Dual Material Gate Technique for Enhanced

Transconductance and Breakdown Voltage of Trench

Power MOSFETs
Raghvendra S. Saxena and M. Jagadesh Kumar, IEEE Trans. on Electron Devices, Vol.56,
pp.517-522, March 2009.

Abstract: In this paper, we propose a new dual material gate trench power MOSFET that
exhibits a significant improvement in its transconductance and breakdown voltage without any
degradation in ON-resistance. In the proposed structure, we have split the gate of a conventional
trench MOSFET structure into two parts for work-function engineering. The two gates share the
control of the inversion charge in the channel. Using two-dimensional numerical simulation, we
have shown that by adjusting the lengths of the two gates to allow equal share of the inversion
charge by them, we get the optimum device performance. Using N+ poly Si as lower gate
material and P+ poly Si as upper gate material, approximately 44% improvement in peak
transconductance and 20% improvement in breakdown voltage may be achieved in the new
device compared to the conventional trench MOSFET.

Index Terms: Trench Gate, Dual Material Gate, Power MOSFET, On-resistance, Breakdown
voltage

Accepted final version of MS# 6437B

The authors are with the Department of Electrical Engineering, Indian Institute of Technology, Hauz Khas, New Delhi 110 016, India (e-mail:
[email protected]).

1
 I. INTRODUCTION

Trench gate technology [1-9] makes the MOSFET a more attractive device for low to medium

voltage power applications as it provides reduced conduction power losses and lower forward

voltage drop due to the absence of parasitic JFET region as compared to the planar power

MOSFET. A trench gate power MOSFET can be designed to have ultra low ON-resistance [1-7]

with high switching speed [2, 3] and can be fabricated easily using the standard Si process

technology [7-9]. Also, since high packing densities can be achieved using trench power

MOSFETs [8], the ON-resistance can be reduced significantly by the parallel conduction of a

large number of unit cells in a given chip area. These features make the trench MOSFETs

suitable for control switching, DC-DC converters, automotive electronics, microprocessor power

supplies etc.

In various power electronic applications, low ON-resistance, higher drive current, low gate-to-

drain capacitance, high transconductance and high breakdown voltage are the desired features

[10-13]. Designing a power MOSFET for a specific application is a compromise among these

parameters because they are linked together by the technology and any attempt to improve one

may adversely affect the other, making the device unsuitable in many other applications. The

most desired performance specifications in all the applications are low ON-resistance and high

breakdown voltage. However, when we attempt to improve the breakdown voltage of a power

MOSFET, the ON-resistance increases drastically [13, 14]. Therefore, to overcome this

difficulty, we propose a dual material gate trench (DMGT) MOSFET structure that shows

improvement not only in breakdown characteristics but also in transconductance without any

degradation in its ON-resistance. A high transconductance makes the device suited for RF power

2
amplification also [11, 15]. Using two-dimensional numerical simulation in ATLAS device

simulator [16], we have analyzed the improved performance of the proposed DMGT MOSFET

by comparing it with a conventional trench gate MOSFET.

II. DEVICE STRUCTURE AND PROPOSED FABRICATION PROCEDURE

Fig. 1(a) shows the schematic top view of the proposed DMGT device and Fig. 1(b) shows its

cross-sectional view along the cut line AA’. As apparent from the figure, DMGT device contains

two sections of its trench gate. The upper section (G1) has a higher work function gate material

whereas lower section (G2) has the smaller work function material. The lengths of the two

sections of the gate are denoted as L1 and L2 in Fig. 1. These sections are electrically connected

by a metal. The lower section of the gate material is chosen to be N+ poly Si (work function, φG2

= 4.17 eV) and the upper gate material is chosen to be P+ poly Si (work function, φG1 = 5.25 eV).

However, to analyze how the work function difference between the two gate materials (ΔφG =

φG1 - φG2) affects the device performance, we have also varied the value of φG1 from (φG2 + 0.25

eV) to (φG2 + 1.25 eV) in our simulations.

The fabrication process of the DMGT structure is illustrated in Fig. 2. First, using some initial

processing steps of a conventional trench gate MOSFET fabrication [7-9], we create the structure

as shown in Fig. 2(a) consisting of the N+ substrate (ND = 1×1019 cm-3) as the drain and 0.1 μm

thick N+ source (ND = 1×1019 cm-3) on the top side, a 2.7 μm thick N-type drift region (ND =

1×1016 cm-3) and a 0.7 μm thick P-type body region (NA = 5×1017 cm-3). It also has a 1.2 μm

wide and 1.2 μm deep trench with a 50 nm thick gate oxide layer grown inside the trench. In the

next step, N+ poly Si is done and CMP is carried out to get the structure shown in Fig. 2(b). A

3
selective etching of the upper part of the poly Si followed by the oxide etching forms the N+ poly

lower gate electrode as shown in Fig. 2(c). Again by growing a 50 nm thick gate oxide followed

by P+ poly deposition and CMP, we realize the structure shown in Fig. 2(d). Finally, to take the

gate contact, we make a contact hole in the middle of the main trench as shown in Fig. 2(e),

having a depth just enough to reach the lower gate material. It is then filled with a metal using

standard metallization process. The final structure is shown in Fig. 2(f).

For device simulation, we have created the DMGT structure in ATLAS and compared it with

the conventional device having the same parameters as above except that the conventional

device has only one gate material of N+ poly Si.

III. SIMULATION RESULTS AND DISCUSSION

The performance of DMGT device depends on the work function difference ΔφG of the two

gates and their lengths L1 and L2. We have first simulated the DMGT device by varying φG1

keeping φG2 constant at 4.17 eV. For this analysis we kept L1 fixed at 0.3 µm. We have also

analyzed the effect of gate length variation of DMGT device by varying L1 but keeping the total

gate length constant (L1 + L2 = L). With this analysis we found the optimum gate length L1 to

be 0.3 μm for achieving the best performance. Then we compared the optimized DMGT device

with the conventional device for current-voltage characteristics, transconductance, ON-

resistance and breakdown voltage.

The difference in the work function of the two gates causes a change in the surface potential as

shown in Fig. 3(a). It is clear from the figure that as the ΔφG increases, the step in the surface

potential profile also increases modifying the channel electric field as shown in Fig 3(b). The

4
channel electric field for different values of L1 is shown in Fig. 3(c) for a fixed ΔφG (1.08 eV,

poly-Si gate case) in comparison with the conventional device. In addition to the peaks at the

body-source junction and at the body-drift junction, the channel electric field has one more peak

in the DMGT device at the location where gate G1 ends, whereas in a conventional device the

electric field increases gradually in the channel. This modified electric field profile results in a

higher acceleration and higher velocity of the charge carriers coming from the source as shown

in Fig. 3(d) in the DMGT device (ΔφG =1.08 eV). Hence, a higher transconductance is expected

in the DMGT device. The additional peak in the channel electric field also provides screening to

the increased drain voltage making the DMGT device better from the hot-carrier suppression

point of view. The value of this peak has almost linear dependence on L1 as shown in Fig. 4.

If the gate length is fixed at L = L1 + L2, for L1 = 0, the device behaves like a conventional

trench gate MOSFET with N+ poly gate length L = L2 and for L2 = 0, it again becomes a

conventional device with P+ poly gate length L = L1. Therefore, the best advantage of the dual

material gate is expected to be in between these two extremes.

Because of increase in peak channel electric field with the increasing L1 (as depicted in Fig. 4),

the carrier transport efficiency from the source to the channel is improved. However, beyond a

certain value of L1, this improvement reduces due to the lowering of the second peak of the

electric field at the body-drift junction. As a result, compared to the conventional single gate

case (having only N+ poly gate), the drive current first increases with increasing L1 and starts

reducing when L1 is increased beyond the limit of 0.3 μm as depicted in Fig. 5(a), which shows

the drive current of DMGT device (for different values of L1) along with the conventional

device as functions of over-drive gate voltage (VGS-VTh).

5
 At higher drain voltages, the electric field peak near the end of the trench is responsible for the

breakdown in both conventional and DMGT devices and may be considered as the main peak of

the electric field. The additional electric field peak in the channel of DMGT device results in the

reduction of the main peak as compared to the conventional device and therefore, we observe an

improvement in the breakdown voltage. Furthermore, as L1 increases the peak channel electric

filed also increases (as depicted in Fig. 4), giving higher breakdown voltage till L1 approaches

the value of L (when it reduces to a single gate device) because at that situation the additional

peak merges with the main peak of the electric field to make it further high. The breakdown

behavior of DMGT device, for various values of L1 along with conventional device (having only

one N+ poly gate) is shown in Fig. 5(b) indicating not only a significant increase in the

breakdown voltage in DMGT device but also a monotonic increase in the breakdown voltage

with increasing L1 (of course till L1 < L), as expected.

The influence of the variation in L1 is summarized in Fig. 6, where various performance

parameters such as threshold voltage, transconductance, ON-resistance and breakdown voltage

are plotted as functions of L1. Fig. 6(a) indicates increase in the threshold voltage in the DMGT

device that can be taken care of in the design by selecting appropriate body doping or gate oxide

thickness. The ON-resistance and the transconductance of the device, evaluated at VDS = 1.0 V,

are shown as functions of L1 in Fig. 6(b) and (c) respectively, indicating that the best values are

achieved at L1 = 0.3 μm. The breakdown voltage variation with L1 is shown in Fig. 6(d) that

indicates a monotonic improvement in the breakdown voltage with increasing L1, as anticipated

earlier.

From the above discussion, it may be inferred that the best performance is achieved when L1 is

kept at about half of the total length of the p-region (~0.3 μm), giving equal control of the

6
inversion charge in the p-region to both the gates G1 and G2 as shown in Fig. 1. When we

compare the performance of the optimized DMGT device (L1 = 0.3 µm) with the conventional

device, we find 44% improvement in peak transconductance, 20% improvement in breakdown

voltage, 6% improvement in drive current and 0.5% improvement in ON-resistance. Usually an

improvement in breakdown voltage adversely affects the ON-resistance [11-13]. However, in the

proposed device, although there is a 20 % improvement in the breakdown voltage, it does not

affect the ON-resistance.

IV. CONCLUSIONS

Using 2D numerical simulations, we have demonstrated that in a trench gate power MOSFET

the sectioning of the gate into two parts gives the flexibility of gate work-function engineering.

Using P+ poly in upper section and N+ poly in deeper section of the gate, we have shown a

significant improvement in the device performance. The optimized performance is achieved

when both the gates have equal control of the inversion charge of the channel. We obtained 20%

improvement in the breakdown voltage and 44% improvement in the peak transconductance in

optimized DMGT device over conventional device. An additional electric field peak obtained in

the channel of the DMGT device also makes it better from hot carrier suppression point of view.

7
 REFERENCES

[1] K. Shenai, “Optimized Trench MOSFET Technologies for Power Devices”, IEEE Trans.

Electron Devices, vol. 39, No. 6, pp.1435-1443, June 1992.

[2] A. Narazaki, J. Maruyama, T. Kayumi, H. Hamachi, J. Moritani and S. Hine, “A 0.35um

Trench Gate MOSFET with an ultra low on state resistance and a high destruction immunity

during the inductive switching”, in Proc. of ISPSD-2OOO, May 22-25, Toulouse, France, pp.

377-380, 2000.

[3] M. Danvish, C. Yue, K. H. Lui, F. Giles, B. Chan, K. Chen, D. Pattanayak, Q. Chen, K.

Terrill and K. Owyang, “A New Power W-Gated Trench MOSFET (WMOSFET) with High

Switching Performance”, in Proc. of ISPSD-2OO3, Apr 14-17, Cambridge, UK, pp. 24-27,

2003.

[4] J. H. Hong, S. K. Chung and Y. I. Choi, “Optimum design for minimum ON-resistance of

low voltage trench power MOSFET”, Microelectronics Journal, vol. 35, No. 3, pp. 287-289,

Mar 2004.

[5] S. Ono, Y. Kawaguchi and A. Nakagawa, “30V New Fine Trench MOSFET with Ultra Low

On-Resistance”, in Proc. of ISPSD-2OO3, Apr 14-17, Cambridge, UK, pp. 28-31, 2003.

[6] X. Yang, Y. C. Liang,G. S. Samudra and Y. Liu, “Tunable Oxide-Bypassed Trench Gate

MOSFET: Breaking the Ideal Superjunction MOSFET Performance Line at Equal Column

Width”, IEEE Trans. Electron Devices, vol. 24, No. 11, pp. 704-706, Nov 2003.

[7] M. H. Juang, W. C. Chueh and S. L. Jang, “The formation of trench-gate power MOSFETs

with a SiGe channel region”, Semicond. Sci. Technol., vol 21, No. 6, pp. 799–802, May

2006.

8
[8] K. S. Nam, J. W. Lee, S. G. Kim, T. M. Roh, H. S. Park, J. G. Koo and K. I. Cho, “A Novel

Simplified Process for Fabricating a Very High Density P-Channel Trench Gate Power

MOSFET”, IEEE Electron Device Letters, vol. 21, No. 7, pp. 365-367, Jul 2000.

[9] M. H. Juang, W. T. Chen, C. I. Ou-Yang, S. L. Jang, M.J. Lin, H. C. Cheng, “Fabrication of

trench-gate power MOSFETs by using a dual doped body region”, Solid-State

Electronics, vol. 48, No. 7, pp. 1079-1085, July 2004.

[10]C. Hu, M. H. Chi and V. M. Patel, “Optimum Design of Power MOSFET‘s”, IEEE Trans.

Electron Devices, vol. 31, No.12, pp. 1693-1700, Dec 1984.

[11]B. J. Baliga, “Power Semiconductor Device Figure of Merit for High-Frequency

Applications”, IEEE Trans. Electron Devices, vol. 10, No.10, pp. 455-457, Oct 1989.

[12]B. J. Baliga, “An Overview of Smart Power Technology”, IEEE Trans. Electron Devices,

vol. 38, No. 7, pp. 1568-1575, July 1991.

[13]R. P. Zingg, “On the Specific On-Resistance of High-Voltage and Power Devices”, IEEE

Trans. Electron Devices, vol. 51, No. 3, pp. 492-499, Mar 2004.

[14]C. Rochefort and R. V. Dalen, “A scalable trench etch based process for high voltage

vertical RESURF MOSFETs”, in Proc. of ISPSD-2005, May 23-26, Santa Barbara, CA, pp.

1-4, 2005.

[15]G. P. V. Pathirana and F. Udrea, “High performance RF power MOSFET”, Electronics Lett.,

vol. 38, No. 21, pp. 1286-1288, Oct 2002.

[16]Atlas User’s Manual: Device Simulation Software, Silvaco Int., Santa Clara, CA.

9
FIGURE CAPTIONS

Fig. 1: Schematic dual material gate trench (DMGT) MOSFET device (a) the top view, (b) the
cross-sectional view along cut line AA’.

Fig. 2: Process steps to fabricate DMGT device.

Fig. 3: (a) Surface potential profile along the channel for different work function differences
between G1 and G2, (b) Electric field profile along the channel for different work function
differences between G1 and G2, (c) Electric field profile for the conventional and DMGT
devices with gate length L1 = 0.3 μm and 0.5 μm while keeping the total channel length, L1 +
L2 constant at 1.1 µm. (d) The corresponding electron velocity in the channel for the
conventional and the DMGT devices with for different values of L1.

Fig. 4: Peak value of the channel electric field in DMGT device as function of gate1 length (L1),

Fig. 5: (a) Transfer characteristics of conventional device and DMGT device and (b) The break
down performance of conventional and DMGT device with different gate1 lengths (L1) and
constant total channel length, L1 + L2 = 1.1 µm.

Fig. 6: The effect of gate1 length (L1) variation in DMGT device on (a) Threshold voltage, (b)
On-resistance, (c) Transconductance and (d) Breakdown voltage with constant total channel
length (L1 + L2 = 1.1 µm)

10
 1.2 µm

A A

Trench

(a)

Source Gate Source

0.0
N+-Si
Depth (μm)

0.2
L1 G1
0.4
P-Si Body
0.6 P+ Poly
L2 G2
0.8
Trench N+ Poly
1.0

1.2 SiO2

N-Si Drift region

Metal

N+-Si Substrate (b)

Drain
Fig. 1

11
 Si3N4

P Body N+
Trench

N Drift N+ poly

N+ Drain (a) (b)

P+ Poly

(c) (d)

Metal

(e) (f)

Fig. 2

12
(a) (b)

(c) (d)

Fig. 3

13
Fig. 4

14
(a)

(b)

Fig. 5

15
(a) (b)

(c) (d)

Fig. 6

16
 R. S. Saxena did his B.E. in Electronics and Communication Engineering from
G. B. Pant Engineering College, Pauri-Garhwal, UP, India, in 1997 and M. Tech.
in Microelectronics from Indian Institute of Technology, Bombay, India, in
2003. He joined Solid State Physics Laboratory (SSPL), Delhi, in 1998 as a
Scientist where he worked on design, modeling and characterization of Infrared
detectors and their read out circuits. He has published about 10 papers in various
journals and conference proceedings. He is currently a PhD research scholar in Electrical
Engineering Department at Indian Institute of Technology, Delhi, India. His current fields of
interest are power electronic devices, nanoscale VLSI devices and Infrared detectors. He is also a
member of Institution of Electronics and Telecommunication Engineers (IETE), India.

M. Jagadesh Kumar (M’95–SM’99) was born inMamidala, Andhra

Pradesh, India. He received the M.S. and Ph.D. degrees in electrical
engineering from the Indian Institute of Technology (IIT), Madras, India.

From 1991 to 1994, he performed postdoctoral research on the modeling

and processing of high-speed bipolar transistors in the Department of
Electrical and Computer Engineering, University of Waterloo, Waterloo,
ON, Canada. While with the University of Waterloo, he also did a research on amorphous-silicon
TFTs. From July 1994 to December 1995, he was initially with the Department of Electronics
and Electrical Communication Engineering, IIT, Kharagpur, India, and then joined the
Department of Electrical Engineering, IIT, New Delhi, India, where he became an Associate
Professor in July 1997 and a Full Professor in January 2005. He has authored three book
chapters and more than 125 publications in refereed journals and conference proceedings. He is
the Editor-in-Chief of IETE Technical Review. He also serves on the Editorial Board of the
Journal of Computational Electronics, Recent Patents on Nanotechnology, Recent Patents on
Electrical Engineering, Journal of Low Power Electronics, and Journal of Nanoscience and
Nanotechnology. His teaching has often been rated as outstanding by the Faculty Appraisal
Committee of IIT Delhi. His research interests include nanoelectronic devices, modeling and
simulation for nanoscale applications, integrated-circuit technology, and power semiconductor
devices.

Dr. Kumar is a Fellow of the Indian National Academy of Engineering and the Institution of
Electronics and Telecommunication Engineers (IETE) India. He is an IEEE Distinguished
Lecturer of the IEEE Electron Devices Society (EDS). He is also a member of the EDS
Publications Committee and EDS Educational Activities Committee. He is an Editor for the
IEEE TRANSACTIONS ON ELECTRON DEVICES. He is the lead guest editor of the joint
special issue of IEEE Transactions on Electron Devices and IEEE Transactions on
Nanotechnology on “Nanowire Transistors: Modeling, Device Design, and Technology”
(November 2008 issue). He has extensively reviewed for different international journals. He was
the recipient of the 29th IETE Ram LalWadhwa GoldMedal for his distinguished contribution in
the field of semiconductor device design and modeling and the 2008 IBM Faculty Award. He
was the first recipient of ISA-VSI TechnoMentor Award given by the India Semiconductor
Association in recognition of a distinguished Indian academician or researcher for playing a
significant role as a Mentor and Researcher.

17