround
 Internal voltage
 Linearity
 Nanosheet
 nEpitaxy
 \(\hbox{Si}_{1x}\hbox{Ge}_x\)
 Switching energy