1ESSDERC 2002 – Florence – 25.09.2002
Universal Test Structure andUniversal Test Structure andCharacterization Method for BiasCharacterization Method for Bias--Dependent Dependent
Drift Series Resistance of HV MOSFETsDrift Series Resistance of HV MOSFETs
C. Anghel1, N. Hefyene1, A.M. Ionescu1, R. Gillon2, S. Frere2, J. Rhayem2
1Institute of Microelectronics and Microsystems,Swiss Federal Institute of Technology Lausanne (EPFL)
2AMIS, Oudenaarde, Belgium
2ESSDERC 2002 – Florence – 25.09.2002
Outline:Outline:- Introduction- HV X-DMOS and Intrinsic Drain (Key) Voltage, VK:impact on modelling and electrical characterization
- New Test Structure and related DC characterization:MESDRIFT
- Experimental investigation of Rdrift with MESDRIFT- Physics and modelling of bias-dependent Rdrift
- HV DC model build-up with MESDRIFT- Extension for AC modelling- Conclusion
3ESSDERC 2002 – Florence – 25.09.2002
IntroductionIntroduction
Modelling Status: • Macro-models → good accuracy, but not physical ⇒ not able to take into account phenomena specific to the extended drain region: quasi-saturation • Compact models → time consuming for calibration and convergence
Interest HV Lateral MOSFETs: +Automotive RF
Hot Topics:• Develop fully-analytical DC/AC accurate models of the HV MOSFETs• Investigation of SOA & HC degradation
Modelling & Characterisation Possible Approach: • Investigate/access/model separately the intrinsic MOS channel region and the bias-dependent characteristics of the drift region
4ESSDERC 2002 – Florence – 25.09.2002
XDMOS Transistor Architecture and Intrinsic Drain Voltage, VXDMOS Transistor Architecture and Intrinsic Drain Voltage, VKK
0
1
2
3
4
5
6
7
0 20 40 60 80 100
VD (V)
ID(m
A)
VG=12V
10V
8V
6V
4V
2V3V
5V
7V
9V
11V
I
II
III
XDMOS – ID(VD) characteristics- I – intrinsic MOS channel pinch-off- II – quasi-saturation → channel pinch-off →
self-heating- III – quasi-saturation
VK allows: • distinguish between different saturation mechanisms → physical modeling!• investigation and modeling of the drift region: bias dependent resistance
Intrinsic MOS
Drain extension
Intrinsic MOS saturation
Self-heating
Quasi-saturation
5ESSDERC 2002 – Florence – 25.09.2002
MESDRIFT MESDRIFT –– a new test structurea new test structure
MESDRIFT:• small contact, same type as drift, close to the boundary MOSFET – drift zone• high impedance voltmeter used to monitor VK
• width of K contact negligible in comparison with W of global transistor → no influence on the global characteristics• physical dimensions of K contact → small shift in VK
MESDRIFT – XDMOS – cross-section
MESDRIFT – XDMOS – upper view
6ESSDERC 2002 – Florence – 25.09.2002
XDMOS vs. MESDRIFT XDMOS vs. MESDRIFT –– DC Characteristics:DC Characteristics:
ID – VG @ #VD
• very good match obtained for transfer characteristics• increased leakage currents observed in the sub-threshold regime due to the influence of the K contact near the metallurgical junction
• low VG: VDBR MESDRIFT < VDBR XDMOS (<5%)
ID – VD @ #VG
7ESSDERC 2002 – Florence – 25.09.2002
Experimental investigation of Experimental investigation of RRdriftdrift with MESDRIFT (1)with MESDRIFT (1)Rdrift extraction at low VD: intrinsic vs. extrinsic characteristics
Transfer Characteristics: Extrinsic/Intrinsic – using MESDRIFT – low VD Rdrift when VG due to:
• (acc.) channel formation in drift• slight reduction of depletion zones
( ) ( ) ( )gext G gint Gdrift G
ox 0
V VR V
(W / L)Cθ −θ
=µ
max int. max extr.
g int. g extr.
g > g < θ θ
( )( )
D m G Tg
G T
I / g V V 1V V
− − θ =−
8ESSDERC 2002 – Florence – 25.09.2002
Experimental investigation of Experimental investigation of RRdriftdrift with MESDRIFT (2)with MESDRIFT (2)VK measurements, @ different biasing conditions
0
5
10
15
0 5 10 15 20 25 30 35
VD (V)
VK (V
)
VG = 2V
3V
4V
5V
6V
12V
VK vs. VG with VD as parameter
0
4
8
12
16
20
1 2 3 4 5 6 7 8 9 10 11 12
VG(V)
VK(V
)
VD=30V
20V
10V
5V
2V
1V
VK vs. VD with VG as parameter
• VK slope-change – transition saturation – linear • VK VG (see explanation on next slide)• VK VD (expected)
9ESSDERC 2002 – Florence – 25.09.2002
Drift resistance variation function of biasing conditionsDrift resistance variation function of biasing conditions
Rdrift vs. VD with VG as parameter
100
1000
10000
100000
0 5 10 15 20 25 30
VD(V)
R drif
t, R c
h, (
k Ω)
VG=2V
4V, 6V, 8V, 10V and 12V
0.1
1.0
10.0
100.0
1000.0
10000.0
1 2 3 4 5 6 7 8 9 10 11 12
VG(V)
Rdr
ift, R
ch (k
Ω)
VD=30V
20V, 10V, 5V, 2V and 1V
Rdrift vs. VG with VD as parameter
• Extraction: straight-forward, no time consuming method• Rdrift Rch VG - drift influence dictates the overall behaviour at high VG
10ESSDERC 2002 – Florence – 25.09.2002
Drift resistance variation induced by accumulation/depletion in Drift resistance variation induced by accumulation/depletion in driftdrift
Depletion zones evolution VD= 20V, #VG Depletion zones evolution VG= 3V, #VD
#VG, VD=const• depletion depth induced by pin diode:slight reduction• channel modulation controlled by gate terminal in the drift zone (JFET effect)
#VD, VG=const• by increasing VD the accumulation channel in the drift pinches off → a depletion zone appears in the drift (JFET - like turn-off)
11ESSDERC 2002 – Florence – 25.09.2002
RRdriftdrift/R/Ronon in all XDMOS operation regimesin all XDMOS operation regimes Temperature influenceTemperature influence
0
25
50
75
100
0 2 4 6 8 10 12
VG(V)
Rdr
ift/(R
ch+R
drift
)(%)
VD=1V
VD=30V
T(°C)=25, 75, 125
0
20
40
60
80
100
0 5 10 15 20 25 30
V D(V)
Rdr
ift/(R
ch+R
drift
) (%
)
V G=1V
2V
3V
4V
12V
Rdrift/Ron vs. VD, #VG, at T=25°C(~ negligible self-heating)
Rdrift/Ron vs. VG, VD and T as parameters
• analogue operation regime → Ron dominated by Rch• medium, high voltage for VD or/and VG→ Ron dominated by Rdrift
• proper modelling of drift region is critical for accurate simulation at high voltages
12ESSDERC 2002 – Florence – 25.09.2002
HV DC model buildHV DC model build--up with MESDRIFTup with MESDRIFT
1. Low voltage MOS model calibrated on intrinsic transfer and output characteristics
2. Quasi-empirical expression tuned on Rdrift experimentally revealed by MESDRIFT
1000
10000
100000
0 10 20 30
VD (V)
Rdrif
t (Ω
)T = 25 C
75 C150 C
VG = 2V
VG = 4V
VG = 12V
o o o o o o o measures
_____ drift expression
( )[ ]1CVBexplnARR D0DD ++⋅+=*
* N. Hefyene, E. Vestiel , S. Frere, C. Anghel, A.M. Ionescu, R. Gillon, SISPAD 2002
Rdrift (VD), #VG and T, solid – quasi-empirical expression, dashed – MESDRIFT
13ESSDERC 2002 – Florence – 25.09.2002
Modelling HV XDMOS with MESDRIFT: transfer characteristicsModelling HV XDMOS with MESDRIFT: transfer characteristics
0.E+00
1.E-03
2.E-03
3.E-03
4.E-03
5.E-03
0 2 4 6 8 10 12 14
VG (V)
ID (A
)
VD = 20V
10V9V
2V
3V
4V
5V
6V7V
8V
1V
ID(VG) black – simulation, red – measurement
MAX. ERR. < 15%
1.E-13
1.E-11
1.E-09
1.E-07
1.E-05
1.E-03
0 2 4 6 8 10 12 14
VG (V)
Lo
g ID
(A
)
VD = 1 to 20V
log ID(VG) black – simulation, red – measurement
0.E+00
2.E-04
4.E-04
6.E-04
8.E-04
1.E-03
0 2 4 6 8 10 12 14
VG (V)
gm (S
)
VD = 20V
10V
1V
gm(VG) black – simulation, red – measurement
• medium and high voltage regimes →optimization using drift resistance parameters
14ESSDERC 2002 – Florence – 25.09.2002
Modelling HV XDMOS with MESDRIFT: output characteristicsModelling HV XDMOS with MESDRIFT: output characteristics
0
2
4
6
8
0 20 40 60 80
V D (V)
I D (m
A)
VG = 12V
VG = 6V
T = 25, 75 & 125°C
Quasi-saturation • good results obtained for overall characteristics including temperature dependence (max. err. < 20% and RMS < 5%)• saturation phenomena, including quasi-saturation well modelled with physical parameters• better results are expected when tuning the drift model on the pulsed measurements (without self-heating)
ID(VD) black – simulation, red – measurement
15ESSDERC 2002 – Florence – 25.09.2002
ConclusionConclusion• a new test structure, MESDRIFT, for characterisation of the drift zone of HV MOSFETs was designed and successfully validated based on the intrinsic drain voltage concept (VK)• for the first time, drift resistance variation for the entire voltage domain was revealed• temperature effects studied from 25°C up to 150°C• a simple VK - based modelling strategy using a low-voltage BSIM3v3transistor module and an empirical expression for drift resistance calibrated using MESDRIFT → validated for the whole operation range.
Acknowledgement• This work was supported by the IST-1999-12257 'Automacs' EC project and the Swiss OFES No. 00-0009• The authors acknowledge M. Vermandel and B. Bakeroot from ELIS TFCG/IMEC, University of Gent and P. Moens from AMIS, Belgium for TCAD simulations