APPLIED PHYSICS LETTERS                                                VOLUME 72, NUMBER 22                                                        1 JUNE 1998

Phases of cobalt-iron ternary disilicides
          I. De´zsi,a) Cs. Fetzer, and I. Szu cs
          KFKI Research Institute for Particle and Nuclear Physics, H-1525 Budapest, P.O. Box 49, 114,
          Hungary
          G. Langouche and A. Vantomme
          Instituut voor Kern- en Stralingsfysika, Katholieke Universiteit Leuven, Celestijnenlaan 200D,
          B-3001 Leuven, Belgium
           Received 14 January 1998; accepted for publication 26 March 1998 
          Cobalt­iron transition-metal disilicides were investigated by Mo¨ssbauer effect and x-ray diffraction
          in order to determine the concentration range of their homogeneous and separate phase formation.
          Except at low Co or Fe concentrations, Co and Fe formed separate CoSi2 and FeSi2 phases. Up to
          10 at % Co was found soluble in  -FeSi2; Fe dissolved in CoSi2 below 1.5 at % and was positioned
          at two different sites of cubic symmetry. The results obtained for the phase formation in thin layers
          of epitaxial CoSi2 on Si implanted with Fe were in agreement with the results obtained for the bulk
          samples. © 1998 American Institute of Physics.  S0003-6951 98 01122-X 


     In the last decade, interest in metallic silicides has in-                       energy at room temperature. The annealings of the samples
creased considerably because of their potential applications                          were made in vacuum of 1 10 4 Pa.
in micro- and optoelectronics.1­4 Ternary disilicides have                                The samples were measured by using Mo¨ssbauer spec-
been less studied than binary ones, although the former may                           trometers in constant accelerator mode. For the thin layer
be of greater importance as doped semiconducting disilicides                          samples a small sized low background proportional conver-
and may have a beneficial surface effect on the formation of                          sion electron counter was used. As single line sources 20­50
perfect epitaxial layers.                                                             mCi 57Co in Rh matrix were used. The measurements were
     Early Mo¨ssbauer studies of 57Co implanted atoms in Si                           performed at room temperature. The spectra were fitted by
revealed the formation of buried CoSi                                                 using a least squares fitting program allowing fittings in the
                                               2 precipitates of fluo-
rite structure.5 It was surprising to observe not only one                            presence of the distribution of the hyperfine splitting param-
single resonance line for iron but, in addition, another                              eters. For x-ray diffraction measurements a Philips vertical
``anomalous'' one.6 Later when implanting 57Fe in epitaxi-                            powder diffractometer with reflected beam monochromator
ally grown CoSi                                                                       was used. Cr K
                     2 on Si, the resonance line with the same                                            radiation was used.
isomer shift value as for the ``anomalous'' line of 57Co ap-                              The Mo¨ssbauer spectra of bulk FexCo(1 x) are shown in
peared with high relative intensity.7 Recently, buried ternary                        Figs. 1 a ­1 e . In Fig. 1 a  the spectrum measured at x
Co, Fe silicide phases with fluorite structure were formed8­10                         0.005 could be fitted by two single lines  SL1 and SL2 .
on implanting Co and Fe in Si at 350 °C. Theoretical calcu-                           For the FexCo(1 x)Si2 bulk samples at low x values the  
lations for the phase diagram of Co­Fe ternary disilicide11
suggested that disordered ternary phase in fluorite form
could easily be grown above 160 K. However, because of the
structural differences of CoSi2 and FeSi2 very low miscibil-
ity was expected.12 In order to learn more about the ternary
phase formation between Co, Fe and Si and about the solu-
bility of Fe in CoSi2, we carried out Mo¨ssbauer and x-ray
diffraction studies on bulk FexCo(1 x)Si2 in a broad concen-
tration range and on thin layers of FexCo(1 x)Si2 formed
upon ion implanting and annealing epitaxial CoSi2 on Si.
     The bulk ternary disilicide samples were synthesized us-
ing 4N metals and Si of very high purity. The elements were
melted together four times in an induction oven in vacuum of
better than 1 10 4 Pa. For the samples of low Fe concen-
trations 57Fe  enriched to 96%  was used. The weight loss
after melting was less than 0.05%. The annealings were
made in vacuo of  1 10 4 Pa. The CoSi2 layers epitaxi-
ally grown on  111  Si with 1000 Å thickness were im-
planted with 57Fe in the Leuven isotope separator at 80 keV

a Also at: Research Institute for Technical Physics and Materials Science;
 Electronic mail: dezsi@rmki.kfki.hu                                                              FIG. 1. Mo¨ssbauer spectra of FexCo(1 x)Si2 .

0003-6951/98/72(22)/2826/3/$15.00                                             2826                                   © 1998 American Institute of Physics



Appl. Phys. Lett., Vol. 72, No. 22, 1 June 1998                                                                         De´zsi et al.        2827

values of the single lines agree with the values formerly ob-
tained for the (57Co)57Fe source5­7   0.08 mm/s   Ref. 13 
and for the 57Fe absorber  0.43 mm/s  in the epitaxial CoSi2
layer on the Si surface. In these two different samples both
lines were present but with opposite line intensities. For 57Co
in CoSi2 and for 57Fe in the CoSi2 layer, the ratios of the
relative intensities of SL1 and SL2 were larger than 3 for
57Co and smaller than 1/3 for 57Fe, excluding the fact that the
resonance lines belong to a quadrupole doublet. The   values
of the singlets indicate covalently bonded iron atoms. The
covalent character of the bonds in CoSi2 was explained
theoretically.14 Both lines can be attributed to iron in sites of    FIG. 2. X-ray diffraction pattern of Co0.5Fe0.5Si2 as-prepared bulk sample:
                                                                      1  CoSi
cubic symmetry and bonded to Si atoms. CoSi                                  2,  2   -FeSi2
                                                   2 has C1 fluo-
rite structure. Because it is a metallic conductor, the appear-      the transition rate slowed down considerably when the
ance of an after effect resulting in anomalous charge state of       sample consisted of  -FeSi
iron after electron cyclotron  EC  decay can be excluded.                                               2 and CoSi2. For x 0.5, com-
                                                                     plete transition could be reached after annealing at 750 °C
Since the SL1 line always appears with very high relative            for 480 h. Therefore, the presence of Co does not decrease
intensity for (57Co 57Fe sources in CoSi2 , either diffused or       the     transition temperature, but the presence of the
implanted and subsequently annealed,5­7 this line can be at-         CoSi
tributed to 57Fe formed after EC decay in the Co lattice po-               2 phase retards the transition. At x 0.9, after long time
                                                                     annealing the spectrum of pure  -FeSi
sition. If Fe were positioned in Si sites, a lower   value                                                           2 appeared. The x-ray
                                                                     diffraction pattern showed only the lines of  -FeSi
would be expected than the value measured in the Co posi-                                                                          2, indicat-
                                                                     ing that Co is soluble at x 0.9.
tion because in this case Co atoms are the nearest neighbors              The as-implanted CoSi
of Fe. Since a larger   value is found, it is highly probable                                           2 thin layer samples showed
                                                                     quadrupole split doublets  one spectrum of the sample im-
that the iron is positioned in the vacant sites. The higher          planted with 6 1016 atom/cm2 dose is shown in Fig. 3 a  
stability for the iron in this position is shown by the increase     with   and  EQ values characteristic of 57Fe in the amor-
of the relative intensity of the SL2 line upon thermal anneal-       phous phase.21 On annealing at 573 K for 10 min the spec-
ing of the as-prepared bulk samples  Fig. 1 b  . It is probable      trum of the sample implanted with the lower 3
that this position is also partially populated by Co atoms in         1015 atom/cm2 dose  Fig. 3 d  , resulted in the spectrum
pure CoSi2 but because of the random and relatively low              with   values obtained in the bulk FexCo(1 x)Si2 sample at a
population, these atoms have not been observed by other              low x value. The spectrum of the sample with 6
methods as yet. Above x 0.015 the spectral shape of the               1016 atom/cm2 dose changed after annealing at 573 K to
as-prepared samples changed more and more to an asymmet-             the asymmetric doublet characteristic of  -FeSi2 and also
ric doublet with different linewidth values. The spectra be-         showed the presence of a fraction of the  -FeSi2 phase  Fig.
came very similar to those earlier measured for  -FeSi2.15,16        3 b  . After annealing this sample at 1023 K for 240 h, the
These spectra could be fitted by a doublet with distribution in      spectrum changed to that of the pure  -FeSi2  Fig. 3 c  . In
the hyperfine interaction parameters  Fig. 1 c  .     0.22 1         the sample implanted with 3 1015 atom/cm2 dose, the av-
mm/s and   EQ  0.54(1) mm/s average values were ob-                  erage number of iron atoms in the 2     is the straggling 
tained.  -FeSi2 has tetragonal structure17 P4/mmm, in                thickness was 0.5 at. %. This value is in the range where the
which Fe is surrounded by eight Si atoms. The distribution
appears because of the structural vacancies in the  -FeSi2
lattice, thereby resulting in the distribution of the electric
field gradient and isomer shift at the iron sites. Only long
time annealing  for 750 h  at 1023 K transformed the quad-
rupole split spectra to the characteristic spectrum of
 -FeSi2-the stable phase below 1210 K  Figs. 1 d  and
1 e  . The transformed spectrum was fitted by considering
the two possible pairings of the resonance lines: either the
1,3 and 2,4 or the 1,4 and 2,3 lines for the two doublets.18,19
In both cases the quality of the fitting was the same. We
chose the 1,3, 2,4 pairing to enable direct comparison of the
data with the earlier published ones.  1 0.02(1),  2
 0.15(1), EQ1 0.44(1) and  EQ2 0.41(1) values  in
mm/s  were found. In the x-ray diffraction patterns of the
as-prepared FexCo(1 x)Si2 samples  shown in Fig. 2 for x
 0.5), two phases  CoSi2 and  -FeSi2) could be identified
consistently with the Mo¨ssbauer spectra of the samples.
    Formation of the separate phases indicates that no homo-
geneous phase is able to form at x 0.015. The     trans-             FIG. 3. Mo¨ssbauer spectra of 57Fe implanted in epitaxial CoSi
formation kinetics was studied in detail.20 We realized that                                                                             2 layer on
                                                                      111  Si.



2828      Appl. Phys. Lett., Vol. 72, No. 22, 1 June 1998                                                                                     De´zsi et al.

iron populates two lattice sites in the CoxFe(1 x)Si2 bulk                        de Potter, J. Phys.  France   France  41, C1-425  1980 .
samples. At the higher dose value, the relative concentration                6 I. De´zsi, H. Engelmann, U. Gonser, and G. Langouche, Hyperfine Interact.
in 2  thickness is in the range where phase separation takes                      33, 161  1987 .
                                                                             7 A. Vantomme, M. F. Wu, I. De´zsi, P. Q. Zhang, and G. Langouche,
place, resulting in  -FeSi2 in the bulk samples. Probably                         Hyperfine Interact. 54, 2133  1990 .
because of the close structural relationship between the cubic               8 Z. Tan, F. Namavar, J. I. Budnick, F. H. Sanchez, A. Fasihuddin, S. M.
fluorite structure and the tetragonal lattice of the  -FeSi                       Heald, C. E. Bouldin, and J. C. Woicik, Phys. Rev. B 46, 4077  1992 .
                                                                 2, the      9
  phase forms first at 573 K and only after annealing at 873                      Z. Tan, F. Namavar, S. M. Heald, and J. I. Budnick, Appl. Phys. Lett. 63,
                                                                                  791  1993 .
K does the orthorhombic  -FeSi2 phase form, which is                        10 A. Vantomme, M. F. Wu, G. Langouche, J. Tavares, and H. Bender, Nucl.
stable at this temperature.                                                       Instrum. Methods Phys. Res. B 106, 404  1995 .
    The results clearly show for Co, Fe and Si ternary phases               11 N. Motta and N. E. Christensen, Phys. Rev. B 43, 4902  1991 .
                                                                            12
that a homogeneous single phase may form only in limited                          A. Wittman, K. O. Burger, and H. Novotny, Monatsch. Chem. 92, 961
concentration ranges.                                                              1961 .
                                                                            13 We use here the absorber convention, the negative sign of   for the 57Co
    The authors thank Dr. B. Molna´r and F. Gazda´cska for                        source is taken to be positive.
                                                                            14 F. Tersoff and D. R. Hamman, Phys. Rev. B 28, 1168  1983 .
their help in preparing the samples. The work was supported                 15 C. Blaauw, F. van der Woude, and G. Sawatzky, J. Phys. C 6, 2371
by Grants Nos. T 4405  OTKA , ERBCIPDCT940021  EU                                  1973 .
and under cooperation agreement between OMFB and the                        16 F. A. Sidorenko, P. V. Geld, V. Ya. Elner, and B. V. Ryzhenko, J. Phys.
Ministry of the Flemish Community  Grant No. B14/96 .                             Chem. Solids 43, 297  1982 .
                                                                            17 NBS Monogr. 25, 73  1984 .
                                                                            18 R. Wandji, C. Le Corre, J. M. Genin, and B. Roques, Phys. Status Solidi
1 S. Mantl, Mater. Sci. Rep. 8, 1  1992 .                                         B 45, K123  1971 .
2 H. von Ka¨nel, Mater. Sci. Rep. 8, 193  1992 .                            19 M. Fanciulli, C. Rosenblad, G. Weyer, A. Svane, N. Christensen, and H.
3 N. Cherief, C. D'Anterroches, R. C. Cinti, T. A. Nguyen Tan, and J.             von Ka¨nel, Phys. Rev. Lett. 75, 1642  1995 .
 Derrien, Appl. Phys. Lett. 55, 1671  1989 .                                20 T. I. Sigfu´sson and O¨. Helgason, Hyperfine Interact. 54, 861  1990 .
4 H. Lange, Mater. Res. Soc. Symp. Proc. 402, 307  1996 .                   21 P. Zhang, L. Urhahn, I. De´zsi, A. Vantomme, and G. Langouche, Hyper-
5 I. De´zsi, R. Coussement, G. Langouche, B. Molna´r, D. L. Nagy, and M.          fine Interact. 56, 1667  1990 .