Professional Documents
Culture Documents
https://doi.org/10.1007/s12633-020-00861-z
ORIGINAL PAPER
Abstract
U-channel ultra-thin body and buried oxide (U-UTBB) Silicon On Insulator (SOI) Metal Oxide Semiconductor Field Effect
Transistors (MOSFETs) present unique features which are simple, high-performance, area efficient, and compatible with CMOS
technology. In this paper, we present the electrical characteristics of a U-UTBB SOI MOSFET with reforming the U-channel.
This proposed symmetrical and planar device is initiated by removing two spacers around the recessed metal gate in the U-
channel MOSFET. Hence, it causes to increase the overall channel length at the firm metal gate space along with saving area. As
it has confirmed by two-dimensional and two-carrier device simulation results, the proposed structure which is termed Reformed
U-UTBB (R-U-UTBB) fully depleted SOI MOSFET exhibits advantages in the device performance focusing on short channel
effects (SCEs) factors including the sub-threshold slope, the drain induced barrier lowering (DIBL), and the threshold voltage
roll-off. Leakage current and on-to-off current ratio are studied as well which all of them show the superiority of proposed
structure compared with the U-UTBB FD SOI MOSFET and a conventional UTBB (C-UTBB) FD SOI MOSFET. In addition,
the effects of process-induced variations have investigated by varying buried oxide thickness, top silicon thickness, channel
thickness, substrate doping, and oxide thickness on threshold voltage, sub-threshold slope, and drain induced barrier lowering
(DIBL).
Keywords U-channel . UTBB . SOI MOSFET . TCAD . Short Channel effects . High-k dielectric
(FD) SOI MOSFETs. It has advantages for future low-cost The schematic cross section of U-channel UTBB (U-
energy-efficient applications [15], excellent for short channel UTBB) structure [15] is shown in Fig. 1(b) which is asserted
effect immunity and ideal subthreshold characteristics [17, by moving up the spacers located between gate and source/
18]. In UTBB device, short-channel effects are usually con- drain and adding two vertical channels (Xj) to the horizontal
trolled by a narrow silicon layer, thus this thin silicon body channel (Lg), DIBL and subthreshold slope are enhanced.
enhances the electrostatic qualities [19], reduces the leakage Source and drain regions in these structures have constructed
current [5, 20], increases on-to-off current ratio because of using uniform doping profiles with Gaussian tails. In the pro-
carrier mobility improvement [21], and allows the channel to posed nMOS structure that is shown in Fig. 1(c) and we have
be un-doped [5, 22]. Furthermore, thinning of the BOX has called it as Reformed U-channel UTBB (R-U-UTBB) struc-
been suggested as a good solution to lower self-heating effect ture, the spacers are eliminated entirely, so the Gaussian dop-
and fringing electric fields through the BOX and substrate ing profile with minimum uniform doping part in an intrinsic
[23]. By using intrinsic channel, higher drive current and silicon film is used.
steeper subthreshold slope are attained. Possessing awesome There is a ration of the gate length to the channel
subthreshold swing and DIBL are critical for good overdrive thickness (Lg /Tch) of UTB MOSFETs that short-channel
and hence device switching speed. Moreover, because of re- effects are sufficiently defeated when this ratio is greater
moving the random dopant fluctuation, significant improve- than 4 [5]. So, the gate lengths of structures have chosen
ment of variability control is achieved [24, 25]. 16-nm. Also, the top silicon film has 15-nm thick and
To derive the most forgoing benefits, an innovative U- reaches to 4-nm beneath the metal gate to satisfy Lg/Tch
channel UTBB (U-UTBB) SOI MOSFET has submitted in ratio. For the reason that thin body devices can control
this work which has no fabrication process complexity in short channel effects with only un-doped channel, the
comparison with FinFET and fully compatible with SOI channel of novel structure has left intrinsic. Significant
CMOS technology as well as owing admissible Technology performance enhancements such as improving carrier mo-
Computer-Aided Design (TCAD) analysis. In the proposed bility and high on-to-off current ratio with low change
U-channel MOSFET device, two spacers around the recessed because of random dopant fluctuation can be expected
metal gate are removed. The proposed structure is called [21]. Although, with un-doped channel in UTBB
Reformed U-UTBB (R-U-UTBB) FD SOI MOSFET. Our MOSFETs, threshold voltage adjustment can be attained
results show the superiority of proposed structure compared by gate work function engineering if necessary [5].
with the U-UTBB FD SOI MOSFET and a conventional According to experimental results reported in [29], a prac-
UTBB (C-UTBB) FD SOI MOSFET. The paper is organized tical limit for SiO2 thickness as a gate oxide is 10–12 Å
as follows: In Section 2, the macroscopic novel device tem- for excellent electrical properties, which a 1-nm film of
plate, simulation details, and modeling are described. And in SiO2 as an interfacial layer has used. An aluminum gate
Section 3, the proposed device compares with its conventional with a 2-nm-thick HfO2 dielectric has specified. So the
counterparts and advantages are discussed individually. effective work function is calculated by simulator and is
given by:
models have activated in simulation. The physical device 3 Results and Discussion
model includes the effects such as band gap narrowing
(BGN) for the carrier statistics, combination of Klaassen 3.1 Output Characteristic and Sub-Threshold Slope
(KLA) and Shiharata (SHI) and parallel electric field depen-
dence are considered for carrier mobility modeling (doping In this paper, we compare the simulation results of R-U-
dependence, high field saturation effects). Also, Shockley– UTBB structure with the U-UTBB and the C-UTBB struc-
Read–Hall (SRH) and Auger (AUG) for recombination as tures. The output characteristics of three structures for two
well as avalanche generation are used. With continued scaling gate-source voltages of 0.4 and 0.8 V are shown in Fig. 2.
into the deep nanometer regime, the energy balance model has
activated (HCTE) to perform accurate simulation. When the
energy balance transport model is applied, the appropriate Table 2 Critical parameters of the proposed structure in the device
impact ionization model is (CONCANNON). This accurate simulation
model calculates the probability of a carrier having sufficient
Device parameter Value
energy to cause impact ionization. In addition, due to take into
account quantum effects, Bohm Quantum Potential (BQP) for Gate length (Lg) 16 nm
quantum modelling has applied. Second interfacial layer (SiO2) thickness (TSio2) 1 nm
high-k gate dielectric (HfO2) thickness (THk) 2 nm
Top silicon thickness (TSOI) 15 nm
Table 1 Dielectric Channel thickness (Tch) 4 nm
parameter defaults Parameter name Value
Buried Oxide (BOX) thickness (TBOX) 20 nm
E·HIGHK 4.81 eV Effective work function (Φmeff) 4.3065 eV
S·HIGHK 0.95 n+ Source & Drain maximum doping 1020 cm−3
WORKF 4.28 eV Substrate doping 1015 cm−3
Silicon
However, the R-U-UTBB device has a lower saturation cur- Leff ¼ Lg þ 2X j ð5Þ
rent than the U-channel and conventional UTBB, but it is
more stable than the C-UTBB and U-UTBB structures as it
can be concluded from slope of the graphs. The values of this parameter is 24-nm for the U-UTBB
The sub-threshold slope is one of the important quality of a and 38-nm for the R-U-UTBB structures which are given
MOSFET’s current–voltage characteristic. The drain current in Table 3. However, by increasing Leff in the U-UTBB
behavior in the sub-threshold region is similar to the exponen- and the R-U-UTBB structures, both the Cdep and Cox are
tially decreasing the current of a forward biased diode. The growing, but the effect of rising C o x is dominant.
sub-threshold slope is linked to the depletion capacitance by Consequently, the sub-threshold slope of R-U-UTBB
the equation: structure improved about 8% and 18% in comparison with
the U-UTBB and the C-UTBB structures, respectively.
KT C dep These values are summarized in Table 3. As transfer char-
S¼ lnð10Þ 1 þ ð2Þ
q C OX acteristics of three structures are shown in Fig. 3, the re-
duction of sub-threshold slope for the proposed R-U-
where T is the temperature, q is the electronic charge, S is the
UTBB device than the C-UTBB and U-UTBB devices
sub-threshold slope, Cdep is the capacitance of depletion re-
are notable. To show the variation of sub-threshold slop
gion, and Cox is the gate-oxide capacitance [33]. It can be seen
versus gate length, this short channel effect has simulated
from Eq. 2 that decreasing the depletion capacitance Cdep, or
for gate length of 8 to 26-nm and plotted in Fig. 4.
increasing Cox will improve the sub-threshold slope.
In fact, with decreasing the gate length, an extra factor is
introduced to account for increasing the sub-threshold slope
Table 3 SCEs parameters of R-U-UTBB SOI-MOSFET compared to
[19], so that:
U-UTBB and C-UTBB devices
T OX el X j 3 T dep
EIS ¼ 1þ ð3Þ parameter C-UTBB U-UTBB R-U-UTBB
Lel 2 4 Lel
" sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi# Xj (nm) 0 4 11
KT C dep εSi V ds Lg (nm) 16 16 16
S¼ lnð10Þ 1 þ þ EIC 1 þ 2 ð4Þ
q C OX εOX φd Leff (nm) 16 24 38
DIBL (mV/V) 137 100 81
where Vds is the source to drain potential and φd is the drain Subthreshold slope (mV/dec) 93 81 76
barrier potential that should be taken to 1 V for simplicity. Threshold voltage (V) 0.27 0.23 0.25
Figure 3 shows the transfer characteristic for Vds = 0.05 V and Threshold voltage roll-off (mV) 46 47 25
subthreshold slope of the R-U-UTBB in comparison with the Ioff (A) 8 × 10−08 2 × 10−08 4.3 × 10−09
U-UTBB and the C-UTBB structures. In fact, in the U-channel Ion/Ioff 4.4 × 103 1.6 × 104 6.4 × 104
structures instead of Lg, Leff should be used as follows [14]:
Silicon
Fig. 5 Id-Vg curves on linear (Vds = 0.05 V) and saturation (Vds = 1 V) Fig. 7 Ion/Ioff ratio of the C-UTBB, U-UTBB, and R-U-UTBB SOI-
threshold voltage for showing DIBL of R-U-UTBB SOI-MOSFET MOSFETs versus gate length which
Silicon
roll-off UTBB SOI-MOSFETs are documented at 16-nm 7-nm with a step of 1-nm. As shown in Fig. 11(a),
gate length in Table 3. despite the gradually increase and decrease, respectively
in the threshold voltage and the sub-threshold slop,
3.5 Process-Induced Variations DIBL behavior in Fig. 11(b) is different. Because of
Quantum effects [15] and strongly dependence of
Due to variations during the device fabrication process, the SCEs on the body thickness for UTBB SOI-MOSFETs
geometry and different parameters of the R-U-UTBB SOI- [5], the amount of DIBL has increased at channel thick-
MOSFET are subject to uncertainties. Process-induced varia- nesses less than 4 nm. On the other hand, as the chan-
tions have investigated by studying the effect of buried oxide nel thickens, due to deviation of the structure from U-
thickness (TBOX), top silicon thickness (TSOI), channel thick- shape and thus reducing the effect of channel separation
ness (Tch), substrate doping, and oxide thickness (Tox) on in U-channel device [16], a significant increment on
threshold voltage, sub-threshold slope, and DIBL. DIBL has been occurred. Therefore, in the simulation
Figure 9(a) shows the effect of BOX thickness variation from results of 4 nm channel thickness, the lowest value of
10 to 30-nm on threshold voltage and sub-threshold slope of DIBL has been observed.
R-U-UTBB device. However sub-threshold slope become al- However, the ballistic transport has ignored in simulations.
most unchanged for thicknesses more than 20-nm, but thresh- It has reported in a 30-nm gate of UTBB MOSFET, about
old voltage is decreasing constantly due to reducing effect of 20% of electrons arriving to the drain terminal are ballistic.
substrate acceptor dopant. The minimum value of DIBL occur Moreover, it is to be emphasized that the requirement of an
at BOX thickness of 20-nm as demonstrate in Fig. 9(b), while un-doped channel will increase this value [19].
for thin BOX it has significant increase. To reduce the doping concentration of channel and as
Figure 10(a) and (b) shows the effect of silicon film varia- a result the Vth variations in the UTBB SOI-MOSFET,
tion on threshold voltage, sub-threshold slope, and DIBL, re- high-impurity-concentration substrate has been proposed
spectively, which has minor impact on them compared with [25]. The effect of this parameter variation on threshold
other parameters. As shown in Fig. 10(b), the minimum value voltage, sub-threshold slope, and DIBL has been inves-
of DIBL belong to SOI thickness of 16-nm and is about tigated for both donor and acceptor dopant of substrate
80 mV/V. and present measured data in Fig. 12(a), (b), and (c). As
The next geometry parameter of novel device is shown in these figures, if the substrate dopes with donor
channel thickness which has been changed from 2 to dopant with density of 1015 atom/cm3, 48 mV/V will be
23. Burignat S, Flandre D, Md Arshad MK, Kilchytska V, Andrieu F, Mocuta A, Womack S, Gribelyuk M, Jones EC, Miller RJ, Philip
Faynot O, Raskin J-P (2010) Substrate impact on threshold voltage Wong H-S, Haensch W (2002) Extreme scaling with ultra-thin Si
and subthreshold slope of sub-32 nm ultra thin SOI MOSFETs with channel MOSFETs. IEEE Electron Dev Meeting:267–270
thin buried oxide and un-doped channel. Solid State Electron 54: 29. Wilk GD, Wallace RM, Anthony JM (2001) High-κ gate dielec-
213–219 trics: current status and materials properties considerations. J Appl
24. Sugii N, Tsuchiya R, Ishigaki T, Morita Y, Yoshimoto H, Torii K, Phys 89:5243–5275
Kimura S (2008) Comprehensive study on Vth variability in silicon 30. Trojman L, Vaca A (2018) Luis Miguel Procel, “on the parameter
on thin BOX (SOTB) CMOS with small random-dopant fluctua- extraction of short channel UTBB FD SOI FET’s with high-κ metal
tion: finding a way to further reduce variation. IEEE Electron Dev gate and TCAD Modelling:1–4
Meeting:1–4 31. Device Simulator Atlas User's Manual, Silvaco International
25. Ohtou T, Sugii N, Hiramoto T (2007) Impact of parameter varia- Software, Santa Clara, USA, January (2017) Silvaco
tions and random dopant fluctuations on short-channel fully deplet- International, Santa Clara 〈www.silvaco.com〉
ed SOI MOSFETs with extremely thin BOX. IEEE Electron Device 32. Md Arshad MK, Kilchytska V, Emam M, Andrieu F, Flandre D,
Letters 28:740–742 Raskin J-P (2014) Effect of parasitic elements on UTBB FD SOI
26. Kim S-D, Park C-M, Woo JCS (2002) Advanced model and anal- MOSFETs RF figures of merit. Solid State Electron 97:38–44
ysis of series resistance for CMOS scaling into nanometer regime-
33. Colinge J-P (1991) Silicon-on-insulator technology: materials to
part II quantitative analysis. IEEE Trans Electron Dev:467–472
VLSI. Kluwer Academic Publishers
27. Doyle B, Arghavani R, Barlage D, Datta S, Doczy M, Kavalieros J,
Murthy A, Chau R (2002) Transistor elements for 30 nm physical
gate lengths and beyond. Intel Technol J Semiconduct Technol Publisher’s Note Springer Nature remains neutral with regard to jurisdic-
Manufact:42–54 tional claims in published maps and institutional affiliations.
28. Doris B, Ieong M, Kanarsky T, Zhang Y, Roy RA, Dokumaci O,
Ren Z, Jamin F-F, Shi L, Natzle W, Huang H-J, Mezzapelle J,