Note: Descriptions are shown in the official language in which they were submitted.
~ 1'796
S P E C I F I C A T I O N S
8ackground of ~he Invention
1 The present invention relates to integrated circuit semi-
2 conductor devices and more particularly to a method for fabricating
3 such devices with self-aligned contacts.
4 Large-scale integrated circuits,often having thousands of
5 MOSFET's on a single semiconductor chip, must have a ~Nl~plicitv of
6 contacts to provide the necessary interconnections between circuit
7 lines, source-drain regions and gate electrodes of individual
8 transistor elements. Using prior art procedures, it is necessary
9 to make oversized conductive areas and larger contact openings in
10 order to accommodate mask alignment tolerances. This generally
11 results in devices requiring a relatively large chip area. With
12 the rapid increase in large-scale integrated circuit devices having
13 even greater numbers of MOSFET elements, efforts have been made to
14 reduce not only the element size but also the size of the required
15 contacts~ With the trend toward reduced design tolexances and
16 narrower interconnect lines this became an increasingly severe
17 problem. One suggested solu~ion described in J. Electrochem. Soc.
18 Solid State Science and Technology, Vol. 125, No. 3, March 1978,
19 pp. 471-472, is to provide a gate material of polycrystalline
20 silicon which was coated on its sides and top with a thin silicon
21 dioxide (Si02) layer; However, t-his proved to be unsatisfactory
22 because it fails to eliminate the problem of shorts due to
23 occasional breakdowns or fractures of the SiO2 layer during subse-
24 quent process steps. The present invention overcomes these and
25other problems and provides several other advantages in addition
i~to providing a means for making a large-scale integrated circuit
~7device with a substantially smaller area per MOS transistor element
~ i
, ~,
r ~ 1 1131796
1 than was heretofore possible. Moreover, the invention makes pos-
21 sible the production of such reduced area devices wherein the
31 source~drain regions of the MOSFET elements can be formed either
41 by standard diffusion or ion implantation techniques.
51
61 Brief Summary of the Invention
I .. __
71 In accordance with the principles of the present invention
8l ~IOSFET elements with self-aligned contacts forming an integrated
9¦ circuit device may be fabricated in a semiconductor substrate
having a first conductivity type by a method wherein a permanent
11 ¦ internal protective layer is formed. Preliminary steps of the
12 ¦ method utilize conventional fabrication techniques. After the
13 ¦ field oxide areas are formed with openings for transistor elements,
14 ¦ polysilicon gate areas are provided within the openings~ Poly-
15 ¦ silicon conductive lines are also simultaneously formed on the
6¦ field oxide close or adjacent to such openings where necessary~
17 I Thereafter, contact areas with the minimum dimensions required are
8¦ formed on opposite sides of each gate area and also where required
19 ¦ on the conductive lines. In one version of the present method, all
20¦ of these poly gates and conductive lines are first provided with a
~1¦ top layer of silicon nitride and thereafter a thin oxide layer on
22¦ their sides. Source-drain regions then are formed by diffusion
231 techniques and thereafter the thin protective layer of silicon
241 nitride is provided over the entire chip, covering the field oxide
areas, the poly silicon areas and the areas surrounding the poly
26 gate areas. Now, a standard layer of phosphorous impregnated glass
271 (PVX) is also applied to the entire chip covering the thin nitride
28¦ layer, and thereafter a contact mask on the PVX layer is used to
29¦ form the necessary contact openings by first etching away the PVX
30~ in the contact openi.ng regions but stopping at the protective
ll -2-
-
11;~1796
1 nitride layer. The thick field oxide and the thin oxide layer on
2 the sides of the poly gate areas are prevented from being attacked
3 during this PVX etch by the thin nitride protective layer. Follow-
4 ing this, the thin protective nitride layer in the contact areas
5 is etched away by an etchant that will not attack the field oxide
6 and the protective poly oxide. The thin layer of gate oxide
7 exposed by the contact opening is then removed. Thereafter, a
8 poly contact mask is used to form contact openings in the PVX layer
9 and the top nitride layer on the poly lines for the contacts on the
10 poly interconnect lines. Both of these latter masks can utilize
11 the relatively large openings to assure registration or self-align-
12 ment with the desired contact areas, because the previously applied
13 thin nitride layer provides protection for the field oxide and the
14 poly oxide on the gate areas and assures against circuit shorts
15 between gates, poly lines and N+ interconnect lines. With this
16 added internal protection the alignment tolerances heretofore
17 required between poly gates, poly lines and contact openings are
18 substantially reduced, yet without requiring unusually close
19 tolerances on the contact masks for forming the contact openings.
20 The invention thus greatly reduces the problems of producing
21 integrated circuit devices with more closely packed elements per
22 unit area and yet a higher yield.
23 In an alternative form of the present method, the poly
24 gates and conductive lines, after being formed, are provided with
25 a thin silicon dioxide layer on their sides and also on their top
26 surface instead of the initial nitride layer. The source-drain
27 regions are then formed by ion implantation techniques with the
28 poly silicon gate serving as a mask in the well-known manner.
29 Thereafter, the thin internal protective nitride layer is applied
30 over the entire chip surface before fabrication is completed. The
31 protective layer again performs its function of preventing internal
32 shorts and any
-- 3 --
;~
1796
1 overetching of the thin oxide layer on the conductive poly gates
2¦ or poly lines and during the formation of oversized holes in the
31 insulating PVX layer.
41 In summary, the objects of the invention are: to provide an
51 improved method for forming semiconductor devices with self-aligned
61 contacts; to provide a method that will allow a reduction in the
7 spacing between contacts and conductive interconnect lines and
8 ¦ thereby facilitate production of more closely packed devices; to
9¦ provide`a method that can be easily controlled with standard
10 ¦ semiconductor production ~acilities; and to provide a method that
11 ¦ will significantly increase the production yield of large-scale
12 ¦ semiconductor devices with self-aligned contacts.
131
14 I Brief Description of the Drawing
1-
15 ¦ Fig. l is plan view of a typical MOS transistor structure
16 ¦ with contacts formed as in the prior art;
~7 ¦ Fig. 2 is a plan view of an MOS transistor structure formed
18 with self-aligned contacts;
19 I Figs. 3a-14a illustrate the steps for forming self-aligned
20 ¦ contacts for a semiconductor device according to the method of the
21 ¦ present invention; and
22 Figs. 3b-14b illustrate the steps for forming self-aligned
231 contacts for a semiconductor device using a somewhat modified
24 method according to the present invention.
26 ¦ Detailed Description of the Preferred Embodiments
271 With reference to the drawing, Fig. 1 illustrates in plan
2~1 view a conventional MOS transistor 10 of the prior art having non-
2g self-~ligned source and drain contacts 12 and 14 and a gate contact
16. Generally accepted design rules for snch transistors
13~796
in a large-scale integrated circuit required each contact on a
source and drain region 18 and on a gate electrode 20 extending
over it to have a minimum area. Because of alignment tolerances
in forming such contacts using conventional fabrication proce-
dures, it was necessary for the underlying source-drain region 18
- to be considerably larger than the minimum contact area in order
to assure proper registration of the contacts~ For example, in
order to have a minimum required contact area, a uniform toler-
ance around all sides of the contacts (shown at Ll and L2) and
predetermined minimum spacing between contact edge and poly sili-
con edge (L3) was required using conventional fabrication
technology. The design requirements resulted in an MOS semi-
conductor device as shown in Fig. 1 in order to prevent shorts
and leakage problems in an integrated circuit comprised of many
such MOS devices.
The reduction in chip area that can be accomplished for
a single MOS transistor lOa with self-aligned contacts is
illustrated in Fig. 2. ~ere, the source, drain and gate contacts
12a, 14a and 16a, all having the minimum area, are automatically
registered with the borders of their Source-drain region 18a or
the gate electrode 20a and the surrounding field gxide~ The
tolerances Ll, L2 and L3 are reduced to zero, and each diffused
region 18a can have minimum dimensions in width and in length
using conventional design rules. Also, because each contact is
self-aligned or completely contained on its respective contact
area, the spacing from an adjacent conductive line can be mini-
mized, thereby further decreasing the overall chip area required
for a semiconductor device.
The more important method steps for making such a
semiconductor device with self-aligned contacts according to the
present invention will now be described relative to Figs. 3a-14a.
' l
` 11;~1796
As shown in Fig. 3a, the method commences with the provision
2 of a semiconductor substrate 22 (e.g. ~ 100~ plane silicon
3 material) that is doped in the suitable range to provide the desir-
4 able characteristics. Tllis su~strate is covered with an initial
oxidation layer 24 of 500-lOOOA on which is deposited a second
6 layer 26 of silicon nitride of approximately equal thickness.
7 Using a field oxide mask, the layers 24 and 26 are removed
8 by etching in the field areas as indicated in Fig. 4a, and these
9 areas are then field implanted as indicated ~y the dotted lines 28
10¦ to adjust field threshold levels in the conventional manner.
11 As shown in ~ig. 5a, a relatively thick field oxide 30 is
12 now grown in the field areas, also in the conventional manner.
13 This drives the field implanted areas 28 further into the substrate
14 22 under the oxide areas. In a typical semiconductor structure,
the field oxide is configured to form holes or openings within
16 which each MOS transistor is to be formed.
17 After the field oxide is formed, the original nitride layer
18 26 and the gate oxide layer 24 are removed by etching. Thereafter,
19 a new gate oxide layer 32 is formed within the hole in the field
oxide.
21 Now, over the entire device surface, including the new oxide
22 layer 32 and the field oxide 30 (as shown in Fig. 6a), a thin (e.g.
23 150-300A) nitride layer 34 is deposited using conventional vapor
24 deposition techniques. In order to insure stability of the pro-
duct, the upper ~urface of this ni~ride layer is oxidized in steam
26 or dry oxygen ambien~ ~not shown~. The ~tep ~llustr~ted in Pig. 6a
27 provides a new nitride/oxide sand~ich that ha~n't undergone the
28 heat treatment which was applied during formation of the field
29 oxide adjusted ~o appropriate thicknesses. The original oxide/
3o nitride ~andwich 24,26, shown in ~ig. 3a, adiusted to appropriate
. -6-
'i,
': ' . , . - .
1131796
1 thicknesses, could be used as the gate dielectric.
21 In the next step of the method embodiment of Fig. 7a, a layer
3 36 of poly-crystalline silicon (poly) is deposited, by a standard
41 vapor deposition process, onto the entire surface of the chip being
51 fa~ricated to a typical thickness of around 3000-5000A.
6¦ A mask is then used to define gate electrodes 38 (as
71 shown in Figure 8a) within the active areas formed in the field
81 oxide and interconnect lines 40 situated on top of the field
91 oxide 30 and adjacent to one or more gate elements. At this
0 point, all portions of conductive poly within a field Gxide
opening and on the field oxide are situated on a nitride/oxide
2¦ sandwich. Using known silicon gate procedures wherein the gate
13l serves as a mask, ion implantation techniques are employed, as
14 represented by the vertical arrows in Fig. 9a, to form source
5l and drain regions 42 and 44 just below the substrate surface
6¦ within the field oxide opening and on opposite sides of the poly-
7¦ silicon gate 38.
8¦ In the next step, as shown in Fig. lOa, a layer 46 of sili-
9 con-dioxide is grown on all sides and also on the top of all
conductive poly areas including the gate poly areas 38 and the
21 adjacent poly interconnect lines 40. The thickness of this cover-
22 ing layer on the poly is generally much greater than the gate oxide
23 32 (e.g. around 3000A), and its purpose is to provide a protective
24 layer on the poly for making the self-aligned contact structure.
In the next step, as also shown in Fig. lOa, a thin protec-
26 tive nitride layer 48 is deposited over the entire structure at
27 this point, including the field areas 30, the source and drain
2~ areas 42 and 44 and the areas 38 and 40 of oxide covered poly.
29 This nitride layer will later serve to provide vital protection
for field oxide and poly oxide during subsequent process steps.
Following this application of the thin nitride layer the entire
7-
,: ,
~:131~96
1 ~ chip, s shown in ~ig. lla, is covered with a relatively thick
2 ¦ layer 50 of phospho-silicate-glass (PVX) in the conventional
3 ¦ manner.
4 ¦ Now, as shown in Fig. 12a, a first mask (not shown) for
5 ¦ the N+ contacts is applied to the PVX and a suitable etchant (e.g.
6 ¦ buffered hydrofluoric acid) is used to etch away the PVX layer 50
7 in the contact area. As is well known, hydrofluoric acid does
¦ not affect nitride layers 34 and 46. A suitable etchant (e.g.,
9 ¦ phosphoric acid) is then used to etch the nitride layers 34 and
lOj 36 in the contact opening. Of importance, this nitride etch does
1¦ not attack field oxide 30, gate oxide 32 or oxide 46 which
12¦ protects poly 38. Gate oxide is then removed in the contact area
13 using a suitable etchant such a~ hydrofluoric acid. Of impor-
14¦ tance, oxide layer 46 protecting poly 38 is significantly thicker
lS than the gate oxide being removed, thus preventing appreciable
6¦ damage to protective oxide 46 during removal of the gate oxide.
7¦ With protective oxide 46 intact, contact metalization is later
18 deposited without forming electrical shorts to poly 38.
9¦ Thereafter, a second contact mask is applied to the chip
201 in the same manner as the first mask and etchants are used (again
21¦ in three steps) to etch away the PVX, nitride and the oxide on
22 the polysilicon lines. These latter two masks for the N+ and
23 polysilicon contacts may be applied in reverse order, if desired.
24 This leaves the chip, as shown in Fig. 13a, with the PVX layer
50 coincident with the thin nitride layer 48 having windows to
26¦ expose the drain contact area 42 and also an exposed contact
271 area 51 on the adjacent poly interconnect line 40 having no
28~ oxide coverino.
291 At this point, standard fabrication method steps can be used
to deposit metal in the contact areas to define metal contacts 52
I! ;
`` ~13~796
1 and 54 as part of a desired metal interconnect pattern on the
2 semiconductor device. Generally, these metallization steps include
3 the evaporation of metal and definition thereof with an appropriate
4 metal mask and thereafter the application of a top protection di-
S electric layer over the entire chip (not shown) for passivation.
6 In a modified version of method according to the present
7 invention, illustrated by Figs. 3b-14b, the initial steps of Figs.
8 3b-6b inclusive are identical to those of Figs. 3a-6a. However,
9 this modified version avoids the necessity for using ion implanta-
tion equipment and procedures for forming the source and drain
11 regions.
12 Thus, as shown in Fig. 7b, a poly layer 36 having a typical
- 8a -
~,
~1~17~6
1 thickness in the range of 3000-5000A is formed on the chip over the
2 gate nitride layer 34 by ~onventional chemical vapor deposition
3 techniques. This poly layer is then doped by diffusing phosphorous
4 into it to make it more conductive. Thereafter, a nitride layer 56
5 having a thickness that is considerably greater than that of the
6 gate nitride layer 34 (e.g. l000-?OOOA) is deposited on the poly-
7 silicon layer 36.
8 As shown in Fig. 8b, the polysilicon layer 36 is defined into
9 gate areas and interconnect lines by using a poly mask (not shown)
10 and standard etching techniques which removes the unwanted material
11 in the poly and nitride layers. This leaves the structure with a
12 doped polysilicon gate or electrode 38 within an area surrounded by
13 field oxide 30 and an adjacent poly interconnect line 40 situated
14 on the oxide, both o~ these poly elements having the nitride layer
15 56 on their top surfaces.
16 In the next step, shown in Fig. 9b, the poly gate elements 38
17 and the poly interconnect lines 40 are provided with an oxide layer
18 46 on their sides having a thickness of around 3000A. This is ac-
19 complished by simple thermal oxidation in a chamber in accordance
20 with well-known procedures.
21 Now, as shown in Fig. lOb, the source and drain regions
22 ~2 and 44 are formed by diffusion techniques. First, using
23 suitable, well~known masking and etching techniques, the gate
24 nitride layer 34 is etched away from every surface except the tops
25 of poly layers 38 and 40. Then, the gate oxide layer 32 is etched
261 in all the areas surrounding the gate poly layer. Standard
27 diffusion procedures are now applied to form the source and drain
28 regions 42 and 44. Following this, a new thin gate oxide layer
29 58 is formed in the diffused regions, to a thickness of around
500A.
31 To the structure shown in Fig. lOb, a thin protective nitride
32 layer 60 (e.g. 150-300A) is applied. This layer 60 is thus much
~,,f~'` _g_
:
`~
1131796
thinner than the nitride layer 56, and as with the previous embodi-
2 ment, layer 60 extends over the entire chip including the field
31 areas 30, the source and drain areas and the nitride covered gate
4 38 and line poly areas 40.
5 ¦ Now, the PVS layer 50 is applied (Fig. llb) and etched
6 (Figs. 12b and 13b), using contact masks in the same manner as
7 ¦previously described in respect to the first embodiment of the
8 method. Nitride layer 60 is then etched in the contact area in
9 ¦the same manner as in the previously described first embodiment,
10 without affecting gate oxide 58 and protective oxide 46. Thin
11 ¦oxide layer 58 is then removed from the contact area. Of impor-
12 ¦tance, gate oxide 58 is significantly thinner than oxide layer 46
13 ¦protecting poly region 38. Thus, oxide layer 46 is not signifi-
14 ¦cantly damaged during the removal of the gate oxide 58. With
15 oversized contact holes formed in the PVX layer as previously
16 ¦described, the metallization of the MOS elements on the chip is
17 ¦performed to form its metal contacts 52 and 54 with accompanying
18¦ interconnect lines in the standard manner as shown in Fig. 14b.
19¦ Because oxide 46 protecting poly 38 is not significantly damaged
201 during the formation of the contact, shorts between metallization
21¦ 52 and poly 38 are prevented.
22¦ By utilization of either embodiment of the method according
23¦ to the invention, it is possible to produce large-scale semiconduc-
241 tor devices with a multiplicity of MOSFET elements having self-
25 aligned contacts and therefore requiring a minimum of chip area in
26 a closely-packed array. For example, in a typical random access
271 memory (RAM) the area required for a single memory cell was 1344
28¦ square microns, whereas, with the self-aligned contacts made po-
2g¦ sible using the present method, the same memory cell has an area of
301 only 950 square microns, reduction in area of approximately 30%.
31 Yet, with the method of the present invention, the yield of such
,, -10-
li;~l796
1 closely-packed devices with self-aligned contacts can be even hi~her
2 than with prior conventional devices because the internal protective
3 nitride layers 48 and 60 maintain circuit integrity during critical
4 process steps by preventing shorts or failures heretofore caused
5 during the various process steps. While silicon nitride is a pre-
6 ferred material for the protective layers, other materials could be
7 used such as silicon car~ide or aluminum oxide.
~ To those skilled in the art to which this invention relates,
g many changes in construction and widely differing embodiments and
- lOa -
. ,,
~ .~,
11;~1~96
1~applications of the invention will suggest themselves without de-
2 parting from the spirit and scope of the invention. The disclosures
3 and the description herein are purely illustrative and are not
6¦intended o be in any sense limiting.
51
lal
223 .
26
301