<title>Virtual phase CCD imagers for scientific applications</title>

Abstract

A family of virtual phase (VP) CCD array image sensors for various scientific applications was designed, fabricated and tested. All of them share the common concept of "2.5-phase" photosensitive cell combining known benefits of "1.5-phase" VP CCDs with extended functionality (bi-directional charge transfer and inherent antiblooming) and less demanding fabrication process. Organization and main photoelectric parameters of sensors are presented. HYSTORICAL BACKGROUNDFirst VP CCDs have been invented in early 80's' as an alternative to all-polysilicon gates devices. This invention marked a great step toward improvements in such crucial CCD parameters as QE and dark current. In first time on a front-side illuminated devices, due to large area free from polysilicon gates, a 50% QE has been achieved, and operation at pin condition over whole device area gave dark current as low as 0.5nA/cm2. However, "classic" virtual phase, i.e. a structure with a single polysilicon level, showed some drawbacks as well, primarily one-directional charge transfer and difficulty of antiblooming. And, taking into account complex process of the fabrication of the "1.5-phase" devices, associated with the multiple high-precision ion implantation required to form four differently doped regions, it is no surprise that large-area scientific VP CCDS are rare at market Further development of VP concept is a new CCD pixel design2 consisting of 3 regions. Two of them have a structure identical to that of conventional 3-phase buried channel COD with polysilicon gates, and the third one is uniformly doped virtual well region with shallow p-type virtual gate layer connected to the stop-channel (Fig. 2). This structure operates in a mode that can be called "2.5-phase" (two gates clocked and one -"virtual"), unlike original VP with the mode corresponding to "1.5-phase" (one electrode clocked and one -"virtual").The structure proposed is free from one of the most biting drawback of 1.5-phase device -inherent onedirectional charge transfer detennined by incorporated potential barriers both in VP region and under clocked gate. In 2.5-phase device, charge transfer direction is determined by gates clocking like in conventional 3-phase CCD. This feature provides a possibility of operative choice of transfer direction and expands device functionality.Another important advantage of two-gate structure is a simplicity of recombination antiblooming3 since there are all necessary components, and no additional gate is required.An important feature of this structure is less demanding fabrication technology which provides higher yield and consequently lower device price. The key moment here is absence of self-aligned potential barriers in clock and virtual regions. Meanwhile this structure preserves high QE and low dark current inherent to VP CCDs, the latter being guaranteed by proper doping of clocked and virtual regions providing potential relief for charge accumulation with surface potential pinned over whole device area.As to QE and spectral range, VP CCDs have no...