Surface Science 53 (lYY1) 173--277
IO nm of Si are ohtaincd conditions. the implant
is mainly dstrrmincd
1. Introduction Ion beam synthesis has become an attractive technique for the formation of buried epitaxially oriented CoSi, layers [ 11. Whereas molecular beam epitaxy of CoSi 1 on ( 11 I) Si and monocrystallinc Si on top of (111) CoSi, has been successful, the formation of the Si/CoSi,/Si heterostructure on (100) Si has up to now not been demonstrated. The ion beam synthesis approach allows formation of CoSi, both in (Ill) and in (100) Si and benefits from the attractive opportunity to thicken the surface Si film by cpitaxial CVD Si. The formation of ultra-thin layers by ion beam synthesis has been achieved previously on (111) Si substrates . Layers with thicknesses down to 16 nm have been demonstrated. The formation of similarly thin layers in (100) Si, was found to be much more difficult. In this paper the formation of ultra-thin buried silicide layers in (100) Si is examined. It will be demonstrated that the density of pin-holes and their growth during annealing is causing an instability of the layer during high temperature trcat0 I hY-4332/Y
Layers air thin as 33 nm under
the dew and energy of the Co implant and continuously
is found to he the temperature
dul-ing Implant. The disintegration
i\ caused by the prcsencc of !mall pin-holes.
layers by ion implantation
buried C‘oSi, layers in (I(K)) Si by ion beam synthe>i< is investigated.
The most critical pat-ametrr
applied surface science
Formation of ultra-thin in (100) Si
by the \uhxqurnt
of the c‘oSiz layers upon
to he directlq dependent
ments. These temperature treatments are mandatory to improve the crystalline quality of the layer and to reduce the defect densities in the upper and lower Si regions.
2. Experimental The cxpcriments were performed on 125 mm wafers of n-type (100) Si with a sheet resistance of 10-20 0. cm. Co implantations were performed using a resistively heated chuck. Doses and energies were varied in the experiment. The beam current was typically held at 4 to 5 @AA/cm’. The temperature of the chuck was measured with a thermocouple. According to our latest results, the temperature of a wafer on a heated chuck in a vacuum ambient is quite different from the chuck temperature. A new temperature measurcment set-up has therefore been developed recently . The temperatures mentioned in this paper still refer to the chuck temperature, as measured by the thermocouple. After implantation the samples were annealed in an N, ambient using an AG-associates rapid thermal processor.
Elsevier Science Puhlishcrs B. V. All rights resrned
The annealing time was held constant at 5 min and the temperature was ranging from 800 to 1100 o C. The samples were analyzed with a four-point probe for sheet resistance measurements, with cross-sectional and plan-view TEM for crystal structure and defect observations and with RBS for composition and thickness of the layers.
3. Results and discussion
The current description of the ion beam synthesis process of CoSi, invokes the existence of a critical dose above which coalcscencc into a single layer occurs after high temperature treatment. Hull et al.  have studied subcritical and supercritical doses for 100 keV implants. According to their conclusions the crucial condition for forming a continuous layer involves formation of a connected silicide layer during the implantation. However, according to the findings in ref.  and also according to our own experience. the idea of having a critical dose (or better critical concentration) does not seem to hold for layers thinner than 40 nm and accordingly lower implant energies. since doses that were by far supercritical did not yield nicely coalesced layers.
Fig. 2. Crowwctional
Fig. I. Sheet r&stance C’o/cm’
I j .I 1100
plot ol samples implantctl
at 50 heV, at various tempcratutw to 100”(‘. after subsequent annealing
with 7 x IO”’
ranging fl-0113.300 for 5 min.
In order to find out which parameters arc most important for thin layer formation. a systematic experiment was set up using the implant temperature and the high tempcraturc trcatmcnt subsequent to the implantation as pnramctcrs. The temperature during implant has also been reported to bc an important parameter for the formation of thick buried layers [Xl. To invcstigate this latter parameter a dose of 7 x 10”‘ Co/cm’ at 50 kcV was chosen. which should yield a 30 nm thick buried CoSi, layer. The top
for a layer formed hy an implant of 7 X IO” annealed
I ..,. 1000
at lOSO”C for 5 min.
50 keV at 1Oo“C and subxqucntly
concentration of the Co after implant was measured by RBS to be 26%, which is distinctly higher than the critical concentration of 17% as defined in refs. [5,6]. A first evaluation of the various synthesized layers was done by sheet resistance measurements. Fig. 1 shows an overview of the sheet resistance of the layers for implant temperatures ranging, nominally from 300 to 4OO”C, and subseranging from quent annealing at temperatures
800 to 1100°C. Two conclusions can be drawn from this graph. The lowest sheet resistance is obtained after a 900°C anneal, for all implant temperatures. The sheet resistance after this anneal temperature is strongly dependent on the implant temperature, as is also the thermal stability of the layer, viz. the sheet resistance of the layer after anneal at higher temperatures. The implant dose at a nominal temperature of 400 o C results in low resistive layers even after 1050°C I
I a Si
1.0 2 5 -u z x 0.5
9OOC, 5 min 300
Fig. 3. (a) RBS spectrum
and (b) cross-sectional TEM micrograph of a sample implanted 400°C after annealing at 900°C for 5 min.
with 5.75 X 10”
40 keV at
anneals. Only an anneal at 1100°C gives a slight increase in sheet resistance. The sheet resistance value is consistent with a resistivity value of the CoSi, of 15 ~0. cm. Fig. 2 shows a cross-sectional TEM micrograph of the layer implanted at a nominal temperature of 400 o C and annealed at 1050 o C. Even after this high temperature treatment, the amount of (111) facets remains very limited. Moreover, it is interesting to note that the defect density in the upper Si layer and in the Si substrate close to the CoSi, interface decreased below the detection level of cross-sectional TEM. By lowering dose and energy further down to 5.75 x 10’” Co/cm’ at 40 keV, CoSi, layers of 23 nm thickness underneath 19 nm of Si have been ohtaincd. The RBS spectrum and the TEM cross-section micrograph arc shown in fig. 3. Again the silicide interfaces arc atomically flat with only a limited amount of (I 11) facets. According to our observations of thin silicide layers, samples with nominally the same implant conditions, show a very different behaviour after identical high temperature treatments. This implies that minor diffcrenccs in implant conditions can result in very different layers. Since the high tcmpcraturcs are required to improve the crystal quality of the CoSi, and to reduce the amount of
Fig. 5. Plan-view TEM IX
of samples implanted
1000 o C (b) and 1050°C (c), exhibiting
a had thermal
at (a). stahillt>
9OOC, 5 min
200 channel Fig.
300 number implanted
at 300°C‘: as-implanted a, 000 o C.
and after annealing
defects in the surrounding Si, we investigated which mechanism is responsible for the disintegration of the layers. Samples with nominally the same implant conditions but behaving differently when subjected to high temperature treatments, were compared by RBS and detailed plan-view TEM investigation. Fig. 4 shows RBS spectra of a
sample implanted with 7 x 10’” Co/cm’, at 50 keV and annealing at 900°C. According to the sheet resistance measurements, the layer in this sample has a very bad thermal stability. However, the RBS spectrum after 900°C shows a nicely coalesced buried CoSi, layer. After lOSOY?, no complete coalescence was observed by RBS. By increasing the temperature treatment, the Si/Co ratio decreases systematically from 2 (at 900°C) down to around 1.5 (at 1050°C). The sample implanted with nominally the same implant conditions but showing a good thermal stability according to sheet resistance measurements yields identical RBS spectra as-implanted and annealed at 900°C. It is only the treatments at higher temperatures that yield different RBS spectra. Similar to MBE-grown epitaxial layers, it was observed that some ion beam synthesized silicidcs suffer from pin-hole formation. Visiualization of these pin-holes is possible by using a TEM in dark field [O]. In the samples with a bad high temperature behaviour pin-holes arc detected (fig. 5). They are rectangular in shape and aligned along the (1 IO) direction. Pin-holes in the 900°C samples cover of the order of 1% of the surface area of the wafer and therefore they cannot possibly be detected by RBS. The pin-holes arc much bigger for the 1050°C anneal temperature. The area covered by the pin-holes in this case is in accordance with the RBS measurement. The density of the pin-holes remains the same, irrespective of the annealing temperature. It should be made clear at this point that in a sample, implanted with nominally the same conditions, but exhibiting a very good thermal stability, no pin-holes were observed, despite explicit efforts to find some. The presence of pin-holes in the CoSi, layer, is therefore directly correlated with the bad thermal stability of the Si/CoSiz/Si heterostructure.
4. Conclusions In this work the formation of ultra-thin buried CoSi? layers has been investigated using ion beam synthesis in (100) Si. By systematic optimization of the implant conditions, layers as thin as 23 nm underneath 19 nm of Si have been obtained. The
interfaces between the Si and the CoSi, are atomically flat and only a limited amount of facets along the (Ill) direction have been found. The thermally stability of the heterostructure is of uttermost importance to reduce the amount of defects in the surrounding Si. It was shown that the disintegration of the CoSi, layers is attributed to the growth of pin-holes. The density of the pin-holes depends critically on the implant conditions, whereas their size is mainly determined by the subsequent temperature treatment. A more extended study, in which temperature during implantation and beam current density will be in situ monitored is underway aiming at a detailed optimization of the implant conditions by minimizing the pin-hole density.
Acknowledgements The authors would like to thank J. Van Laer, J. Vandesande, C. Alaerts, R. Verbeeck, F. Loosen and 1. Callant for their diligent technical help and P. Vandenabeelc for invaluable discussions on temperature measurement. The RBS measurements were performed by B. Brijs. K. Maex is a research associate of the Belgian Fund for Scientific Research. Part of this work has been funded by the ESPRIT program Tip Base.
50 (1987) YS. G.
and Y. Bruynscraede.
[41 P. Vandenabecle
and W. Jager,
and K. Maex,
Res. Sot. Proc.
(61 AH. heim
107 (1988) 3. Bulle-Lieuwma,
(71 R. Hull.
Phys. 68 (1990)
[Xl E.H.A. drnhoudt, ers. Nucl.
Dekempeneer, C.W.T. Meth.
[91 J. Vanhellemont.
to be published.
et al.. to be published.