Note: Descriptions are shown in the official language in which they were submitted.
~737~
BACKGROU~D OF THE INVENTION
lo Field of the Invention
This invention relates in general to imaging
devices, and in particular to solid state color imaging
devices.
2 Background Relative to the Prior Art
The prior art, say, for color cameras, involves
electron beam scanned tubes: A three color signal is derived
by either utilizing three tu~es with a beam splitter and
optical filters or one tube with a color stripe filter
affixed to the image receiving surface of the targetO The
former method requires the maintenance of registration of
the image on the three separate tubes and the latter method
suffers from loss of resolution, at least in part, because
the stripe filter must be separated from the target b~
approximately 1¢0 microns~
Recent U.S. Patents3 namely, 3,860,956 issued 1/14/75
to Kubo et al and 3,576,392 issued 4/27/71 to Hofstein, describe
single beam scanned color image tubes which do not utilize color
filters. The target of each tube is comprised of a plurality
of photodiodes. The color imaging capabi-ity arises from the
intrinsic wavelength dependent optical absorption of the target
material, which in both cases is silicon. Blue light is more
strongly absorbed than green light which is, in turn, more
strongly absorbed than red light. q'his is termed differential
~ptical absorption. The imagers described in '956 and '392
have their photodiodes grouped into pixel triads and are so
constructed that each member of a triad has a different spectral
sensitivity. For '956, the pixels are sensitive to blue (B )
blue plus green (B+G), and blue plus green plus red (B+G+R)o
For '392, the pixels are sensitive to (R), (R+G) and (R+G+B)o
. ~
~ 2-
73~
Various techniques ~or providing a solid state
color imaging device have started to appear in the litera
ture. These solid state devices are based upon arrays such
as charge coupled devices (CCD's), charge injection devices
(CID's), photodiodes, and phototransistors, which are self
scanned as opposed to beam scanned image tubes.
U.S. Patent 3,971,065 issued 7/20/76 to Bayer
discloses one approach to implementation of such solid state
arrays. The general approach of Bayer is by the use of a
special arrangement of triads o~ color filters overlaying the
imaging sites. The color ~ilter mosaic optimizes the resolution
for a fixed number of image sites. A CCD imager incorporating
this concept was reported by Dillon et al, International
Electron Devices Meeting~ Washington, DC, December 1976.
Published U.S. Patent Application B-502,289
published 1/13/76 by Choi describes another solid state imager,
such imager employing a color coding filter af~ixed to a
solid state, sel~ scanned array.
A third approach to solid state color imaging,
which approach utilizes the di~ferential optlcal absorption
of the silicon substrate to provide the three color signal,
is described in U.S. Patent 3,985,449 issued 10/12/76 to Patrin.
This approach employs ad~acent pixel triads. As a result of
different voltage biasing conditions the three pixels of a
trlad are sensitive to (B), (B+G) and (B+G+R), respectively~
SUMMARY OF THE INVENTION
The invelltion resides in a buried or bulk channel
charge coupled device (bccd) employing, typically3 three channels.
Bccd's have been described in the literature (U.S. Patents
3~739,240 issued 6/12/73 to Krambeck and 3,792~332 issued 2/12/74
to Boyle et al) and, as may be know~n~ a threë channel bccd
would employ six semiconductor
:`
~ -3-
.~
73~7~
layers o~ alterna-tely different dopant types. By so setting the
thicknesses of the first and second layers that a first
color~ because of differential absorption, is prevented from
appreciably entering the third and subsequent layers -- and
by so setting the thicknesses of the first through fourth
layers that a second color, because of absorption, is pre-
vented from appreciably entering the fifth and sixth layers --
a three channel color sensitive bccd is provided: assuming
the first, third and fifth layers are p~doped (acceptor
doped), and the second, fourth and s:ixth layers (the sixth
layer may comprise the semiconductor subætrate) are n-doped
(donor doped)~ a ~irst signal channel extends from the
surface o~ the charge coupled device to somewhere within the
n-doped second layer, although the p-doped first la~er
ca.rries signal charges, i~ any, similarly, a second signal
channel extends from somewhere within the n-doped second
layer to somewhere within the n-doped fourth layer, the
sandwiched p-doped third layer, however, being a second
signal-carrying layer, and, finally, a third signal channel
extends from somewhere within the n-doped fourth layer to
the n-doped sixth layer, the sandwiched p-doped ~ifth layer
being a third signal-carrying layer. Although each of the
three signal channels has a width that includes adjacent
non-signal-carrying layers, photon-generated signal carriers
which occur within the non-signal-carrying layers selec
tively d~ift to, and are processed by, respective signal-
carrying layers.
~ ssuming, for example~ the first~ second, and
third colors are respectively blue, green, and red, all
photo-generated ca:rriers produced within the first channel
-4-
'.~
7~37~
by blue, green, and red radiation drift to the first layer for
processing by gate electrodes on the surface of the device.
Similarly, all photon-generated carriers produced within the
second channel by green and red radiation drift to the third
layer for processing by the gate electrodes. And all photon-
generated carriers produce~ within the third channel by red
radiation drift to the fifth layer for processing by the gate
electrodes. Thus, the gate electrodes of the bccd are common
to all three channels (i.e., triads comprise superpositioned
-- as opposed to side-by-side-- regions of the device) and
simultaneously process all three color slgnals in proper phase
wi~th each other.
Thus, in accordance with the present teachings, there
is provided a solid state imaging device which comprises a chip
of semiconductor material comprising at least six layers of
alternately different dopant types wherein the thicknesses of
the layers are such that, in response to incident white light
falling on the first layer, substantially no blue light pene-
trates to the third layer and substantially no green light pene-
trates to the fifth layer. Means are provided for scavenging
-` mo~ile majority charge carriers from the first, third and fifth
layers to form respective buried charge transporting channels
~ in those layers with nonconductive transparent means covering
- the surface of the first iayer and transparent electrode means
on the transparent nonconductive means.
In accordance with a urther teaching an imaging
sensor device is provided which comprises a wafer of silicon,
a transparent oxide of silicon on the wafer and a plurality of
rows of transparent electrod means on the oxide. The wafer has
at least six contiguous layers wherein the layers of the device
are such that first, second and third colors are absorbed within
the combination of the first and second layers and second and
: :'
~ 37~
third colors are absorbed within the combination of the second,
third and fourth layers with the third color being absorbed with-
in the combination of the fourth, fifth and sixth layers, each
being doped with impurity atoms and each being doped with a type
impurity different than any of its contiguous layers with the
layers being disposed so tha-t the oxide layer is contiguous with
the first layer and respective oh~lic row contacts to the firs-t,
third and fifth layers is provided for removing mobile majority
carriers from those layers.
By means of the teaching of the invention, side-by~
side "color triads" are obviated. As a consequence, an imaging
array according to the invention possesses higher spatial resolu-
tion since only one pi~el (or image site) provides all color
~nformation; this is to be contrasted with solid state array
schemes utilizing color filter overlays which require three
pixels for the same color information. Furthermore, the incident
radiation is, by means of the invention, more efficiently utilized
since all "visible spectrum" photons which are incident upon a
pixel will generate a signal charge in one of the three channels.
This is to be compared with those color filter overlay schemes
wherein two-thixds of the incident photons are wasted since,
for example r green and blue photons incident upon a red
sensitive pixel or image site will not contribute to the
output signal.
- In addition to the advantages noted above, since
all the color signal information from a given pixel arrives
simultaneously a-t the output of the array, decoding and
. ~
` C
'
- . . -
~ . - ... . . . . . .. .
37~
delay circuitry is unnecessary. Thus, discrete color sig-
nals may be processed directly~ for example, by well known
linear matrix methods to achieve proper color balance for a
particular display mode, such as television.
The invention will be described further in connec-
tion with the figures, of which:
Figs. la and lb are diagrams useful in describing
the invention,
Figs. 2 is a plan view of an embodiment of the
invention;
Fig. 3 is a generally schematic elevational view
of the invention embodiment of Fig. 2;
Fig. ~ is a view of the embodiment of Fig. 2 taken
generally along lines 4-4 thereof; and
Fig. 5 is a schematic showing of an area array
according to the invention.
Construction of a multiple channel bccd according
to the invention will be described with reference to the
energy band diagram of Fig. la: Starting with an original
wafer (6th layer) that contains 2 x 104 donor impurities
per cm3, a 1 um thick p-doped (boron) region (5th layer) is
ion-implanted into the wafer, the dopant level of the p-
region being .6 x 1016 impurities per cm3. A 2 um thick n-
doped epitaxia~ layer is then grown atop the p-doped (5th)
~ layer by heating the wafer in an atmosphere of arsenic-doped
; silane. The dopant level of the epigrown layer is .8 x 1016
impurities per cm30 Then, a 1 um thick p-doped (boron)
region (1 x 1016 impurities per cm3) is ion-implanted into
the epigrown n-doped layer to form two 1 um thick layers~
i.e. the third and ~o~xth layer~. Again~ an epitaxially
grown n-doped layer is formed atop the p-doped third layer
.
-6-
,,: ':
by heating the wafer in an atmosphere of arsenic-doped silane,
this epigrown layer being 1.3 um in thickness. By lon-
implanting to a depth of .3 um (3.5 x 1016 b~ron impurities
per cm3) into the 1.3 um -thick layer, such layer is
converted into a pair of layers, one of .3 um thickness and
one of 1 um thickness (i.e., the first and second layers of
the device). A gate oxide 10 is then grown or deposited atop
the device, after which a transparent gate electrode (6)
12 is applied over the oxide.
The fabrication of the gate oxide and gate structure
is determined by the type of CCD imager: that is, two phase3
three phase, four phase, or interline transfer. This aspect
of the structure is well known in the art.
Suitable electrical contact must be established
with the layers. This is accomplished away from the trans-
fer gate area, namely~ at the input or output end of a line
of photoelements or transfer gates. With electrical contact
so made, the p-doped first, third and fifth layers are
reverse-biased with respect to the second and fourth layers
and substrate. [The substrate, second and fourth layers are
held at ground potential and the first~ third and fifth
layers are held at negative voltage.] The unbiased energy
band diagram is shown in Fig~ lb. Application of such
reverse bias causes all mobile charges to be drained from
the layers3 resulting in the energy band profile shown in
Fig. la. The exact shape o~ the energy band diagram depends
critically upon the doping levels of the various layers, the
substrate doping, the gate oxide thickness and the voltage
applied to the charge draining electrode. Once these parameters
are known~ the energy band diagram is obtained by solution
~ o~ the Poisson equation.
:
-7-
.
. .
3~
The layer thicknesses and doping levels of Fig.
la, with an oxide thickness of .2 um, and with small negative
"biasing" voltage, produce relative minima in the band
diagram at approximately .7 um and 2.6 um below the oxide.
The first photosensitive channel is approximately .7 um
wide being bounded by the oxide layer,10 interface and the
first energy band minimum, i.e., the minimum nearest the
oxide. The second photosensitive channel is approximately
1.9 um wide being bounded by the -two potential minima. The
third photosensitive channel is more than 10 um wide belng
bounded, in Fig. la, on the lef-t by the second energy band
minimum and on the right, several mlcrons into the substrate,
depending mostly on the minority carrier dlffusion length.
The imager is irradiated from the gate side. Both
the gate insulator and gate electrode are virtually trans-
parent to visible light. Photons in the visible spectrum
will be essentially completely absorbed in the layered
structure since the penetration depth lies between .2 um and
5 um for the wavelength range .4 um to .7 um. Blue radia-
tion is substantially absorbed within the .7 um wide channel
nearest thè oxide. Green radiation is substantially absorbed
~ , .
within the two channels closest to the oxide. Only red
radiation penetrates deeper than the boundary between the
second and third channçls at 2.6 um, and is therefore ab-
sorbed within the third channel.
For a p-channel devlcej an absorption event gen-
erates a hole as:the signal charge. The hole is produced at
the depth or location in the semiconductor at which the
absorption event occurs. If a signal hole 14 is created in
the- first channel (by a red, green or blue photon), it
.
~ 8-
~ . :
737~
drifts to the potent:lal well (for holes) 16 of the first
channel; similarly, a signal hole 18 cr-eated in the second
channel, by a green or red photon, drifts to the potential
well 20 of the second channel; and a signal hole 22 created
in the third channel (by a red phot;on) drifts to the poten-
tial well 24 of the third channel. The signal charge accumu-
lates in the channels according to the radiation exposure
incident upon the gate.
The electrostatic potential of the three potential
wells in which the'signal charge acccumulates may be manipu-
lated by the gate voltage. It should be appreciated that
the potential wells associated wikh all three color channels
are controlled by a single gate voltage, and therefore the
signal holes may be manipulated simultaneously, ~for example,
transferred from the region beneath one gate to the region
beneath an adjacent gate, just as for a conventional single
channel CCD as is well known ln the art.
Referring now to Figs. 2-4, a three phase linear
bccd imaging device according to the invention comprises an
: . . . .
n-doped silicon wafer'(chip) 2~6 into which a p-doped layer
28 is ion-implanted. An epigrown n-doped layer 30, formed
over the layer ?8, has a p-doped layer 32 ion-implanted into
it; and an epigrown n-doped layer 34 has a p-doped layer 36
'ion-implanted into it. As taught in connection with Fig.
la, the ion-implanted layers 28, 32 and 36 are 1 um, 1 um,
and .3 um thick, respectively; and the epigrown layers 30
and 34 are 2 um thick.
Transparent siO2 38 covers the face of the device,
and overlaying the oxide covering is a linear array of ' .-
transparent gate electrode(s'? 40 appropriately intercon-
nected for.purposes of charge transfer.
.
_ 9 _ -
:
,: . .
. : ~ ' - , :
: : ~ - . .
-.: '.'- ' ."',, ' ' .' '`'' '. ' .' ' , '
The lon-irnplanted layer 36 fans out, at either end
of, and to the ~lde X O r the device. Similarly, the ion-
implanted layer 28 fans out, at either end of, and to the
side Y Or the device. And the ion--implanted layer 32 ex-
tends, at either end Or the device~ toward the extremities
Z-Z .
Heavily p-doped diffusions 42, 44, and 46 extend
from windows in the nonconductive oxide layer 38 to, respec-
tively, the signal-processing p-channels 28, 32, and 36,
ohmic metal contacts 48, 50, and 52 being made, respec-
tively, to the diffusions 42, 44, and 46. A channel stop
diffusion 47, shown only in Fig. 2, confines photon-gen-
~ erated charges to processing by the gate electrode(s) 40.
; A typical e~vironment in which the device of Figs.
?--4 would find use would be in the line scanning of images...
and typical operation of the device would have reverse-
biasing negative voltages applied to the contacts (ter-
minals) 48, 50, and 52. Such voltages would deplete mobile
carriers from the signal handling channels 28, 32, and 36, .
.
~ and create the energy band profile of ~ig. la. After a
.. . . , ~
clocking period during which photon-produced holes have
eollected in the channels 28, 32, and 36, say under the gate
electrode 40A (to which, nominally, a zero voltage is applied)
a negative voltage would be appIied to the electrode 40A,
while the electrode 40B is caused to go to (or remain at)
zero volt~s. This would cause the signal holes in each of
the channels 28 ? 32, and 36 to sh~ft simultaneously from
under the gate 40A to under gate 40B. Further processing
would be in accordance with techniques known to the art.
As noted heretofore, ~he present invention offers
; ~many improvements over previous sol~id state color imagers,
: , .
: : ,: - , ;
'~' ' ' , -~ ' '
'~ ' ' ., ' ., ' . ' ' '' . , , ' ' ' ' ' . ,:
namely, improved spatial resolution, higher effectiYe quantum
efficlency and the elimination of the need for signal decoding
and delay circuits.
As the timed "superposed" color signals simul-
taneously exit the device they are applied to a matrlxing
circuit encompassing appropriate coefficients for the dis-
crete colors as is known in the art. One such matrixing
circuit, simply depicted, is indicated in connection with
Fig. 2.
The invention has been desc~ribed in detail with
particular reference to a preferred embodiment thereof, but
it will be understood that variations and modifications can
be effected within the spirit and scope of the invention.
For example, while a linear imaging device is depicted in
Figs. 2-4, the concepts of the invention may be incorporated
into an area imaging array, say in the manner depicted in
~ig. 5 And, while a p-channel device has been discussed in
connection with Figs~ 1-4, an n-channel device accordlng to
the invention would be the same as that shown in Figs. 1-4,
except that all impurity types noted in Figs. 1-4 would be
reversed, and gate and bias voltages would become positive.
Also, while a three channel device has been described, a
similar such device having any number of superposed channels
greater than one would be within the scope of the invention,
provided, of course, that the channels are selective of
color. And, if desired, filters may be applied over the
device to limit the response of the device, ~ay, to the
visible spectrum. Furthermore, although a three phase
device is shown in Figs. 2-4, both two or four phase con- -
Pi~urations, as well as the interline transfer type imager,
may incorporate the invention.
'~ .
