C-DLTS

Principle of operation and limits of application

E. Fretwurst

Institute for Experimental Physics, University of Hamburg

 Principle of operation and basics

 The C-DLTFFT-System at Hamburg
Different hardware tools
Principle of operation
Methods of defect parameter evaluation, limits and systematic errors
High Resolution option, basics, an example and practical limits
 Summary

1 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006

 Principle of Operation
Traps in the space charge region of a
p+-n diode
Left: Electron trap Right: Hole trap
[1] Constant reverse bias (V R)
traps empty traps filled

[2] Carrier injection (Vp)

electron capture hole capture

[3] Thermal emission of trapped carriers (VR)

electron emission hole emission

Bias pulse
Vp < 0 Vp > 0

Capacitance transients C(t) = C(t) – CR

for t > tp
negative positive

2 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006

 Transient analysis
Capacitance transient:

C(t) = C(t) – CR = C0·exp(-(t+t0)/e)

Emisson time constant: 1/e = en + ep
for en » ep  1/e = en
en,ep emission rates for electrons, holes
From measured transients as function of T:
 e(T) values are extracted
Different DLTS techniques:  From Arrhenius plot activation energy E a,n,p and capture
cross section n,p can be extracted using:
 Analog signal processing:  Ea ,n, p 
double boxcar integrator en , p  n , p th, n , p N C ,V exp  
 k BT 
lock-in amplifier
analog correlator
assuming n,p independent on T
 Digital signal processing:
Ea ,n, p
various correlator functions    
ln  e th, n , p N C ,V  ln  n , p 
k BT
Fast Fourier Transformation FFT 
Laplace Transformation
Refolding of “period scans”

3 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006

 Determination of Defect Concentration
Band bending diagrams for deep acceptor:
[2] during filling pulse
[3] during transient phase
2 0 ( E F  Et )
Transition region: 
q 02 N D
Defect concentration Nt:
Amplitude of the C-transient C0  Nt

 2 
1
C  C  2C  C 
0 1   R R  
Nt  2 N
D C     1  R  
 Cp   A  C 
p 
R    0 
 

For  << WR simplifies to:

C
Nt 2 N 0
D C
R

4 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006

 Requirements, Limitations
C-DLTS requirement:
Exponential behavior of capacitance transient if
C  CR or Nt  Ns
Trap concentration  shallow doping concentration
 This implies a limitation for the maximal particle fluence range which can
be investigated
E.g. for Ns = 1012 cm-3, Nt/Ns = 0.1 and a defect with an introduction rate of
g = 1 cm-1 the maximal fluence would be max ≈ 1011 cm-2

Lower limit for detectable trap concentrations:

Depends on the sensitivity of the C-bridge and S/N ratio
E.g. for C0,min ≈ 5 fF, CR ≈ 50 pF  (Nt/Ns)min ≈ 2(C0,min/CR) ≈ 2·10-4

Limitations in the detection of trap levels:

Very shallow trap levels could not be measured due to freeze-out of free charge carriers
(wR  d = diode thickness; CR = Cd = constant)
Detection of very deep trap levels might be difficult since the change of the occupation
might be very small
Minority carrier trap levels could only be detected by forward biasing if c p >> cn ,
otherwise optical injection of minority carriers

5 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006

 DLTFFT-System in Hamburg
from PhysTech GmbH

PULSE VOLTAGE ANTI-ALIAS. TRANSIENT PROCESSOR

CONTROL AMP FILTER RECORDER
Optical Bias: +/- 20 V Rin = 1 M Bessel 8. order 32 K d.p. 18 correl.
Injection HV: +/- 100 V gain: 1-128 1 Hz to 100 64xoversampl. functions
tp min: 1 µs kHz 12 bit FFT processing
resolution C compensation
FAST PULSE
Bias:+/- 16 V
CRYOSTAT tp min: 10 ns

DUT
BOONTON 72B
Capacitance Meter

PC

T-sensor
DT-470 SD

LakeShore 340
Temperature
Controller

6 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006

 Digital signal processing
using 18 correlator functions

b1 a1

b1(Tw/2) a1(Tw/2)

b1(TW/4) a1(Tw/4)

b1(Tw/8) a1(Tw/16)

b1(Tw/16) a1H

b1(Tw/32) a1M

b1H a2N

b1M DR
1 1
b2N DR (Tw/4)
-1 -1
0 Time TW 0 Time TW

7 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006

 DLTS spectra and maximum analysis

0.3
b1 b1 (TW/2)
b1 M 58
b1 (TW/4)
DLTS - signal [pF]

0.2 b1 (TW/8)

ln(e vth,n NC)

b1 (TW/16) 57

b1H b1 (TW/32)
0.1
56

DR

0 55
180 200 220 240 4.4 4.6 4.8 5
temperature [K] 1000/T [K-1]

DLTS spectra obtained with different correlators (left).

Arrhenius plot (right) contains data obtained with all 18 correlators.
( transition V(-/0), 60Co irrad. 10 Mrad; VR=-10 V, Vp=0 V, Tw=200 ms, t0=6 ms)

8 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006

 DLTFFT spectra and direct analysis

0.3
Direct evaluation:
b1 The emission time constant
58
can be evaluated from the
DLTS - signal [pF]

0.2 correlator signals an(T), bn(T)

ln(e vth,n NC)

b2 at all temperatures where the
a1 signals are above
57 a given threshold
0.1 E.g.
a2
e (a1,b1) 1 bn
e (a2,b2)  e an , bn   
n a n
0 56
180 200 220 240 4.6 4.8 5
-1
temperature [K] 1000/T [K ]

DLTS spectra obtained with sine and cosine correlators a 1, b1, a2 and b2.
Same measurement as shown before. Arrhenius plot contains data obtained
from the direct evaluation method.

9 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006

 Variable time window method
Time window Tw is changed with temperature T
The ratio e/Tw is kept constant (≈ 0.2)  optimal signal
e values extracted from an and bn for first T-steps
Program produces Arrhenius plot  allows calculation of optimal Tw for
the next temperature
Restriction: transients have to be exponential (estimated by the program)
Advantage: large T-range for the Arrhenius plot  accurate defect parameters
Draw back: no DLTS spectrum is produced

64
62
ln(e vth,n NC)

60
58
56
54
52
4 4.5 5 5.5 6
1000/T K-1]

Example: Arrhenius plot for same defect as shown before. T-range: 170 – 245 K

10 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006

 Comparison of the three methods
 Defect parameters obtained for the 3 different methods for V (-/0):

Type of evaluation Ea [eV] n [cm2]

Maxima with 3 Tw 0.427 2.610-15

Direct with 3 Tw 0.420 1.710-15

Variable time window 0.424 2.210-15
Tw = 40, 200, 2000 ms

Recalculation of time constants for different temperatures:

Temperature 180 K 200 K 230 K

e - maxima evaluation 3.05 s 157 ms 4.70 ms

e - direct evaluation 2.97 s 160 ms 5.05 ms

e - variable Tw 2.97 s 156 ms 4.77 ms

11 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006

 High resolution method
Principle of operation:
Period scan at constant temperature: Transients measured as function of T w near the DLTS peak max.
Calculated correlator coeff. an(Tw), bn(Tw) are normalized and transformed  a’n(), b’n()
Refolding of a’n(), b’n()  distribution function f() (Gaussian-like) of the involved trap levels

1 1

0.8 0.8

norm. b1 [pF]
0.6 0.6

0.4 0.4

0.2 0.2

0 0
0.001 0.01 0.1 1 10 100 1000
/TW
Simulated DLTS spectrum for two levels with
similar properties: E1 = 0.410 eV,  = 1·10-15 cm2 Normalized b1 coefficient versus /Tw with constant t0/Tw = 0.25 (t0:
N1 = 4·1010 cm-3; E2 = 0.400 eV,  = 2·10-15 cm2 delay time after fill pulse, Tw: time window for recorded transient)
N2 = 6·1010 cm-3

12 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006

 Refolded spectra

Refolding of the normalized coefficient a’1() with order N1= 40 (left) or N2 = 60 (right)
[N is related to the width of the distribution function of the  values (Gauss-like)]
Squares: transformed data points of measured transients during Tw scan
Solid lines: refolded function f() for two levels for different order N
Vertical lines: indicate the peak maxima

Results from simulations at different temperatures:

For both trap levels the parameters Ea, n and Nt are well reproduced
Ea/Ea  0.5 %, n/n  10 %, Nt/Nt  1 %
Limitations: Time constants of transitions should differ by a factor of > 2 and the
ratio of the concentrations should be  0.1

13 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006

 Summary
C-DLTS is a very powerful tool for:
Evaluation of defect parameters (majority and minority carrier traps):
- Activation energy Ea and capture cross sections n,p
- Accuracy of Ea and n,p depends on S/N of transients, accuracy of T
measurement, extent of temperature range and evaluation method
- Direct measurement of  via variation of filling pulse duration, with fast pulse
option  ≈ 10-12 cm2 detectable (e.g. for TD)
- Separation of closely spaced trap levels possible by Laplace- or High Resolution-
DLTS (limited by minimal  difference and ratio of trap concentrations)

Evaluation of trap concentrations Nt:

- Nt/Ns  C/CR  sets lower and upper limit for detectable Nt,
- (Nt/Ns)min  10-4, (Nt/Ns)max  0.1 (C « CR), for higher values up to 0.4 CC-DLTS
- Accurate Nt evaluation needs  correction
- Nt depth profiles could be measured by variation of fill pulse and reverse bias

14 E. Fretwurst, University of Hamburg Workshop on Defects, Hamburg, August 23-2006