METHOD TO REDUCE SEEDLAYER TOPOGRAPHY
IN BICMOS PROCESS
The present invention relates generally to the fabrication of a BICMOS device, and
5 more specifically relates to a method for reducing the seedlayer topography in a BICMOS
The addition of germanium to silicon technologies to form silicon germanium
(SiGe/SiGeC) devices has created a revolution in the semiconductor industry. Germanium
is added to silicon to form high-performance heterojunction bipolar transistors (HBTs),
10 which can operate at much higher speeds than standard silicon bipolar transistors. SiGe
HBTs can operate at speeds previously attainable only with gallium-arsenide and enjoy the
advantage of being manufactured in existing silicon fabs using standard tool sets. SiGe
devices can also be integrated into standard CMOS logic technologies, resulting in the
integration of high-performance analog and RF circuits with dense CMOS logic.
15 The performance and integration capability of SiGe devices has enabled the
production of a wide range of new products for wireless and wired communications, high- .
speed-test, and disk-drive applications. SiGe products include chip sets for wireless cellular
handsets and base stations, as well as wireless local area network and high-speed/high-
capacity wired network applications.
20 One technique for growing epitaxial SiGeC base layers in bipolar devices is
referred to as Differential Epitaxial Growth (DEG) or Non-Selective Epitaxial Growth
(NSEG). One of the issues with DEG involves uniformous nucleation. SiGe/SiGeC does
not nucleate well on the field isolation oxide, i.e., on the shallow trench isolation (STI).
One way to address this problem is to use an amorphous or polysilicon seed layer on the
field isolation oxide. For instance, a stack of silicon oxide, silicon nitride and amorphous
silicon or polysilicon can be deposited on an STI patterned wafer.
In a BiCMOS process, the seed layer also serves as a protection layer for the
CMOS (i.e., active) part of the wafer, during the bipolar processing steps. After the CMOS
5 processing, the seed layer stack can be opened in the active area to expose the silicon
substrate where the DEG process is performed and the bipolar transistor is processed. At
the end of the bipolar processing, the seed layer can additionally be patterned to be used as
a silicidation protect layer.
Depending on the design rules, the seed layer is either patterned inside or outside
10 the active area. In the case where the seed layer is patterned outside the active area, the
topography of the seed layer is added to the topography of the STI ditch. This additional
topography on the STI ditch (e.g., 90 nm) has a negative influence on the subsequent
processing. Accordingly, a need exists for a method that improves (i.e., reduces) the
topography to avoid the negative processing effects.
15 The present invention addresses the above-mentioned problems, as well as others,
by providing a stepped seed layer to smooth out the topography on the STI ditch. In a first
aspect, the invention provides a method for forming an epitaxial base layer in a bipolar
device, comprising: providing a structure having a field isolation oxide region adjacent to
an active silicon region; forming a silicon nitride/silicon stack above the field isolation
20 oxide region, wherein the silicon nitride/silicon stack includes a top layer of silicon and a
bottom layer of silicon nitride; performing an etch to the silicon nitride/silicon stack to
form a stepped seed layer, wherein the top layer of silicon is etched laterally at the same
time the bottom layer of silicon nitride is etched; and growing an Si/SiGe/Si stack over the
stepped seed layer and active region.
In a second aspect, the invention provides a structure for forming an epitaxial base
layer in a bipolar device, comprising: a silicon nitride/silicon stack formed above a field
isolation oxide region, wherein the silicon nitride/silicon stack includes a top layer of
silicon and a bottom layer of silicon nitride, and wherein the top layer of silicon is laterally
5 stepped back from the bottom layer of silicon nitride to form a stepped seed layer; and a
smeared Si/SiGe/Si layer formed above both the stepped seed layer and an adjacent active
In a third aspect, the invention provides a method for forming an epitaxial
base layer in a bipolar device, comprising: providing a structure having a field isolation
1 0 oxide region adjacent to an active silicon region; forming a silicon nitride/silicon stack
above the field isolation oxide region, wherein the silicon nitride/silicon stack includes a
top layer of silicon and a bottom layer of silicon nitride; substantially covering the field
isolation oxide region with a photo-resist mask; and performing an etch to the silicon
nitride/silicon stack to form a stepped seed layer, wherein the top layer of silicon is etched
15 laterally at the same time the bottom layer of silicon nitride is etched vertically.
These and other features of this invention will be more readily understood from the
following detailed description of the various aspects of the invention taken in conjunction
with the accompanying drawings in which:
Figure 1 depicts a structure having seed layers deposited on an STL
20 Figure 2 depicts the structure of Figure 1 being etched.
Figure 3 depicts the resultant structure after etching.
Figure 4 contrasts cross-sections of a seed layer structure formed using a prior art
technique and the current technique.
Figure 5 depicts the cross-sections of Figure 4 after a DEG process.
Referring now to the drawings, Figures 1 -3 depict the steps involved in an
exemplary method for reducing the seedlayer topography in a BICMOS fabrication
process. As depicted in Figure 1, a portion of an STI patterned wafer is shown comprising
5 an active silicon region 10 and a field isolation oxide region or STI region 12. Also shown
is the STI ditch 34, which is formed where the active silicon region touches the oxide
during the process of making the STI regions. More specifically, the STI ditch 34 is formed
when hydrofluoric acid (HF) is used. Because HF etching is applied several times in
various processes before the epitaxial growth, the STI ditch 34 is one of the features that
10 introduce topography.
In the first step, a seedlayer comprised of silicon nitride 16 and silicon 14 (or
polysilicon) is placed onto a silicon oxide layer 17. . As is known in the art, the silicon
oxide layer 17 is required since the silicon nitride or the silicon/silicon nitride stack 14, 16
has to be patterned, and the dry etch to do this is not allowed on the active silicon because
15 it would damage the silicon lattice. The silicon oxide layer 17 can be grown or deposited.
If deposited, layer 17 is present everywhere, and if grown, layer 17 is only present on the
active silicon 10 where it is grown. In the latter case, all silicon that is exposed will ■
oxidize during a thermal treatment. In Figure 1, this would mean that only the top surface
of the active Si area 10 would be covered with a thin silicon oxide layer. The silicon
20 nitride layer 16 is deposited and the silicon layer 14 can be amorphous or polycrystalline.
Typical thicknesses for these layers are 50 nm for the silicon layer 14, 20nm for the
silicon nitride layer 16, and 20nm for the silicon oxide layer 17. The thickness of the seed
layer 14, 16 increases the topography (i.e., the step) after patterning, which makes it
difficult for dry etches later on in the process, to make precise patterning. To address this,
an etch technique in accordance with the present invention is described below.
As shown in Figures 2-3, a "smearing" process is carried out to achieve the desired
result. Specifically, an etch chemistry is utilized that laterally etches the top polysilicon
5 layer 14 fast at the same time the bottom silicon nitride layer 16 is etched. In an exemplary
embodiment, this can be achieved with an SF 6 based dry etch in a TCP9400 LAM
RESEARCH™ reactor. The recipe is composed of 2 sub-steps: The first sub-step is an
anisotropic polysilicon etch, meaning that the polysilicon is etched vertically down (as
shown by arrows 36 in Figure 2) with respect to, e.g., photo-resist mask 32, and the second
10 sub-step is an anisotropic nitride etch with an isotropic polysilicon etch (as shown by arrow
38 in Figure 3).
In general, plasma etching involves a model in which positive ions are accelerated
by a bias voltage to the substrate and by sputter materials (i.e., referred to as ion
sputtering). Chemical etching occurs when neutral reactive species (called radicals) react
15 with the surface to make a chemical a volatile component.- For the polysilicon etch,
Cb/HBr chemistry may be used. The etching mechanisms are ion sputtering driven, so the
etch is anisotropic.
When the silicon layer 14 is etched, the chemistry is changed to a SF6 plasma. The
nitride etch is ion-sputtering driven, but free fluorine radicals are present in the plasma and
20 fluorine atoms are highly reactive with pure silicon (accordingly, the chemical etching is
very important). Thus, while the nitride is etched with ions, the polysilicon is laterally
etched (there is still some photoresist on top) with radicals. With this technique, the
topography is reduced because the polysilicon 14 is "pushed" away.
This method could work with other fluorine gases, although care must be taken
while using fluorocarbon gases (CHF 3 , CH 2 F 2 , C4F8, C 3 F 6 , and NF 3 for example), because
fluorine atoms react chemically, but carbon atoms consume fluorine by forming polymers
so that it prevents chemical etching by creating a protective layer on the sidewall, so only
5 ion sputtering etch is possible. Thus, care must be used in selecting an appropriate fluor-
to-carbon (F/C) ratio.
The lateral etching of the silicon layer 14 is self-aligned to the nitride etch of the
silicon nitride layer 16. This self-alignment provides an extra benefit, because if two litho
operations (resist spin, exposure, development) were required, the two lithography steps
10 could be easily misaligned. In such a case, the exposed nitride surface could then be larger
at the left than at the right (in cross-section). Typical best in class alignment specs are on
the order of -30-50 nra.
As can be seen in Figure 3, a "stepped" structure is formed in which the silicon
layer 14 is laterally etched far from the active silicon region 10, and the silicon nitride layer.
15 16 is etched between the active silicon region 10 and the silicon layer 14. The result is a
topography having two small steps or tiers instead of one big one. The fact that part of the
silicon nitride layer 16 is left uncovered in this manner has no significant consequences on
the differential epitaxial growth (DEG) process. The distance from the edge of the silicon
layer 14 to the edge of the silicon nitride is typically 100 nm or more. The distance of the
20 nitride layer 16 to the silicon seed edge is typically of that same distance (e.g., a few
Figure 4-5 depicts a cross-section of the structure (before and after the DEG
process takes place) in which the left side depicts the prior art technique 22 and the right
side depicts the current technique 24 in accordance with the present invention. (Note that
these figure are used solely to contrast the current technique 24 versus the prior art
technique 22, and are not meant to suggest that both techniques should be used to form a
single device.) As can be seen in Figure 4, the silicon layer 14 on the right side, using the
current technique 24, has been etched laterally as indicated by arrow 28 to create a stepped
5 seed layer.
Figure 5 depicts the structure after the DEG process has taken place and EPI stack
20 has been formed. Reference numeral 26 shows the point between where the grown
Si/SiGe/Si stack 20 transitions from epitaxial (mono-crystalline) over the active region 10
to polycrystalline over the STI. As can be seen, using the prior art technique 22, the
10 topography 32 of stack 20 comprises a severe slope. Conversely, the topography 30 of the
EPI stack 20 using the current technique 24 has been smeared out to provide a much more
The foregoing description of the preferred embodiments of the invention has been
presented for purposes of illustration and description. They are not intended to be
15 exhaustive or to limit the invention to the precise form disclosed, and obviously many '
modifications and variations are possible in light of the above teachings. Such
modifications and variations that are apparent to a person skilled in the art are intended to
be included within the scope of this invention as defined by the accompanying claims.