Bipolar transistor having lightly doped epitaxial collector region constant in dopant impurity and process of fabrication thereof

Abstract

A bipolar transistor has a lightly doped n-type single crystal silicon layer epitaxially grown in a recess formed in a heavily doped n-type impurity region after a selective growth of a thick field oxide layer, a base region, an emitter region and a collector contact region are formed in surface portions of the lightly doped n-type single crystal silicon layer, and the single crystal silicon layer is not affected by the heat during the growth of the thick field oxide layer, and has a flat zone constant in dopant concentration regardless of the thickness thereof.

Description

FIELD OF THE INVENTION

This invention relates to a semiconductor device and, more particularly, to a structure of a bipolar transistor and a process of fabrication thereof.

DESCRIPTION OF THE RELATED ART

A bipolar transistor is an important circuit component of a semiconductor integrated circuit device used for a communication network in the giga-hertz band. The switching speed of the bipolar transistor is mainly dominated by the thickness of the base region where the carrier passes through. The thinner the base region is, the faster the switching action is. The resistance of the emitter, base and collector regions and the parasitic capacitances coupled to the emitter/base and collector regions affect the switching speed of the bipolar transistor. These factors strongly relate to the miniaturization and the accuracy of patterning technologies used in the fabrication process of the bipolar transistor. However, a self-aligning technology between the emitter region and the base contact region makes the improvement in switching speed free from the accuracy of patterning technologies. The self-aligning technology is disclosed by Tak H. Ning et al. in “Self-Aligned Bipolar Transistors for High-Performance and Low-Power Delay VLSI”, IEEE Transactions on Electron Devices”, vol. ED-28, No. 9, September 1981, pages 1010 to 1013.

FIGS. 1A to 1 G illustrate a typical example of the process of fabricating the self-aligned bipolar transistor of the n-p-n type. The prior art process starts with preparation of a p-type silicon substrate 1 . A photo-resist ion-implantation mask (not shown) is prepared on the major surface of the p-type silicon substrate 1 by using lithographic techniques, and an area is uncovered with the photo-resist ion-implantation mask. Arsenic is ion implanted into the area, and the photo-resist ion-implantation mask is stripped off. The ion-implanted arsenic is activated in nitrogen ambience at 1000 degrees to 1200 degrees centigrade for 2 to 4 hours, and forms a heavily doped n-type buried layer 1 b.

A photo-resist ion-implantation mask (not shown) is patterned on major surface of the p-type silicon layer 1 a by using the lithographic techniques, and another area around the heavily doped n-type buried region 1 b is uncovered with the photo-resist ion-implantation mask. Boron is ion implanted into the exposed area, and the photo-resist ion-implantation mask is stripped off. The ion-implanted boron is activated in a the nitrogen atmosphere at 900 degrees to 1100 degrees centigrade for 30 minutes to an hour, and forms a heavily doped p-type buried region 1 c as shown in FIG. 1 A. The heavily doped p-type buried region 1 c electrically isolates the self-aligned bipolar transistor from another circuit component.

N-type silicon is epitaxially grown to 2 microns thick on the major surface of the p-type silicon substrate 1 a , and the p-type silicon substrate 1 a is overlain by an n-type epitaxial silicon layer 2 a . A p-type channel stopper region 2 b is formed in the n-type epitaxial silicon layer 2 a , and is merged with the heavily doped p-type buried layer 1 c as shown in FIG. 1 B.

A thick field oxide layer 3 is selectively grown to 600 nanometers thick by using LOCOS (local oxidation of silicon) technology. The growth of the thick field oxide layer 3 is carried out at 1000 degrees centigrade, and consumes long time. While the heat is growing the thick field oxide layer 3 , the boron and the arsenic are diffused from the heavily-doped p-type buried layer/p-type channel stopper region 1 c / 2 b and the heavily doped n-type buried layer 1 b , respectively, and the n-type buried layer 1 b expands as shown in FIG. 1 C. As a result, the expansion of the n-type buried layer 1 b decreases the thickness of the n-type epitaxial layer 2 a inside of the thick field oxide layer 3 .

Subsequently, phosphorous is thermally diffused into a narrow area of the n-type epitaxial layer 2 a , and reaches the heavily doped n-type buried layer 1 b . The phosphorous forms an n-type collector contact region 4 a merged into the heavily doped n-type buried layer 1 b.

Silicon oxide is deposited over the entire surface of the resultant semiconductor structure by using a chemical vapor deposition, and forms a silicon oxide layer. A photo-resist etching mask is patterned on the silicon oxide layer through the lithography, and the silicon oxide layer is selectively etched away. The silicon oxide layer is patterned into a silicon oxide mask 5 a . The n-type collector contact region 4 a is covered with the silicon oxide mask 5 a ; however, the n-type epitaxial silicon layer 2 a is exposed to an opening of the silicon oxide mask 5 a as shown in FIG. 1 D.

Subsequently, polysilicon is deposited over the entire surface of the resultant semiconductor structure by using a chemical vapor deposition, and p-type dopant impurity is introduced into the polysilicon layer. In this instance, boron is introduced into the polysilicon through an in-situ doping technique, or boron is ion implanted into the amorphous silicon layer. The boron-doped polysilicon layer is used for a base electrode as described hereinlater.

In order to isolate the base electrode from an emitter electrode, silicon nitride is deposited over the boron-doped polysilicon layer, and the boron-doped polysilicon layer is overlain by a silicon nitride layer. A photo-resist etching mask (not shown) is patterned on the silicon nitride layer, and the silicon nitride layer and the boron-doped polysilicon layer are selectively etched away so as to form a base electrode 4 b covered with an inter-level insulating layer 5 b as shown in FIG. 1 E.

A photo-resist etching mask (not shown) is patterned on the inter-level insulating layer 5 b , and has an opening over a central area of the n-type epitaxial silicon layer 2 a . Using the photo-resist etching mask, the inter-level insulating layer 5 b and the base electrode 4 b are selectively etched away so as to form an opening 5 c over the central area of the n-type epitaxial layer 2 a.

The resultant semiconductor structure is treated with heat, and the boron is diffused from the base electrode 4 b into the central area of the n-type epitaxial layer 2 a . The boron forms a graft base region 4 c beneath the base electrode 4 b . Boron or boron difluoride (BF 2 ) is ion implanted into the central area of the n-type epitaxial silicon layer 2 a , and forms an intrinsic base region 4 d as shown in FIG. 1 F.

Silicon oxide is deposited over the entire surface of the resultant semiconductor structure, and forms a silicon oxide layer topographically extending over the resultant semiconductor structure. The silicon oxide layer is anisotropically etched away without a photo-resist etching mask, and side wall spacers 5 d / 5 e are left on the inner and outer side surfaces of the base electrode 4 b . The side wall spacer 5 d on the inner side surface covers a peripheral area of the intrinsic base region 4 d , and a central area of the intrinsic base region 4 d is still exposed.

Heavily arsenic-doped polysilicon is grown on the entire surface of the resultant semiconductor structure, and a heavily arsenic-doped polysilicon layer is held in contact with the central area of the intrinsic base region 4 d . A photo-resist etching mask (not shown) is patterned on the heavily arsenic-doped polysilicon layer, and the heavily arsenic-doped polysilicon layer is patterned into an emitter electrode 4 e.

The arsenic is thermally diffused from the emitter electrode 4 e into the central area of the intrinsic base region 4 d by using a lamp annealing, and forms an emitter region 4 f.

Finally, a collector contact hole is formed in the silicon oxide layer 5 a , and a collector electrode 4 g is held in contact with the corrector contact region 4 a through the collector contact hole as shown in FIG. 1 G.

Thus, the side wall spacer 5 d causes the emitter region 4 f to be exactly nested into the intrinsic base region 4 c , and the emitter region 4 f never enters into the graft base region 4 c . However, the n-type epitaxial silicon layer 2 a is too thick to improve the switching speed. In detail, it is important to reduce the collector resistance for a high speed switching action, and the reduction of the collector resistance is achieved by a thin n-type epitaxial layer 2 a . However, if the n-type epitaxial layer 2 a is thin, the n-type dopant impurity is diffused from the heavily doped n-type buried layer 1 b into the thin epitaxial layer 2 a during the heat treatment for the thick field oxide layer 3 , and increases the dopant concentration of the n-type epitaxial layer 2 a . A lightly doped n-type region called a “flat zone” is necessary for the collector region, and the n-type dopant impurity diffused from the heavily doped n-type buried layer 1 b damages the flat zone. This results in deterioration of the bipolar transistor. Thus, the diffusion from the heavily doped n-type buried layer 1 b does not allow the manufacturer to make the n-type epitaxial layer 2 a thin, and the thick n-type epitaxial layer 2 a sets a limit on the switching speed of the prior art bipolar transistor.

A problem of the prior art process is the lithographic step is repeated twice for the heavily doped n-type buried layer 1 b and the heavily doped p-type buried layer 1 c . The prior art process is complex, and increases the production cost of the prior art bipolar transistor.

SUMMARY OF THE INVENTION

It is therefore an important object of the present invention to provide a bipolar transistor which is improved in switching speed.

It is also an important object of the present invention to provide a simple process of fabricating the bipolar transistor.

To accomplish the object, the present invention proposes to grow a lightly doped epitaxial silicon layer in a recess formed after a growth of a field oxide layer.

In accordance with one aspect of the present invention, there is provided a bipolar transistor fabricated on a silicon substrate of a first conductivity type, comprising: a heavily doped impurity region formed in a surface portion of the silicon substrate and having a second conductivity type opposite to the first conductivity type, a recess being formed in a surface portion of the heavily doped impurity region; a lightly doped epitaxial silicon layer of the second conductivity type filling the recess and having a flat zone substantially constant in dopant concentration below a first surface portion thereof; a base region of the first conductivity type formed in the first surface portion of the lightly doped epitaxial silicon layer; a heavily doped collector contact region of the second conductivity type formed in a second surface portion of the lightly doped epitaxial silicon layer contiguous to the flat zone; and an emitter region of the second conductivity type formed in a surface portion of the base region.

In accordance with one preferred embodiment of the above described bipolar transistor, the surface portion of the silicon substrate is formed by a (100) crystal plane or a crystal plane equivalent to a (100) crystal plane, and the heavily doped impurity region has a first surface defining a bottom of the recess and formed by a (100) crystal plane or a crystal plane equivalent to a (100) crystal plane. In another embodiment, the heavily doped impurity region further has a second surface defining a side of the recess and formed by a (111) crystal plane, a (110) crystal plane or a crystal plane equivalent to a (111) crystal plane or a (110) crystal plane.

In accordance with another aspect of the present invention, there is provided a process of fabricating a bipolar transistor, comprising the steps of: a) preparing a silicon substrate of a first conductivity type; b) introducing a first dopant impurity into a surface portion of the silicon substrate so as to form a heavily doped impurity region of a second conductivity type opposite to the first conductivity type; c) thermally growing a field insulating layer occupying at least an outer peripheral area of the heavily doped impurity region; d) selectively removing a central portion of the heavily doped impurity region for forming a recess therein; e) epitaxially growing a single crystal silicon in the recess so as to form a lightly doped epitaxial silicon layer of the second conductivity type; and f) forming a base region in a surface portion of the lightly doped epitaxial silicon layer and an emitter region in a surface portion of the base region.

BRIEF DESCRIPTION OF THE DRAWINGS

The features and advantages of the bipolar transistor and the process according to the present invention will be more clearly understood from the following description taken in conjunction with the accompanying drawings in which:

FIGS. 1A to 1 G are cross sectional views showing the prior art process of fabricating the self-aligned bipolar transistor;

FIGS. 2A to 2 H are cross sectional views showing a process of fabricating a bipolar transistor according to the present invention;

FIG. 3 is a graph showing an impurity profile of a collector region of the bipolar transistor;

FIG. 4 is a plan view showing the layout of another bipolar transistor according to the present invention; and

FIGS. 5A to 5 H are cross sectional views taken along line A—A of FIG. 4 and showing a process of fabricating the bipolar transistor.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

First Embodiment

FIGS. 2A to 2 H illustrate a process of fabricating a bipolar transistor embodying the present invention. The bipolar transistor described hereinbelow is assumed to be an n-p-n type. However, a p-n-p type bipolar transistor may be fabricated on an n-type silicon substrate by exchanging the conductivity types of dopant impurities.

The process starts with preparation of a p-type silicon substrate 11 . The p-type silicon substrate 11 has a major surface 11 a with crystal plane (100). Boron is ion implanted through the major surface 11 a into the p-type silicon substrate 11 at dose of 1×10 13 to 3×10 13 ion per cm 2 under acceleration energy of 250 KeV to 350 KeV. The ion-implanted boron forms a heavily doped p-type impurity layer 11 b of 0.4 micron to 1.0 micron in depth. Although the ion-implantation is carried out, no heat treatment follows the ion-implantation. The resultant semiconductor structure is illustrated in FIG. 2 A.

A photo-resist solution is spread over the upper surface of the heavily-doped p-type impurity layer 11 b , and is baked so as to form a photo-resist layer on the heavily doped p-type impurity layer. A pattern image is optically transferred from a photo-mask to the photo-resist layer by using a lithography, and forms a latent image in the heavily doped p-type impurity layer 11 b . The latent image is developed, and the photo-resist layer is patterned into a photo-resist ion-implantation mask 12 . Using the photo-resist ion-implantation mask 12 , phosphorous is ion implanted into the heavily doped p-type impurity layer 11 b at dose of 4×10 14 ions per cm 2 to 6×10 14 ions per cm 2 under acceleration energy of 550 KeV to 650 KeV, and forms a heavily doped n-type impurity region 11 c as shown in FIG. 2 B. The heavily doped n-type impurity region 11 c is as deep as the heavily doped p-type impurity region 11 b , and is corresponding to the heavily doped n-type buried layer 1 b and the heavily doped n-type impurity region 4 a.

The photo-resist ion-implantation mask 12 is stripped off. A thick field oxide layer 13 a is selectively grown to 300 to 400 nanometers thick on the major surface 11 a by using LOCOS technology. While the thick field oxide layer is being grown, the heat activates the n-type dopant impurity of the heavily doped n-type impurity region 11 c and the p-type dopant impurity of the heavily doped p-type impurity layer 11 b.

Subsequently, silicon oxide is deposited over the resultant semiconductor structure, and a silicon oxide layer topographically extends over the entire surface of the resultant semiconductor structure. A photo-resist etching mask (not shown) is provided on the silicon oxide layer by using the lithographic techniques, and the silicon oxide layer is patterned into a mask layer 13 b.

The resultant semiconductor structure is dipped into an etchant containing hydrazine or potassium hydroxide. The mask layer serves as an etching mask, and a recess 14 a is formed in the heavily doped n-type impurity region 11 c . (100) crystal plane defines the bottom of the recess 14 a . The etchant causes (111) crystal plane of single crystal silicon or the equivalent crystal plane, which are hereinbefore referred to as (111) crystal plane, to form the inner surface 11 d defining the recess 14 a . The recess 14 a ranges 0.2 micron to 0.8 micron in depth, and is shallower than the heavily doped n-type impurity region 11 c . The outer periphery of the recess 14 a is terminated at the lower surface of the thick field oxide layer 13 a . The resultant semiconductor structure is illustrated in FIG. 2 D.

The resultant semiconductor structure is dipped into an etchant containing hydrazine or potassium hydroxide. The mask layer serves as an etching mask, and a recess 14 a is formed in the heavily doped n-type impurity region 11 c . {100} crystal plane/defines the bottom of the recess 14 a . The etchant causes (111) crystal plane of single crystal silicon, which are hereinbefore referred to as {111} crystal plane, to form the inner surface lid defining the recess 14 a . The recess 14 a ranges 0.2 micron to 0.8 micron in depth, and is shallower than the heavily doped n-type impurity region 11 c . The outer periphery of the recess 14 a is terminated at the lower surface of the thick field oxide layer 13 a . The resultant semiconductor structure is illustrated in FIG. 2 D.

Lightly doped n-type single crystal silicon is grown in the recess 14 a by using a selective epitaxial growing technique, and forms a lightly doped n-type single crystal silicon layer 11 e as shown in FIG. 2 E. In this instance, the selective epitaxial growing technique is a chemical vapor deposition, and the lightly doped n-type single crystal silicon is grown from gaseous mixture containing SiH 2 Cl 2 and HCL at 700 degrees to 800 degrees centigrade. Water vapor or oxygen in the gaseous mixture is minimized, and is less than 10 −7 torr. For this reason, the deposition temperature is lowered. The lightly doped n-type single crystal silicon is grown on (100) crystal plane forming the bottom surface of the recess 14 a , and is hardly grown on {111} crystal plane. For this reason, the lightly doped n-type single crystal silicon is flat and good in crystal. The dopant concentration of the lightly doped n-type single crystal silicon layer 11 e is of the order of 1×10 15 atoms per cm 3 .

Polysilicon is grown on the resultant semiconductor structure to 150 nanometers to 300 nanometers thick by using the chemical vapor deposition, and p-type dopant impurity is introduced into the polysilicon layer at 10 18 atoms per cm 3 . Insulating material is further deposited to 100 nanometers to 200 nanometers thick over the p-type polysilicon layer, and the p-type polysilicon layer is overlain by an insulating layer. A photo-resist etching mask (not shown) is provided on the insulating layer, and the insulating layer and the p-type polysilicon layer are patterned into a base electrode 15 a and an inter-level insulating layer 13 c.

Subsequently, a photo-resist etching mask (not shown) is formed on the inter-level insulating layer 13 c , and the inter-level insulating layer 13 c and the base electrode 15 a are selectively etched away so as to form a primary emitter contact hole 14 b in the lamination of the inter-level insulating layer 13 c and the base electrode 15 a.

Heat is applied to the base electrode 15 a , and the p-type dopant impurity is diffused from the base electrode 15 a into an outer peripheral area of the lightly doped n-type single crystal silicon layer 11 e . The p-type dopant impurity forms a graft base region 11 f.

P-type dopant impurity such as boron or boron difluoride is ion implanted into the central area of the lightly doped n-type single crystal silicon layer 11 e exposed to the primary emitter contact hole 14 b , and forms an intrinsic base region 11 g inside of the graft base region 11 f . The resultant semiconductor structure is illustrated in FIG. 2 F.

Silicon oxide is deposited over the resultant semiconductor structure, and a silicon oxide layer topographically extends over the entire surface of the resultant semiconductor structure. The silicon oxide layer is anisotropically etched without a photo-resist etching mask, and a side wall spacer 15 c is formed on the inner side surfaces of the base electrode/inter-level insulating layer 15 a / 13 c . The side wall spacer 15 c defines a secondary emitter contact hole 14 c , and only a central area of the intrinsic base region 11 g is exposed to the secondary emitter contact hole 14 c.

Heavily arsenic-doped polysilicon is grown to 100 nanometers to 200 nanometers thick over the resultant semiconductor structure, and the arsenic concentration is of the order of 10 19 to 10 21 atoms per cm 3 . The heavily arsenic-doped polysilicon fills the secondary emitter contact hole, and swells into a heavily doped arsenic-doped polysilicon layer. The heavily doped arsenic-doped polysilicon layer is held in contact with the central area of the intrinsic base region 11 g exposed to the secondary emitter contact hole 14 c.

A photo-resist etching mask (not shown) is formed on the heavily arsenic-doped polysilicon layer, and the heavily arsenic-doped polysilicon layer is patterned into an emitter electrode 15 b . The arsenic is thermally diffused into the central area of the intrinsic base region 11 g by using a lamp annealing, and forms an emitter region 11 h as shown in FIG. 2 G.

The base electrode 15 a is further patterned by using the lithographic techniques, and becomes small. A photo-resist etching mask (not shown) is provided on the resultant semiconductor structure, and has an opening over the heavily doped n-type impurity region 11 c on the right side of the lightly doped n-type single crystal silicon layer 11 e . The silicon oxide layer 13 b is selectively etched away, and a collector contact hole 14 d is formed in the silicon oxide layer 13 b . The heavily doped n-type impurity region 11 c is partially exposed to the collector contact hole 14 d.

Doped polysilicon is deposited over the entire surface of the resultant semiconductor structure. The doped polysilicon fills the collector contact hole, and swells into a doped polysilicon layer. The doped polysilicon layer is patterned into a collector electrode 15 c held in contact with the heavily doped n-type impurity region 11 c as shown in FIG. 2 H.

Thus, the graft base region 11 f and the intrinsic base region 11 g are formed in the lightly doped n-type single crystal silicon layer 11 e which was grown in the heavily doped n-type impurity region 11 c . The lightly doped n-type single crystal silicon layer 11 c and the heavily doped n-type impurity region 11 c serve as a collector region of the bipolar transistor.

The lightly doped n-type single crystal silicon layer 11 e is completed during the growth of the thick field oxide layer 13 a , and is free from undesirable out-diffusion inherent in the prior art bipolar transistor. For this reason, the impurity profile in the collector region is stable, and the collector region has a clear flat zone.

The present inventor confirmed the flat zone formed in the collector region. The present inventor measured the dopant concentration in the collector region of the bipolar transistor according to the present invention and in the collector region of the prior art bipolar transistor, and plotted the impurity profiles in FIG. 3 . The lightly doped n-type single crystal silicon layer 11 e and the n-type epitaxial silicon layer 2 a were corresponding to “lightly doped layer”, and were 0.5 micron thick. The heavily doped n-type impurity region 11 c and the heavily doped n-type buried layer 1 b were represented by “heavily doped layer”.

The impurity profile of the prior art bipolar transistor was represented by plots PL 1 , and a flat zone was not observed. On the other hand, plots PL 2 represented the impurity profile of the present invention, and a flat zone was clearly formed in the lightly doped layer around 10 15 atoms per cm 3 . Thus, even if the lightly doped n-type single crystal silicon layer 11 e was only 0.5 micron thick, the flat zone was clearly observed, and the extremely thin lightly doped n-type single crystal silicon layer 11 e drastically decreased the collector resistance without sacrifice of the transistor characteristics.

Moreover, the p-type impurity region 11 b is formed through the ion-implantation of the p-type dopant impurity without lithography, and the fabrication process becomes simple.

Second Embodiment

FIG. 4 illustrates the layout of another bipolar transistor embodying the present invention, and FIGS. 5A to 5 H show a process of fabricating the bipolar transistor. A thick field oxide layer 21 a is selectively grown on a p-type silicon substrate 21 , and has an inner edge 21 a ′ defining a recess 22 a filled with a lightly doped n-type single crystal silicon layer 21 b (not shown in FIG. 4 ).

A heavily doped collector contact region 23 a is formed in the peripheral area of the lightly doped n-type single crystal silicon layer 21 b along the inner edge 21 a ′, and a collector electrode 24 a is held in contact with the collector contact region 22 a.

A graft base region 23 b is formed inside of the heavily doped collector contact region 23 a , and a base electrode 24 b is held in contact with the graft base region 23 b . The base electrode 24 b is electrically isolated from the collector electrode 24 a by means of an inter-level insulating layer 25 a (not shown in FIG. 4 ).

An emitter region 23 c is formed inside of the graft base region 23 b , and an emitter electrode 24 c is held in contact with the emitter region 23 c . An inter-level insulating layer 25 b (not shown in FIG. 4 ) electrically isolates the emitter electrode 24 c from the base electrode 24 b . The collector electrode 24 a , the base electrode 24 b and the emitter electrode 24 c are self-aligned with the thick field oxide layer 21 a , the collector electrode 24 a and the base electrode 24 b , respectively.

The bipolar transistor shown in FIG. 4 is fabricated as follows. The p-type silicon substrate 21 is firstly prepared, and (100) crystal plane forms the major surface 21 b of the p-type silicon substrate 21 .

Boron is ion implanted through the major surface 21 b into the p-type silicon substrate 21 , and forms a heavily doped p-type impurity layer 21 c . The heavily doped p-type impurity layer 21 c is 0.4 micron to 1.0 micron in thickness. The ion-implantation of the boron is carried out under the same conditions as the first embodiment.

Subsequently, phosphorous is ion implanted into the heavily doped p-type impurity layer 21 c , and forms a heavily doped n-type impurity region 21 d as deep as the heavily doped p-type impurity layer 21 c . The ion-implantation of the phosphorous is carried out under the same conditions as the first embodiment.

The thick field oxide layer 21 a is selectively grown to 300 nanometers to 400 nanometers thick by using LOCOS technology, and the ion-implanted boron and the ion-implanted phosphorous are activated with heat during the growth of the thick field oxide layer 21 a . The resultant semiconductor structure is shown in FIG. 5 A.

Using the thick field oxide layer 21 a as an etching mask, the heavily doped n-type impurity region 21 d is selectively etched away by using an anisotropic dry etching technique, and forms a recess 22 a in the heavily doped n-type impurity region 21 d . The anisotropic dry etching does not control the crystal plane of the inner surface 21 e of the heavily doped n-type impurity region 21 d . The recess 22 a is shallower than the heavily doped n-type impurity region 21 d by 0.2 micron to 0.8 micron, and the resultant semiconductor structure is shown in FIG. 5 B.

The lightly doped n-type single crystal silicon is epitaxially grown in the recess 22 a , and the dopant concentration of the lightly doped n-type single crystal silicon is of the order of 1×10 16 atoms per cm 3 . The selective epitaxial growth is carried out by using chemical vapor deposition as similar to the first embodiment. However, the lightly doped single crystal silicon is grown on not only the bottom surface but also the inner surface 21 e , and the lightly doped single crystal silicon layer 21 b has a convex portion 21 f along the inner edge of the thick field oxide layer 21 a as shown in FIG. 5 C. However, the height of the convex portion 21 f is not greater than 0.1 micron.

Phosphorous-doped polysilicon is deposited to 100 nanometers to 200 nanometers thick over the resultant semiconductor structure by using a chemical vapor deposition, and the phosphorous concentration is of the order of 10 19 atoms per cm 3 . The phosphorous-doped polysilicon is patterned into the collector electrode 24 a , and the inter-level insulating layer 25 a of 200 nanometers thick is deposited over the resultant structure. Subsequently, the collector electrode 24 a is heated, and the phosphorous is thermally diffused from the collector electrode 24 a into the peripheral area of the lightly doped n-type single crystal silicon layer 21 b . The phosphorous forms the collector contact region 23 a as shown in FIG. 5 D.

Subsequently, a part of the inter-level insulating layer 25 a is etched away, and forms a base contact hole 22 b to which the lightly doped n-type single crystal silicon layer 21 b is exposed. P-type doped polysilicon is deposited to 150 nanometers to 300 nanometers thick over the resultant semiconductor structure, and the dopant concentration of the p-type doped polysilicon contains the p-type dopant impurity of the order of 10 18 atoms per cm 3 . The p-type doped polysilicon layer 26 a is held on contact with the lightly doped n-type single crystal silicon layer 21 b through the base contact hole 22 b . The resultant semiconductor structure is covered with an insulating layer 26 b of 100 nanometers to.200 nanometers thick as shown in FIG. 5 E.

The insulating layer 26 b and the p-type doped polysilicon layer 26 a are patterned into the inter-level insulating layer 25 b and the base electrode 24 b , and a primary emitter contact hole 22 c is formed in the lamination of the base electrode 24 b and the inter-level insulating layer 25 b.

Heat is applied to the base electrode 24 b , and the p-type dopant impurity is diffused from the base electrode 24 b into the lightly doped n-type single crystal silicon layer 21 b , and forms the graft base region 23 b.

Boron or boron difluoride is ion implanted through the primary emitter contact hole 22 c into the lightly doped n-type single crystal silicon layer 21 b , and is activated through a heat treatment. As a result, an intrinsic base region 23 d is formed inside of the graft base region 23 b as shown in FIG. 5 F.

Silicon oxide is deposited over the resultant semiconductor structure, and a side wall spacer 25 c is formed from the silicon oxide layer on the inner surface of the lamination of the base electrode 24 b and the interlevel insulating layer 25 b by using an etch-back technique. The side wall spacer 25 c defines a secondary emitter contact hole 22 d . Heavily arsenic-doped polysilicon is deposited to 100 nanometer to 200 nanometers thick, and arsenic concentration is of the order of 10 19 to 10 21 atoms per cm 3 . The heavily arsenic-doped polysilicon layer is patterned into the emitter electrode 24 c , and the arsenic is thermally diffused from the emitter electrode 24 c into the central area of the intrinsic base region 23 d through a lamp annealing. The arsenic forms the emitter region 23 c as shown in FIG. 5 G.

Finally, the base electrode 24 b and the inter-level insulating layer 25 b are partially etched away, and becomes small.

Thus, the collector contact region 23 a , the graft base region 23 b and the intrinsic base region 23 d are formed in the lightly doped single crystal silicon layer 21 b , and the lightly doped n-type single crystal silicon layer 21 b is formed after the growth of the thick field oxide layer 21 a , and the n-type dopant impurity is less diffused from the heavily doped n-type impurity region 21 d into the lightly doped n-type single crystal silicon layer 21 b . As a result, a flat zone takes place in the lightly doped n-type single crystal silicon layer 21 b , and collector resistance is decreased without sacrifice of the transistor characteristics.

The collector electrode 24 a , the base electrode 24 b and the emitter electrode 24 c are respectively self-aligned with the thick field oxide layer 21 a , the collector electrode 24 a and the base electrode 24 b . For this reason, the bipolar transistor implementing the second embodiment is suitable for an ultra large scale integration.

If the edge of the recess is directed to [110], a facet takes place in the periphery of the lightly doped n-type single crystal silicon layer, and cancels the convex portion. Therefore, a flat surface is created in the second embodiment.

Although particular embodiments of the present invention have been shown and described, it will be obvious to those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the present invention.

For example, if the recess 14 a is formed in the heavily doped n-type impurity region 11 c through a dry etching, (110) crystal plane or the equivalent crystal plane forms the inner surface 11 d , and the single crystal silicon layer 11 e is similarly grown.

The heavily doped n-type impurity region 21 d may be formed after the growth of the thick field oxide layer 21 a.

The heavily doped n-type impurity regions 11 c / 21 d may be thicker than the heavily doped p-type impurity layer 11 b / 21 c.

A p-n-p type bipolar transistor may be formed through one of the processes described hereinbefore.

Claims

1. A bipolar transistor fabricated on a silicon substrate of a first conductivity type, comprising:a heavily doped impurity region formed in a surface portion of said silicon substrate and having a second conductivity type opposite to said first conductivity type, a recess being formed in a surface portion of said heavily doped impurity region; a lightly doped epitaxial silicon layer of said second conductivity type formed in and filling said recess and having a flat zone substantially constant in dopant concentration in a direction of a depth of said lightly doped epitaxial silicon layer below a first surface portion thereof; a base region of said first conductivity type formed in said first surface portion of said lightly doped epitaxial silicon layer, said base region including a graft base region formed in a region in said lightly doped epitaxial silicon layer, and an intrinsic base region formed inside of said graft base region; a heavily doped collector contact region of said second conductivity type formed in a second surface portion of said lightly doped epitaxial silicon layer contiguous to said flat zone; and an emitter region of said second conductivity type formed in a surface portion of said base region.

2. A bipolar transistor fabricated on a silicon substrate of a first conductivity type, comprising:a heavily doped impurity region formed in a surface portion of said silicon substrate and having a second conductivity type opposite to said first conductivity type, a recess being formed in a surface portion of said heavily doped impurity region; a lightly doped epitaxial single crystal silicon layer of said second conductivity type formed in and filling said recess and having a flat zone substantially constant in dopant concentration below a first surface portion thereof; a base region of said first conductivity type formed in said first surface portion of said lightly doped epitaxial silicon layer, said base region including a graft base region formed in a region in said lightly doped epitaxial single crystal layer, and an intrinsic base region formed inside of said graft base region; a heavily doped collector contact region of said second conductivity type formed in a second surface portion of said lightly doped epitaxial silicon layer contiguous to said flat zone; and an emitter region of said second conductivity type formed in a surface portion of said base region.

3. The bipolar transistor as set forth in claim 1, in which said surface portion of said silicon substrate is formed by (100) crystal plane or a crystal plane equivalent to said (100) crystal plane, and said heavily doped impurity region has a first surface defining a bottom of said recess and formed by said (100) crystal plane or said crystal plane equivalent to said (100) crystal plane.

4. The bipolar transistor as set forth in claim 3, in which said heavily doped impurity region further has a second surface defining a side of said recess and formed by (111) crystal plane, (110) crystal plane or a crystal plane equivalent to said (111) crystal plane or said (110) crystal plane.

5. The bipolar transistor as set forth in claim 1, in which said lightly doped epitaxial single silicon crystal layer has an outer periphery substantially aligned with an inner periphery of a field insulating layer selectively grown on said silicon substrate.

6. The bipolar transistor as set forth in claim 5, in which said collector contact region is formed in an outer peripheral area held in contact with a collector electrode, and said base region is formed in a central area inside of said outer peripheral area.