Zero Loss Transfer Across Gaps in a CCD by RH. KRAMBECK, Wen a chorye-coupled device in made with a single layer of metlica tion, anjasent eleroos ute placed seeral micrometer apart. Ae @ rant tore ming be some dfety Sn moving the chase fam one eetode to another thi poper it wil be whos ta, for any aubsrte atrial Moran wie range of srkrface chars far which complete traafer can te aohieead repartee of elaine swparoion. 11 sell further be shown ‘at fir poubseate woth doping of Tew than 10"/ou, the natwrat Chara Jour in a goed guaity thermally grow layer of SiO, fa alvaye ff the appropriate aie awd motte ts enoure epee anaes. Thne- fora, fo nip fabrewsion of e CCD ih ne ayer of ste te substrate alert fe te approprite choi. Tn chargeenpied daviee,' chorge is transferred from a potential ‘yall under me wealistion to @ deeper potontal well under an djscent Iota, Sie the mitals eanot ton one rufber, his poees involves Passing through an interscetrce yop IT the potential pattor is such that there isa potential baron ct pocencial wel in the epueo wt th cof transfer, then wompiele eharge teanefer i impossible. Tf bvo a of mctalzction are ved (while mpatatel by an insulting layer of LOOD 33, ro boner o wel ns: (om This 2reuies mro eom= plex ceenology eo it would be devine make a CCD with ore layer of retaliztion, Tm ehis paper ths efecto jotarfars charge and substrate doping potenti we eae barr furan sil be analyzed, H wil be shown hat, for any’ given substrate ping and drive voltages, there is an Sitesi change Tv whish neither well nor Bari frm. "The caloula- tions in this payer will aime a infntelslome gap whi wil ensure ‘hat no yin lose will eral fom small variations in gap width or sre sarge magi S170 ce ott avwsmne rROUNICAL JURA, ROR aH ‘The abjective of thi analyst wil he to find che eonditons whieh will omit compte ehangs transi, Tis therfore neeesany to resting {he alate of the CCD at thy sl af the eraser of only the ieyeat ONE. “het con be atooramodatod. Tht i, i the hewest ONE. is tenner “sithout lore, than any’ pie of information could be transfered with Zo lows uf change. Our objective then ie to find frcecurice density ws postion at Ue snd of tonsfer. 1 ne nother pone nora minim 2t free charge density iv Die spaoe, then the space x wot Pteroring Fish toaster "To ind fce-camienderaity un he stave, we must Sind val euuee charge donsiy an enbaat fe large density. he Ast pat ofthe problem is to ind chatge densi ve eeilon at the end of trensfor. {This simplified by the fact det ws tie ud of transfor ofthe larga posible ONE, the poiieeadustur sichuor is an equipotental (the Iavement of foe eange will na stop uni an equipotent achive) "The quometry of the inerleele spaoe is shown in ig. 1. Jost below the senicondsetor sirius the field ie wafer beouuse the ‘Surfer sa enuipotersial, The dxplcement felis ven by b= ViNeFa @ where i fa eeaoondura dpiog, «(8 the cemioondustor deters Constant, ¥ Sa the wurlaye ewventil, and g is tee leetnnie charg. Tost alae the steoe he Fl De = Dg On @ where ne Hho frewelectron devsity (en) nl Qn, is the densi of Charge! states ether inthe ovide ur ab the semiocnduetor surieoe fprispe subsirwe has bron assumed with eletrone ge froc enti) torn equation (1) and @) an = Py, — VIREVG+ On 8 Since Qu, vied Fats constants difereseos in elation density from fone poi to another are areetly vated to changes in. by Du an a ‘This means that if Py, watts monotonically from de region under cone sleet to the region wider the ajaoeat electrode, tes ellen dni aso varies monotonieslly andr harvicr orwell ests, Tn Fig. 2 Fig 1 Coes mse fw channel rae hg Ue ites sp possible vations of B,, rth pstiow are eon, Rast, £3 Fig. 20 VF and VV, whens Py and Fae the voagen epic tu th exo metals. Ia this ase, Dy it pine uedae Bota rete’ hit drops ta zor in betwern for alent ree separation). This mrs Uloetron donsty 3 lower inthe spare between tke eletees Ti i is unver oir of them, and 96 a reel a baie wt bs holding ale aege. Seeund, in Pig. BoP > Vand P > P. Hore Bi menace ‘usdet ear locale bs again any in tke ep. This means leeteon lens ie higher in Se ace thaa under ithe lest, m0» potential acct gp oe hay aie US aS me See well mut be present. Hinally, in Fig, 20. & Vy and ¥ = Wa. Hore Da, 2.0 under tote t, Dy, % 0 under melul 2, nnd D, ‘where in haven, Thi isthe tation desiced. Tes therfore noe ‘ofl ly what wake of Q, will pane le susfac potential at the end bf trols Lo be in betroen Ye to mata voltages Tat secure rth 1 i giving ils charge to mst 2. The te ‘end of taanaer freecootney Menrty rer mcta 1 i geo. ‘Therefore rows equation 20. ® Wa VENGPS. a ee VINE Pa ® Aven evap, wits f= Fo, ¢ 10" Demand Fy = 1 vol, "Thi seriion ee Any Teer value would give ine Coa barren ‘To reson oration ofa rl, we hove V2. Unde levis 2 Day BO ‘Theneiore @. 2 VINGE— o fee largest aloe Qa ing used, Ba, — Onn F = Vy Ale, frum enoetion (8 ‘Thereore where it the oxide diletve wonetant, The maximum value of Qi, is Chron 0. 5 Venton + 50 — 19 an Tring Vy = 10 volts and the sare munbene as above forthe other parameters giver 520 10"em 7 equations (8) and (00) speci the allowable range for (2, wich & the res nseced, ‘The provi setion show thal eosin ange of Qu, ppp for unt" given setrate doping wri pair 0 devine valle Tes section tis abtasnability ofthis range wil be discussed The range 3 hewn graphisely 1g. 3 for tho deving vecagae V0 — 1 volt and ¥) = Lovolt 1 tures out that for a teemmaly grown oxide on silicon Qi in bP a “EE SGAGaNOT md mei tae sae ate sir = SI7A sme mE svsniae Teenwteat Format, BRCRUSEE Ent the oxide ie normally in tho ring 10" to 6 x 10)"fom. Therefore for a ype eubetate any dosing wake 10°/ea? wil ensure compote tanter, or oiher sulairaes and oxidee come doping must be ine troduced, Probsbly dhe eat wy’ would he & uniform ion implant fhceca 20 the total charge including oxide charge lcs in the alioned. range “An important pont ubuul Fig 8 isthe width of the allowed range For drive voltage of J volt sod 10 yile (he allowed rango of cherge fetends over st lewit ono ner of magnitude for any substrate doging ‘blow 19!"/om which ensures thal » proper amant of change can be btained cyen with rulnluatil slcetoalioe veriatiors in charge. Te ‘Goo tne Thal aging eect axe unllaly to ebange the amount of Charger ton vals tide of te desired an, Tin the uualyie it wae aasoned there i sisiont ten for complet crane tv agear. Aa appropriate quewion is Iw rue Tonge transfor takes wo separates the rata, RJ. Strain rl N. 1 Searynr neve how that uw CCD te rata of eenie low inversely proportional 2 The ses of dhe enpatannes betoeen surfyce and wibetrate an between sunfeo and meta. Sing shit um iz roue higher when the suri it covered hy tinels, Uw low eapacitares gap ‘ill no guionatly Ineente umsafer time a€ Jong oe the gyp ix shorter shan the metel irom the abo kesion$t am bp en chat the simplest sey to beatin complete trurfer ia charge-coupled davies, with w thermal SiO, layer used inmiator, ie tne a8 Pubetzate caatril p-type sileon vith a doping of les than 10" im ‘e ean be forehercopeluded thw for wher subszate zeatevals and for other inmuasvors a apprepriale chorgn mange alvays exis, and that this charge monee nw abuaized by ion impntstie eer the titi sarfuor af (veal ithout use of « phatogmibie nakag rep. Selec npr of th region betroen the motala and the regions tnder te eager ef the mszaleeosld ep eliminate porential bor ind wells, However, beatse of fringe Telds, thie ean be scbioved tly by exten of He kl of toe change denoibution for a spectie Aistiner under each metal, but thie would be 4 mone dient wy to nohiowe complete tne. Summavng, ik lnwld lear that potential barvore and eli in the spaces botucen elerandes in ebargeovvplel deviees ean be "inate msl henefve the possibility of incomplete oharge rane ran tase macs aun should sob be cited as a reat far ming lle abornally doeely on wing fen layers a metalization. Por mort apoliasions future CCD should be roade on peigie sbeteates with Tess chan Wem ding snes the proper chany neatly obtain 2, Sui Be a eyo No pe wk