Thick-film Photosensors
J. N. Ross
(c) University of Southampton 1994
Contents
- Introduction
- Ink formulation
- Screen printed photoresistors
- Photoconductive position sensors
- Conclusions
- References
The thick-film techniques used in the fabrication of hybrid electronic circuits may be readily
adapted to the production of low cost sensors for physical variables [1]. Temperature sensors
may be fabricated using commercial thermistor pastes while, strain gauges using standard
thick-film resistor materials on a variety of substrate materials have been thoroughly
investigated and shown to be practical and versatile [2], [3]. Thick-film materials developed
specifically for sensor applications have included magneto-resistive and piezoelectric
inks [1], [4]. In all these cases the printing and processing techniques are compatible with
conventional thick film hybrid circuit fabrication techniques, which provides a convenient and
flexible technology for manufacture and enables the sensors to be integrated with hybrid
electronic circuits.
It is well known that cadmium sulphide can be screen printed and sintered to form films that
display photosensitivity. Such sintered films have been used to produce photoconductive
sensors [5] and more recently for photovoltaic solar cells [6]. In this paper photoconductive
sensors based on screen printed cadmium sulphide and cadmium selenide thick films printed
over standard silver-palladium conductors are described. The process is such that the sensor
could be readily integrated into a hybrid circuit. Simple photoconductive arrays and a
potentiometric position sensor have been fabricated.
Inks have been made from CdS, CdSe and mixtures of these two materials. To produce an
ink finely powdered CdS and, or CdSe is mixed with a small quantity of
CdCl
and then fired
in a covered ceramic boat at a temperature of about 600C. Firing is done in a continuously
moving belt furnace with a total firing time of about 30 minutes, and only a few minutes at the
peak temperature. This initial sintering produces a soft ``ingot'' which is then ground down to a
powder and mixed with a suitable organic vehicle ready for screen printing. The ink so formed
may be screen printed in the usual way and fired in air at a temperature between 560 and
600C.
The amount of CdCl that is required to ensure that the film formed after printing and firing
shows reasonable adhesion varies from about 10% by weight for CdS to 3% by weight for
CdSe. For inks made with mixtures of CdS and CdSe intermediate percentages of
CdCl have
been used. The CdCl is a donor in CdS or CdSe and so effects the conductivity of the film,
but is also volatile at temperatures above 450C and so the conductivity will be modified by the
firing temperature and time. For films printed over silver palladium conductors it is found that
excessive amounts of CdCl will corrode the conductors under the film, but if the percentage
of CdCl is just sufficient to produce a sound film, then corrosion is not a problem.
Photoresistors have been produced by printing silver palladium electrodes on an alumina
substrate and over-printing the electrodes with the photoconducting ink. The conductor ink
used, (ESL 9633B) is an ink formulated for low silver diffusion. The conductor tracks are
printed and fired in the usual way, and have a have a thickness of approximately
15
m.
The photoconducting ink is screen printed over the conductor pattern, dried, and fired in a
belt furnace at a temperature of 590
C. The samples pass through the 1m long furnace at a
speed of 25mm/min. The resultant film is relatively soft, it shows reasonable adhesion to the
alumina substrate, but can be scraped off. The film has a rather porous structure and a
thickness of about 50-100m.
Where the photoconducting film is printed over the silver palladium electrodes the film is
darkened, indicating diffusion of material from the conductor into the film. This darkening
extends up to 0.5mm beyond the edge of the conductor. For photoresistors printed over an
interdigitated array of pitch 1mm or less the film looks almost uniformly darkened over the
electrode pattern. This diffusion of material (probably silver) into the CdS/CdSe film is
significant in controlling the dark resistance and photoconductivity of the film.
Test resistors were fabricated using an interdigitated array with conductors 0.4mm in width on
a pitch of 0.6mm. The array was overprinted with the photoconducting film with an active area
of about 9mm by 9mm. Several different inks were used ranging in composition from CdS
only, through mixtures of CdS and CdSe to CdSe only. The photosensitivity of these resistors
was tested by illuminating them with an LED with its emission centred at 660nm. The CdS
resistors showed an almost linear variation of conductivity with illumination, but with relatively
low sensitivity (about 2 x 10
S m
W
). The greatest sensitivity (about 1.5 x
10
S m W)
was obtained with a film comprising a 1:1 mixture, by weight, of CdS and CdSe, again the
response was approximately linear. For a CdSe film the conductivity rose approximately as
the square of the illumination. The total change of resistance may be very large. For the
photoresistor with the 1:1 CdS/CdSe mix the dark resistance exceeded
20M
, reducing to
about 75 in bright sunlight.
The spectral response of the sensors was measured using a white light source and a
monochromator. The spectral sensitivity was calibrated against a silicon diode of known
spectral response. Results are shown in Figure 1 for sensors using a CdS film (a) and a 1:1 mix
of CdS and CdSe (b). The CdS sensor shows a spectral response which peaks at about
620nm and which shows a sharp dip at about 500-520nm, close to the absorption edge of
CdS. This agrees well with a spectrum published by Wright [7] for CdS activated with silver
and a halogen. The addition of CdSe to the ink shifts the peak sensitivity to the red and
improves the relative sensitivity in both the near infrared and the blue parts of the spectrum.
A CdSe film shows a peak sensitivity at 780nm.
Figure 1: The spectral response of photoresistors (a) CdS (b) CdS/CdSe 1:1
mixture.
The response time of the photoresistors has been measured for a step rise in illumination
using an LED source centred at 660nm. The rise time decreases with increasing illumination
and with the fraction of CdSe in the film. For the photoresistor with 1:1 mix of CdS and CdSe
the measured rise time (10-90%) ranged from 130ms at an illumination level of about
0.17Wm
to 16ms at an illumination of about 10Wm. The rise time varied approximately as the optical
power to the power 
. For a CdS film the measured rise times were in the range of
0.2-2s,
while for CdSe they were of order 2.5-12ms.
An undesirable feature shown by the CdSe films is that in response to a step rise in
illumination the photocurrent rises to a maximum and then falls off. The decrease in
photocurrent is initially rapid, but while the rate of fall of current slows down, it continues for
some minutes. The photocurrent reaches a steady value of only about 66% of its initial peak.
This effect was not observed for either the CdS film or the mixtures.
All the photoresistors printed over the silver palladium conductors showed good linearity
between the current and the applied voltage at constant illumination.
Three types of optical position sensor have been fabricated using screen printed
photoconductive films: an array sensor, a four quadrant sensor and a potentiometric sensor.
The structures of these sensors are shown in Figure 2. The simplest is the array sensor with a
series of photo-conductive cells with one common terminal. As fabricated two parallel arrays
were formed. Each cell has a width of 1mm, with a conductor line width of about 0.2mm.
When illuminated with a thin bar of white light (of width 0.14mm), parallel to the conductor
strips, the response of one cell of the array shows a double peak, as would be expected
since the bar of light has a width less than that of each conducting strip. Increasing the width
of the bar of light to about 0.4mm removes the double peak.
Figure 2: Position sensor structures.
The four quadrant detector was designed to detect movements of a spot of light. Its
performance was very much as expected. For a spot size of 2.5mm diameter the relation
between photocurrent in each segment and the spot position closely followed the area of
overlap between the spot and the segment. The separation of the conductors of the
interdigitated array was about 0.5mm (between the centres of the lines), so that for small spot
sizes, this simple relation will break down.
The third position sensor design is based on a potentiometer. A spot, or bar of light induces
photoconduction which bridges the gap between a conductor electrode and a resistance. The
resistance was formed using a standard thick-film resistor ink of either
10k per square or
100k per square (Dupont 8039 or 8049). The photoconductive film was printed over the
linear resistor and the conductor running parallel to the resistor. For both resistor inks it was
found that the contact between the resistor and the photoconducting film was non-ohmic. For
sensors fabricated with the 10K per square, a rectifying junction was formed, with a
forward bias threshold voltage of about 0.4V. When illuminated with ambient room light the
reverse current (metal electrode positive) was about 300 times smaller than the forward
current with an applied voltage of 5V. For a resistor made with the
100k per square ink the
non-linearity was much less marked.
The non-ohmic junction also showed a small photovoltaic effect. In bright sunlight and with a
10M load between one end of the resistor and the conductor electrode, a voltage of a few
10s of mV was observed. At lower levels of illumination this voltage decreased rapidly. The
higher resistivity resistor ink showed a smaller effect.
In order to avoid errors due to the photovoltaic effect when using the sensor to measure the
position of a light spot it is better to use the sensor as a current divider rather than a potential
divider. Each end of the resistance is connected to the virtual earth input of a transimpedance
amplifier and the silver palladium electrode is supplied with a constant current (in the
appropriate direction to forward bias the semiconductor junction). The difference in output of
the two transimpedance amplifiers is proportional to the distance of the light spot from the
centre of the resistor. The sensor used this way showed good linearity, limited by the
uniformity of the resistor. Changes in spot brightness or size may produce changes in
apparent spot position due to changes in the potential distribution in the resistor these effects
are most marked at either end of the sensor, but are very small.
Photoconductive sensors based on CdS and CdSe have been fabricated using conventional
thick-film processing methods used for hybrid electronic circuits. The sensors could be
integrated on the hybrid circuit with other components, either printed before the photoresistor,
or subsequently surface mounted (soldered).
The sensors show a broad spectral sensitivity across the visible part of the spectrum, a large
dynamic range and good sensitivity. A variety of ink compositions have been investigated, but
the best overall performance has been obtained with an ink based upon CdS and CdSe in
approximately equal proportions. This material gave the highest sensitivity, a response time of
the order of tens of milliseconds and a reasonably linear response between photocurrent and
illumination.
The possibility of producing simple, low cost position sensors using arrays of photoresistors
has been demonstrated.
- M. Prudenziati and B. Morten, 1992 The state of the art in thick-film sensors
Microelectronics Journal 23 133-141.
- J. E. Brignell , N. M. White and A. W. J. Cranny, 1988 Sensor applications of thick-film
technology IEE Proc 135 77-84.
- N. M. White and J. E. Brignell, 1991 A planar thick-film load cell
Sensors and Actuators
25-27 313-319.
- N. M. White and V. T. K. Ko, 1993 Thick-film acoustic wave sensor structure
Electron Letts 29 1807-1808.
- F. H. Nicoll and B. Kazan, 1955 Large area high-current photoconductive cell using
cadmium sulphide powder J Opt Soc Am 45 647-650.
- S. Vojdani, A. Sharifanai and M. Doroudian, 1973 Photovoltaic junctions formed on silk-
screened cadmium sulphide layers Electron. Letts 9 128-129.
- D. A. Wright, 1958 Photoconductivity Brit J Appl Phys
9 205-214.