United States Patent (19)
`Jastrzebski et al.
`. . . . . . . .
`.
`.
`
`11
`45
`
`Patent Number:
`Date of Patent:
`
`4,481,522
`Nov. 6, 1984
`
`54 CCD IMAGERS WITH SUBSTRATES
`HAVING DRIFT FELD
`75 Inventors: Lubomir L. Jastrzebski, Plainsboro;
`Peter A. Levine, Trenton, both of
`N.J.
`73) Assignee: RCA Corporation, New York, N.Y.
`21 Appl. No.: 361,228
`22 Filed:
`Mar. 24, 1982
`51
`int. C. ..................... H01L 29/78; H01L 27/14;
`HOL 31/00; H04N 9/07
`52 U.S. C. ........................................ 357/24; 357/30;
`358/44
`58) Field of Search ................ 357/24 LR, 30; 358/44
`56)
`References Cited
`U.S. PATENT DOCUMENTS
`3,860,956 1/1975 Kubo et al. ........................... 358/44
`4,00,878 1/1977 Weimer.......
`... 357/24 LR
`4,155,094 5/1979 Ohba et al. ..................... 357/24 LR
`
`
`
`A
`
`A 357/30
`
`4,160,985 7/1979 Kamins et al. ..
`4,210,922 7/1980 Shannon ...
`4,229,754 10/1980 French .................................. 357/30
`4,242,694 l2/1980 Koike et al. .
`... 357/24 LR
`4,247,862 l/1981 Klein ........
`... 357/24 ER
`4,282,547 - 8/1981 Morishita .............................. 358/44
`4,348,690 9/1982 Jastrzebski et al. ............ 357/24 LR
`Primary Examiner-Gene M. Munson
`Attorney, Agent, or Firm-Joseph S. Tripoli; George E.
`Haas; Allen LeRoy Limberg
`57
`ABSTRACT
`A CCD imager is made to have an internal drift field
`tending to force further into the bulk any charge carri
`ers outside the potential wells induced adjacent to inte
`grating electrodes, so those charge carriers recombine
`in the bulk. This reduces background striations, reduces
`crosstalk, can be used for improving blooming control,
`and can be exploited to control the wavelengths of light
`to which the imager is responsive.
`
`V
`
`16 Claims, 6 Drawing Figures
`
`NNNNN Ele ls s 'lls s 2
`
`ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ
`
`
`
`LOG DOPING
`CONCENTRATION
`
`Dell Ex. 1010
`Page 1
`
`

`

`U.S. Patent Nov. 6, 1984
`
`Sheet 1 of 4
`
`4,481,522
`
`
`
`S.
`
`----
`t---0--
`2
`
`s
`
`LOG DOPNG
`CONCENTRATION
`
`
`
`
`
`
`
`
`
`
`
`W
`
`
`
`00
`
`80
`
`60
`
`40
`
`W/2-ts
`
`R
`
`DIFFUSION DISTANCE
`Aig. 2
`
`Dell Ex. 1010
`Page 2
`
`

`

`U.S. Patent Nov. 6, 1984
`
`Sheet 2 of 4
`
`4,481,522
`
`N
`S.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`20
`
`Dell Ex. 1010
`Page 3
`
`

`

`U.S. Patent
`
`Nov. 6, 1984
`
`Sheet 3 of 4
`
`4,481,522
`
`
`
`
`
`
`
`
`
`
`
`
`
`Dell Ex. 1010
`Page 4
`
`

`

`U.S. Patent Nov. 6, 1984
`
`Sheet 4 of 4
`
`4,481,522
`
`CHANNEL STOPS
`
`(P. a
`low I C
`w I C B it w if C
`B
`B
`A2
`R
`If W C
`B F W C B F W C
`B
`(3 H Élow - C - B - W - C - B - W - C : B XA REGISTER
`H Bit Wil C
`B
`
`W II C B -it
`
`S
`
`
`
`(Ph3 H -
`H H H H H H H H
`B REGISTER
`F.
`.
`.
`.
`.
`.
`.
`.
`.
`
`
`
`C
`
`
`
`CH.
`STOPNE
`(Pole
`
`r
`
`D
`
`f
`3: E.
`B
`PPP
`
`(P3 C
`
`Aig. 5
`
`Dell Ex. 1010
`Page 5
`
`

`

`O
`
`15
`
`25
`
`1.
`
`CCD IMAGERS WITH SUBSTRATES HAVING
`DRIFT FELD
`
`The invention concerns charge-coupled device
`(CCD) imagers and, more particularly, ones having
`drift fields in their semiconductive substrates to im
`prove the operation of the CCD.
`Charge-coupled imagers may be constructed using
`either surface or buried charge transfer channels. The
`better transfer efficiency of buried charge transfer chan
`nel imagers allows them to provide better image resolu
`tion than imagers having surface charge transfer chan
`nels can. However, imagers with buried charge transfer
`channels tend to be more costly to construct because of
`the additional processing steps needed to obtain buried
`channel operation with anti-blooming control. The in
`vention to be described, is particularly well-adapted to
`solving problems associated with surface channel imag
`ers operated in the so called anti-blooming mode, but is
`20
`also applicable to suppressing cross-talk on buried chan
`nel imagers. Suppression of cross-talk between charge
`transfer channels is of particular interest when they are
`arranged to respond to different wavelengths of light,
`since such cross-talk gives rise to undesirable cross
`color phenomena easily discerned in a television display
`generated from video samples from the CCD imager. In
`the present application, the invention is described with
`particular regard to surface channel CCD imagers.
`A basic problem in such CCD imagers is the appear
`30
`ance of so-called striation patterns in the background of
`the detected television image. L. Jastrzebski, P. A. Le
`vine, A. D. Cope, W. N. Henry and D. F. Battson dis
`cuss the origins of these striations in their paper "Mate
`rial Limitations which Cause Striations in CCD Imag
`35
`ers' appearing pp. 1694-1701 of IEEE TRANSAC
`TIONS ON ELECTRON DEVICES, Vol. ED-27, NO.
`8, Aug. 1980. In surface channel CCD imagers operated
`in the anti-blooming accumulation mode, some part of
`the background striations is attributable to variations in
`40
`the resistivity of the semiconductor material in which
`the charge transfer channels are induced. The resistivity
`variations give rise to electric field intensity variations
`in the semiconductor with components parallel to the
`surface along which charge is transferred, which com
`45
`ponents alternately aid and hinder charge transfer to
`give rise to background striations.
`The predominating cause of background striations in
`surface channel imagers is the non-uniformity of the
`density of charge recombination centers in the bulk
`50
`-i.e., those portions of the semiconductive substrate
`outside the depletion regions induced in the semicon
`ductive substrate adjacent to the gate electrodes of the
`CCD. (Regions depleted of electrons are induced by
`relatively positive potentials applied to the gate elec
`55
`trodes adjoining a p-type semiconductive material in
`which electrons are minority carriers. Analogous phe
`nomena to those described take place when regions are
`depleted of holes induced by relatively negative poten
`tials being applied to gate electrodes adjoining an n-type
`semiconductive material in which holes are minority
`carriers.) In the depletion regions the Fermi level is
`sufficiently shifted to make majority charge carriers
`unavailable for recombination with the minority charge
`carriers, which have been generated by photoconver
`65
`sion of radiant energy in the image projected into the
`semiconductor. Majority carriers are available for re
`combination in the bulk, however, and minority carriers
`
`4,481,522
`2
`can recombine with them at recombination centers asso
`ciated with irregularities in the crystalline lattice struc
`ture of the semiconductive material. The charge carri
`ers generated in the bulk by photoconversion diffuse,
`since the imager is operated at temperatures above Zero
`Kelvin.
`In prior art CCD imagers this thermal diffusion is
`sufficiently random in direction that an appreciable
`number of the charge carriers diffuse from the bulk
`towards the depletion regions induced under the gate
`electrodes. The fraction of these charge carriers which
`reach the depletion regions varies, according to the
`probability of their being recombined in the bulk in
`stead. Decreased density of charge recombination cen
`ters in portions of the bulk with lesser density of lattice
`irregularities increases the likelihood of charge carriers
`reaching nearby depletion regions, causing greater col
`lection of charge carriers in those depletion regions and
`leading to relative brightness in the pixels supplied in
`response to integrated charge accumulated from those
`depletion regions.
`The invention, in one of its aspects, is directed to
`suppressing these sources of background striations, de
`scribed in the preceding paragraph, by preventing all or
`nearly all of the charge carriers generated in the bulk
`from reaching the depletion regions induced under the
`gate electrodes. This is done by creating a field, such as
`a drift field, in the semiconductor substrate to sweep
`minority charge carriers generated in the bulk further
`into the bulk, away from the electrode-bearing surface
`of the substrate, there to recombine with majority carri
`ers. This reduces the likelihood of charge carriers gen
`erated closest to one CCD register gate electrode diffus
`ing to a depletion region under another gate electrode
`to cause various cross-talk phenomena, as will be de
`scribed in the detailed description that follows, as well
`as suppressing background striations. A drift field can
`be generated by grading the concentration of doping
`impurities in the semiconductor substrate such that the
`concentration decreases with increasing distance from
`the surface which the charge transfer channels of the
`CCD shift registers adjoin.
`In the drawing:
`FIGS. 1a and 1b, drawn to a common exaggerated
`vertical scale, are a profile view of a portion of a CCD
`imager image register and a plot of the doping concen
`tration in such an image register constructed in accor
`dance with the invention;
`FIG. 2 is a graph of the probability of recombination
`of minority charge carriers in a semiconductive solid
`versus their diffusion distanced plotted on a roughly
`ln(d/l) scale, where l is the mean diffusion distance;
`FIGS. 3 and 4 are block diagrams of alternative color
`camera arrangements using CCD imagers in accor
`dance with the invention, which figures omit the con
`ventional zoom and relay lens portions of the cameras
`used to project images into the CCD imagers; and
`FIG. 5 is a plan view of a respresentative CCD
`imager constructed according to the invention for use in
`a color camera arrangement of the sort shown in FIG.
`2.
`FIG. 1a shows the profile of a portion of one charge
`transfer channel in a three-phase CCD imager image, or
`A, register. The substrate, 10 of semiconductive mate
`rial has a first surface 11 adjoined by a dielectric layer
`12. (For clarity, sectioning lines are omitted on the cut
`through substrate 10.) Where the substrate 10 is silicon,
`dielectric layer 12 may be silicon dioxide formed from
`
`Dell Ex. 1010
`Page 6
`
`

`

`10
`
`20
`
`4,481,522
`4.
`3
`region will migrate to the charge depletion regions,
`an oxidizing atmosphere, for example. Atop dielectric
`owing to normal diffusion processes in a semiconductor
`layer 12 are electrodes 13-18, which may be metallic
`substrate operated at temperature above zero Kelvin.
`electrodes formed by selective etching from a film de
`These charge carriers collected in the depletion regions
`posited by evaporation or sputtering. Or the electrodes
`form, in aggregate, the charge packets which transfer to
`may be polysilicon, supposing the substrate to be silicon
`right during transfer intervals. The CCD imager with
`and the dielectric layer to be silicon dioxide. (Referring
`three-phase clocking during charge transfer intervals is
`to the first plane surface 11 of the substrate 10, atop
`representative of a class of CCD imagers which may
`which the electrodes 13-18 are deposed, as its "top"
`differ from it in using uni-phase, two-phase, four-phase,
`surface and referring to an opposed second plane sur
`and other multiphase clocking schemes; the mode of
`face 19 of the substrate 10 as its “bottom" surface is a
`clocking is incidental to the invention, which is applica
`convention adopted for facilitating description of the
`ble to any of these CCD imagers.
`processes used to manufacture the CCD imager. The
`A problem in prior art imagers of this class is associ
`orientation of the charge coupled device in physical
`ated with the carriers generated in the bulk of the sub
`space may be otherwise. As utilized in a television cam
`strate 10 migrating through the substrate lattice to add
`era the top and bottom surfaces of the CCD imager are
`15
`to charge carriers generated in the depletion regions
`conventionally referred to as its "front” side and “back"
`themselves, to augment the charge packets transferred
`side, respectively, particularly in connection with speci
`to right during transfer intervals. This, as described at
`fying which of these surfaces is illuminated by the light
`length above, gives rise to background striations owing
`image.)
`.
`.
`.
`.
`to the non-uniform concentration of charge recombina
`Electrodes 13 and 16 are shown connected to receive
`a first phase (b1 of the three-phase clocking signals;
`tion centers in the substrate 10 bulk. The more remote
`from the depletion regions the site where a charge car
`electrodes 14 and 17, to receive a second phase db2 of
`rier is generated is, the more likely the carrier is to
`those signals; and electrodes 15 and 18, to receive a
`diffuse to a region other than the one closest to its gen
`third phase db3 of those signals. The successive applica
`eration site. However, the number of charge carriers
`tion of the clocking signals di, db2, and d3 in order of 25
`generated in the substrate 10 decreases further into the
`their ordinal number subscripts will, as known, cause
`substrate 10 from the surface through which the image
`shift of charge packets in potential wells induced under
`is projected. So, in a CCD imager receiving an image
`the electrodes in a rightward direction on each clocking
`through the top surface 11 most of the background
`signal transition. These potential wells are associated
`striations will be caused by charge carriers generated in
`with the charge depletion regions induced below the
`regions between the depletion regions. Charge carriers
`electrodes receiving relatively positive clocking signal
`generated in the middle portions of these regions be
`phases. In FIG. 1 the depletion regions are shown in
`tween depletion regions will be most likely to experi
`dashed outline being induced under electrode pairs 13,
`ence varying rate of recombination and will be primar
`14 and 16, 17 presumed to be receiving b1, db2 clock
`ily responsible for background striations. In a thinned
`voltages relatively positive to the d3 clock voltage ap
`35
`substrate CCD imager receiving an image through its
`plied to electrodes 15 and 18.
`-
`to be one of a plurality of
`bottom surface 19, the charge carriers generated at
`The shift register is assumed
`depths from top surface 11 greater than that over which
`column or row registers that are components of the
`the depletion regions extend will have a greater effect
`complete image register of the optical imager. So shift
`on the generation of unwanted backgound striations.
`register. operation takes place only during recurrent
`40
`No matter which of its top and bottom surfaces a CCD
`transfer intervals when the d3, b1, d2 phases, are cycli
`imager receives radiant energy from the image through,
`cally clocked "low" or relatively negative relative to
`the charge carriers generated at depths from top surface
`the other two. These transfer intervals are interspersed
`11, greater than the channel stops extend, will give rise
`with charge integration intervals when clocking is
`to some cross-talk between adjacent charge transfer
`halted to leave two of the clock phases "high" or rela
`45
`tively positive relative to the other. (b1, d2 will be as
`channels.
`-
`CCD imagers are often used after optical filters that
`sumed to be the high clock phases in the field of video
`provide color selection of the image in patterns. In such
`samples being generated. Where field interlacing is used
`CCD imagers, certain of the cross-talk phenomena de
`to increase vertical resolution, the phases selected to be
`scribed in the preceding paragraph will give rise to
`high during charge integration intervals will change
`undesirable cross-color in a television display originat
`from field to field, assuming the CCD imager to be a
`ing from video samples generated by the CCD imager.
`vertical-field-transfer type.
`-
`Some of the charge carriers generated in the bulk in
`During each charge integration interval a radiant
`response to wavelengths of light selected by the color
`energy image (which may be in the visible light spec
`filter, will diffuse to depletion regions that are not the
`trum supposing the CCD imager detecting elements to
`55
`closest depletion regions to their respective generation
`be silicon) is projected into the substrate 10 through its
`sites. Charge packets in these non-closest regions are
`top surface 11 or its bottom surface. 19. This energy
`properly associated with an adjacent color in the color
`interacts with the substrate 10 to generate charge carri
`filter, so undesirable cross-color conditions develop.
`ers in the stratum next to dielectric layer 12. The minor
`For example, a CCD imager may be used in conjunc
`ity charge carriers that are to be collected as image
`60
`tion with a color-stripe filter having its stripes in regis
`samples are electrons in the case where the substrate 10
`tration with the charge transfer channels in the image,
`has p-type doping. Charge depletion regions induced
`or A, register. Cross-talk between the charge transfer
`under selected ones of the electrodes 13-18 usually
`channels will often give rise to cross-color. A CCD
`extend only part way into the stratum of substrate 10
`imager may be used in conjunction with a color stripe
`next to surface 12. Charge carriers generated in charge
`filter perpendicular to the charge transfer channels in
`depletion regions are retained in these regions, owing to
`the image register if the resulting color samples are
`the potential wells associated with those regions. Other
`subsequently commutated at correct spatial frequency.
`charge carriers generated in the bulk portions of the
`
`30
`
`50
`
`65
`
`Dell Ex. 1010
`Page 7
`
`

`

`4,481,522
`5
`Cross-talk along the length of the charge transfer chan
`nels in the image and transfer registers will then give
`rise to cross-color. If the color filter is of checkerboard
`pattern both types of cross-color will arise.
`These background striation and cross-talk problems 5
`are surmounted by constructing CCD shift registers, as
`exemplified by the FIG. 1a structural segment, in accor
`dance with the present invention, so that there is an
`electric field normal to the surface 11 that tends to force
`carriers deeper into the bulk away from the depletion 10
`regions induced adjacent to surface 11. This electric
`field can be a drift field created by a graded concentra
`tion of doping in the substrate. In the FIG. 1a structural
`segment the gradient of doping concentration would be
`normal, or perpendicular, to the surface 11 of substrate 15
`10, with the doping concentration decreasing with
`depth from surface 11. The drift field is accordingly
`directed to force most of the charge carriers generated
`outside a depletion region deep into the substrate 10
`bulk. In these portions of the bulk remote from surface 20
`11 of substrate 10 the charge carriers recombine. Ar
`rangements can also be made to place deep drain'struc
`tures to dispose of the charge carriers driven into these
`remote portions of the bulk.
`FIG. 1b is a graph of the logarithm of doping concen- 25
`tration versus depth into the first plane surface 11 of
`substrate 10 as will result in a uniform drift field. To
`prevent carrier diffusion over a given vertical distance
`of arbitrarily small value y normal to the first plane
`surface 11 of substrate 10, the doping concentration 30
`should change over distance y by a factor of at leaste,
`the base of natural logarithms. That is, in excess of 2.7
`times. This estimate is obtained by choosing the
`strength E of the electric field gradient to be such that
`the time tE for a charge carrier to tend to drift a distance 35
`y downward into the bulk owing to the drift field is less
`than the time td for a charge carrier to tend to diffuse
`upward the same distance.
`The time td for a charge carrier to diffuse a distance
`y is approximated by equation (1), following:
`tp=y/D. .
`.
`.
`.
`:
`D=(kT/g),
`In equation (2),
`k is Boltzmann's constant,
`-
`.
`.
`. .
`.
`. .
`T is the absolute temperature of the semiconductor,
`q is the unit electron charge, and
`u is the mobility of the minority charge carrier (elec 50
`tron or hole) of concern.
`.
`The time tE for a charge carrier to drift a distance y
`owing to the action of an electric drift field gradient of
`strength E is given by equation 3, following.
`te=y/E
`(a) 55
`E is determined, aS follows, preceeding from inequality
`(4).
`
`: (1)
`(2)
`
`40
`
`... :
`...
`
`45
`
`(4) 60
`tD>t E.
`Substitution is made into inequality (4) from equations
`(1) and (3).
`y/Day/u.E.
`
`(5) 65
`
`w
`
`Multiplying both sides of the inequality by the common
`factor DE/y2 results in inequality (6), following.
`
`(6)
`EcD/uy
`The value of D per equation (3) substituted into inequal
`ity (6) results in inequality (7), following, after elimina
`tion of u.
`El kT/qy
`
`.
`
`.
`
`(7)
`
`The average strength of electric field gradient between
`two points in a semiconductor with respective doping
`contentrations N1 and N2 can, if they are closeby each
`other, be satisfactorily approximated per the following
`equation.
`(8)
`E=(kT/qy) in (Ni/N.)
`To satisfy inequality (7), then, by substitution from
`equation (8), the conditions on change in doping con
`centration can be found to be as follows.
`In (Ni/N2)> 1.
`. .
`.
`
`(9)
`
`..N1/N2>e.
`,
`; :
`(10)
`Estimation of the distancey is the remaining thing to do
`in order to determine the doping gradient required to
`establish a drift field of the desired strength.
`Consider the choice of y insofar as reducing the visi
`bility of striation patterns in front-side illuminated CCD
`imagers is concerned. The vertical scale of FIG. 1a, as
`has been noted, is substantially expanded as compared
`to its horizontal scale. The depletion regions under
`“high" gate electrodes usually have depths in the range
`of 750 nm to 5um, while the breadth of spacing under
`the “low” electrodes between these depletion regions is
`of the order of 20 um. The channel stops separating
`adjacent charge transfer channels are about 5um wide.
`The diffusion length for charge carriers is normally
`much longer, 100 um or so, and varies inversely as the
`concentration of recombination center in the semicon
`ductor material of the substrate 10.
`FIG. 2 is a sketch graphing the probability of recom
`bination of a charge carrier as ordinate versus its diffu
`sion distance as abscissa. Although thermal diffusion of
`a charge carrier in a semiconductive solid is random in
`direction at any particular time, so the actual path taken
`by the carier is irregular, the diffusion distance is ex
`pressed in straight-line terms between the point the
`charge originates and a point at which the charge is at
`some point later in time. The diffusion distance, d, is
`plotted on a scale that is increasingly non-linear with
`increasing diffusion distance. The abscissa is roughly lin
`(d/l) d, which makes the scale roughly linear up to half
`the 20 um gate length and roughly logarithmic through
`the region of mean diffusion length 1. The mean diffu
`sion length l is that diffusion distance which would be
`associated with 1/e probably of recombination in an
`ideal semiconductive material having uniform average
`density of recombination centers, such that the proba
`bility of recombination would be an average PR-Av. In
`a substrate of actual semiconductive material probabil
`ity of recombination may range between PR-Lo and
`PR. Hiin various portions of the substrate.
`As pointed out previously, most of the charge carri
`ers responsible for visible striation patterns in a CCD
`imager illuminated through top surface 11 originate
`close to that surface between depletion regions. This is
`particularly so if the depletion regions extend deeper
`
`Dell Ex. 1010
`Page 8
`
`

`

`20
`
`4,481,522
`8
`7
`portion of the remaining charge cariers to be considered
`into the substrate to depths beyond which few photons
`are generated far enough away from the surface 11 that
`penetrate the substrate. There is less photoelectric gen
`even though the straight-line direction of their diffusion
`eration of charge carriers in the channel stops than in
`has a component towards the depletion region they are
`the charge transfer channels-that is quantum effi
`swept far away from surface 11 before they can be
`ciency is lower. Further, the channel stops have much
`collected in a depletion region. This effect becomes
`smaller area adjacent to the illuminated surface than the
`significant as y is reduced to dimensions comparable to
`charge transfer channels do. The means the spacings
`depletion region depth and becomes increasingly more
`between depletion regions in the same charge transfer
`significantasy is further reduced to be smaller than that
`channel are particularly of interest insofar as striation
`depth. Owing to the convergence of PR-Lo and PR-HI
`pattern generation is concerned. The fringing fields
`10
`limits on probability of recombination, there is substan
`surrounding potential "wells' associated with the de
`tially less variation in the percentage of the remaining
`pletion regions flanking such a spacing tend to make
`charge carriers collected in depletion regions than there
`charge carriers migrate towards the closer of the two
`would be in charge carriers from closer to the center of
`depletion regions, this tendency being stronger the
`the spacing between depletion regions, swept into the
`closer the charge carrier is to that depletion region. So
`bulk to recombine in the CCD imager with drift field.
`most of the charge carriers collected in a depletion
`Thus, choosing y to be some not too large fraction of
`region which are not intially generated in the depletion
`half the spacing between depletion regions will reduce
`region come from no more than half w, w being the
`background striations appreciably in a front-side illumi
`distance between depletion regions in the same charge
`nation CCD imager.
`transfer channel-i.e., half a gate electrode length
`Consider now the choice of y from the point of view
`where the imager is operated with only one clock phase
`of preventing cross-talk that can give rise to cross
`low during charge integration intervals.
`color. Cross-talk arises in front-side illuminated CCD
`Looking at FIG. 2, one notes that the probability of
`imagers from charge carriers generated further from
`recombination within such distances, 10 um or so, is
`surface 11 than the depletion regions extend. Any value
`low where diffusion lengths are an order of magnitude
`25
`ofy small enough to suppress background striations will
`larger, 100 p.m or so. The portion of the charge carrier
`population generated in the spacings between depletion
`prevent cross-talk between successive stages in a charge
`transfer channel. The primary source of cross-talk to be
`regions is at small diffusion distances from the depletion
`considered, then, is between depletion regions abreast
`regions. Where PR-Lo, PR-Av and PR-HI are con
`each other in adjoining charge transfer channels. This
`verging to zero as the diffusion distance, d, approaches
`30
`cross-talk causes undesirable cross-color where the
`zero. This means that the range of variation in probabil
`adjacent charge transfer channels are made to respond
`ity of recombination of a charge carrier before its col
`to different portions of the light spectrum. The drift
`lection in a depletion region decreases as distance of the
`field, then, should be sufficiently strong that charge
`site of its generation from the depletion region de
`carriers cannot migrate across some fraction-say, hal
`creases. That is, the contribution to background stria
`35
`f-the distance between these abreast depletion regions.
`tion of charge carriers generated at sites various dis
`Making y this fraction of this distance will accomplish
`tances from a depletion region increases with distance.
`this.
`The charge carriers generated near the middle of the
`All in all, y's of about 1 um are sufficiently small to
`spacings between depletion regions are chiefly responsi
`significantly reduce background striations and cross
`ble for the background striations, the charge carriers
`40
`talk that can give rise to cross-color in CCD imagers of
`generated closer to the depletion regions contribute
`the dimensions described. This is true in back-side illu
`much less to background striations.
`minated CCD imagers as well, where background stria
`So, then, suppose y is chosen smaller than ww2. The
`tions as well as channel-to-channel cross-talk originate
`rate of diffusion owing to the semiconductive substrate
`from charge carriers generated further from surface 11
`being above zero Kelvin is essentially uniform in all
`45
`than the depletion regions extend. Making y smaller
`directions. So, charge carriers more than distance y
`than necessary results in reduction in the number of
`from the depletion region will be, no matter what
`charqe carriers collected in the depletion regions,
`straight-line direction they tend to diffuse in overtime
`which will reduce imager sensitivity.
`tD, forced away from surface 11 deeper into substrate 10
`The drift field for sweeping charge carriers outside
`than the depletion regions extend, before they can be
`50
`the depletion regions into the bulk to recombine, can
`collected by the nearest depletion region. These charge
`improve anti-blooming for a surface-channel CCD
`carriers are 1-(2y/w)-100 percent of the charge carri
`imager operated in the accumulation mode of anti
`ers generated in the space between the depletion re
`blooming. Blooming is the condition where charge
`gions. They will comprise 50% of the charge carriers
`carriers generated in the depletion regions responsive to
`generated between the depletion regions if Y= w/4;
`intense illumination fill those depletion regions and the
`75% if y = w/8 and 87.5% if y = w/16. And, as noted
`excess carriers spil out of those depletion regions to
`above, these charge carriers swept into the bulk are the
`other less intensely illuminated depletion regions in the
`ones chiefly responsible for background striations.
`CCD imager. The drift field pushes substantial portions
`Half of the remaining charge carriers to be consid
`of the spilled over charge carriers into the bulk to re
`ered diffuse in straight-line directions away from the
`combine, rather than permitting them to reach other
`closest depletion region and will be swept deep into the
`depletion regions.
`substrate by the drift field, not to be collected in a deple
`A CCD imager of the vertical-field-transfer type
`tion region. Of the remaining charge carriers, still oth
`embodying the present invention has been built using a
`ers have diffusion directions that have such relatively
`standard thick-oxide p-MOS process to make the sur
`small components parallel to surface 11, as compared to
`65
`face-channel charge transfer channels of its A and B
`components perpendicular to that surface, so they too
`registers. The starting wafer was 100-oriented silicon
`will be swept deeper into the substrate to recombine
`uniformly doped with p-type atoms (more particularly,
`before they can diffuse to a depletion region. Some
`
`5
`
`55
`
`Dell Ex. 1010
`Page 9
`
`

`

`4,481,522
`10
`boron) in a concentration of about 1.4.101 atoms/cm3.
`In an imager with semiconductor substrate illumi
`Prior to the steps of the standard thick-oxide p-MOS
`nated on the opposite surface-i.e., with "back-side
`process the drift field normal to the top surface of the
`illumination'-the deepest depletion regions can be
`made responsive to the full visible light spectrum; the
`wafer was created by ion implantation of further p-type
`shallowest depletion regions, primarily red responsive;
`atoms. A dose of about 1.1012 atoms/cm2 with implanta
`and the intermediate-depth regions, primarily green and
`tion energy between 150 and 200 keV was followed by
`red responsive.
`72 hour heating of the wafer at 1100' C. in an inert
`atmosphere. Subsequent to this period of heating to
`One can introduce various patterns of depletion re
`diffuse the dose into the substrate provided by the wa
`gion depth variations into the semiconductor substrate
`fer, slow cooling was permitted. (This wafer prepara
`of a CCD imager to cause it to supply output video
`10
`samples differing in color response in a prescribed se
`tion technique also reduces background striations attrib
`quence. For example, the depths of the depletion re
`utable to variations in the resistivity of the semiconduc
`tive material near the top surface of the wafer along
`gions in each column of the image register in a vertical
`field transfer type of CCD imager may vary cyclically
`directions parallel to that surface, as described by A. M.
`in a pattern recurring every three successive cyclic
`Goodman in U.S. Pat. No. 4,396,438 issued 2 Aug. 1983,
`15
`groups of electrodes. Or the depths of the d