FIELD OF THE INVENTIONThe invention relates to semiconductor devices and the fabrication thereof and, more particularly, to isolation and well regions formed in a substrate and the fabrication thereof.[0001]
BACKGROUND OF THE INVENTIONIn one method of the related art P and N-channel MOS transistors are fabricated in a semiconductor substrate in a dual polysilicon gate process. In a dual polysilicon gate process, N-channel and P-channel gates are defined with separate individual masks, in contrast to a single polysilicon gate process wherein both the N-channel and the P-channel gates are formed using a single photolithographic process. The semiconductor substrate is doped with negative impurity atoms and positive impurity atoms to create negative and positive active regions respectively. These active regions can be thought of as having opposite polarities or opposite conductivities. One of the regions is created in the substrate while the other region is created in a well. The well is formed by doping a portion of the substrate to have a conductivity opposite to that of the original substrate.[0002]
FIGS.[0003]1-3 show a cross sectional portion of a semiconductor wafer following process fabrication steps used in a method of the related art to form P- and N-channel MOS transistors. In FIGS.1-3 an N-well5 has been created in a P-substrate10 by conventional fabrication means.
The N-well is a counterpart to the P-substrate in that it functions as a region in which to form P-channel MOS transistors, while the P-substrate functions as a region in which to form N-channel MOS transistors. A P-type region and an N-type region are opposite of each other with respect to energy bands. An N-type region has many electrons in its conduction band, while a P-type region has relatively few electrons in its conduction band; and a P-type region has many holes in its valence band while an N-type region has relatively few holes in its valence band. Micro Chip, A Practical Guide to Semiconductor Processing, by Peter Van Zant and Electronic Principles, third edition, by Albert Paul Malvino are herein incorporated by reference in order to determine a minimal knowledge of someone skilled in the art.[0004]
[0005]Thick oxide25 is grown to form field oxide regions to electrically isolate active regions from each other, and athin gate oxide30 is grown overlying the active regions. The formation of thethick oxide layer25 and thegate oxide layer30 are well known to those skilled in the art.
Referring to FIG. 1, the conventional fabrication means is continued, and a[0006]polysilicon layer31 is masked with photoresist to define an N-channel gate polysilicon and interconnect and an N-well tie. The polysilicon is etched andspacers33 are formed on opposing sides of the polysilicon remaining after the etch.
The in-process wafer is bombarded with negative ions to form N-type regions in the active regions not covered with[0007]polysilicon31. N-typeactive regions35 function as source/drain regions of an N-channel MOS transistor, andpolysilicon layer31 interposed between theregions35 functions as the gate of the N-channel MOS transistor thus formed. N-typeactive region45 is an N-well tie. An N-well tie is a region formed in the surface of the substrate in the N-well region that provides ohmic contact of the N-well to an external supply potential.
In FIG. 2 a second[0008]photoresist mask50 defines P-channel gate polysilicon and interconnect and protects the N-channel gate polysilicon and N-typeactive regions35 and45 previously defined.
In FIG. 3 the[0009]polysilicon layer31 remaining exposed at this juncture are etched. The substrate is now bombarded with positive ions and P-typeactive regions60 and65 are formed in the surface of the substrate. It is important to note that thethick oxide regions25 also function as masks during both the positive and negative ion bombardments that form the N-channel and P-channel source/drain regions and well/substrate ties.
P-type[0010]active regions60 function as the sources/drain regions of a P-channel MOS transistor.Polysilicon layer31 interposed between source/drain regions60 functions as the gate of the P-channel MOS transistor thus formed. P-typeactive region65 is a P-substrate tie and provides ohmic contact to the substrate from an external supply potential.
Although it would seem that the N-[0011]well tie45 shown in the cross section comprises two portions, the N-well tie45 may actually be a continuous ring surrounding the P-channel transistors, and the P-substrate tie65 may actually be a continuous ring surrounding the N-well5. In further fabrication steps, contacts (not shown) are formed with the P-substrate tie65 and the N-well tie45 as well as the source/drain and gate terminals of the MOS devices. The P-substrate tie65 is connected to a potential having a low voltage, typically ground, and the N-well tie45 is connected to a potential having a high voltage, typically VCC. The P-substrate tie and N-well tie help prevent latch up of the device when interposed between the N-MOS and P-MOS device.
Latchup occurs when two parasitic cross coupled bipolar transistors are actuated and essentially short a first external supply potential, V[0012]CC, to a ground potential, VSS. When the fabrication of the transistors is completed the N-channel source/drain regions35 form the emitter, P-substrate10 forms the base, and the N-well5 forms the collector of a horizontal parasitic NPN transistor; and the P-channel source/drain regions60 form the emitter, the N-well5 forms the base, and the p-substrate10 forms the collector of a vertical parasitic PNP transistor. The parasitic PNP and NPN bipolar transistors thus formed are actuated by the injection of minority carriers in the bulk of the substrate or the N-well. To prevent latch up, the lifetime of these carriers must be reduced, or the resistance of the substrate must be decreased. The latter method may force compromise in device performance by increasing junction capacitance and body effect.
The N and P substrate ties formed in the related art are gaurdbands which reduce the lifetime of the minority carriers. The gaurdbands act as a sink for the minority carriers that are injected into the substrate or N-well when the emitter/base junction of either parasitic bipolar device is forward biased. The gaurdbands also increase the distance these minority carriers must travel thereby increasing the probability that they will recombine with majority carriers. The gaurdbands are typically formed between the N-channel MOS transistor and the P-channel MOS transistor. The gaurdbands are strips of P+ active regions in the P-substrate and N+ active regions in the N-well. The gaurdbands tie down the substrate and well potentials and prevent latchup by collecting any injected minority carriers from forward biasing the MOS device source or drain regions.[0013]
OBJECTS OF THE INVENTIONIt is an object of the invention to achieve improved latchup immunity without increasing the complexity of the process steps in a dual-polysilicon process. It is a further object of the invention to reduce the lifetime of the minority carriers, and increase the distance minority carriers must travel to reach a parasitic bipolar collector.[0014]
SUMMARY OF THE INVENTIONThe invention features a method for forming a trench in an active region of a substrate, and features the trench thus formed. The substrate surface reserved for trench formation is doped to have a first type conductivity. Portions of the substrate are protected, and the substrate is etched in the unprotected areas to form the trench. At the same time the trench is etched a layer overlying the substrate is also etched to form a gate region. Thus, the trench is formed without increasing processing steps. A substrate region at the bottom of the trench is doped to create a substrate region having a second type conductivity. The trench is positioned between a first and second active device and directly in a desired path of the minority carriers. The trench reduces minority current, thereby reducing the possibility of latchup.[0015]
In one embodiment of the invention, unannealed damage caused by the trench etch functions as a recombination medium for minority carriers.[0016]
In a further embodiment of the invention the trench is filled with polysilicon. The polysilicon functions as a recombination medium.[0017]
In addition the polysilicon may function as an electrical interconnect between the doped substrate and an external potential. In this embodiment the polysilicon filled trench is also a substrate tie.[0018]
BRIEF DESCRIPTION OF THE DRAWINGSFIG. 1 is a cross section of a portion of a related art semiconductor wafer following the formation of an N-well and N+ active regions.[0019]
FIG. 2 is a cross section of the related art semiconductor wafer of FIG. 1 following the masking of the N+ regions.[0020]
FIG. 3 is a cross section of the related art semiconductor wafer of FIG. 2 following the formation of a P+ active regions.[0021]
FIGS.[0022]4A-C are cross sections of a portion of a semiconductor wafer of the invention following the formation of N+ regions in the substrate.
FIGS.[0023]5A-C are cross sections of a portion of the semiconductor wafer of FIGS. 4A and 4B, respectively, following the masking of the substrate to protect selected N+ regions.
FIGS.[0024]6A-C are cross sections of a portion of the semiconductor wafer of FIGS. 5A and 5B, respectively, following the formation of a trench and P+ regions in the substrate.
DETAILED DESCRIPTION OF THE INVENTIONThe invention is a method for creating a portion of an integrated circuit on a semiconductor wafer. In the method of the invention an isolation trench and an isolation region are formed in order to electrically isolate two regions of the integrated circuit from each other. The isolation trench and the isolation region minimize undesired effects, such as latch up caused by the actuation of parasitic bipolar transistors.[0025]
Although the method of the invention pertains to the fabrication of the isolation trench and the isolation region of the integrated circuit, the method will be described in conjunction with the formation of transistors in order to provide a clearer picture of the function of the invention.[0026]
Referring to FIGS.[0027]4A-C, an N-well66 is formed in a P-substrate67. Creating an N-well within a P-type substrate is well known to those skilled in the art and involves doping the substrate with impurity atoms to create a negatively doped region in which to fabricate a P-channel metal oxide semiconductor (MOS) transistor. Since a P-type substrate is positively doped the P-type substrate and the N-well can be thought of as having opposite polarities or opposite conductivities.
Still referring to FIGS.[0028]4A-C, the substrate has been patterned and athick oxide70 has been formed in isolation regions using a conventional local oxidation of silicon (LOCOS) process. Optionally other isolation schemes, such as trench isolation and poly buffered LOCOS, may be used. These alternate isolation schemes are also well known to those skilled in the art.
Following the removal of the mask used to grow the field oxide,[0029]thin gate oxide75 is formed in future active regions. The creation of thin gate oxide is also well known to those skilled in the art.
Following the formation of the gate oxide[0030]75 apolysilicon layer80 is deposited. A photoresist mask, not shown, is used to pattern thepolysilicon layer80 thereby exposing a selected first portion of the future active regions. The unprotected portion of the polysilicon is etched exposinggate oxide75 in the selected first portion of the future active regions. The polysilicon etch also exposes portions of thethick oxide70.Spacers86 are created on the opposed exposed ends of the etchedpolysilicon layer80. Thespacers86 may be electrically insulative and are created with methods well known to those skilled in the art. Thespacers86 are typically oxide.
Next the substrate is bombarded with ions having a sufficient dose to form highly doped N+ regions in each of the selected first portion of exposed active regions. Typical ions used to fabricate N+ regions are arsenic, antimony, or phosphorous. N+[0031]active regions90 function as source/drain regions of an N-channel MOS transistor. The gate of the N-channel MOS transistor is thepolysilicon layer80 interposed between the N+active regions90. N+active region95 is an isolation region which functions as an N-well tie when connected to an external supply potential, such as VCC. N+active region95 may be a contiguous annular ring. It should be noted that an annular ring is not necessarily circular. N+active region97 fabricated during this step will be removed during the formation of a substrate trench. N+active region97 may form a contiguous annular ring surrounding the N-well66 and basically concentric to the annular ring formed by N+active region95. The distance between the tworegions95 and97 may vary due to variations in the width of thethick oxide layer70 interposed between the twoactive regions95 and97.
In addition to the[0032]polysilicon layer80 withspacers86, thethick oxide70 functions as a mask during the ion bombardment.
In FIG. 4B a[0033]segment98 ofpolysilicon layer80 withspacers86 functions as a mask to defineactive region97. It can be seen by studying FIG. 4B thatsegment98 extends beyond thethick oxide70 and actually overlies a portion of thegate oxide layer75. In addition it is possible to use asecond segment99 ofpolysilicon layer80 withspacers86 to mask the other side ofactive area97 as shown in FIG. 4C.
During spacer formation, all oxide overlying exposed active regions is typically removed (FIGS.[0034]4A-C), although this is not a requirement of this invention.
Referring now to FIGS.[0035]5A-C, aphotoresist mask100 is formed to protect the N+active regions90 and95 during the formation of a second portion of active regions. If remainingoxide75 has not yet been removed from active region92 it is now removed in exposed areas during a short oxide etch. This oxide etch is selective over the exposedpolysilicon80.
Referring now to FIGS.[0036]6A-C, an etch selective to silicon over oxide is performed to create atrench105 in thesilicon substrate67. The etch substantially removes theactive area97 to create a recess and then removes the substrate underlying the recess to complete the trench formation. In FIG. 6A thetrench105 is self-aligned to thethick oxide70. In FIG. 6B thetrench105 is self aligned on one side to thethick oxide70 and is aligned on the other side to the edge of thespacer86. In FIG. 6C the trench is aligned on both sides to the edge of thespacers86. During the etch unmasked portions ofpolysilicon layer80, includingsegments98 and99, are also etched. However, thesubstrate67 underlying the etchedpolysilicon layer80 is protected during the etch by thegate oxide layer75. In addition, thespacers86 protect thesubstrate67 during the etch. In a preferred embodiment thetrench105 is created to have a depth greater than the depth of the N-well66.
Following the trench formation the wafer is bombarded with positive ions having a sufficient energy to form highly doped to P+ regions in the unmasked portions of the substrate. Typical ions used during the ion bombardment are aluminum, boron, or gallium. P+[0037]active regions110 are source/drain regions of a P-channel MOS transistor. The gate of the P-channel MOS transistor is thepolysilicon layer80 interposed between the P+active regions110.P+ region115 is formed in thesilicon substrate67 at the base of thetrench105. Thetrench105 with itsP+ region115 and the N+active region95 functioning as an N-well tie separate the P-channel MOS transistor from the N-channel MOS transistor thus formed. In FIGS. 6B and 6C an additional P+active region116 is formed adjacent to a top portion of the trench. Standard metalization contacts can readily be formed overlying the P+active region116.
A P+ region and an N+ region are opposite of each other with respect to energy bands. An N+ region has many electrons in its conduction band, while a P+ region has relatively few electrons in its conduction band; and a P+ region has many holes in its valence band while an N+ region has relatively few holes in its valence band.[0038]
Once the[0039]trench105 has been fabricated according to the steps of the invention alternate embodiments may be employed at the discretion of the designer. In one option a high temperature anneal is performed subsequent to the trench formation in order to create regions at the bottom of thetrench105 which have recombination centers even when high temperature anneals are performed later in the fabrication process. These regions function as recombination centers for the minority carriers and further reduce the concentration of the minority carriers in the device.
An additional option comprises filling the[0040]trench105 with polysilicon. The polysilicon provides additional recombination centers. In the present embodiment the polysilicon is either P+ type or undoped.
In a further embodiment the substrate adjacent to the sidewalls of the trench are positively doped. In this embodiment the trench may be filled with polysilicon or a dielectric following the trench etch and sidewall doping. Alternately the substrate adjacent to the sidewalls may be doped automatically during the doping of a previously deposited filler material such as the polysilicon.[0041]
In a further embodiment the[0042]P+ region115 may be connected to an external potential having a low voltage, typically ground via polysilicon deposited in the trench. In this case theP+ region115 is a substrate tie.
In addition the shape of the trench is not important, although one skilled in the art would be aware of the importance of using suitable design rules to prevent shorting the trench to the boundary of the N-well or an adjacent N-channel region.[0043]
Although the detailed description has described the formation of one[0044]trench105, the invention is applicable when forming a plurality of trenches.
Although the invention has been described in terms of forming the trench in a P-type substrate with an N-well formed therein, the invention is equally applicable to forming a trench in an N-type substrate with a P-well formed therein and to creating an N-type region at the base of the trench thus formed.[0045]
There are many methods that may be employed to dope a substrate in addition to bombarding the substrate with dopant ions. These alternate methods are well known to those skilled in the art. One alternate method is diffusion.[0046]
Although the method is more effective when the trench is formed in the substrate rather than the well, the invention can be performed for either case, and in either case the trench is positioned to reduce minority current caused by parasitic bipolar transistors.[0047]
No additional mask steps are required in this invention over those typically used in a dual poly gate patterning process. However, the invention is not limited to MOS devices. Other options include using the method and or apparatus of the invention to improve the isolation of bipolar devices one from another.[0048]
Although the invention has been described as a dual polysilicon process the method could also be applied to MOS or bipolar process that etch layers of polysilicon doped and patterned at different steps. Such a process could use[0049]polysilicon1 and polysilicon2 etch steps to form the trench.
While this invention has been described with reference to an illustrative embodiment, this description is not intended to be construed in a limiting sense. Various modifications as well as other embodiments for fabricating a trench in a previously doped active region or for the trench formed therein will be apparent to persons skilled in the art upon reference to this description. It is, therefore contemplated that the appended claims will cover any such modifications or embodiments as fall within the true scope of the invention.[0050]