Photocapacitive effects at silicon-collodion-gold contacts

Photocapacitive effects at silicon-collodion-gold contacts

Solid-State Electronics Pergamon Press 1968. Vol. 11, pp. 301-304. PHOTOCAPACITIVE EFFECTS AT Printed in Great Britain SILICON-COLLODION- GOL...

339KB Sizes 0 Downloads 14 Views

Solid-State

Electronics

Pergamon

Press 1968. Vol. 11, pp. 301-304.

PHOTOCAPACITIVE

EFFECTS

AT

Printed in Great Britain

SILICON-COLLODION-

GOLD

CONTACTS

S. LEE

H. K. HENISCH

and

Materials Research Laboratory, The Pennsylvania State University, University Park, Pennsylvania, U.S.A. (Received

22 June 1967; in revised form 20 October 1967)

Abstract-The

paper describes the photocapacitive response of silicon-collodion-gold contacts in the visible and near-infrared regions. The principal light effects are shown to be associated with the lateral diffusion of light-generated minority carriers under the gold electrode. They are therefore essentially edge effects. Factors governing spectral response and sensitivity are discussed. R&sum&-Cet article decrit la reponse de la photocapacite des contacts silicium-collodion-or dam les regions visibles et infra-rouges. Les effets principaux de lumiere sont dimontrb comme &ant associes a la diffusion latdrale des porteurs minoritaires sous l’electrode en or. 11s sont done essentiellement des effets de bords. Les facteurs gouvernant la ‘reponse spectrale et la sensitivite sont discutb. Zusammenfassung-Die Arbeit beschreibt die Kaparitatsanderung von Silizium-KollodiumGold-Kontakten unter Lichteinwirkung aus dem sichtbaren und dem nahen ultraroten Bereich. Die beobachteten Effekte sind bedingt durch seitliches Diffundieren der durch Lichtstrahlung erzeugten MinoritBtstrager unter der Goldelektrode. Es handelt sich deshalb wesentlich urn Randeffekte. Die Ursachen fiir die spektrale Empfindlichkeit werden diskutiert.

INTRODUCTION

The present light-sensitive capacitors are thus quite different from those described by SIHVONEN et uZ.(~) which rely on the series connection of a fixed capacitor and a photoconductor. It has been shown that photocapacitive light detection can be much more sensitive than photoconductive detection with the same material. Ratios of (AC/C) : (AC/D) as high as 10 : 1 have been observed on the existing systems. Two photocapacitive effects have been noted; a small effect (capacitance decrease) which depends on the area of the gold electrode and a large effect in the opposite direction which depends on the electrode edges. The results are interpreted on the basis of a model which envisages an influx of minority carriers from the brightly illuminated surroundings into the region under the semitransparent test electrode.

VOLTAGE sensitive

capacitors based on space charge layers in series with thin insulating layers have been investigated by MoLL,(~) PFANN and GARRETT,@) TERMAN, FRANKL(~) and others, and have been used for a variety of purposes, such as parametric amplifiers and harmonic generators. Moll observed that illumination of a siliconsilicon oxide layer results in an increase of surface capacitance. The present investigation is concerned specifically with capacitance changes resulting from the illumination(photocapacitance) of silicon-collodion-gold systems. Collodion was chosen as the insulating layer because of its ease of preparation, lack of pinholes, high dielectric strength and transparency. The photocapacitance arises from the minority carrier collection properties of the inversion layer which is part of the barrier region in the semiconductor adjacent to the insulator. The thickness of this barrier can be modulated by the creation of electron-hole pairs in the vicinity.

EXPERIMENTAL PROCEDURE Specimens were prepared from 1300 Q-cm 301

302

S. LEE and H. K. HENISCH

n-type silicon wafers (lifetime 400psec) about l-mm thick, lapped and polished and Ni plated. Three Pt wire electrodes were then applied to one side as shown in Fig. l(a), using Sn as solder. The central electrode served as an ohmic connection to the capacitor; two outside electrodes were used for a rough assessment of photoconductive effects in the specimen. Once applied in this way, the electrodes were given a protective coating of polystyrene which enabled the Ni to be removed from the rest of the specimen by means of an HNO, etch. The free surface was then polished with 0.5 p grit, etched (10 HNO, : 1HF following CP4-A) and dried in air. A thin layer of collodion was applied by allowing a solution in amylacetate to dry. Film thicknesses of about 0.3 p were used without incurring dielectric breakdown during the experiments. A semitransparent film of gold was evaporated on to the collodion and contact to this film was made by means of a gold ‘fuzz ball’ which consists of very fine gold wire. Figure l(b) shows the experimental arrangement. Light chopped at 1 kc/s and with a 20 percent duty cycle was incident on the specimen. The capacitance was measured at 1 MC/S by means of a Boonton 71A instrument with CRT balancing read-out, using a 15 mV alternating signal. Dissipation factors were determined by means of a WayneKerr B201 bridge. At lower frequencies a GR716C instrument was used.

Typical

EXPERIMENTAL RESULTS light effects as a function of contact

r

,OOt IL a g

600

z 5 b 2 : u

400

2001’ IO

FIG.

frequency

r

gold film

9

Pt wire \O)

L collodlon

layer

FIG. 1. Experimental arrangements. (a) Structure specimen. (b) Measurement enclosure (schematic).

of

bias are shown in Fig. 2 for two incident light intensities. Results of this kind are obtained when the whole electrode area and its edge are illuminated. When only the interior of the semitransparent electrode is illuminated, AC is always negative and the capacitance changes are too small to be apparent on the same scale, even when due allowance is made for the amount of light absorbed by the electrode itself. This suggests that the large and positive capacitance change is essentially an edge effect, a conclusion which is further supported by the results in Fig. 3. These refer to strip and circular electrodes of different sizes. Here and elsewhere,

: I me/s

.

dork capacitance

q

310 p w/cm’

0 2mw/cm2

-4 -2 -8 -6 BIAS OF GOLD ELECTRODE RELATIVE (VOLTS)

2. Photocapacitive response of collodion-gold system as a function

0 TO BASE

a typical siliconof bias voltage.

2

PHOTOCAPACITIVE (al

AT

303

CONTACTS

slr~p electrodes

I

1 I

I

I

I

ormalized electrode

1!

,

-lb

circular

I

0.8

dark

s: :

area

I

-6

OF

I

j;,q,jn t”,”j

,

-8

B!dS

(b)

EFFECTS

GOLD

-4

,

ELECTRODE

-8

0.4

-2

0

BIAS

-6

VOLTAGE

-4

OF GOLD

-2

ELECTRODE

0 (VOLTS

1

(VOLTS)

electrodes 0

450

0

2mw/cm2

pw/cm2

(bl freq:

0

I

I

-16 BIAS

1 -12

OF

GOLD

I

I

I

-8 ELECTRODE

-LO.8

mm dlo.

a 2 mm

I

I

dia.

-4

c 5

I

I me/s

E

-12

-10 BIAS

-6

VOLTAGE

-6 OF GOLD

-4 ELECTRODE

-2

0 (VOLTS1

0 (VOLTS)

FIG. 3. Photocapacitive response of a typical siliconcollodion-gold system as a function of electrode shape and size.

the capacitance changes are referred to the maximum capacitance (C,,,) measured under darkness. In general terms and within the dimensional limits so far explored, the photocapacitive light sensitivity continues to increase as the electrode size diminishes. Maximum light sensitivity is obtained for a definite d.c. bias voltage, as shown by Fig. 4(a). Between 5 kc/s and 1 MC/S, these results are not significantly frequency sensitive. Figure 4(b) gives the dissipation factor as a function of light intensity, and the low tan 6 values show that the measurements are not affected by superimposed photoconductive effects to any significant extent. A photoconductive change between the two collodion-free base electrodes can indeed be detected but it is always less than l/lOth of the magnitude of the relative capacitance change. The spectral response of a typical photocapacitive system is shown in Fig. 5 in comparison with

FIG. 4. Typical photocapacitive performance data. (a) Response as a function of bias and frequency. (b) Dissipation factor as a function of bias and light intensity.

the photovoltaic response of a silicon surfacebarrier without insulating layer. The collodion layer transmits more than 95 percent of the light over the entire range. DISCUSSION

AND

CONCLUSIONS

On the basis of a simplified model discussed by GROVE and coworkers,@) one would expect the illumination of a contact inversion layer to result in a small and positive change of space charge thickness. This corresponds to a small diminution of barrier capacitance and (in the case of n-type bulk material) involves also a diminution in the number of excess positive holes in the inversion layer. The barrier becomes slightly flattened. This is, indeed, the kind of change observed when only the interior of the electrode is illuminated. When light is also incident on the surrounding areas, lateral concentration gradients are set up. As long as the light intensity is small, the situation is not essentially changed, but when it is large, a stage is

S. LEE and H. K. HENISCH

304

I

I

I

I

I

x PERCENT

I

0

28

optical

absorption

24

ct system; Ahlstrom

20

- 6 volts

A

-8volts

0

-9volts

IO4

photovoltoge. and Gtirtner11962

F

i 103 =

16

z

12

;: tm tocopactive

system

102

0

% -

IO’ 0.60

0.70

0.00

0.90

WAVELENGTH

FIG. 5. Comparison

1.00

I. IO

(~1

of spectral characteristics.

reached when the concentration of non-equilibrium holes outside the contact area is actually greater than the concentration of holes in the inversion region. Holes then diffuse into the area under the gold electrode. The result is a sharpening of the barrier and a corresponding diminution of barrier width which shows itself as a substantial capacitance increase. In their present form, these systems have been found capable of detecting light intensities of 10e5 W/cm2 at a wavelength of 0.9 p. This compares favorably with the sensitivity of conventional silicon surface-barrier photocells.“) Moreover, the present photocapacitive systems have not yet been optimized from a device point of view and it is reasonable to expect that their performance could be substantially improved, e.g. by using a thinner dielectric, an anti-reflection coating on the silicon and possibly a silicon substrate of higher carrier lifetime. To a first order of approximation, it must be supposed that the additional holes diffuse under the gold electrode to the extent of one diffusion length. The optimum electrode configuration would thus be a matrix of square dots, one diffusion length apart and of the same edge length. For the silicon used, this amounts to about 1 mm. The electrodes need not, in fact, be transparent.

As may be seen from Fig. 5, the spectral response in the near infrared is better than that of conventional silicon detectors, though the detailed reasons for the peak below 1 p are not well understood. Its dependence on bias suggests that some of the absorption involves electrons in surface states which are filled to an extent which is bias dependent. The matter is in need of further investigation. Results which are similar in principle to those reported above have also been obtained with germanium. Acknowledgements-The work here described was supported by contract DA-44009-AMC-581(T) of the U.S. Army Engineering Research and Development Laboratories, Fort Belvoir, Virginia. Thanks are due to Professor D. R. FRANKL for much useful advice. REFERENCES 1. J. 2. 3. 4. 5. 6. 7.

L. MOLL, IRE WESCON Conv. Rec. 3, 32 (1959). W. G. PFANN and C. G. B. GARRETT, Proc. Inst. Radio Engrs. 47,201l (1961). L. M. TERMAN, Solid-St. Electron. 5, 285 (1962). D. R. FRANKL, Solid-St. Electron, 2, 71 (1961). Y. T. SIHVONEN, D. R. BOYD and E. L. KITS JR., Proc. IEEE 53, 378 (1965). A. S. GROVE, E. H. SNOW, B. E. DEAL and C. ‘I‘. SAH, J. appl. Phys. 35,2458 (1964). E. AHLSTROMand W. W. GARTNER, J. appl. Pkys. 33, 2602 (1962).