The effect of annealing on the properties of
`silicidized molybdenum thin films
`
`Cite as: Journal of Applied Physics 52, 6331 (1981); https://doi.org/10.1063/1.328575
`Published Online: 04 June 1998
`
`T. P. Chow, A. J. Steckl, and D. M. Brown
`
`ARTICLES YOU MAY BE INTERESTED IN
`
`Rapid thermal annealed Molybdenum back contact for Cu2ZnSnS4 thin film solar cells
`Applied Physics Letters 106, 131110 (2015); https://doi.org/10.1063/1.4916994
`
`Thickness dependence of the properties of highly c-axis textured AlN thin films
`Journal of Vacuum Science & Technology A 22, 361 (2004); https://doi.org/10.1116/1.1649343
`
`Growth of highly c-axis textured AlN films on Mo electrodes for film bulk acoustic wave
`resonators
`Journal of Vacuum Science & Technology A 21, 1 (2003); https://doi.org/10.1116/1.1521961
`
`Journal of Applied Physics 52, 6331 (1981); https://doi.org/10.1063/1.328575
`
`52, 6331
`
`© 1981 American Institute of Physics.
`
`Micron Ex. 1030, p. 1
`Micron v. Godo Kaisha IP Bridge 1
`IPR2020-01008
`
`

`

`The effect of annealing on the properties of silicidized molybdenum thin films
`T. P. Chow
`General Electric Corporate Research and Development, Schenectady, New York 12301, and Rensselaer
`Polytechnic Institute, Center for Integrated Electronics, Troy, New York 12181
`A. J. Steckl
`Rensselaer Polytechnic Institute, Center for Integrated Electronics, Troy, New York 12181
`D. M. Brown
`General Electric Corporate Research and Development, Schenectady, New York 12301
`
`(Received 13 April 1981; accepted for publication II June 1981)
`
`The effect of isothermal and isochronal annealing on the structural and electrical properties of
`silicidized Mo thin films is reported. The silicidation of Mo with SiH4 resulted in Mo films with
`increasing hexagonal MoSi2 content as the reaction time increased. Post-reaction annealing was
`performed in various ambients (hydrogen, nitrogen, and vacuum) at temperatures from 800 to
`1000 ·C for times up to I h. Annealing in H2 at 1000·C for only 5 min results in the virtual
`disappearance of the original components and the formation of intermediate silicide phases
`(M05Si3 and M03Si) and the tetragonal MoSi2 phase. This structural transformation leads to a
`significant increase in resistivity. The reaction kinetics are considerably slower in the case of N2
`ambient where even after a 6O-min anneal at 1000·C a substantial percentage ofMo remains. This
`slower rate of phase change is reflected in a more gradual increase in the sheet resistance. The
`characteristics of vacuum-annealed films followed the N2 case for short-term anneals (t:S 10 min)
`but resembled the H2 case longer-term anneals (t > 15 min). The dependence of reaction
`mechanisms on various ambients is discussed.
`
`PACS numbers: 81.40.Rs, 81.30.Hd, 64.70.Kb, 73.60.0t
`
`I. INTRODUCTION
`3 and metal silicides4-7 are under
`Refractory metals l
`-
`active investigation as highly conductive interconnect and
`electrode materials for very-Iarge-scale-integration (VLSI)
`circuits. One disadvantage of the refractory metals is their
`poOr oxidation resistance. Since their oxides are volatile at
`high temperatures, surface passivation schemes usually in(cid:173)
`volve deposited dielectrics, as shown in Fig. I(a). On the oth(cid:173)
`er hand, the silicides can generally form a high-quality pro(cid:173)
`tective oxide, resulting in a self-passivating structure, as
`shown in Fig. I(b). To avoid excessive consumption of the
`silicide layer during oxidation, a polysilicon pad can be
`placed8 between the gate dielectric and the silicide [Fig. I(c)).
`This, however, is at the expense of increased process
`complexity.
`Generally, refractory metal silicide films have been pre(cid:173)
`pared by evaporation or sputtering from either the alloy4.7 or
`the individual constituents. 5
`6 Recently, it has been shown
`•
`that MoSi2 films can also be prepared by the reaction of Mo
`thin films with SiH4.9 By this technique, it is hoped that one
`can combine a refractory metal with an overcoat of its own
`silicide. The result, shown in Fig. I(d), is a self-aligned metal(cid:173)
`oxide-semiconductor (MOS) electrode structure, the so(cid:173)
`called" heart-of-Moly," which has both the high conductiv(cid:173)
`ity of the metal and the oxidation resistance of the silicide.
`The silicidation process resulted in a combination of hexag(cid:173)
`onal and tetragonal MoSi2, MosSi3 , and M03Si phases.
`Among the molybdenum silicide phases formed, MoSi2 is
`the most oxidation resistant and the only silicide phase suit(cid:173)
`able for self-passivating oxide growth. In addition, in order
`to realize the advantages of the heart-of-Moly structure, its
`electrical characteristics must be stable with subsequent
`
`high-temperature processing steps. In this paper, we report
`on the effect of post-reaction annealing conditions (e.g., am(cid:173)
`bient, temperature, time) on the structural and electrical
`properties of silicidized molybdenum thin films.
`
`II. EXPERIMENTAL PROCEDURE
`The sample preparation and the silicidation process
`have been previously described. 9 The Mo films used for this
`study were nominally 3000 A thick and were silicidized at
`800·C for either 60 or 120 sec. Following silicidation, the
`samples were annealed under various conditions. Three am(cid:173)
`bients-hydrogen, nitrogen, and vacuum-were used. For
`the first two a flow rate of 151 per min was maintained. In the
`vacuum anneal cases, a pressure of 2 X 10-7 Torr was used.
`The anneals performed were either isothermal (1000 .C) or
`isochronal (5 min). For isothermal anneals, the duration was
`varied from 5-60 min. For isochronal anneals, the tempera(cid:173)
`ture was varied between 800-1000·C. In all cases, the an(cid:173)
`neals were not cumulative.
`To determine the thickness of the Mo film after silicida(cid:173)
`tion, a step was created in the films by etching in HN03/5%
`NH4F solution with photoresist masking, and its thickness
`was measured with a profilometer. The sheet resistance of
`the films was measured using a four-point probe system. The
`structural composition was studied using x-ray diffraction
`performed on a Siemens 0500 automatic power
`diffractometer.
`The surface morphology of the films was observed us(cid:173)
`ing a scanning electron microscopy (SEM). In Fig. 2, a typi(cid:173)
`cal SEM microphotograph of an as-silicidized Mo film is
`shown. This film was obtained for a 60-sec silicidation at
`800 ·C. The surface is somewhat rougher than a typical as-
`
`6331
`
`J. Appl. Phys. 52(10), October 1981
`
`0021-8979/81/106331-06$01.10
`
`@ 1981 American Institute of Physics
`
`6331
`
`Micron Ex. 1030, p. 2
`Micron v. Godo Kaisha IP Bridge 1
`IPR2020-01008
`
`

`

`~:g;RMAL r ~O,
`ZZ~171717
`
`ZZ
`
`(a)
`
`(b)
`
`MoGATE
`
`MoSi2 GATE
`
`(e)
`
`MoSi21POL Y Si GATE
`POLYCIDE
`
`(d)
`
`MoSiiMo GATE
`M.LY
`
`FIG. I. Schematic cross sections of various refractory gate structures (a) Mo
`gate, (b) MoSi, gate, (c) MoSVpoly-Si (polycide) gate, and (d) MoSi,/Mo
`("Heart-of-Moly") gate.
`
`sputtered MoSiz film (picture not shown) obtained by dc
`magnetron sputtering from a stiochiometric alloy target.
`The overall adhesion of the silicidized films was good. How(cid:173)
`ever, some peeling was observed at the wafer periphery. This
`stress-induced effect was aggravated for reactions per(cid:173)
`formed at higher SiH4 flow rates or longer silicidation times.
`
`III. RESULTS
`The surface morphology observed after the high-tem(cid:173)
`perature anneals indicated no significant changes. This in(cid:173)
`cludes the edge lift-off effect observed after silicidation.
`Films were reacted at 800·C with 10% SiH4 in Hz. The 10%
`SiH4 and main Hz flow rates were 244 cc/min and 30 l/min,
`respectively. For a 60-sec reaction, the composition of the
`un annealed films, shown in Fig. 3(a), consists of mainly Mo
`oriented in the (110) direction. 10 A small hexagonal I I MoSiz
`component oriented in the (111) direction is also observed.
`This latter component becomes comparable to the Mo peak
`for a l20-sec reaction [Fig. 7(a)]. In either case, no other
`silicide phase was observed within the detection limit of the
`diffractometer.
`The structural composition of films isothermally an(cid:173)
`nealed in Hz at lO00·C is shown in Figs. 3(b) and (c). As can
`be seen, even a S-min anneal in this case results in a drastic
`change in constituent phases. The Mo component is totally
`absent, while intermediate silicide phases [M05Si~ (Ref. 12)
`and M03Si (Ref. 13)] are now dominant. No further structur(cid:173)
`al change was observed for anneals up to 60 min.
`Isothermal annealing in a Nz ambient at the same tem(cid:173)
`perature (Fig. 4) results in a more gradual Mo to silicide
`transformation. In this case, the Mo component still domi(cid:173)
`nates after a 20-min anneal and remains significant after 60
`
`PRE-ANNEAL CONDITIONS:
`3300 A INITIAL Mo FILM
`SILICIDIZED AT aoo'c
`144cc/min 10% SiH4 FOR 60 sec.
`
`5 MIN
`
`60 MIN
`
`Ihl
`
`FIG. 2. SEM microphotograph of a typical as-silicidized Mo thin film.
`Silicidation was done at 800 "C for 60 sec.
`
`FIG. 3. Relative x-ray intensities (peak values, not corrected for back(cid:173)
`ground and normalized to the highest peak) ofa 3300-A.-thick Mo film
`which was first silicidized for 60 sec at 800 'C and then isothermally an(cid:173)
`nealed in H, at 1000 'C for (a) 0 min (as-reacted), (bl 5 min, and (c160 min.
`
`6332
`
`J. Appl. Phys., Vol. 52, No. 10, October 1981
`
`Chow, Steckl, and Brown
`
`6332
`
`Micron Ex. 1030, p. 3
`Micron v. Godo Kaisha IP Bridge 1
`IPR2020-01008
`
`

`

`100
`
`50
`
`Mo
`(110)
`\
`
`PRE-ANNEAL CONDITIONS:
`INITIAL Mo ALM 3300 A THICK
`SILICIDIZED ATBOO·C
`144cc/min 10% SiH4 FOR 60sec
`h-MoSi2
`/ (III) AS REACTED
`
`(a)
`
`Mo PRE-ANNEALCONDITION5:
`(I/O) INITIAL Mo FILM 3300 A THICK
`I 51 LlCIDIZED AT 800'C
`
`144 cc/min 10% 5iH4 FOR 60 sec
`
`100
`
`(b)
`
`.1
`
`II
`" 60 MIN lid:
`510 515 7b 7S 80
`4b
`2 THETA (DEGREES)
`
`FIG. 4. Same as Fig. 3, except the film was isothermally annealed in N, at
`1000 'C for (a) 0 min (as-reacted), (h) 5 min, (c) 20 min, and (d) 60 min.
`
`FIG. 6. Relative x-ray intensities of a 3300-A-thick Mo film which was first
`silicidized for 60 sec at 800 'c and then isochronally annealed in H, for 5
`min at (a) 20 'C (as-reacted), (h) 850 'C, (c) 900 'C, and (d) 1000 'c.
`
`PRE-ANNEAL CONDITIONS:
`INITIAL Mo FILM 3300 A THICK
`r ° I
`SILICIDIZED AT 800·C
`h-MOSi 2UI0),I M ( 10)
`144cc/mIO 10"10 SiH4 FOR 60sec
`Mo~Si3(202)
`+
`
`5 MIN
`
`r Mo Si (222)
`t-MoSi2003)
`r 5 3
`
`10 MIN
`
`15MIN
`
`Mo 3Si
`(400)
`j
`
`(a)
`
`1
`
`1
`
`1
`
`..,Ib)
`
`Mo 5i(200) r M03S'(210)
`1 l ,---,--t"
`~05Si3(o02)
`CD~. log ~'
`~ 50
`~ 0 ~-~ -T--,----l"
`~ 100 f
`
`100
`
`+
`l
`50 3.
`l. h- MoSI 2(IOI) I
`. j
`
`(f)
`
`mm.
`The effect of annealing in vacuum (Fig. 5) results in a
`structural change process which is between those observed
`in the H2 and N2 cases. For a 5-min anneal, the relative peak
`intensity ofMo to silicides is nearly identical to the N2 case .
`On the other hand, after 60 min, the phase composition is
`very close to that obtained in the equivalent H2 case, result(cid:173)
`ing in the complete absence of Mo. While the H2 ambient the
`Mo disappears after less than 5 min, this transformation oc(cid:173)
`curs between 10 and 15 min in the vacuum case. In contrast,
`in the N2 case, this complete transformation was not ob(cid:173)
`served for anneals up to 60 min.
`To further examine this Mo to silicide transformation,
`isochronal anneals were performed. To accelerate the trans(cid:173)
`formation rate, H2 ambient was used, thus allowing for a 5-
`min anneal time. The results of isochronal anneals of a film
`silicidized for 60 sec are shown in Fig. 6. For an 850°C an(cid:173)
`neal, the (110) Mo peak is still dominant. It is interesting to
`point out that the relative ratio of various phases are similar
`to those observed for anneals of the same duration in N2 and
`vacuum, but at the higher temperature of 1000 DC. Increas(cid:173)
`ing the temperature of isochronal anneal by only 50 to 900 °C
`results in making the intermediate silicides the dominant
`phases at the expense of the Mo component.
`To evaluate the effect of a larger initial percentage of
`MoSi2 on the structural transformation, a film silicidized for
`120 sec was also subjected to isochronal anneals. In Fig. 7(a),
`the as-reacted film is shown to be composed of approximate-
`
`~ 50
`>
`~ 0 L,---'--,---'-_J.--'--'---;'''I--'--'-,I (e)
`
`I, ~~~~30--lMIN
`g 1:000 l
`Ll~
`
`'I"r··-,--1, (d)
`
`60MIN
`
`l
`
`I
`I
`5'0 515 7'0 7'5 8'0
`4'0
`2 THETA (DEGREES)
`
`11
`
`(e)
`
`FIG. 5. Same as Fig. 3, except the film was isothermally annealed in vacuum
`at 1000 'C for (a) 5 min, (h) 10 min, Ie) 15 min, (d) 30 min, and (e) 60 min.
`
`6333
`
`J. Appl. Phys., Vol. 52, No.1 0, October 1981
`
`Chow, Steckl, and Brown
`
`6333
`
`Micron Ex. 1030, p. 4
`Micron v. Godo Kaisha IP Bridge 1
`IPR2020-01008
`
`

`

`PRE-ANNEAL CONDITIONS:
`INITIAL Mo FILM 3300 l THICK
`Mo SILICIDIZED AT800·C
`(110) 144cc/min 10% SiH4 FOR 120 sec
`t-MoSi2 j h - MoSi2
`(110)
`,/llil
`(222) (400)
`/103)
`I I I
`AS REACTED
`----,-JLL--,--r~--, (a)
`M05Si3
`(2~2J) h-MoSi2
`MO~i
`(III)
`(210)
`800.C
`I,
`
`I t-MoSi2 Mo5Si3 M03Si
`
`II
`
`1 (b)
`
`~ 150000 tWlO)
`:z r::t '-----'----',-11 -L1-'-1----'r--'I'-r~I,~I,~-· -L~ ----'1 (c)
`
`Mo5Si 3
`(002)
`Mo~ Si
`
`:z 100
`=>
`~
`«
`~
`V5
`
`50
`a
`
`0:: 100
`
`50
`
`035
`
`IOOO·C
`Ii? I
`45
`2 THETA (DEGREES)
`
`ab (d)
`
`FIG. 7. Same as Fig. 6, except the film was first silicidized for 120 sec and
`then isochronally annealed in H2 for 5 min. at (a) 20 'C (as-reacted), (b)
`800 'C, (c) 900 'C, and (d) 1000 'c.
`
`ly equal amounts of Mo and hexagonal MoSi2 . For a subse(cid:173)
`quent anneal at 800 ·C [Fig. 7(b)), it is observed that the hex(cid:173)
`agonal MoSi2 phase is greatly diminished while the Mo
`component is still dominant. Intermediate silicide phases are
`now also detected. Following an anneal at 900·C [Fig. 7(c)],
`the Mo component is seen to be greatly reduced. In its place,
`large amounts of both intermediate silicides and tetragonal
`MoSi2 are present. By comparison, the composition of a film
`silicidized for only 60 sec and annealed under the same con(cid:173)
`ditions [Fig. 6(c)) shows no significant amount of tetragonal
`MoSi2• Further increase in anneal temperature to 1000·C
`[Fig_ 7(d)) essentially eliminates the Mo peak and results in a
`slight reduction in the t-MoSi 2 peaks.
`While our results were obtained for silicidized Mo thin
`films it is also interesting to point out that the same transfor(cid:173)
`mation has been observed in the bulk. 14 Thick films
`(> 40,um) of MoSi2 grown on bulk Mo are transformed into
`M05Si3 and M03Si upon high-temperature (1200-1900·C)
`annealing.
`Furthermore, from data on heat offormation of various
`silicides, the change in enthalpy for the change in the reac(cid:173)
`tion of MoSi2/Mo to M05Si3 or to M03Si is calculated to
`be - 26 or - 29 Kcal/g mole of intermediate silicide
`formed. To a first approximation, this indicates that M05Si 3
`and M03Si have an approximately equal probability of
`formation.
`Turning now to the electrical properties, the sheet resis(cid:173)
`tance (R s) of silicidized films isothermally annealed at
`l000·C is shown in Fig. 8. The data in Fig. 8 were taken on
`
`"~~
`f
`AMBIENT f :~~
`
`12
`II
`10
`0.9
`!2. 08
`CI
`.
`~O.7i
`
`/
`
`/
`
`TEMP 1000.C
`
`05
`OAL
`o
`
`_L_
`10
`
`loVACUUM
`SILICIDATION TIME
`60 SEC
`_~_ 1.
`I
`30
`40
`20
`ANNEALING TI ME (MIN)
`
`L ____ ...L_
`50
`60
`
`I
`
`~
`
`FIG. 8. Sheet resistance ofsilicidized Mo films as a function ofpost-silicida(cid:173)
`tion annealing time in various ambients (H2' N" and vacuum) at 1000 'c.
`
`films silicidized for 60 sec. The film thickness measured was
`7000-8000 A. The as-reacted R, is between 0.4-0.5 n /D. For
`H2 anneals, R, increased rapidly after only a 5-min anneal to
`a value greater than 1 n /D. Longer term H2 anneals, up to
`60 min, do not significantly affect R,. On the other hand,
`
`o
`~ 0.9
`
`o::(/) 0.8
`
`0.7
`
`0.6
`
`SILICIDATION TIME
`~ 60 SEC
`·120 SEC
`
`900
`0.5 25(800
`TEMPERATURE (OC)
`(a)
`
`1000
`
`110
`
`UN ANNEALED
`+
`'I-.
`
`100
`E 90
`...,
`a.. 1
`~ao
`
`SIUCIDATION TIME
`y 60 SEC
`A 120 SEC
`
`70
`
`60
`
`(h)
`
`50
`45~~~-------L------~~
`25 aoo
`900
`1000
`TEMPERATURE (OC)
`FIG. 9. Isochronal anneal in H2 for 5 min. (a) Sheet resistance and (b) resis(cid:173)
`tivity of silicidized Mo films as a function of isochronal annealing tempera(cid:173)
`ture at silicidation times of 60 and 120 sec.
`
`6334
`
`J. Appl. Phys., Vol. 52, No. 10, October 1981
`
`Chow, Steckl, and Brown
`
`6334
`
`Micron Ex. 1030, p. 5
`Micron v. Godo Kaisha IP Bridge 1
`IPR2020-01008
`
`

`

`anneals in N2 result in a more gradual increase in Rs with
`anneal time. After the 5-min anneal, Rs increased by - 50%
`to 0.62 fl /0. After the 60-min anneal, Rs reaches 0.95 fl /0
`approaching the value obtained in the H2 case. In the vacu(cid:173)
`um anneal case, the initial increase of Rs resembles that of
`N2, to -0.65 fl /0 after 5 min. Longer term annealing, R,
`quickly increases to > 1 fl /0 after 15 min and no significant
`change was observed up to 60 min. The Rs values obtained
`for tan neal > 15 min are very close to those obtained in the H2
`anneal case.
`The dependence of sheet resistance and resistivity on
`isochronal anneal temperature is shown in Figs. 9(a) and (b),
`respectively, for films silicidized for 60 and 120 sec. The
`anneal conditions were 5 min in a H2 ambient. While the
`films were 7000-8000 A. thick after 60-sec silicidation, those
`si1icidized for 120 sec had a thickness of 1-1.1 f-l. The as(cid:173)
`reacted films had anRs of 0.45 fl /0 and 1.6fl /0 for the 60-
`and 120-sec silicidation, respectively. Annealing the film
`reacted for 60 sec at temperatures up to 900 °C increases Rs
`to 1.05 fl /0. Anneals at 1000 °C results in only marginally
`higher R,. The films reacted for 120 sec exhibit a different
`dependence. For anneals up to 900 °C, Rs decreases down to
`0.62 fl /0. Higher-temperature anneals result in a slight in(cid:173)
`crease in Rs.
`
`IV. DISCUSSION
`Before correlating the structural and electrical data, it
`is instructive to consider the resistivity of the various molyb(cid:173)
`denum silicides, as shown in Fig. 10. In bulk form, three
`silicide phases have been identified ls : MoSi2, MOsSi3, and
`M03Si. The MOsSi3 phase has the higher bulk resistivity,
`- 50 f-lfl cm, while the other two are in the 20-25 f-lfl cm
`range. By comparison, the bulk resistivity of Mo is - 5
`f-lfl cm. In annealed thin films the resistivity is generally
`higher due to reduced carrier mobility. For example, in the
`case of MoSi2, the thin film values are between 2 to 4 times
`higher than the bulk value. 4.16 While actual resistivity values
`of various silicides depend on the deposition method, the
`
`103~-
`
`J 100
`I
`o BULK VALUES REF.11
`• THIN { CO - EVAPORATED AEF. 41
`• FILM
`CO- SPUTTERED REF. 16
`SPUTTERED, THIS WORK
`~
`
`102
`
`1
`I
`
`10
`
`~
`
`E
`<..>
`I
`c:
`::I..
`
`0
`
`0
`::t:
`~
`
`....J ~ Vi
`
`E
`<..>
`I
`~
`::I..
`
`(f)
`l.LJ
`a
`Q
`
`~
`
`general trend is one of decreasing resistivity with increasing
`Mo content. In the extreme, for the sputtered Mo thin films,
`a resistivity of - 8-10 f-lfl cm was measured.
`As-sputtered 3300-A.-thick Mo films had an R, of
`-0.35 fl /0 prior to the actual silicidation reaction. The Mo
`films undergo high temperature in situ cleaning9 (1000 °C, 5
`min in H2) which results in a decrease in R, to 0.25 fl /0.
`Following silicidation for 60 sec, R, increases to -0.45
`fl /0. This increase is attributed to the formation of a small
`amount ofMoSi2 [see Fig. 4(a)) which has a higher resistivity
`than Mo. This process is enhanced when the silicidation time
`is increased to 120 sec. In this case, Rs is approximately 1.6
`fl /0 and a substantial amount ofMoSi 2 is detected [see Fig.
`7(a)].
`Let us consider first isothermal annealing. In the H2
`case, the rapid formation of intermediate silicides [Fig. 4(b)]
`is the cause of the sharp increase in resistivity within the first
`5 min of anneal (Fig. 9). Longer term anneals result in no
`further structural change [Fig. 4(c)] and this is reflected in a
`constant sheet resistance. The more gradual effect ofN2 an(cid:173)
`neals is seen in both a slower structural transformation (Fig.
`4) and a slower increase in R,. After as-min N2 anneal, the
`film-still consists mainly of Mo, unlike the H2 case where
`MosSi 3 and M03Si dominate. This is consistent with a much
`lower R, measured in the N2 case. After the 60-min anneal,
`the transformation is still not complete [Fig. 4(d)] with sub(cid:173)
`stantial Mo still present. This is reflected in the R, data
`which show that saturation is not reached. Vacuum anneal(cid:173)
`ing results in characteristics which initially resemble the N2
`case in both structure and resistivity. For 5-min anneals, the
`R, obtained in the two cases is the same, with an almost
`identical structural composition [Figs. 4(b) and 5(b)]. Longer
`term anneals (> 15 min) result in characteristics similar to
`those obtained in the H2 case. In both cases, R, remains
`constant with anneal time and no structural change is ob(cid:173)
`served, with the intermediate silicide phases dominating.
`Isochronal annealing also results in correlated changes
`in structural composition and electrical resistivity. Initially,
`the shorter (60 sec) silicidation time results in smaller
`amounts of h-MoSi2 formed [compare Fig. 6(a) with Fig.
`7(a)] and hence smaller increase in sheet resistance or resis(cid:173)
`tivity [Figs. 9(a) and (b)] than the longer (120 sec) silicidation
`time. Also, the increase in film thickness is less for the former
`than for the latter (from 3300 to 8600 A. and 1.05 f-l,
`respectively).
`As the postsilicidation annealing proceeds, various sili(cid:173)
`cide phases become the resistivity-determining components,
`instead ofMo in the as-reacted films. This leads to the differ(cid:173)
`ent dependence on isochronal annealing temperature for the
`two silicidation times. Firstly, for the silicidation time of60
`sec, the rapid increase in sheet resistance or resistivity [Figs.
`9(a) and (b)] between 850 and 900 °C is due to the structural
`transformation in which the intermediate silicides replace
`Mo and MoSi2 as dominant components [Figs. 6(b) and (e))
`and thus increase the overall resistivity. Anneals performed
`at higher temperatures (900-1000 0c) result in little change
`in either resistivity or structural composition. On the other
`hand, for the silicidation time of 120 sec, a decrease of Rs or
`Ps is observed between 800 and 900 °C [Figs. 9(a) and (b)],
`
`10
`
`0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
`Mo/Si ATOMIC RATIO
`
`Mo
`
`FIG. 10. Bulk and thin-film resistivities of Mo and its various silicides of
`different Mo/Si ratio.
`
`6335
`
`J. Appl. Phys., Vol. 52, No. 10, October 1981
`
`Chow, Steckl, and Brown
`
`6335
`
`Micron Ex. 1030, p. 6
`Micron v. Godo Kaisha IP Bridge 1
`IPR2020-01008
`
`

`

`together with a significant increase in various silicides [Fig.
`7(c)], including t-MoSi2 that compensates for the loss ofMo
`and consequently lowers the overall resistivity. Also, anneal(cid:173)
`ing between 900 and 1000 °C does not affect the resistivity
`but eliminates all and not some of the Mo. Furthermore, it
`may be noted that for the two silicidation times, there is a
`large difference in sheet resistance (1.1 n ID vs 0.65 n ID
`reached after annealing at the highest temperature (1000 °C,
`as shown in Fig. 9(a), but that is mostly due to the difference
`in film thickness, as verified in Fig. 9(b), where the differ(cid:173)
`ence, in terms of sheet resistivity, is only 10 f-Ln cm.
`In general, H2 appears to act as a catalytic agent during
`the annealing process, resulting in the rapid transformation
`of Mo into the silicide phases. This fast reaction rate is prob(cid:173)
`ably indicative of grain boundary diffusion of Si. It has been
`shown 17 that in the Mo/Si thin film interaction, Si is the
`main diffusing species. Because of the reducing nature ofH2,
`any trace of oxygen in the ambient is removed, thus prevent(cid:173)
`ing any Mo/0 2 or Si/02 interaction. In addition, the H2 also
`can break up any Mo-O (though not Si-O bonds) within the
`film, thus further increasing the rate of Mo/Si interaction.
`On the other hand, the N 2 ambient can be easily contaminat(cid:173)
`ed with traces of oxygen, which, by tying up Si or Mo atoms,
`would slow down the reaction rate. The vacuum anneal re(cid:173)
`moves the possibility of this additional Si/02 and Mo/0 2
`interactions, but does not eliminate those Mo-O or Si-O
`bonds already formed. In this respect, one can see how the
`transformation rate in vacuum falls in between those of H2
`and N2 ambients. Clearly, more work needs to be done to
`elucidate the exact mechanisms responsible for the Mo to
`silicide transformation.
`The silicidation reaction has also been attempted with
`other refractory metals: Nb and Ti. However, we were un(cid:173)
`successful due to the fact that both Nb and Ti react with the
`underlying Si02 at the high temperature (1000 0c) used for
`the in situ cleaning prior to silicidation. This is consistent
`with free energy calculations which indicate that in the case
`ofNb, Ti, and Ta the metal/Si02 reaction is favored, while in
`the case of Mo and W no reaction is expected.
`
`v. CONCLUSIONS
`The effect of annealing on the structural and electrical
`properties of silicidized Mo thin films has been studied. The
`
`as-reacted films have Mo and h-MoSi2 as the dominant
`phases. Upon annealing they are replaced by the intermedi(cid:173)
`ate silicide phases, such as MosSi3 and M03Si, and t-MoSi 2•
`The kinetics of this structural transformation is strongly de(cid:173)
`pendent on annealing ambient and temperature. Among the
`three ambients (H2' N 2, and vacuum) investigated, the reac(cid:173)
`tion proceeds fastest in H2 and slowest in N 2, whereas in
`vacuum, it initially resembles that in N2 but follows that in
`Hz case for longer-time annealing. A significant increase in
`resistivity was observed with this structural transformation
`and has been correlated well with the dissolution on forma(cid:173)
`tion of Mo and its various silicides.
`
`ACKNOWLEDGMENTS
`
`The authors would like to thank S. P. Murarka for the
`vacuum anneals and useful discussions, R. Goehner for the
`x-ray diffraction, C. Ludwin for sample sputtering, and G. J.
`Charney and K. J. Lanning for technical assistance.
`
`'D. M. Brown, W. E. Engeler, M. Garfinkel, and P. V. Gray, Solid-State
`Electron. 11, 1105 (1968).
`'P. L. Shah, IEEE Trans. Electron Devices ED·26, 631 (1979).
`'F. Yanagawa, K. Kiuchi, T. Hosoya, T. Tsujimani, T. Amazawa, and T.
`Mano, IEEE Trans. Electron Devices ED-27, 1602 (1980).
`4T. Mochizuki, K. Shibata, T. Inoue, and K. Ohuchi, Jpn. J. App!. Phys.
`Supp!. 17-1, 37 (1978).
`'B. L. Crowder and S. Zirinsky, IEEE Trans. Electron Devices ED·26, 369
`(1979).
`oS. P. Murarka, IEEE-IEDM 1979 Technical Digest, 454 (1979).
`'T. P. Chow and A. J. Steckl, App!. Phys. Lett. 36, 297 (1980).
`'S. Zirinsky, W. Hammer, F. d'Heurle, and 1. Baglin, App!. Phys. Lett. 33,
`76 (1978).
`"T. P. Chow, D. M. Brown, A. J. Steckl, and M. Garfinkel, J. App!. Phys.
`51, 5981 (1980).
`'''Standard Diffraction Pattern 4-0809.
`"Standard Diffraction Pattern 17-917. Correction has been made to the
`spacing given for the (110) planes to 2.304 1>...
`"Standard Diffraction Pattern 17-415.
`"Standard Diffraction Pattern 4-0814.
`'4R. W. Bartlett, P. R. Gage, and P. A. Larssen, Trans. Metal!. Soc. AIME
`230, 1528 (1964).
`"G. V. Samsonov, Silicides and their Uses in Engineering (Acad. Sci.
`UkrSSR, Kiev, 1959).
`"'S. P. Murarka, D. B. Fraser, T. F. Retajczyk, Jr., and T. T. Sheng, J. App!.
`Phys. 51, 5380 (1980).
`"J. Baglin, J. Dempsey, W. Hammer, F. d'Heurle, S. Petersson, and C.
`Serrano, J. Electron Mater. 8, 641 (19791.
`
`6336
`
`J. Appl. Phys., Vol. 52, No.1 0, October 1981
`
`Chow, Steckl, and Brown
`
`6336
`
`Micron Ex. 1030, p. 7
`Micron v. Godo Kaisha IP Bridge 1
`IPR2020-01008
`
`