Elevtrochimica Ana . 1916, Vol . 21, pp 45-49. P°rgamon Press.

Printed in Great Britain

THE fcc-hcp PHASE TRANSITION IN

ELECTRODEPOSITED EPITAXIAL COBALT FILMS*
H . L . GAIcHER and N . G . VAN DER BERG
Department of Physics, University of Pretoria, Pretoria . South Africa

(Received 17Apri11974)

Abstract-Thin (50-500 A) films of cobalt were electrodeposited from sulphate solutions having pH
values in the range 1 .9-5 onto thin (001) single-crystal copper substrates . Bicrystals and stripped
deposits were examined by transmission electron microscopy and electron diffraction . Deposits from
solutions with 45 gIl boric acid were essentially fcc for all pH values . Stacking faults, associated with
imperfect misfit dislocations, converted only a very small fraction of the fcc cobalt into the hcp
structure . Deposits from lightly-buffered (6 g) I boric acid) solutions contained small crystallites of hop
cobalt provided the solution pH z 3 . Growth of the hcp phase is probably associated with the presence
of colloidal hydroxide at the cathode surface .

INTRODUCTION Unless otherwise stated the copper substrates were

The structure of polycrystalline electrodeposited immediately afterwards transferred to the electro-
cobalt was shown to depend on the pH of the lyte surface such that deposition occurred on the
plating solution[1,2] . For values of pH't 1 .9[2] clean side of the substrate, ic, the side which was in
both hcp or-cobalt as well as fcc 13-cobalt, normally contact with the salt . Electrical contact with the
only stable above 425°C, were observed . At higher floating substrates was made by means of a strip of
pH values the deposits consisted of hcp cobalt aluminium foil[5] .
only . Goddard and Wright[3] investigated the struc- In one series of experiments the substrates,
ture of single crystal films of cobalt grown epitaxi- floating on water, were scooped onto clean glass
ally by electrodeposition onto the (WI) and (110} slides and allowed to dry in the atmosphere for
planes of electropolished copper single crystal sub- 1-2 h . The edges of the substrates, which adhered
strates . The phase formed during deposition was to the glass slides, were subsequently painted with
found to be primarily dependent upon the pH of the insulating lacquer such that -0-5 em' of the sub-
plating solution : For pH values less than -2-4 only strate surface remained exposed . The substrate on
fcc cobalt was deposited, but for pH values c- 3 the glass slide was suspended vertically in the
only the hcp structure occurred . In the pH range electrolyte . Subsequent deposition occurred on the
2 . 4-2 . 9 the deposits consisted of varying propor- side which was exposed during the entire prepara-
tions of fcc and hcp phases. These results were also tion procedure and therefore slightly oxidized .
confirmed by Wright[4] who investigated the elec- High purity cobalt served as anode . Two differ-
trolytic growth of cobalt onto {110] copper and ent electrolytes, made up from `AnalaR' chemicals,
nickel substrates . It was found that the phase were used : (1) Solution A[6] consisted of 312g/l
transformation not only depends on the pH, as CoSO4 . 7H,O, 45 g/l H,BO, and 20 g/l NaCl . (2) .
previously found, but also on the thickness of the Solution B, used by Goddard and Wright[3], con-
deposit . At a critical thickness of 350 .k the com- sisted of 300 g/l CoSO4 . 7H_O, 6 g/l H,BO, and 3 g/l
plete fcc-hcp phase transformation occurred, pro- NaCl . Deposition was carried out at -20°C and,
vided of course the pH exceeded -3 . unless otherwise stated, at a current density of
In the present work thin films of cobalt were 5 mA/cm' . Average deposit thickness were deter-
clectrodeposited onto thin (001) single crystal cop- mined by application of Faradays' laws .
per substrates which were prepared by vacuum Following deposition the Co-Cu bicrystals were
evaporation . Two electrolytes. which differed washed, mounted on grids and examined in the
mainly in the degree of buffering, were used . The electron microscope.
structure, in particular the fcc-hcp phase change, Stripped Co-films (thickness --200,k) were ob-
of deposits obtained from these electrolytes were tained by dissolving the copper substrates in a
examined as a function of deposit thickness solution of 100 g/l KCN at --40°C .
(50-500 .k) and pH (1 . 9-5) by means of transmis-
sion electron microscopy and electron diffraction . RESULTS AND DISCUSSION
Deposits from solution A
EXPERIMENTAL . For the ranges of solution pH and film thickness
Thin (--1500 A) (001) single-crystal substrates investigated all the cobalt films deposited from this
were prepared by evaporating high purity (>99-9%) electrolyte were essentially fcc single crystals in
copper in a vacuum of 10 - ' torr onto air-cleared parallel orientation with the substrate, ie,
sodium chloride held at ---320°C . The copper films (001)[100])-CoQ(00l)[1W]Cu . A typical transmis-
were removed from the sodium chloride by lower- sion electron diffraction pattern of a stripped cobalt
ing the latter, copper side upward, into water . film is shown in Fig. I . In addition to the normal fee
spots due to the (001) orientation, twin spots,
* Part of a project of The Institute of Microstructures . streaks and pseudospots were observed . The

45
46 H . L. GAIGHER and N . G . VAN DER BERG

to depend on the pH of the plating solution .

However, it should be pointed out that a really
reliable comparison of dislocation densities was
ruled out due to (a) uncertainty as to the exact local
deposit thickness and (b) the difficulty of resolving
misfit dislocations in the fields of overlapping
contrast associated with extensive twinning, espe-
cially in deposits plated from high pH solutions .
Diffraction contrast experiments[10, 11] revealed
that almost all of these misfit dislocations are
imperfect edge dislocations with Burgers vectors of
the type 1-' (112) . The imperfect misfit dislocations
are generated by glide of partial dislocations on the
{111} planes of the deposit[l0, 11] . Ribbons of
E stacking faults are thus formed on the {I III planes
Fig . 1 . Electron diffraction pattern from a 400 A thick of the deposit . Figure 3 shows such a cross-grid of
stripped cobalt film prepared from solution A with pH of stacking faults in a stripped cobalt film . The
4 . 5 . Beam nearly parallel to [001] . creation of these stacking faults may of course be
regarded as introducing a thin layer (two close-
streaks are in (I11) reciprocal lattice directions and packed planes) of hcp material . Misfit dislocations
are due to stacking faults or very thin twins. therefore not only relieve misfit strain, but also
Pseudospots occur where the sphere of reflection convert some of the fcc cobalt into the stable hcp
cuts the streaks[7, 8] . On tilting the specimen the structure[11] .
positions of these pseudospots changed as the
points of intersection of the Ewald sphere and the
streaks moved[8] . No spots which could conclu-
sively be attributed to crystallites of hcp material
were ever observed in these diffraction patterns. In
particular, hcp a-cobalt in the orientation (001)a-
CoIj(001)Cu reported by Goddard and Wright[3] for
pH > 2 . 7, was not observed .
Electron microscopy of the bilayer films revealed
ripplelike contrast effects for very thin (thickness <
100 A) deposits of cobalt . These contrast effects are
similar to those previously obtained for thin nickel
deposits on copper[9] and are due to local strain
fields in the bicrystal films .
Misfit dislocations first appeared at average
deposit thicknesses of --100A . The misfit disloca-
tions were parallel to the mutually perpendicular
(110) directions in the (001) specimen plane. Fig . 3 . Crossed grid of stacking faults in a (001) stripped
Increasing the deposit thickness resulted in a cobalt film . Deposit thickness 400 A . Solution A, pH 1 . 9 .
rapidly increasing misfit dislocation density . Figure x49300 .
2 shows a typical crossed grid of misfit dislocations
in a Co-Cu bi-crystal . The average thickness of the However, only a very small fraction of the fcc
deposit at which misfit dislocations first appeared, cobalt can in this way be transformed into the
as well as the density of misfit dislocations for a hcp structure . For example, if all the misfit were
particular average deposit thickness, did not seem accommodated by imperfect dislocations the frac-
tion transformed is only --0 . 2[10] . A significant
fraction may be transformed by martensitic-type
transformations[12, 10] . The hcp phase thus gener-
ated will be oriented with (0001) hcp 11{111} fcc .
For specimen orientations close to (001) it is
difficult, if not impossible, to distinguish unambig-
ously between pseudo-spots, twin spots and
spots due to the above-mentioned hcp
orientations[] 3, 14] . The specimen was therefore
tilted until the electron beam was incident in the
exact [011] direction . Complications due to
pseudospots do not arise for the exact (011)
orientation since the streaks, as well as spots due to
fcc, twins and hcp, all lie exactly in the (011)
plane[ 131 . Figure 4 shows a typical (011) diffraction
2 pattern . Long streaks, parallel to (I11), extend from
Fig. 2 . Transmission electron micrograph of misfit dislo- one matrix point to another and are due to stacking
cations in a (001) cobalt-copper bicrystal. Deposit faults and probably also very thin twins . Twin spots
thickness 150 .A . Solution A, pH 1 .9 . X49300. appear at certain positions (Fig . 4) displaced by
 The fec-hcp phase transition in electrodeposiled opitaxial cobalt films 47

Fig . 4 . Electron diffraction pattern from a (001) stripped Fig . 6 . Dark field micrograph of a (001) stripped cobalt
cobalt film with the beam parallel to [011] . Arrows show film, formed with a 111 secondary twin reflection . Deposit
some twin spots . Deposit thickness 500 A . Solution A, thickness 500 A . Solution A, pH 5 . x49300
pH 3 1 .
twinning . When multiple twinning occurs on {111}
;(111) from the matrix spots[131 . No spots charac- planes of a (001) film some {lll] planes of the
teristic of the hcp phase were observed . Some of secondary twins are nearly parallel to the electron
these spots should for example have been in beam[l5,161, and thus give rise to the observed
positions halfway between the matrix spots and the {111} reflections, lying on (110) reciprocal lattice
associated twin spots of Fig . 4[13] . vectors, Fig . 5 .
These results indicate that the extent of the fcc to Whereas pronounced {111} spots were charac-
hcp transformation was quite small, being limited to teristic of the diffraction patterns obtained from
the occurrence of stacking faults and perhaps very deposits grown on contaminated substrates, these
thin twins . spots were by no means a unique feature of these
The possibility that slight contamination of the deposits . Similar spots were also observed sporadi-
substrate might aid the formation of hcp material cally in diffraction patterns of deposits grown on
was next investigated . Cobalt was plated onto the clean substrates from relatively high pH (z3)
substrate surfaces which were slightly oxidized by solutions .
exposure to the atmosphere . Figure 5 shows a The occurrence of multiple twinning appears to
typical electron diffraction pattern obtained from be a consequence of the mode of growth . Multiple
such deposits for all the values of solution pH and twinning was most prevalent in those deposits for
deposit thicknesses investigated . Of particular which growth began with the formation of definite
interest are the innermost spots w hich. at first sight, three-dimensional island nuclei . For deposition
might appear to be {1010} spots due to (0001) onto clean substrates from pH < 3 solutions, for
hcpjl(00l)fcc . Careful d-spacing measurements which monolayer-like growth occurred,, multiple
showed that these spots are due to { 111} fee cobalt twinning was practically absent .
reflections and did definitely not arise from the hcp
phase . Dark field images formed by these fill} Deposits from solution B
reflections revealed a high density (-10tt lc-m') of Electrolyte B is the one originally used by
small (5(l-200 A) crystallites, Fig . 6. These results Newman[17] and subsequently by Goddard and
can be satisfactorily explained in terms of multiple Wright[3] .
The results below were obtained for deposition
onto floating substrates and a cd of 10 mA/cm',
which was also the cd used by Goddard and
Wright[3] . Lowering of the cd to 5 mAlcm', the
value previously used in the present study, did not
have a noticeable effect on the structural features
investigated .
The results obtained for the structure of deposits
plated from solutions with pH < 3 were in all
respects similar to those of deposits grown from
solution A . However, in contrast with the above
results, hop cobalt was observed for solutions with
pH ? 3 . Figure 7 shows a typical diffraction pattern
from a cobalt film after deposition from solutions of
pH Ze 3 . Measurements showed that the extra spots
correspond to those expected from hcp cobalt
Fig. 5 . Electron diffraction pattern from a 100A thick crystallites oriented such that (001)(210] hcp
stripped cobalt film prepared on an oxidized substrate CoII(001)(tl0)Cu . This observation is, except that
surface from solution A with pH of 1 9 . Beam parallel to complete transformation was not observed . consis-
tool l . The innermost spots have Ill indices and arc due to tent with the results of Goddard and Wright[3] and
multiple twinning . Faint oxide spots are also visible . Wright[4] .
48 H . L, GAIGHER and N . G . VAN DER BERG

Fig. 7 . Electron difraction pattern from a 500 A thick Fig . 9 . Bright field micrograph of a stripped cobalt film
stripped cobalt film prepared from solution B with pH of prepared from solution B with pH 22 . Deposit thickness
4, showing hcp reflections in addition to the (001) fcc 400 A . x49300
reflections.

A dark field image, Fig . 8, formed with a (100)

n-Co reflection shows that the a-Co phase is in the
form of small crystallites (100-500 A) with a density
of -10 10 per cm' .
Figures 9 and 10 are bright field transmission
electron micrographs typical of deposits obtained
from solutions with pH < 3 and pH > 3 respec-
tively . The relatively long, straight misfit disloca-
tions and associated stacking faults characteristic
of deposits grown from pH < 3 solutions, Fig . 9,
were never seen, at any thickness, in the films
which contained the hcp phase, Fig . 10 .
The two electrolytes used differed primarily in
their H,BO, and NaCI contents, Solution A had, as
compared to solution B, much higher H,BO, and
NaC1 contents . The question arises as to which of Fig- 10 . Bright field micrograph of a stripped cobalt film
these bath constituents is . of primary importance in prepared from solution B with pH 4 . Deposit thickness
determining the structural properties in question . 400 A. x49300 .
Both H,BO, and NaCI could influence the structure
of the deposit : Boric acid through its buffering 45 g/l H,BO, and 3 g/l NaCI . but did occur with
effect and NaCI through the effect of the sodium solutions containing 6g/l H,BO4 and 20g/I NaCl .
and chloride ions on the amount of colloidal, as These, together with the previous results suggest
opposed to flocculated, basic material at the that the hep phase is primarily associated with the
cathode surface[l2] . degree to which the solution is buffered .
Additional experiments have been initiated to For the case of nickel electrodeposition it has
determine the effect of these electrolyte con- been suggested[18] and convincingly cur-
stituents on the structure of the deposits : The hcp roborated[19] that the catholyte pH increases
phase was not observed with solutions containing during deposition . The local rise of pH can be
sufficient to result in the formation of colloidal
hydroxides or basic salts . This should also apply to
cobalt deposition since the cobalt and nickel
potential-pH diagrams are very similar[20] .
Thus, in lightly-buffered solutions an appropriate
rise in catholyte pH will result in the formation of
colloidal cobalt hydroxide at the cathode surface .
The cobalt hydroxide will interfere with the normal
process of crystal growth and can conceivably
result in the formation of the hcp phase . This is in
agreement with a suggestion that for vapour-
deposited epitaxial cobalt films substrate contamin-
ants promote the occurrence of hcp cobalt in
preference to fcc cobalt[21] . The presence of
colloidal hydroxide at the cathode surface may
furthermore induce a loss of coherency between
Fig . S . Dark field micrograph of a stripped cobalt film, overgrowth and substrate and hence explains the
formed with a 1010 hcp reflection. Deposit thickness absence of misfit dislocations .
500 A . Solution B, pH 4 . x49300 The present results implicate that the observed
 The fcc-hcp phase transition i n electrodeposiled epitaxial cobalt films 49

a-Co crystallites were formed by nucleation and REFERENCES

growth and not by phase change mechanisms 1 . H . Kersten, Physics 2, 274 (1932) .
involving the generation and movement of misfit 2- Yu . M . Polukarov, Russ . J. phys . Chem . 34,68 (1960)-
dislocations at a relatively advanced stage of 3 . J . Goddard and J. G . Wright, Br. J. app!. Phys . 15 . 807
growth . (1964).
4. 3 . G . Wright . Thin Solid Films 22, 197 (1974) .
5 . E . R . Thompson and K . R . Lawless, Appl. Phys . Left.
CONCLUSIONS 9, 138 (1966) .
6 . H .Fischer,ElektrolytischeAbscheidungundElektrok-
Transmission electron microscopy and electron ristallisation von Metallen, p . 626 . Springer, Gottingen
diffraction studies of the structure of thin cobalt (1954) .
films, electrodeposited from sulphate solutions 7 . P . B . Hirsch, A . Howie, R . B . Nicholson, D . W .
onto vapour-deposited (001) single-crystal copper Pashley and M . J. Whelan, Electron Microscopy of
substrates, have shown that : Thin Crystals, p . 132. Butterworths, London (1965).
8 . M . I . Abdalla and D . B . Holt, Phys . star. sot. (a) 17,
1 . Deposits from solutions with 450 H 3BO, 267 (1973).
9. H . L. Gaigher and G . N . van Wyk, Thin Solid Films
were essentially fcc with twins and stacking faults .
15, 163 (1973) .
The stacking faults converted only a very small
10. W . A . Jesser and J . W . Matthews, Phil . Mag . 17,461
fraction of the fcc cobalt into the stable hcp
(1968) .
structure . 11 . W . A . Jesser and J . W . Matthews, Ada Met. 16, 1307
2 . In deposits onto oxidized substrates multiple (1968) .
twinning was prevalent . i2. A . Seeger, Z. Metailk 44, 247 (1953) .
3 . Deposits from lightly buffered (6g/l H,BO,) 13. K . Otsuka and C . M . Wayman, Phys. star . sot. 22 .579
solutions with pH r 3 contained crystallites of (1967) .
a-cobalt in the orientation (001)[2101¢-COl(001) 14. D. W. Pashley and M. J . Stowell, Phil. Mag. g, 1605
(I10)Cu . (1963) .
15 . S . Inn, J . phys. Soc . Japan 21, 346 (1966) .
4. The occurrence of the a-cobalt crystallites
16. J . G. Allpress and J . V . Sanders, Surf. ScL 7, 1 (1967) .
appears to be a growth phenomenon associated 17. R . C . Newman, Proc . phys . Soc. B 69, 432 (1956).
with the presence of colloidal hydroxide at the 18 . J . MacNaughtan, G . E . Gardam and R . A . F .
cathode surface . Hammond, Trans . Faraday Soc . 29, 729 (1933),
l9. A . G. Ives, J . W. Edington and G . P . Rothwell,
Electrochim . Aria 15, 1797 (1970) .
20. M. Pourbaix, Atlas of Electrochemical Equilibria in
Acknowledgements-We wish to thank the University of Aqueous Solutions, p . 322. Permagon Press, London
Pretoria for financial support and Prof . J . H . van der (1966) .
Merwe, Head of the Department of Physics, for his help 21 . K . Otsuka and C . M . Wayman, Phys . star . sot . 22 .559
and encouragement . (1967) .