Method for fabricating a charge coupled device

The invention relates to formation of charge coupled devices and, more particularly, the invention relates to fabrication of two-phase charge coupled devices.

Owing to the cost competitive nature of charge coupled device (CCD) manufacture, it is desirable to provide a high yield, low cost process. Accordingly, two-phase CCDs, having the advantage over three-phase CCDs in manufacturability, speed and resolution, are being manufactured.

One conventional structure of a two-phase CCD is shown in FIG. 1 . FIG. 1 depicts a cross-sectional view of a portion of a semiconductor device assembly 20 for a two-phase CCD coupled to signal sources 23 and 24 . Signal source 23 and signal source 24 are phase one and two, respectively, and are coupled to gates 21 and 22 with lines 18 and 19 , respectively.

Conventionally, semiconductor device assembly 20 is formed on a p-type single crystalline Silicon substrate 10 . A n-type implant is used to form buried channel layer (bccd) 11 in substrate 10 . Next, a dielectric layer 14 is grown or deposited, after which a conductive layer 16 is deposited over dielectric layer 14 . Layers 14 and 16 are etched to form spaced-apart device stacks. After forming gate stacks 21 , n-minus wells 13 are formed by implanting a p-type material. Accordingly, bccd 11 comprises n regions 12 and n-minus wells 13 in an alternating sequence. Next, a dielectric layer 15 is formed, conventionally by thermial oxidation whereby a portion of layer 21 is consumed. Next, conductive layer 17 is deposited, and gates 22 are formed. Unfortunately, in the process of removing selected portions of conductive layer 16 , remnants, known in the semiconductor industry as “stringers” or “slivers”, are sometimes left behind. These remnants can cause gates 21 to be electrically shorted to one another. Because respective gates 21 are to be electrically separate from one another for a two-phase CCD of the configuration shown in FIG. 1, and because shorting due to stringers or slivers is not typically repairable after forming gates 22 , CCD yield is adversely affected.

To address this problem, an alternative structure and process for fabrication of a two-phase CCD is described with reference to FIG. 2 . FIG. 2 depicts a cross-sectional view of a portion of a semiconductor device assembly 30 for a two-phase CCD. In the alternative structure, gates 21 are connected to one another and to signal source 23 by lines 18 A, and gates 22 are connected to one another and to signal source 24 by lines 19 A, as illustratively shown in FIG. 2 . Accordingly, shorting together of gates 21 from layer 16 remnant formation is not at issue in device assembly 30 .

Marks, not shown, may be made on a semiconductor wafer for registration between the wafer and lithographic equipment, such as a stepper or a step and scan. However, this approach is often dependent on metrological limitations and may require having machine associations with respect to a particular piece of lithographic equipment, and this limits fabrication throughput when the associated piece of lithographic equipment is not readily available. Accordingly, if gates 21 are misaligned, yield is reduced possibly owing to charge trapping or malformation of the device, the CCD may operate in a less than optimal manner.

Therefore, a need exists in the art for a more robust process for forming this alternative structure for a two-phase CCD.

The present invention provides a process for forming a portion of a semiconductor device on a substrate. More particularly, a buried channel layer is formed on the substrate, and a sacrificial layer is deposited over the buried channel layer and patterned to provide spaced-apart rows. A mask is formed extending part way between the rows, and wells are formed in the buried channel layer between the rows using facing sides of the rows and the mask. The mask is removed, and gate stacks between the rows are formed prior to removing the rows. Another mask extending part way between the gate stacks is formed, and other wells are formed in the buried channel layer between the gate stacks using facing sides of the gate stacks and the other mask prior to removing the other mask. Other gate stacks are formed between the existing gate stacks.

Accordingly, it should be appreciated that an aspect of the present invention provides for topological alignment or “self-aligned” formation of wells in a buried channel layer. Because these wells are formed in a self-aligned manner, there is less chance for misalignment.

The teachings of the present invention can be readily understood by considering the following detailed description in conjunction with the accompanying drawings, in which:

FIG. 1 depicts a portion of a conventional semiconductor device assembly in cross section;

FIG. 2 depicts an alternative embodiment of a portion of a conventional semiconductor device assembly in cross section;

FIGS. 3 through 6 depict progressive cross-sectional views of exemplary portions of a process for fabricating a semiconductor device assembly in accordance with aspects of the present invention; and

FIG. 7 depicts a three-dimensional cross-sectional view of an exemplary portion of the in process semiconductor device assembly with line drawn contacts and conductive lines.

To facilitate understanding, identical reference numerals have been used, where possible, to designate identical elements that are common to the figures.

Referring to FIG. 3, there is shown a cross-sectional view of an exemplary portion of an in process charge coupled device (CCD) assembly 100 . Substrate 101 , which may be formed of an n- or p-type formed slice of single crystalline silicon, has a channel layer 102 . Channel layer 102 will be of an opposite type of conductivity from substrate 101 , namely either n- or p-type

For purposes of clarity, a p-type substrate will be described with an n-type channel layer; however, those of ordinary skill in the art will appreciate that a n-type substrate with a p-type channel layer may be used.

Channel layer 102 is implanted to provide a buried channel layer (bccd). This process implants an n-type material with a concentration and power sufficient to make bccd layer 102 n-type to a depth sufficient to accommodate n-type regions for operation of CCD gate stacks.

Sacrificial layer 103 is formed over bccd layer 102 . Notably, a native oxide layer may form on bccd layer 102 during processing. Thus, it should be understood that sacrificial layer 103 need not be in direct contact with bccd layer 102 . Sacrificial layer 103 is patterned and etched to remove portions leaving spaced-apart sacrificial layer portions 103 . These spaced-apart portions should be made of a material selective to a material employed for subsequent gate formation, namely to facilitate removal of sacrificial layer portions 103 while leaving gates, and should be sufficiently tall to block implantation, a conventional height to block implants is approximately 0.2 to 1.0 microns (2,000 to 10,000 Angstroms) dependent on implant material and power and sacrificial layer material. Sacrificial layer 103 may be grown or deposited single layer or a combination of layers for this purpose; for example sacrificial layer 103 may be an oxide, nitride, carbide, polycrystalline silicon (“poly” or “polysilicon”), or a combination thereof.

After patterning and etching of sacrificial layer 103 , remaining sacrificial layer portions bracket or mark channel regions 110 . Channel regions 110 have a middle section as indicated by dashed lines 109 . This middle section is limited by width needed form-type regions for operation of CCD gate stacks, as well as width needed for forming a barrier to limit electrical conductivity between n-type regions.

Referring to FIG. 4, resist 104 is deposited and patterned to provide a barrier mask for implantation 150 . A light concentration of a p-type species, namely a p minus species, is used for implantation 150 to form wells 105 . This implantation may be through any native oxide formed in channel region 110 . Wells 105 provide electrical conductivity barriers between n-type regions of bccd layer 102 and have an n-minus type electrical conductivity characteristic. By way of example and not limitation, wells 105 may be approximately 1 to 1.5 microns deep.

Side or edge 112 of resist mask 104 faces side or edge 111 of sacrificial layer 103 . Edge 112 extends into a middle section shown in FIG. 3 . Edge 111 is an important edge for alignment, as space currently occupied by sacrificial layer 103 will be occupied by at least a portion of a gate stack. Accordingly, formation of well 105 is self-aligned to edge 111 on one side and to edge 112 on an associated side. After implantation 150 , resist mask 104 is removed.

Referring to FIG. 5, gate dielectric layer 106 is formed over wells 105 in channel region 110 shown in FIG. 3 . Dielectric layer 106 material may be selective to sacrificial layer material 103 , so it remains after removal of sacrificial layer 103 . An oxide or a nitride may be used for dielectric layer 106 . For example, an oxide may be thermally grown or deposited, and more particularly may be thermally grown or deposited on a native oxide, if one exists, in channel region 110 . For substrate 101 a silicon, thermal growth means formation of silicon dioxide.

Gate conductive layer 107 is formed over sacrificial layer 103 and gate dielectric layer 106 . Though layer 107 is shown as being a substantially conformal deposition, it need not be. For example, a conductive fill material may be used to eliminate the need to remove portions of layer 107 . However, in this illustrative embodiment, gate conductive layer 107 is a substantially conformal conductive material, such as conductive polysilicon.

As illustratively shown in FIG. 5, a layer of resist 108 may be spun on or otherwise deposited and then patterned. Gate conductive layer 107 is isotropically etched using etch mask 108 to remove portions of gate conductive layer 107 . It is important to clear gate conductive layer 107 such that it does not overlap sacrificial layer 103 . Accordingly and alternatively, to remove portions of gate conductive layer 107 over sacrificial layer 103 , chemical-mechanical polishing or mechanical polishing may be used. After removal of portions of gate conductive layer 107 , etch mask 108 may be removed.

Sacrificial layer 103 is removed by selective etching, namely layer 103 is removed while leaving at least a portion of a remaining portion of gates 107 . During this etching process, a portion along edge 116 of gate dielectric layer 106 may be removed undercutting edge 151 of conductive gate layer 107 .

Referring now to FIG. 6, remaining portions of layers 106 and 107 mark or bracket channel regions 120 . Resist layer 113 is deposited and patterned to provide an implant mask for implantation 150 . Implantation 150 forms additional wells 105 . Because substrate 101 is bccd layer 102 is n-type in this example, a p minus type implant is used to form n-minus type wells 105 . Notably, edges 117 of resist mask 113 extend into middle sections, as indicated by dashed lines 109 , of channel regions 120 . Accordingly, these more recently formed wells are aligned to at least a portion of edges 151 of gate conductive layer 107 , marking channel region 120 on one side and to at least a portion of edges 117 marking channel region 120 on an associated side.

Accordingly, at this point in this description it should be appreciated that wells 105 are aligned to respective sides of gate stacks 132 , though those wells 105 under gates stacks 132 were aligned using sides of sacrificial layer 103 . Wells 105 bracket n-type buried channel regions 125 of bccd layer 102 , as shown in FIG. 7 .

Referring to FIG. 7, dielectric layer 126 is formed over bccd layer 102 in channel regions 120 (indicated in FIG. 6 ). Layer 126 may be a thermally grown oxide, such as silicon dioxide, and may consume a portion of bccd layer 102 or a portion of gate conductive layers 107 or portions of both. Gate conductive layers 207 are formed over, as well as between, dielectric layers 126 . Gates 207 may be a polysilicon conformally deposited and then chemically-mechanically polished back to provide gate stacks 134 . Gates of stacks 132 and 134 are electrically separate from one another by dielectric layers 126 .

Rows of gate stacks 132 and 134 may be connected to conductive lines 133 and 124 for coupling to signal sources 23 and 24 , respectively. Contacts and conductive lines may be formed in a single metal level or a double metal level in a known manner.

A CCD formed in accordance with the present invention may be formed in an array. An image may be focused on this array. Such a CCD array is useful to sense this image for converting it from one form of energy into an electrical representation thereof. Accordingly, such a CCD array may be used in a digital camera, digital video camera, and like devices used for capturing one or more images.

Although various embodiments which incorporate the teachings of the present invention have been shown and described in detail herein, those skilled in the art can readily devise many other varied embodiments that still incorporate these teachings. For example, an n-type substrate 101 may be used. In which embodiment, bccd layer 102 is formed by implanting a p-type species, and wells 105 are formed by implanting an n-minus species to form p minus wells 105 .