Remote contact LBIC imaging of defects in semiconductors

1 70

Journal of (‘r~staI (1 rowt h 103 (1 990) 1 70— 1 75 North—Il oil and

REMOTE CONTACT LBIC IMAGING OF DEFECTS IN SEMICONDUCTORS

J. BAJAJ and W.E. TENNANT Rockwell International Science Center. Thousand Oaks, California 913(d), (ISA

A high resolution and nondestructive optical technique, called “laser-beam-induced current (LBIC)”. has been developed for spatial imaging of electrically active defects in liquid phase epitaxial (LPE) HgCdTe alloy semiconductor. The technique consists of mapping the induced current between two remote contacts on the sample as a function of the incident-focused laser beam position. The induced current is a result of the charge separating effect of built-in fields in the vicinity of defects in semiconductors. The by, laser power does not damage the sample, and the resolution of the technique is limited by the diffusion length of the carriers in the material. Also, since device structures such as p—n junctions are special eases of electricalls’ active regions. LBIC imaging has been utilized to study the opto-eleetronic properties of these structures in a nondestructive manner, without requiring ans electrical contact to the active elements. In addition. LBIC has been utilized to obtain electrical nonuniforn’iities at the HgCdTe semiconductor surface near its interface with a ZnS passivation layer.

1. Introduction

2. Description of LBIC technique

Spatial inhomogeneities in a semiconductor starting material severely affect the uniformity in performance of any opto-electronic devices fabricated on that material. As a specific example. indtvidual elements of a HgCdTe-hased infrared focal plane array can vary significantly in their performance. This variation is most likely caused by spatial inhomogeneities in the starting material itself. Therefore, it is important to develop techniques that yield spatially-resolved information on defects and impurities in semiconductors. We have developed a high resolution technique called. “laser-beam-induced current (LBIC) [1,21, which maps those regions in the semiconductor that are electrically active. This technique is nondestruc tive in that this evaluation does not alter the material which can therefore be passed on for further device processing. Various scanning techniques described in literature focus on a specific aspect of defects. Our contention is that if the concern is electro-optical devices, electrically active defects should have the maximum effect. 0022-0248/90/$03.50

1990

—

The LBIC technique is applied in a manner shown schematically in fig. 1. Two electrical contacts are made at two opposite remote ends of the sample. A focused laser beam is then scanned spatially across sampleisand the induced current between thethe contacts measured as a function of the incident light beam position. For measurement at each point, the focused laser generates electron—hole pairs in the localized area. If this localized region has electrical fields associated with

‘~

INCIDENT, BEAM

H

~

AMPLIFIER TACTIVE LAYER “

~U1PUT

-______________________

SUBSTRATE

~CHARGE SEPARATING DEFECT in IN P. P IN n, n IN n, p IN STRAIN, Eg VARIATION, OTHER INCIDENT BEAMS, SURFACE POTENTIAL VARIATION, ETC I

STEPS IN DETECTION PROCESS: IMPINGES NEAR CHARGESEPARATING REGION ,,~) BEAM CHARGES INDUCED BY INCIDENT REAM ARE SEPARATED TO RECOMBINE WITH EMITTED CHARGE OF OPPOSITE TYPE, CHARGES MUST FLOW AWAY FROM POINT OF INCIDENCE SOME FLOW THROUGH EXTERNAL CIRCUIT, GENERATING THE SIGNAL

Fig. 1. Schematic representation of LBIC detection.

Elsevier Science Publishers By. (North-Holland)

J. Bajaj, W.E. Tennant

/ Remote contact LBIC imaging of defects in semiconductor.c

it caused by a junction or an electrically active defect, the charges will be separated. This separation produces a redistribution of charge over the surface of the region causing a differential current (LBIC) to flow between the two remote contacts. The direction of current flow will depend on the position of the laser beam relative to the local field configurations and the measuring contacts. As a result, LBIC signal shows a characteristic bi-modal behavior as the laser beam is scanned across an electrically active region. Because most defects in semiconductors create some local fields in their vicinity, LBIC reveals the detailed electrical structure of the material; a spatial scan gives a map of electrically active defects. A p—n junction is a special case of an electrically active region in a semiconductor and is a simple structure whose charge separating effects are well understood. For illustrative purposes, therefore, LBIC signal from a p—n junction is discussed below. This discussion highlights the unique characteristics of the technique, and also contrasts it with standard photocurrent scans of a diode. Fig. 2 demonstrates the differences in sample configuration for the normal photovoltaic response, herein referred to as optical-beam-induced current (OBIC), and the LBIC response. In a standard photovoltaic response measurement, contacts are made on each side of the junction

INCIDENT BEAM

~

-

7:

‘A,C

____________

I

1

D

\.,,~

PHOTOVOLTAIC

RESPONSE SCAN I

‘A B :

LBIC

x

SCAN

INCIDENT BEAM POSITION —.

Fig. 2. Optoelectronic response of a p—n junction contrasting photovoltaic response and LBIC.

171

(contacts A and C) in fig. 2a and the current is measured between these contacts as a function of the position of the laser beam. A line scan of this current, as the beam is scanned across the p—n junction, is shown in fig. 2b. By contrast, for LBIC measurement, both the contacts are made on the p-side (contacts A and B) and a line scan of this current, as the beam is scanned across the p—n junction, shows the characteristic bipolar behavior depicted in fig. 2c. The bipolar signal is a result of no bias being applied between the contacts; therefore, there is no preferred direction of current flow. If the carriers are generated in the center of the p—n junction, equal currents will flow in both directions, resulting in zero net current. As the beam is scanned away from the center in both directions, the net current, though equal in magnitude, flows in opposite directions. This bipolar response is an especially distinctive and advantageous feature of LBIC in contrast to conventional photovoltaic response where contact made directly to the device results in a unipolar response. If there was an array of p—n junctions in the sample, the photovoltaic response measurement would require a contact to each element of the array. In contrast, the LBIC measurement requires only two contacts at two remote ends to scan all the individual elements of the array. Therefore, LBIC offers a unique way of studying the optoelectronic behavior of each element in the array without requiring contacts to individual elements of the array. The currents ‘AC and ‘A,B are related to each other; the model used to obtain the relationship is discussed in fig. 3. Electron—hole pairs are generated at the illumination spot and are separated by the built-in field of the junction. This results in a steady-state condition in which most of the in.

.

+

jected electrons are found in the n region, most of the injected holes in the p region. If the conductivity of n + region is much larger than in the p region, as is the case in HgCdTe n’~/pphotodiodes, the electrons will redistribute themselves uniformly over the n~region. This leads to reinjection of electrons back into the p region as shown in the figure. Therefore, a field is set up between the majority carriers (holes) near the ii-

J. Bajaj, 141 L’. Tennant ,/ Remote contact LB/C imaging of defects in .seoiii’o,,ducrors

172

,~L

A

greater than that of p region, we assume the diode dimensions are larger than the diffusion length of

B

_____________________________________

~

REINJECTED

-

ELECTR 1S PRIMARY DIRECTION OF CURRENT FLOW

I

.~ ,

.

“

____________

PRIMARY DIRECTION OF CURRENT FLOW

SIDE VIEW

P

—

ILLUMINATION SPOT

\

‘‘

minority carriers and the voltage drop across the p region is much less than k T/q or the drop across . . the Junction. In the region of the illuminated spot. a surface charge appears due to electrons flowing into the n~ region. This is matched by an opposite

~2

surface charge over the diode area due to electrons re-emitted over this surface to maintain steady

VIEW

state. The two surface charges may he treated as line charges perpendicular to the plane of the sample and the problem can he analyzed by ele-

~

AZi~

Fig. 3. Computation of LBIC signal from an n-On-p

Linction.

mentary electrostatics. In this limit, a line dipole P lumination spot and the reinjected electrons which

is created of magnitude

are spatially separated from the holes. This lateral field gives rise to the LBIC signal. In the figure. the illumination spot is to the left of the center of the diode. When the illumination spot is on the right side of the center, lateral field will he set up in a direction opposite to that shown in the figure.

p

This results in bipolar LBIC signal. If the il

lumination spot is at the center, equal number of electrons are reinjected on either side of the center andTotheestimate fields cancel to yield zero netLBIC field. signal. the magnitude of the we make several assumptions. In addition to assuming that the conductivity of n~region is much

~

(RX

where x is the separation vector between the center of the p—n junction and the center of the illumination spot. R is the resistance of the sample and is the photocurrent discussed in fig. 2. The LBIC current between the two contacts (A and B) /,\(.

is therefore 1t BR’ ~.it

~

=

=

cos eR, 0,

+

=

P cR. cos

-

-

v

=

“c

250

0~ + cos 02 R~ R 2

LOS

‘ ,

‘T~

~—---

_________

240 YAG LASER EXCITATION 220

20~ I

I

I

I

I

~‘T”T”T””~~T~7’

a

YAG LASER EXCITATION

01

02

0.3 0.4

05

56 R1

200’

~

180

~

16O~— 1401-

<

Il~~IIII~ 0

— IS I—

07 62’

08

1)

1.0

1.1

1.2

1,3 14

‘‘

-~

40’~

I:J

0.9

H

1,5

0/’ 0

10

20 LASER POWER 30 5W)

40

50

Fig. 4. LIIIC signal from a p—n Junction as a function of (a) contact separation and (h) excitation laser power. The solid straighi line in (a) IS drawn for visual guidance only.

J. Bajaj, W.E. Tennant

/ Remote contact LBIC imaging of defects in semiconductors

where VAB is the potential drop between contacts A and B, R~ and R2 are the distances between the center of diode and the two contacts A and B, and 0, and °2 are the angles between dipole direction and the lines joining the illumination spot to the respective contacts. The above expression predicts the dependence of LBIC signal on excitation intensity (‘A,c is proportional to laser intensity) and the separation distance between contacts. We verified these dependencies by designing a special mask which allowed fabrication on one single diode in the center of a sample and several sets of contact pads (with varying separations) around it. A line scan was taken across the diode and the resulting LBIC signal was measured between different combinations of contacts, This generated LBIC signal as a function of 1/R1 + l/R2 (the sum of the inverse of distance between the center of diode and each contact) is shown in fig. 4a. Also, for a fixed set of contacts around the diode, the signal was measured as a function of laser power; the observed variation is shown in fig. 4b. These observed dependencies are consistent with the expression derived above.

3. Experiments We have used a commercial nonconfocal Scanning Laser Microscope, Model 1000 from Waterloo Scientific, Inc., to perform LBIC experiments on a variety of Hg1 ECdXTe epitaxial layers. A schematic of this microscope is shown in fig. 5. Two translation stages with a scanning resolution of up -

MIRROR

j~”~

~UTRAL.DENSIT0

________ACOUSTOOPTIC

FILTER WHEEL

MODULATOR

LJ I

LASER

I TAG 1 OBp...

I__________________ BEAM

___________

EXPANDER

BUAN~

RE~~ED DETECTOR

SPLITI

J

HOSTCOMPUTER

I

IMAG~IiPLAV

FOCUSING STAGE CONTROL _____ N.D.F CONTROL

SPUTTER

CAMERA

II

STAGE

I

LASER CONTRO~J

LCSS AMPLIFIER

___________________________ nv STAGP

Fig. 5. Schematic of a scanning laser microscope.

I I

173

to 0.5 ~tm provide the scanning capability. The sample is placed in a cryogenic dewar that is fastened to the translation stage. The microscope can house up to three lasers, any one of which can be selected remotely for experiments. For most of the LBIC data presented here, we have used a diode pumped Nd : YLF laser (1.047 ~.im)focuses to approximately 3 ~tm diameter spot. This laser can operate in a cw mode with more than 100 mW at the sample or in Q-switched mode to give 25 nS pulses with a peak power of up to 20 W at the sample and a maximum repetition rate of 10 kHz. To obtain a LBIC image, induced current is measured at each point of a raster scan. This digitized current value is displayed as it is acquired and is simultaneously stored in a file for subsequent processing. The data is displayed in real time as a false color plot obtained by mapping each data point onto a color palette. Apart from monitoring the electrical signal as a function of position, another channel monitors the reflected light and a reflected light image is displayed, in false color plot, simultaneously with the LBIC image. The reflected light image is just an optical image that highlights features at the surface. The microscope is equipped with a powerful image processing software for analyzing the LBIC images. We have applied the LBIC technique to study: (1) Spatial distribution of electrically active defects in LPE HgCdTe semiconductor material; (2) Uniformity in optoelectronic behavior of p—n junction arrays fabricated on LPE HgCdTe; and (3) HgCdTe surface near its interface with ZnS passivation by utilizing MIS structures. Discussed below are a few examples from each of these three applications. Fig. 6 shows the room temperature reflected light and LBIC images of LPE Hg07Cd03Te grown on lattice mismatched CdTe/sapphire substrate. The images were re. . corded simultaneously with a step size of 10 ~.Lmin both x and y directions and a laser power of approximately 50 fiW. The reflected light image shows the morphology the sample few features arising of from visible surface defects and anda damage at the surface. The LBIC image is very rich in structure and consists of two kinds of features a uniform structure and superimposed on it several large scattered features. Some of —

J. Bajaj, 1+1 E. Tennant / Remote contact I.BI(” imagIng

174

T

=

of defects in I’eohiconductors

300K

REFLECTED LIGHT

LBIC

61”

6

,~., ~

I

51~

~‘

4’

5 I

it

1~B;~

~

~

~: ~

iz

4, E E

E E

id ‘,

1

1

0

0 I

1

2

I

I

3

4

~#

—,

A”

?

,

~

~ ~

~

~

~ ~

—

~

~‘IE

. .

—9

B~ •.“Ai

~ ~.v: ~,

.

. ,,

~_

~fl ~

J,

4~,

~

(~B

.y,,

‘4. $~

,‘..

•

~

•~‘

I I

5

6

mm Fig. 6. Reflected light and [BIC’

‘ ~‘

‘B

~

HB.

‘3

.‘ .~

~

1Hk

~

JA.

I,

‘.~

~,,.

2

0

.“‘; ‘.

~4~’;..A~’

O

2

-

-. ~

~

7

0

1

2

3

4

5

6

7

mm Images of an [P11 Hg11 ~.‘d11 1Te/CdTe:sapphire at room temperature.

these scattered features correlate with features in the reflected image. hut many of them have no indication of their presence in the reflected light image. Since the measurements are made in a steady state, the resolution is limited by the diffusion length of carriers. In the sample of fig. 6. the diffusion lengths can be of the order or larger than the approximately 10 /.sni layer thickness. Therefore, the signal in fig. 6 is coming from the entire depth of the sample. This is confirmed by turning the sample upside down and bringing the excitation beam through the substrate. This yields the same large scattered features observed with frontside illumination. The scattered features are therefore a result of electrically active defects buried inside the layer. Fig. 7 shows the details of the LBIC’ scan of fig. 6. Fig. 7 (top) is repeat of LBIC seen in fig. 6, and fig. 7 (middle) is a 2 mm scan with the origin at the same location as in fig. 6 and a step size of 10 p.m in .v and v directions. Fig. 7 (bottom) is a high resolution 40 p.m X 400 p.m scan with the same origin as in fig. 7 (middle) and with a step size of 2 p.m in .v and i’ directions. As expected.

ihe bipolar nature of the signal is seen clearly (red color corresponds to large positive signal. hlLic io large negative signal and yellow/green to near i.ero signal). A comparison of high resolution [BR.’ seen in fig. 7 (bottom) with the corresponding reflected light image shows that LBIC features are not related to the surface morphology. The LBIC’ image from an LPE HgCdTe sample grown on lattice matched CdTe substrate looks very different and is shown in fig. 8 along with the reflected light image. While there at-c the large scattered features, the uniform structure seen in ihe samples on sapphire is not preseni. ftc dislocation density in (‘dTe is more than an order of magnitude lower than that in samples grown on lattice mismatched sapphire suhsiraies. ‘Flits (IiIference and sonic other experimental ohservaiions lead us to believe that the uniform structure is related to the dislocations in the maierial or perhaps to a combination of dislocaiions and the Impurities gettered on theni. As discussed earlier, the LBI(’ iechnique lends itself to a study ot p n junction arrays in a nondestructive manner without having to make an

J. Bajaj, W. E. Tennant

/ Remote contact LB/C

300K LBIC MAPS

electrical contact

to

175

individual elements of the

array. This remote contacting is a tremendous

6

advantage in terms of the time required to test a large array. Fig. 9 shows an LBIC scan on a portion of p—n junction array fabricated on LPE Hg07Cd03Te. The contacts used to measure the

5

E

imaging of defects in semiconductors

________________________________

E 2 1 ______

0

rn 0

1

3mm ~

~I

6

~~

2

LBIC signal are located on the left and right, well outside the area scanned. The bipolar behavior in the scan is again evident. The rectangular shape of the LBIC plot is a result of the location of the electrical contacts; the line joining the contacts is parallel to two opposite sides of the diodes. The signal from each diode matched of light and dark rectangles. In is thisa figure, all set diodes behave uniformly. This is so because these measurements were made at room temperature, where all

E

0 ~

E ‘

_____________________________

devices behave uniformly. At low temperature. we observe a variety of nonuniformities which show up as abrupt changes in the brightness or color compared to the neighboring diodes. The LBIC signal from a diode can also he used to determine quantum efficiency which is a perfor-

mance parameter for diodes. This is done using the expression that relates the LBIC and photocurrent, derived earlier in reference to fig. 3. The LBIC current allows determination of the photoo o

0.5

1 0

400

200

0 200

Fig. 7. Details of the [BIC image in fig. 6.

400

current; ratio of this efficiency. current to The the power of the laser the is the quantum quantum efficiency of the diode can therefore he determined without actually making contact to the two sides of the diode. The LBIC technique has also been used to obtain spatial maps of electrical nonuniformities at the ZnS passivated Hg 07Cd01Te surface. ZnS was deposited by electron beam evaporation on freshly etched samples of HgCdTe; and Ti Au metal dots, 250 p.m in diameter, were deposited to form metal insulator semiconductor (MIS) capacitors. Two ohmic contacts were made on opposite ends of the semiconductor for LBIC measurements. MIS capacitors allow modulation of the HgCdTe surface potential resulting in accumulation. depletion or inversion, depending on the applied bias on the metal. The depleted and inverted surface conditions set up charge separating

REFLECTED LIGHT

LBIC

-

3-

~3

2

.tOs~ $11

___________

2

“~

~

Fig. S Reflected light and Hilt’ i~n.iec’ol 1117 1FF. Hg

cd,

I Ic ( LI Ic it 77 K.

T — 300K

tI~

400!Am

.

(a) ____

-- 400 gm--- -.

(b)

-~

-

LINE PROFILE

~TA\[~TT~ (c)

_::i

0

50

100 150 200 250 300 (MICRONS)

}“~g, L~, Reflected light (a) and [BIC scans )h) of a portIon of p—-n unctIon array fabricated on [FE Hg ( sl~ [c temperature. Also shown isa line scan of [BIC signal showIng the expected bipolar behavior ) c)

II

01(1111

J. Bajaj, W. E. Ten nant 5-6V

0,06 cm

/

Remote contact LB/C imaging ofdefects in semiconductors 6

III

Fig. 10. Detailed [BIC

7V

177

7—By

II

0,06 cm

0.06 cm

scan of the same MIS capacitor for different bias conditions. T= 77 K.

fields in the vicinity of the surface and are detected by LBIC measurements. The LBIC scan for

4. Summary

an ideal circular MIS capacitor should give a bipolar image that is spatially antisymmetric along

LBIC imaging is demonstrated to he a powerful tool to map electrically active defects in HgCdTe material. LBIC can also be used nondestructively

the diameter directed between the contact. The onset of depletion, which can vary from point to point, can be monitored by observing LBIC scans for a series of biases. This measurement will mdicate spatial variation in the surface condition,

to identify optoelectronic defects in p—n junction arrays, without having to make electrical contacts to active elements. This allows testing at various stages of device processing. and isolation of par-

Fig. 10 shows an example of surface uniformity, as determined by LBIC, for an MIS capacitor on p-type Hg0 7Cd0 3Te. The capacitor bias was

ticular processing steps that need optimization. In addition, we have demonstrated the use of LBIC

modulated between two different voltages, V~and

surface near its interface with a passivating layer.

V2 (separately adjustable), to separate surface effects from bulk effects. The voltage window ~ — V2) was set a I V. In fig. 10. as the voltage window is moved to higher voltage (towards depletion

We have presented a model that allows calculating the LBIC signal due to a p—n junction. This calculation is equally valid for any charge separating region provided the conductivity of one side of the Junction is much greater than that of the other

conditions), a signal first appears around a “hot spot” before the rest of the surface is depleted.

There is no indication of this behavior in the C— V characteristics which looks similar to the C— V from several other capacitors that do not show any such hot spots. The implication is that the LBIC scan is more sensitive to variations in the surface potential than the C— V characteristics. LBIC should be a valuable tool to distinguish homogeneous from inhomogeneous surface preparation and passivation methods.

to map electrical nonuniformities at HgCdTe

side and the dimensions of the region are larger

than the diffusion length of minority carriers. The derived expression gives an estimate of the sensitivity of this (LBIC) method in terms of the photocurrent. This expression is a maximum signal for a given size of electrically active defect. The magnitude of the signal will be modified by the nature and size of the specific charge separating region and on the location of contacts with respect to this region. For charge separating regions whose di-

J. Bajaj, l4’l LI Tennant

t7S

/ Remote contact

mensions are smaller than the minority carrier diffusion length, a rough estimate shows that the LBIC signal should reduce by the ratio of diffusion length to the size of the region.

/_BJ( (Imaging oJ de/ec’ts in .scflhic’ondto’tor.I

References

[II J. 121

Acknowledgement

We are grateful to Dr. Chris Moore of Waterloo Scientific. Inc.. for many fruitful discussions.

Bajaj. [0.

Bubulac. P.R

Newman. WE. Teniiarii and

P.M. Raccali. J. Vacuum Sci. Technol. AS (l9S7) 31S6. C.J.L. Moore, J. Hennessey. J. Bajaj and WE. ‘l’ennLIIlt. Phoionics Spectra (September 1988).