US 5,234,846 AGrant
Method of Making Bipolar Transistor With Reduced Topography
Issue Date:1993-08-10
•10 Claims
•3 Drawing Sheets
Abstract
A vertical bipolar transistor is constructed with reduced step height by codeposition of a polysilicon base contact member and an epitaxial device layer, thereby placing the base contact below the device surface, and by the use of a doped glass layer as a dopant source for the base contact and as a dopant source to provide a continuous conductive path to the base, and as the dielectric separating the base contact from the emitter contact, and as an etch stop when forming the base implantation aperture.
Metadata
Assignee
- International Business Machines Corporation
Inventors
- Shao-Fu S. Chu
- Kyong-Min Kim
- Mei Shaw-Ning
- Victor R. Nastasi
- Somnuk Ratanaphanyarat
Application Information
Application Number:US 8765979
Filing Date:1992-04-30
Priority Date:1992-04-30
Art Unit:117
Classifications
IPC:
H01L 21265H01L 2970
Field of Search:
43725731;33515;518
Patent Drawings (3 sheets)
Description
Field of the Invention
The field of the invention is that of making bipolar transistors having reduced topography, or step height.
Background of the Invention
Prior art bipolar transistors, such as that illustrated in FIG. 8, have a number of layers of conductors and insulators in the vicinity of the active elements of the transistor. Interconnection lines must cross a step to enter an aperture above an emitter to make contact with the transistor electrode at the bottom of the aperture. The smaller the dimensions of the transistor, the greater the aspect ratio or ratio of height to width of the emitter aperture. The greater this aspect ratio, or the greater the step, the more prone the transistor is to suffer failure for a number of reasons.
One reason is that of metal penetration and stress-induced failure due to nonconformal metal/dielectric deposition on severe topography (meaning relatively high steps). Another reason is that of device degradation of transistors having small emitters due to non-uniform doping of the emitter because of interference from a sidewall in the aperture used for the implantation. Another reason is the formation of a short circuit between the emitter and the base due to nonconformal sidewall deposition. Yet another problem is variability in the size of the emitter opening recess due to the lack of an etch stop of the material (poly in the prior art) covering the base.
The art has employed a number of approaches to reduce the topography of bipolar transistors. Considerations include the interplay between chemical and mechanical considerations in the assembly of a set of layers and electrical considerations in the operation of the transistor. To improve planarity, the basic problem to be addressed is the reduction of thickness of dielectric and/or a reduction in the number of layers. One example of a prior art transistor is illustrated in FIG. 8, showing a structure formed by a two layer polysilicon (poly) process, in which a base connector is formed in poly 1 above the local surface of the wafer. The poly has to extend over the field oxide because of alignment considerations, so that the set of layers is: field oxide, poly 1, and two layers of dielectric.
Another approach is illustrated in U.S. Pat. No. 4,824,799, in which a step of codeposition of polysilicon and epitaxial monocrystalline silicon (epi) is used, with the epi filling a trench etched in a field insulator to provide the starting material for the transistor and the poly being oxidized to provide an interlayer dielectric.
Summary of the Invention
The invention relates to an improved method of forming a bipolar transistor in which greater planarity is achieved by codeposition of an epitaxial layer that will form the base and a polysilicon layer that will provide electrical contact to the base; and by the use of a doped glass layer above the base region that is used for a number of purposes. The glass layer provides a source of dopant to dope the edges of the base area; it provides a source of dopant for the polysilicon connector that forms the base contact; it contributes to the dielectric that separates the base contact from the emitter contact; it provides an etch stop during the formation of the base aperture; and it reduces the variability in the emitter opening recess due to the lack of an etch stop in the step of removing the layer of polysilicon that forms the base contact.
Brief Description of the Drawings
FIGS. 1 through 5 illustrate intermediate steps in forming the transistor. FIGS. 6A and 6B show a late step in the process and a detail of the structure formed by the step. FIG. 7 illustrates in cross-section a transistor constructed according to the invention. FIG. 8 illustrates in cross-section a prior art transistor.
Detailed Description of the Invention
Referring now to FIG. 1, there is shown in cross-section a portion of a silicon wafer used in fabricating the invention. A P.sup.- monocrystalline silicon substrate indicated by the numeral 5 has formed in it a buried conductor layer 10 that is doped N.sup.+ in the case of an NPN transistor. A top layer in the figure, denoted with the numeral 15 is an N.sup.- epitaxial (sometimes referred to as epi) layer in which the transistor is formed. The areas denoted with the numerals 20 are silicon oxide layers formed by the conventional process of etching trenches in epi layer 15 by any convenient method and then filling the trenches in with silicon dioxide (referred to as oxide) to form a substantially planar surface 3. Transistor isolation is provided by deep trenches 40, which extend down through buried layer 10 into substrate 5.
Transistors constructed according to the invention may be used with silicon on insulator wafers (SOI) in which there is a layer of insulator such as oxide below the transistor areas. In that case, the deep trenches would extend down to the buried insulator, while the shallow trenches stop short of the insulator in order to form the collector extension.
The general area in which the transistors are formed is denoted with numeral 100 and indicated by the bracket labeled 102. On the left of area 100, base region 50 is a portion of epitaxial layer 15 having an initial concentration of N.sup.-. On the right of the transistor area epi region 30, doped N+.sub.in a preliminary doping step, is the electrical contact between the collector and the buried conductor region 10. As is conventional, the emitter and base of transistor 100 will be in the top of area 50 and the collector will be formed by the combination of the bottom of area 50 (below the emitter and the base ) and a conductive path extending down to buried connector 10 and then up through vertical collector extension 30.
Referring now to FIG. 2, there is shown the results of a step of co-depositing a layer of undoped silicon that comes down as a 2000.ANG. layer of polysilicon 60 (indicated by the speckles) above oxide 20 and a second layer of epitaxial silicon above the existing epitaxial layer 15 (indicated by the numerals 55 and 35 in the drawing) and also having a thickness of 2000.ANG.. A new surface 7' has been formed above the old surface 3, which is indicated by a dotted line. The new surface is substantially planar, having a difference in thickness between the epi and poly of about 10% for both atmospheric pressure CVD and ultrahigh vacuum CVD. The temperature during deposition should be below 650.degree. C.
Referring now to FIG. 3, a layer of oxide 65, (referred to as BSG for boro-silicate glass since the dopant in this case is boron for an NPN transistor), has been formed over the entire wafer. Poly 60 and glass 65 have been patterned (by conventional photolithography and etching) and then removed above the collector region. Poly 60 has left a region on either side of epi layer 55, in a geometry in which the base surrounds the emitter. In a second step, a layer of silicon nitride (nitride) 70 has been deposited over the entire wafer covering both the base region 50 and collector region 30.
A process option is the removal only of the BSG, in order to avoid counter doping the collector region 30, and retaining the epi layer in that region in order to have a smaller step between the base region 50 and the collector region. In that case, it is necessary to have an additional mask in order to open the collector aperture, so that the use of the option depends on a tradeoff between reduced topography and process complexity.
As will be explained below, the BSG plays several roles in this method. It is put down doped with 13% boron before nitride 70 is deposited. The dopant concentration is made high enough that the BSG serves as a source of dopant for poly 60 and a portion of epi 55 below it. Since the BSG is a dielectric, it serves as part of the dielectric between the first layer of poly 60 and any conductive layer that is placed over it. An additional layer of nitride 70 will be placed on top of the BSG to prevent boron outdiffusion from the BSG.
In the prior art, it was necessary to use a timed etch with a considerable amount of overetch to ensure the complete removal of the base poly when forming the aperture for the emitter because the process did not provide for an etch stop layer. This overetch, which made a deeper step height in the emitter area, is eliminated in this invention because the BSG in this process provides such an etch stop for the process of opening an aperture in nitride 70.
In FIG. 4, a first base aperture 52 has been RIE etched (reactive ion etch) through nitride 70 to define a first base aperture through which the base dopant will be implanted in a subsequent step. BSG 65 serves as an etch stop for this step, using a conventional RIE with CHF.sub.3 /O.sub.2 as the etch gas. Because of alignment considerations, there will always be some tolerance that needs to be allowed between the base aperture and the boundaries of base region 50. The two brackets denoted with the numeral 51 indicate the extent of this "link" region, which will not be implanted in the base implantation step because of the alignment overhang of nitride 70 and must be doped by some other process.
In FIG. 5, a second base aperture has been etched through BSG layer 65 to expose the top surface 59 of the base area. The first and second base apertures have a combined sidewall that is denoted by numeral 63. The lower second base aperture is shown as being somewhat wider because the BSG is preferably removed in a wet etch step (7:1 BHF dip) that undercuts the oxide below the nitride. The area receiving the implantation is denoted by the dotted lines 57, and will become the transistor base. Those skilled in the art will appreciate that the amount of undercut may be controlled by the use of a partial RIE followed by a shorter wet etch or by the use of a RIE that stops on the silicon after penetrating all the way through the BSG.
After the BSG has been etched away, a heating step is performed in which the wafer is heated to a temperature of 900.degree. C. for a period of 20 minutes, so that the boron in the BSG diffuses out of the oxide into the polysilicon, thus forming the doping step of doping polysilicon 60 and also at the same time forms link areas 51. The BSG was put down with a nominal volume density of 5.times.10.sup.20 /cm.sup.3 and the diffusion step dopes link 51 and poly 60 to a P.sup.+ concentration.
Also, in FIG. 5, there is shown an implantation step that implants a dose of 2.times.10.sup.-- of boron at or below 10 KeV through aperture 52 to form base 57. This base implant is made after the BSG heating step and can be activated in a separate annealing step or by a combination of several high temperature steps that combine to have the same effect as a single step. Preferably, the high temperature activation step is a few seconds at a temperature of about 1000.degree. C. The nominal volume density of dopant in the link areas is 5.times.10.sup.17 /cm.sup.3 to a depth of 1500.ANG. below the BSG and the nominal volume density of dopant in base 57 is 1.times.10.sup.18 /cm.sup.3 to a depth of 1500.ANG..
The implanted region 57 is thermally reoxidized in order to ensure good surface passivation and also to prevent direct contact between the nitride and the silicon surface, which would cause high stress in the silicon.
The sidewalls 63 of aperture 52 are exposed, so that, over time, the boron dopant in glass 65 could diffuse into the base or emitter area. Therefore, in the next step a second layer of nitride is deposited over nitride 70 and is removed by a following RIE process using CF.sub.3 /O.sub.2. This step forms nitride sidewalls 75 on either side of aperture 52 that serve as a diffusion barrier to prevent the diffusion of boron out of BSG 65 into the transistor and as an electric insulator to separate the emitter and base electrodes. The sidewalls are also necessary to reduce the base aperture diameter to permit the doping of the emitter of the transistor without danger of a short circuit from the emitter contact poly 80 to base contact poly 60.
Referring now to FIG. 6A, there is shown one of the later steps in the process, in which a second layer of polysilicon 80 has been deposited filling the aperture 52. This poly is implanted with an N type dopant which is then diffused, doping the polysilicon and also diffusing out from the polysilicon 80 into base 57 in epitaxial layer 55 to form a counterdoped layer that is the emitter 56. The implant dose is a nominal 1-2.times.10.sup.16 of As at 40 KeV. and the heating cycle is a nominal 900.degree. C. for 5 minutes, resulting in a shallow emitter 56 that has a nominal thickness of 500.ANG..
FIG. 6B shows in more detail the corner of aperture 52 showing BSG 65 isolated from polysilicon 80 by sidewall 75. The area denoted by the numeral 53 is a link area that extends downwardly from BSG 65 by a greater depth because of the direct downward diffusion of boron from the BSG into the epi and inwardly by a smaller amount because of the secondary diffusion from the BSG to the poly and then to the epi 55. The amount of dopant in the BSG and the diffusion cycle are set together with the alignment tolerance for the position of aperture 52 so that there is a path from base 57 through the link area 53 into base conductor 60 that has a P.sup.+ concentration of 2.times.10.sup.20 /cm.sup.3.
FIG. 7 illustrates in cross section a completed transistor constructed according to the invention. Old surface 3 is shown with a dotted line passing horizontally through the vertical epi regions--base epi region 50 and collector extension 30. Regions 50 and 30 are separated by a dielectric member 20, formed from a deposited dielectric (SiO.sub.2). Poly 60 is disposed above insulator 20, also above the old surface 3, and abutts a base portion of epi 50 containing base 57 and emitter 56. Poly 60 has a top surface 67 that is coplanar with an epi top surface that is also the top surface of the emitter, so that it is also the device top surface. A composite dielectric member comprising BSG 65 and nitride 70 is disposed above poly 60 and also above a link region of the base portion. There is a base aperture in the composite dielectric member that is used for the implantation of the base and the diffusion of the emitter. Base aperture sidewalls ring the base aperture and provide for horizontal separation of the base and emitter. Aperture 85 on the left is cut at any convenient time, preferably at the same time as epi 90, to permit the connection of the base electrode. In this process, the connection is made with a metal interconnect, but it may be poly 1 or poly 2 if convenient. Similarly, aperture 90 on the right has been cut to provide for connection to the collector conductive lead 30, in electrical contact with the N.sup.- collector below base 57. BSG 65 is heavily doped with boron and is separated from emitter connector 80 and from emitter 56 to prevent diffusion of boron out from the BSG.
The dopant concentration in the base region is characteristic of an implanted dopant, with a peak concentration at a nominal depth of 8000.ANG. below surface 59. The dopant concentration of the emitter is characteristic of a diffused layer, having a peak concentration at surface 59, since the emitter has been formed by diffusion across that surface from poly connector 80. In the link areas, on either side of the base, the dopant gradient is also characteristic of a diffused dopant, since this dopant also was diffused across surface 63, from BSG 65 in this case. There is a continuous P.sup.+ region extending from base 57 through link 53 and into poly 65.
The emitter topography of transistor 100 is conventionally measured from surface 59 to the top surface 58 of nitride 70, and is 0.1 microns for a BSG thickness of 0.05 .mu.m and a final nitride thickness of 0.05 .mu.m. The nominal width of base aperture 52 between sidewalls is 0.5 .mu.m, giving an aspect ratio of 0.2.
For comparison, FIG. 8 illustrates a prior art transistor as illustrated in T. H. Ning, et al "Self-Aligned N-P-N bipolar Transistors", 1980 IEDM Digest pp 823-824, in which the comparable step height is 0.7 micron, provided by a base poly 60' of 0.25 .mu.m, an intermediate oxide 73 of 0.2 .mu.m that does not provide the same functions as oxide 65, and a nitride 70' (0.15 .mu.m) plus a recess formed in the emitter aperture of 0.1 .mu.m. The nominal width of the base aperture between sidewalls is 0.5 .mu.m, giving an aspect ratio of 4.
For convenience, the parameters of the main steps in the process are set out in Table I. The left column indicates parameters of the process for the formation of an NPN transistor and the right column indicates corresponding parameters for a PNP transistor. Since the geometry is the same for both types of transistor, there is no need for additional figures. Phosphorous (P) replaces boron as the dopant for the oxide, forming PSG instead of BSG and arsenic is used for the base implant.
Those skilled in the art will readily be able to devise additional embodiments of the invention, and the scope of the claims is not intended to be limited to the embodiments disclosed.
Claims
We claim:
1. A method of forming a bipolar transistor comprising the steps of: providing a planar silicon substrate having an epitaxial surface layer, with a predetermined epitaxial dopant concentration of a first polarity, disposed above and in contact with a buried conductive layer, doped with a predetermined conductive layer concentration of a dopant of said first polarity; removing a portion of said epitaxial layer to form a set of shallow trenches; depositing insulating material in said shallow trenches to form a set of insulating members extending downwardly from a wafer surface and separating predetermined portions of said epitaxial layer, whereby a base region and a collector region of said epitaxial layer, separated by one of said insulating members, are formed in a transistor region of said epitaxial layer; codepositing substantially undoped polysilicon and epitaxial silicon on said wafer surface forming thereby a layer of undoped polysilicon over said set of insulating members and a layer of undoped epitaxial silicon above said base region and said collector region and having a new substantially planar wafer surface; forming a layer of doped oxide doped with a predetermined oxide dopant of a second polarity opposite said first polarity above said polysilicon and epitaxial silicon; patterning said doped oxide to remove said doped oxide above said collector region; depositing a layer of nitride on said wafer; patterning said nitride to form a first base aperture above a base implant area in said base region, separated from said insulating members by at least one link region within said undoped epitaxial layer in said base region; etching a second base aperture through said doped oxide below said first base aperture to expose said base implant area having a base aperture surface in said undoped epitaxial layer above said base region, whereby said first and second base apertures have a combined base aperture sidewall; heating said wafer for a predetermined time at a predetermined temperature to diffuse said oxide dopant from said doped oxide downwardly into said polysilicon and horizontally into said link region, whereby said polysilicon and link region together form a conductive base connector for a transistor base; implanting a predetermined base dose of a predetermined base dopant of said second polarity in said base implant area having a transistor base aperture surface in said base aperture; oxidizing said base aperture surface of said base implant area; forming an insulating sidewall abutting said base aperture sidewall within said base aperture, whereby said insulating sidewall forms a diffusion barrier between said doped oxide and material disposed above said base region; forming an emitter contact member of doped polysilicon doped with an emitter dopant of said first polarity in said base aperture above said base region between said insulating sidewalls; and heating said wafer for a predetermined time at a predetermined temperature to diffuse said emitter dopant into said transistor base through said transistor base aperture surface to form an emitter.
2. A method according to claim 1, further including the formation of a set of deep isolation trenches overlapping a subset of said set of shallow trenches and extending a predetermined distance downwardly to provide isolation between transistors.
3. A method according to claim 1, in which said substrate is an SOI substrate having a buried insulator layer below said epitaxial surface layer and said deep trenches contact said buried insulator layer.
4. A method according to claim 1, in which said step of patterning said doped oxide is followed by a step of removing said epitaxial surface layer above said collector region, whereby both said doped oxide and said epitaxial surface layer are removed above said collector region.
5. A method according to claim 2, in which said step of patterning said doped oxide is followed by a step of removing said epitaxial surface layer above said collector region, whereby both said doped oxide and said epitaxial surface layer are removed above said collector region.
6. A method according to claim 3, in which said step of patterning said doped oxide is followed by a step of removing said epitaxial surface layer above said collector region, whereby both said doped oxide and said epitaxial surface layer are removed above said collector region.
7. A method according to any of claim 1-6, in which said steps of forming said first base aperture and said second base aperture are a RIE step.
8. A method according to any of claim 1 - 6, in which said steps of forming said first base aperture and said second base aperture are a RIE step followed by a wet etch step.
9. A method according to claim 8, in which said RIE step etches through said layer of nitride and partly through said polysilicon layer, followed by a wet etch step that etches through the remainder of said polysilicon layer.
10. A method according to any of claim 1-6, in which said step of codepositing polysilicon and epitaxial silicon is performed at a temperature of less than 650.degree. C.
Patent Citations (23)
| Patent | Date | Inventor | Cited By |
|---|---|---|---|
| UST104803 | 1984-11-01 | Horng | |
| US3647581 | 1972-03-01 | Mash | |
| US3929528 | 1975-12-01 | Davidson et al. | |
| US4157269 | 1979-06-01 | Ning et al. | |
| US4177094 | 1979-12-01 | Kroon | |
| US4554059 | 1985-11-01 | Short et al. | |
| US4693782 | 1987-09-01 | Kikuchi et al. | |
| US4731341 | 1988-03-01 | Kawakatsu | |
| US4824799 | 1989-04-01 | Komatsu | |
| US4897362 | 1990-01-01 | Delgado et al. | |
| US4902641 | 1990-02-01 | Koury | |
| US4908324 | 1990-03-01 | Nihora et al. | |
| US4952521 | 1990-08-01 | Goto | |
| US4970175 | 1990-11-01 | Haisma et al. | |
| US4971925 | 1990-11-01 | Alexander et al. | |
| US4975381 | 1990-12-01 | Taka et al. | |
| US4980305 | 1990-12-01 | Kadota et al. | |
| US4983538 | 1991-01-01 | Ito et al. | |
| US5034343 | 1991-07-01 | Rouse et al. | |
| US5037768 | 1991-08-01 | Cosertino | |
| US5051378 | 1991-09-01 | Yagi et al. | |
| JPX63-21631 | 1988-01-01 | ||
| JPX63-50791 | 1988-03-01 |