I. INTRODUCTION
353
TABLE I
COMMERCIAL PHOSPHORS SELECTED FOR CONVERSION EFFICIENCY TESTING
Phosphor Technology has a product JL 49 that is supposed to be the fastest phosphor in the world. It met all the compliance criteria but was
too expensive to be selected for testing.
United Mineral and Chemical Corporation, Lyndhurst, NJ USA.
Osram Sylvania, Ltd., Towanda, PA USA.
(1)
is the fraction of light lost due to scattering,
where
is the fraction of light reflected,
is the fraction reflected
is the fraction absorbed by
at the filmsensor interface,
the phosphor, is the incident wavelength, and is the emitted
wavelength. The fraction of light emitted back toward the light
354
source (the radiation that is not captured by total internal reflecis defined as
tion)
(2)
COMPOSITION
OF THE
355
TABLE II
TWO DIFFERENT PHOSPHOR COATINGS TESTED AND THE PURE PLASTIC COATING THAT WAS USED
ALONG WITH COATING 2212-2 TO MAKE A THIRD COATING
A new settle deposition method was also developed in an attempt to improve coating uniformity, which directly impacts the
PRNU and image resolution. The suspension is formed as described in above; however, instead of immediately applying it to
the substrate the suspension was allowed to settle. This is effectively a simple method of filtering the crystal that has a size distribution of 1 to 8 m with an average particle size of 3.6 1.7
m. The solution is then transferred to another container containing the substrate and allowed to evaporate, leaving behind
the uniform coating. It was not necessary to add a subsequent
layer of plastic, as described above, since a layer of plastic settled on top of the phosphor.
IV. CONVERSION AND QUANTUM EFFICIENCIES
In this work, the conversion efficiency is retrieved from measurements of the luminescing efficiency of a coated fused-silica
slide, while the quantum efficiency involves measurements with
a coated CCD sensor. To measure the conversion efficiency,
the experimental setup given in Fig. 3 was used. The UV light
source is an Oriel 200 W Hg bulb. The light was first collimated and then passed through a liquid IR filter. Interference
filters were used to create a narrow-band spectrum made up of
0
atomic mercury lines. A visible light filter (transmission
from 400 to 1000 nm) was used to provide a much better defined UV spectrum. A pinhole slide was used so that the converted light could be considered a point source. This is necessary
because the isotropic luminescence of the phosphor means that
light is emitted equally in all directions, so therefore some will
not be captured by the optical power meter. To account for this
loss, a solid angle correction factor was applied to the optical
power meter reading. The correction factor was determined as
the solid angle made by the pinhole and the circumference of the
hemispherical
optical power meter as a fraction of the entire
solid angle. Fused silica was selected as the coating substrate
because it is transparent at the illumination wavelengths. The
Fig. 4. Conversion efficiency versus wavelength for the three coatings tested.
As in Fig. 1, the coating has been deposited onto a fused silica slide.
long pass filter was used to block out any of the incident UV
light that may have passed though the coating. To measure the
sensor quantum efficiency, the same light source as described
above was used along with a DALSA CA-D1-0256 camera. The
camera was clocked in time delay and integration (TDI) mode
at a line rate and data rate of 18.3 kHz and 5.0 MHz, respectively. For PRNU measurements, the camera was operated in
the normal area imaging mode.
Fig. 4 contains a plot showing the conversion efficiency as
a function of wavelength for the three coatings tested. Inspection of the figure reveals that the additional pure plastic layer
almost doubles the efficiency at 265 nm, from 19% to 35% for
the 2212-2 and two-layer 2212-2 coatings, respectively [13]. An
SEM analysis performed on the two-layer coating shows that the
coating is 30 2 m thick.
The two-layer coating was deposited onto DALSA IA-D10256 area array sensors. The coating was applied as a thick
and thin coating with an estimated thickness of less than 20
m and greater than 50 m, respectively. These estimates were
based on the measured absorption coefficient of a coated fused
silica slide, which was found, using Beers law, to be 15
1 m at 265 nm [13]. A simple calculation of the predicted
coated sensor quantum efficiency is given as the quantum efficiency of the sensor at the phosphor emission wavelength, 0.40,
multiplied by the coating conversion efficiency, 0.35, to give
14%. A plot of the quantum efficiency of both the thick and
thin two-layer coating, and the uncoated sensor as a function
of wavelength is given in Fig. 5. The figure shows that the measured coated sensor quantum efficiency is 12% 0.8% at an
illumination wavelength of 265 nm. This improvement represents a 3.5x improvement in sensor quantum efficiency. (Note
that the nonzero quantum efficiency of the uncoated sensor is
due to the residual blue response of the uncoated sensor.)
356
V. PRNU
PRNU is a measure of pixel-to-pixel variations in sensor responsivity. Ideally the PRNU is zero, meaning that under constant flat field illumination every pixel will have the same responsivity. However, in reality, the PRNU of an uncoated sensor
is typically on the order of 5%10% [15]. The PRNU degradation due to the coating is a simple, nondestructive test that can be
used to quantify nonuniformities in the coating thickness. This
parameter is of particular importance for image sensors where
coating nonuniformity will contribute to resolution degradation.
In this investigation, PRNU is defined as the standard % rms calculation:
PRNU
where
is the standard deviation of the difference between
pixel output and the average pixel output across the entire array
is the average output at
as measured at 50% of saturation,
is the output at zero illumination.
100% of saturation, and
For comparison purposes, the PRNU of the uncoated sensor was
first measured at the coating emission wavelength. The PRNU
of the sensor coated with the two-layer 2212 coating was 7.0 and
1.6 times larger than the uncoated sensor for the thick and thin
coating, respectively (see Table III). Note that the signal output
of each pixel of the coated sensor was found to have a linear relationship with light intensity indicating that post-data processing
could be used to correct PRNU degradation. An improved deposition technique was developed to reduce the amount of PRNU
degradation; these results are described in Section VIII.
VI. CONTRAST TRANSFER FUNCTION
AND PHOTODEGRADATION
The contrast transfer function (CTF) quantifies the degree to
which a given spatial input frequency can be reproduced by the
and output
are the maximum and minimum
where output
and input
are the
output signal, respectively, and input
maximum and minimum input signal. A CTF equal to one indicates that the sensor can perfectly image the target (assuming
an optical transfer function equal to unity).
The CTF is one of the most difficult imaging parameters to
quantify. The following experiment was performed in an attempt
to determine an approximate CTF value for a thick and thin
two-layer coating 2212-2. A precision 100- m-wide slit was
imaged onto a coated DALSA IL-P1 line scan sensor at an illumination wavelength of 265 nm. A lens of magnification 0.5
was used to image the 100- m slit down to 50 m at the sensor.
The line rate and light intensity were adjusted to ensure that the
pixels did not reach their full-well capacity and spill charge into
neighboring pixels. A digital oscilloscope was used to measure
the analog signal generated at the illuminated pixels. Six pixels
were actually illuminated (four fully and two partially); an approximation that five pixels (with a pixel pitch of 10 m) were
illuminated was used for this analysis.
To quantify the CTF degradation of the coated sensors, the
normalized output signal was plotted as a function of the pixel
number. The CTF data of the coated sensors had a Gaussian
shape due to the scattering of light by the coating. A Gaussian
distribution was fitted to the observed data to simulate the response of the coated sensor to an alternating bar target with
50- m-wide light and dark regions. Fig. 6 gives the simulated
response for two thicknesses of coatings; coating 1 is estimated
to be less than 20 m thick and coating 2 greater than 20 m.
If the optical transfer function is assumed to be 1.0, meaning a
difference in the normalized input signal is 1.0 (a valid assumption given that the spatial frequency is 100 lp/mm), then the CTF
for the thin and thick coating is 0.9 and 0.4, respectively. These
findings confirm that the thinner coating is better for imaging
applications.
The photostability of the 2212 phosphor was tested. The twolayer 2212-2 coating was deposited onto a fused silica slide and
357
(a)
REFERENCES
(b)
358
Martin J. Kiik (M96) received the B. Appl. Sci. degree and the M.S. and Ph.D. degrees from the University of Toronto, Toronto, ON, Canada, in 1986,
1989, and 1994, respectively, where his research involved the investigation of the dynamics of chemical
processes in rare-gas discharges in supersonic jet expansions.
Since 1994, he has been working in the Image
Sensor R&D group at DALSA Corporation, Waterloo, ON, Canada.
Dr. Kiik is a member of the Association of Professional Engineers of Ontario, Canada.