Introduction to DLTS(Deep Level Transient Spectroscopy)

III. Our DLTS System

O. BreitensteinMPI MSP Halle

Outline:

1. Basic principles

• Application field of DLTS• Principles of DLTS• Basic measurement techniques

2. Advanced techniques

• Advanced DLTS measurement techniques

3. Our DLTS system• - Philosophy• - Hardware• - User surface

Recapitulation: DLTS routine (repeating!) :

t

Vr

e-

e-

e-

e-

e-

t

C

0

0

bias

banddiagram

RF-capacitance

reverse reducedor forward

reverse

Generation of the DLTS signal

t1 t2t t1 t2 t

Cmeas

t1 t2t

T

Tpeak

Cpeak

DLTS signal = C(t1)-C(t2)

low T opt. T high T

If T is slowly varying, at a certain temperature a DLTS peak occures

12

1

2

1peakp;n tt

t

tln

se

"rate window":

DLTS measurements at different rate windows allow one to measure Et

e01

e02

e03

T1 T2 T3T

ln(en)

1000/TT1T2T3

e01

e02

e03

DLTS kT2

T

1000n*198

meVE t

This "Arrhenius plot" allows an identification of a deep level defect

Advanced techniques

• DLTS on Schottky diodes only reveals majority carrier taps• DLTS on pn junctions also reveals minority carrier traps• Optically excited DLRS (MCDLTS) also reveals minority carrier traps in Schottky diodes• There are special DLTS procedures for measuring:- concentration depth profiles (Vimp scan)- electric field dependence of en;p (Vimp scan)- capture cross sections for electrons and holes (timp scan)• Extended defects are usually characterized by a logarithmic capture behavior and often show non-exponential emission (broadened DLTS peaks)

Philosophy of our DLTS system

1. We don’t save DLTS data but transient data

• Conventional approach: On-line conversion of transient data to DLTS data, saving DLTS(T) (1-dimensional vector of data).Advantage: Small file sizes. Disadvantage: No flexibility with respect to different correlation techniques (see below)

• Our approach: C(t, T) is saved as a 2-dimensional data fileAdvantage: Flexible DLTS correlation. Disadvantage: Larger file sizes (see below).

2. We have both linear and logarithmic time scale at choice

• Linear time scale: time resolution for large times is the same as immediately after the filling pulse. May be advantageous for software-based multiexponential transient deconvolution

tmin0 tn

tn = n tmin

• Logarithmic time scale (base 2, also 1.1 possible): Time resolution proportional to elapsed time, drastic savings in file size

averaging over differently sized periods!

tmin0

tn = (2n-1) tmin

tn

3. We have three different kinds of DLTS correlation at choice

3.1. Modified 2-Point correlation: DLTS = C(t1)-C(t2)

Mathematical formulation:

3

1

3

2

2

1

t

t

t

t23

t

t12dt)t(K)t(Cdt)t(C

tt

1dt)t(C

tt

1DLTS

K(t)

good compromise between resolution and sensitivity, many rate windows

t1 t2 t3

C(t)

3.2. Exponential correlation: High sensitivity, but less resolution

K(t)

t

DLTS

T

C(0) 2-point

3.3. High resolution correlation: Low sensitivity

K(t)

t

DLTS

T

C(0) 2-point

Hardware

• C-Meter working at 1 MHz, made at our electronic workshop• applied HF signal: 100 mV (pk-pk) or 1 V (pk-pk) at choice• electronic C- and G- (conductivity) compensation• manual or automatic compensation• sample bias range: 0 ... 15 V• pulse bias range: 0 ... 15 V. If pulse bias > bias: injection !• preamplifier separated, connected with main unit by 1 m cable• computer controlled via 2x16 bit ADC / DAC interface card

•T control unit, controlled via RS232 by computer• linear or exponential T-ramp at choice, speed adjustable

C Meter and T controller

DLTS system wiring scheme

bias pulse bias Delta C C-compens computer preamp.ext. ext. out out

rear side

C-meter

front side

delta C bias out

Computer

preamplifier

+ Probe

- Probe

Probeextern

INPUT CH.1 INPUT CH.2 EXT.TRIG.

Oscilloscope

Cryostat

T-controller

RS232

"Trig."

ADC 0 ADC 2 DAC 0 ADC 2

sample

2 different cryostatsat choice:

1. Bath cryostat• only for samples mounted on TO5 transistor holders• manually immersing in liquid nitrogen (cool down), measurement after lifting above LN2 level, quick measurement• not optimum for very slow T-ramps or constant T measurements

2. Evaporator cryostat• for samples mounted on TO5 or TO18 holders or bare samples• fully automatic cooling down and heating up (software controlled)• slower measurement, larger LN2 consumption

Software Made by MSC Technik Halle (http://www.msc-technik.de/)

“settings” menu:

“display” menu

What this system can do:

• DLTS measurements from 78 K to 400 K• sample capacitance < 500 pF• sample parallel resistance > 500 • bias and pulse bias range: 0 ... 15 V• samples mounted on transistor holders or as raw chips• linear and logarithmic sampling (to base 2 or 1.1)• rate window range from < 1 s-1 to 104 s-1

• monitoring and storage of C0(T) (basic capacitance)• sensitivity < 10-4pF for 0.1V HF (pk-pk), < 10-5pF for 1 V HF• “batch” measurements for bias, pulse bias, tmin, and timp

• display of up to 10 DLTS traces• export of C transients, C0(T) and DLTS traces as ASCII files• system is available in room B.2.05, to be used only after personal introduction by O.B. !

Plans for the future:

• Establishment of Minority Carrier DLTS (optical excitation)• DLTS peak evaluation software (parameter fitting etc.)

3 ppt Files of this introduction and the DLTS operation manual are available on-line