Epitaxy of Nickel Electrodeposits on a Copper (1 10) Face, from a Sulphamate Bath, in Relation to Rate of Deposition, Deposit Thickness, Degree of Stirring, and Bath Temperature BY S. IS. VERMA AND H. WILMAN" Applied Physics and Chemistry of Surfaces Laboratory, Department of Chemical Engineering and Chemical Technology, Imperial College of Science and Technology, London, S. W.7 Received 5th June 1972 Grazing-incidence electron diffraction at 50-60 kV shows the surface structure of smooth electro- polished Cu (110) faces and of the epitaxial Ni deposits on them from the bath : The Ni deposits were all f.c.c. The effects of c.d. (up to 600 mA/cm2), thickness (100-60 0008,), temperature (50 and 20"C), and vigorous stirring, are shown in detail. The effects of codeposition of Ni(OH)2 at high c.d.are also shown. Special points of interest are : (1) at 50°C although (1 11 1 twinning of the " parallel " epitaxial Ni was strong near the interface (100-200A Ni deposits) at up to -150 mA/cm2, it was weaker at 200 and absent at 300-600 mA/cm2 (at least up to thousands of 8, thickness), and this is similar to our results for Ni on (100) and (111) Cu; (2) at up to -150mA/cm2 at 50°C with unstirred bath the strongly twinned Ni at up to lo00 8, thickness showed no " directed disorientation " though accom- panied by an extremely small trace of widely disoriented (random?) Ni, possibly associated with the epitaxial misfit and contacting crystal nuclei ; (3) at 10 OOO 8, or more, at up to -150 mA/cm2 at 50°C with unstirred bath, additional weak arcs showed a smaIl amount of (100) Ni, and at 30 000- 60 OOO8, these arcs became strong and showed a (100) Ni orientation but limited to azimuths associa- ted with those of the (111) twins of the " parallel " Ni, arising from the twins by a " directed dis- orientation " (a range of rotation round the horizontal [lTO] Ni direction, presumably due to contacts between the twins, as they grow, with misfit stresses and strains at their interface), followed by cube face development leading to preferred growth of crystals in { 100) orientation.Ni(S03*NH2)2 -4Hz0 350 gll., NiClz -6H20 5 gll., H3B03 35 g/l. In epitaxial crystal growth on an atomically highly smooth substrate, twinning in the initial very thin deposits is usually attributed to the " pseudomorphic " tendency and the associated stresses arising from the misfit between the substrate and the deposit at their interface.In addition to twinning, disorientation effects are also sometimes observed, evidently arising from the same causes, and presumably involving dislocation movements and slip processes. A " directed disorientation ", consisting of a range of rotational displacements about a particular densely-populated lattice row of the interface plane, away from the ideal epitaxial orientation, was observed by Evans and Wilman first in the case of ZnO formed as a reaction product on a zinc- blende (110) cleavage face heated in air at 520°C. No reaction occurred at 500°C; thus at 520°C the reaction must have been less rapid than in several previous studies at about 650"C, where also recrystallization must have been occurring and promoting development of the single orientation then observed.Such " directed disorienta- tions" have also been observed in this laboratory in epitaxial electrodeposits on highly smooth single-crystal faces, for Zn and Cd on Cu Cu on Ag(100) and (1 and in Fe and Cr on Cu(lll), (100) and (110).4 Kumar's observation that 185186 EPITAXY OF NICKEL ON A COPPER (110) FACE such a disorientation occurred in Cu on Ag, but not in Ag on Cu, appears to corres- pond to the Cu deposit with axial length 3.61 A being in tension and thus less stable when growing on Ag (a = 4.08 A), whereas for Ag on Cu the Ag is under compres- sion and thus more stable when constrained to fit on to the smaller Cu lattice. In the (21 1)-oriented Fe and Cr deposits on Cu (1 lo), Reddy and Wilman observed a " one-sided directed disorientation ", and various angles of rotational displacement from the ideal (211) orientation, as well as extensive continuous ranges of rotation.In the present system of Ni on Cu (1 lo), no directed disorientation was found in the initial thin epitaxial (parallel) Ni, but one occurs in the upper regions of thick layers ( N 10 000 A and higher), which is associated with the octahedral twins of the " parallel " Ni and leads to the development of a moderately high proportion of an azimuthally limited { 1001 Ni orientation. The present study provides an interesting comparison with our results for Ni on Cu (100) and Cu (111) (to be published), and on mechanically polished polycrystalline copper,6 from the same sulphamate bath.This bath is known to give relatively stress-free ductile nickel deposits, and our electron diffraction observations showed that the nickel is uniformly f.c.c. in structure, free from any close-packed hexagonal nickel such as is partly obtained when the usual sulphate bath with a higher propor- tion of NiC12 is used. This makes the consideration of the epitaxy correspondingly simpler. The results are obtained up to the unusually high current density of 600 mA/cm2, corresponding to N 1500 A/s rate of deposition. EXPERIMENTAL The experimental details were as described by Verma and Wilman.5* The electrolyte consisted of nickel sulphamate, Ni(S03NH& -4Hz0 350 g/l. ; NiClz -6H20 5 g/l. ; H3B30 35 g/l.; it was supplied by British Insulated Callender's Cables Ltd., who obtained it from Messrs. Albright & Wilson Ltd. The electrodeposition was carried out in 60ml of this electrolyte in a Pyrex beaker of 5 cm diam., surrounded by a water jacket at the desired bath temperature. The freshly electropolished copper crystal was mounted with its (1 10) face 3 mm below the surface of the electrolyte and 3 cm above the horizontal nickel-sheet anode. Before deposition from the bath at 50"C, the crystal was immersed for 5 min in some of the solution at 50°C so as to attain this temperature. After nickel was deposited at a desired c.d. for the desired time and thus thickness, the crystal was removed without switching off the current ; it was washed immediately with distilled water followed by acetone to remove the water, and then covered with isopropyl alcohol to minimize oxidation during transfer to the Finch electron diffraction camera, which was then evacuated.Electron diffraction photographs were obtained with the -55 kV electron beam at low grazing incidence and with a specimen-to-plate distance of about 48 cm. The deposit thickness was estimated from Faraday's law, allowing for the current effi- ciency of the deposition, which was determined by a series of experiments on a larger sheet- copper cathode, by weighing the deposit on an Oertling semi-microbalance. The current efficiency at 50°C was virtually 100 % up to about 300 mA/cm2 when there was no stirring of the bath (and up to 360 mA/cmZ when vigorous stirring was used), falling to about 75 % at 600 mA/cm2.6 Nickel deposits were removed by boiling in orthophosphoric acid, and the copper (1 10) face was then re-electropolished and used again as substrate.RESULTS The electron diffraction patterns from the electropolished copper (1 10) face were similar in type to fig. 8 and 13, showing the surface to be highly smooth and flat, on the atomic scale of dimensions (cf. Verma and Wilman 5 ) . The nickel electro- deposits were in all cases f.c.c. Above a certain c.d. and thickness range, some nearly amorphous greenish nickel hydroxide was co-deposited (see fig. 1, 14, 15).FIG. 2.-Electron diffraction from 200 8, Ni at 100 mA/cm2 at 50°C unstirred ; beam along [lYO] of Cu ; note twinning spots. FIG. 3.-As fig. 2 but 10008, Ni.FIG. 4.-As fig. 2 but 10 000 8, Ni ; note faint 200 arc in plane of incidence. FIG. 5.-(a) (6) (4 extending from spots, FIG. 5.-As fig. 2 but 61 000 8, Ni : (a) beam IICu[lTO] ; (b), (c) beam11 Cu(211). Note strong arcs [To face page 1 86FIG. S.-5008, Ni at 300 mA/cm2, 50°C, unstirred bath ; note absence of twinning. FIG. 9.-As fig. 8 but 51 000 8, Ni ; strong twin- ning but smooth horizontal surface. FIG. 10.-10 200 8, Ni at 600 mA/cm2 ; 50°C, FIG. 11.-30 600 8, Ni at 100 mA/cm2 ; 5OoC, bath stirred vigorously ; surface region mainly in azimuthally limited (100) orientn. unstirred ; after Ni(OH)2 etched away. FIG. 12.-Optical micrograph ( x 800) of Ni de- FIG. 13.-34 000 8, Ni at 400 mA/cm2 ; 50°C, posit of fig. 11. bath stirred vigorously.S . K .VERMA AND H . WILMAN 187 NICKEL ELECTRODEPOSITS AT 50°C WITH THE BATH UNSTIRRED Fig. 1 shows the nickel orientations indicated by the electron diffraction photo- graphs, in the surface region (to < about 100 A depth) of nickel deposits prepared at various c.d. and thicknesses. d T T d T Region I(A) Region D(B) %t I o n I3 rl 0 100 2 0 0 3 0 0 400 5 0 0 600 current density/(m A /cm2) FIG. 1 .-Electron diffraction observations on the surface structure of nickel electrodeposits from the sulphamate bath on Cu (110) at 50°C with bath unstirred. 0 = parallel epitaxial single-crystal Ni (region I) ; t , T, a trace of, or much, octahedral twinning of the " parallel " Ni ; r, R, a small amount of, or much, random polycrystalline Ni ; / = one-degree {loo} orientation but azimuthally limited, with a cube edge near to a <211> direction of the Cu (110) substrate ; 0 = Ni(OH)2 surface layer obscuring the Ni (region 11)-the symbol inside the circle shows the Ni orientation after etching away the hydroxide, and region II(A) is where this Ni is still entirely epitaxial (parallel), while in region II(B) twinning and/or random Ni is found.In region I, parallel epitaxial crystal growth of the nickel occurs, and in regions I(A) and I(B), i.e., up to about 250 mA/cm2 at thicknesses up to 25 OOOA, and up to higher c.d. at larger thicknesses, the nickel was mainly in epitaxial orientation with its cube axes parallel to those of the copper, but with also strong { 11 11 twinning even in deposits only about 200 A thick, at up to about 150 mA/cm2 (cf.fig. 2-5). At 200 mA/cm2 the twinning in the very thin initial deposits was much less strong, and at -250-600 mA/cm2 it was absent up to a certain thickness (region I(C) in fig. l), as is seen from fig. 8 ; though twinning developed eventually at large thicknesses ; e.g., it was present at 51 000 A at 300 mA/cm2, as is shown by fig. 9-or else co- deposition of nickel hydroxide occurred (region II in fig. 1). The codeposition of nickel hydroxide (region I1 in fig. 1) was also soon accom- panied by strong twinning of the nickel and a progressively increasing proportion of randomly-oriented polycrystalline nickel, until at 60 000 A deposit thickness the nickel grew entirely as randomly oriented crystals. In region I(A) also, a barely visible trace of the nickel diffraction-ring pattern was present in photographs such as fig.3, from deposits up to a few 1000 A thick at up to188 EPITAXY OF NICKEL ON A COPPER (1 10) FACE 100 mA/cm2 or slightly more. This indicates that a very small trace of randomly oriented nickel was present, denoted by r in fig. 1. Patterns such as fig. 4 from the surface of deposits 10 000 A thick, showed a faint 200 diffraction arc centred on the plane of incidence. This arc became strong as the thickness was increased, as in fig. 5 from a 61 000 deposit at 100 mA/cm2. In fig. 5(a), obtained with the electron beam along the [liO] direction of the copper substrate, the spots from the " parallel " single-crystal nickel are strong, but are limited to a narrow [lTO] zero-order circular Laue zone which passes through the undeflected-beam spot and is centred about 24- 3 cm above the undeflected-beam spot (this copper face was then about 2" from a (1 10) plane, and the 220 Ni Bragg reflection is absent).Most of the strong spots are due to the (111) twins of this nickel lattice, forming component patterns of centred- J2-rectangle type (see fig. 6). \ I A \ FIG. 6.-Diagram showing (not to scale) the component centred-d2-rectangle spot patterns in fig. 5(a) (those of the two twins ABCO, AB'C'O, and the vertical-rectangle pattern from the " parallel " Ni), and the 200 arc E between A and A'. The arc pattern also present in fig. 5(a) appears to correspond to a preferred (100) nickel orientation but not an azimuthally random one which might have been expected to arise from the early small trace of random nickel if these crystals developed cube faces (nickel deposits on mechanically polished copper polycrystalline substrates were epitaxially random, like the copper substrate, but a preferred ( 100) orientation developed as the thickness increased).For example, fig. 5(a) does not show a strong arc on the 220 ring position, at the same level as the strong 200 arc (E in fig. 6), as would be expected from an azimuthally random (100) orientation. Fig. 5(b) and (c) were obtained at a (211) type of azimuth of the copper (and the parallel Ni), from neighbouring parts of the surface. They show that at this azimuth the electron beam was not far from along a cube axis of some of these (100)-oriented nickel crystals, thus giving rise to the square pattern of hkO arcs seen prominently in fig.5(c) and less strongly in fig. 5(b), where the vertically elongated nickel spots in J(8/3) vertical-rectangle array are strongest and confirm that this azimuth is indeed of (21 1) type, relative to the copper. The (100)-oriented nickel crystals present in these discrete azimuths must evidently be associated with the (Ill)-twin lattices of the " parallel " nickel lattice. Two of these twins, on (1 11) planes perpendicular to the Cu (110) substrate, have still a (110) plane parallel to the Cu (110) substrate, but the other two have a cube face at 19" 28' to the substrate and have a cube edge in thatS . K . VERMA A N D H . W I L M A N 189 cube face in a vertical plane through an azimuth of the Cu (1 10) substrate only a few degrees from a (211) type of direction, as is seen from fig.7. At this azimuth these two twins thus evidently give rise to the two spots of 200 type in fig. 5(b) and (c), one on each side of the plane of incidence on which the central 200 arc lies, with which they form a group similar to AEA' of fig. 6 (cf. fig. 5(a)). PLAN FIG. 7.-Side-view of cubic unit cell seen along [lYO] in the (horizontal) (1 10) plane, together with the { 11 1)-twin unit cell ; and plan view of these showing azimuthal relation of (21 1> and the cube edges of the twin. Since the arcs are longest in fig. 5(a), it seems probable that this arcing corresponds to a directed disorientation consisting of a range of rotation from the above two twin lattices, about the [lTO] lattice row which is parallel to the beam in fig.5(a). The strong spots from the twins in fig. 5(a) (cf. fig. 6 ) indeed show a noticeable tailing-off along the arc, on the side corresponding to a rotation in the sense which would bring the cube face nearer to the specimen surface (Cu (110) substrate) than their initial position at 19" 28' to it. The two component arc patterns from the two mirror- symmetrical twins, in fig. 5(a), are rotated in opposite sense, and the 200 arcs overlap across the plane of incidence. The strong (100) orientation (in these azimuths) must then be due to a development of cube faces on these twin crystals and thus a prefer- ential lateral growth and predominance of those twin crystals which (within the rotational range present) have the cube face most nearly normal to the incoming ion stream (cf.Verma and Wilman 6). The refractive drawing-out (downwards) of the strong 200 arc on the plane of incidence in fig. 5(a), (b), (c) is indeed evidence of such cube-face development. This one-sided directed disorientation from the twin orientations seems likely to190 EPITAXY OF NICKEL ON A COPPER (1 l o ) FACE be due to contacts between these two twins during their growth, with consequent generation of dislocations at their interface, and stresses in the interface region. At 300 mA/cm2, in spite of the very high rate of deposition (1000 A/s) and the general tendency for cube faces to be formed on the deposit crystals in the nickel deposits from this bath on polycrystalline copper substrates,6 the photographs such as fig.8 and 9 show by the long vertical elongation of the diffraction spots that the atomically highly smooth surface of the deposit (parallel to the Cu (1 10) substrate) at the early stages of growth is maintained up to large thicknesses. The { 1 1 1) twins in the thick deposits (51 OOOA for fig. 9) are also evidently bounded by a smooth face parallel to the Cu (1 10) substrate face, which is a (1 10) face of two of the twin lattices but a (1 14) face of the other two twin lattices giving the spots present in fig. 9 addi- tional to those from the " parallel " nickel in fig. 8 (cf. fig. 6 and 7). At 600 mA/cm2 (about 1500 A/s rate of nickel deposition), 1000 A deposits gave patterns similar to fig. 8, showing only the parallel epitaxial nickel with smooth (1 10) surface.Fig. 10, from a lO2OOA deposit, was obtained after etching away the amorphous codeposited Ni(OH)2, and shows that the epitaxial nickel and its (111) twins still tend to be bounded by a surface parallel to the copper substrate, though the vertical spot elongation is now shorter and the spots broader, indicating that these elements of surface are of smaller extent, i.e., the surface is less continuously flat. The rings also present in fig. 10 show the presence of much randomly oriented poly- crystalline nickel. At 61 000 A thickness, only a ring pattern from random nickel was observed when the obscuring Ni(OH), was etched away. NICKEL ELECTRODEPOSITS AT 50°C WITH THE BATH STIRRED VIGOROUSLY Fig. 14 shows the electron diffraction results.The locus to the right of which the codeposition of nickel hydroxide occurs is now displaced to much higher current densities, above about 500 mA/cm2. In region I to the left of this locus, the epitaxial " parallel " nickel continues during growth up to 60 000 A or more, accompanied by some (111) twinning, and also, as fig. 11 shows, by much (100) orientation, azimuth- ally limited, at 100 mA/cm2 and about 30 000 A thickness (similarly at 400 mA/cm2 and 50 000 A or more). Fig. 12 is an optical micrograph ( x 800) of the deposit which gave fig. 11, and it shows the rough topography of the deposit on the microscopic scale, as distinct from the submicroscopic form of the tips of these projections shown by fig. 11. Fig. 13 shows the highly smooth (1 10) surface of the " parallel " nickel, and the very small trace of twinning, in the surface region of a 34 000 A deposit at 400 mA/ cm2.A 25 500 A deposit at 500 mA/cm2 showed a similarly smooth surface but a larger proportion of twinning, both the parallel nickel and the twins being bounded by smooth faces parallel to the Cu (110) substrate surface. NICKEL ELECTRODEPOSITS AT 20°C WITH THE BATH UNSTIRRED Fig. 15 shows the electron diffraction results, which were made for comparison with the 50°C results of fig. 1. As in fig. 1 the region I where epitaxial nickel is obtained is bounded by a steeply rising locus on its right-hand limit, which now is at about 120mA/cm2. To the right, in region 11, there is codeposition of Ni(OH)2, accompanied by increasing twinning and random polycrystalline nickel.In region I there is again strong twinning of the " parallel " nickel, even at a few 100 A deposit thickness, at the lower current densities, and it is still considerable at 100 mA/cm2.S . K . VERMA A N D H . WILMAN 6 0 - 5 0 - 40- 2 2 -5 2 2 0 - 0" i? 3 0 - * 8 v1 44 .C( rn 10- 191 dt Region I t t t 1 I 0 100 2 0 0 3 0 0 400 5 0 0 6 0 0 current density/(mA/cm2) FIG. 14.-Electron diffraction observations on the surface structure of nickel electrodeposits from the sulphamate bath on Cu (110) at 50°C with the bath stirred vigorously ; symbols as in fig. 1. Region I r I 0 100 2 0 0 current density/(mA/cm2) FIG. 15.-Electron diffraction observations on the surface structure of nickel electrodeposits from the sulphamate bath on Cu (110) at -20°C ; bath unstirred.192 EPITAXY OF NICKEL ON A COPPER (110) FACE DISCUSSION There appear to have been no previous electron-diffraction results on the structure and growth of nickel electrodeposits from the sulphamate bath, apart froin our recent observations on nickel deposits on Cu (100) and on mechanically polished poly- crystalline copper.6 We find that the deposits from this bath are uniformly f.c.c., and this simplifies consideration of the conditions of growth at the interface with the highly smooth electropolished Cu (1 10) face. The strong twinning of the initial nickel deposits only a few lOOA thick at c.d.up to 100, or possibly 150 mA/cm2, seems to be attributable to the misfit of -2.5 % between the nickel and the copper lattices. This twinning, and possibly the barely detectable trace of random polycrystalline nickel present (cf. fig.l), may perhaps originate in the Ni-Cu interface regions where neighbouring initial nickel crystal nuclei meet together as they grow larger. The reduction in twinning at higher c.d. -200 mA/cm2, and its eventual absence in the initial deposits at 300-600 mA/ cm2, is an observation of a type which has not apparently been observed before, and indeed most studies have not extended up to such high c.d., i.e., high rates of deposition. It seemed likely hitherto that the probability of twinning would increase with increase in rate of deposition, there being then more chance of some atoms being trapped in the alternative set of potential troughs on the growing surface to nucleate a twin lattice, and being overlaid by other atoms and thereby to become built in.However, at 300-600mA/cm2 or more (1000-1500 A/s), the rate of arrival of energy at the cathode must be so high that there will be appreciable rise in temperature in the solution near the cathode, and in the growing cathode surface itself, so that there will be an increase in atomic (or adion) mobility on the growing cathode. In a sense, this can be said to be of the nature of an annealing effect. This decrease in twinning at these high c.d. is matched in our results for epitaxial nickel growth on electropolished Cu (1 11) faces (to be published), and also on the Cu (100) face where, in the latter case, we observed only an occasional faint trace of twin- ning at the lower c.d., and none at the higher c.d. range.On the Cu (100) face, we observed a greatly reduced tendency of the deposit to develop undulating or obliquely facetted surface in the higher c.d. range, at least up to large thicknesses; and this is also attributable to a rise in temperature at and near the cathode surface, with a corresponding diminution of the " outward-growth " tendency.' These results therefore reinforce our earlier conclusions as to the appreciable temperature rise in the surface region of growing deposits.s-lo The absence of any " directed disorientation " of the initial epitaxial nickel in the interface region is interesting, since the nickel is in tension; however, the misfit is only - 2.5 %, thus much smaller than for Cu on Ag, where it is - 11.5 % and a directed disorientation is obser~ed.~ In our results on the epitaxy of Fe and Cr on C U , ~ the misfit was only -2.7 % along the most densely populated atom rows, but about + 12 % in the perpendicular direction ; and extensive directed disorientations were observed. The observed axis of rotational disorientation is not always easily explained in terms of normal slip systems and processes, and it is not clear whether the rotations are of multiple translational slip type, or flexural translational slip (about an axis parallel to the slip plane and normal to the slip direction), or of lamellar rotational- slip type 11-15 with the slip lamellae perpendicular to the axis of rotation.' The directed disorientation in the thicker epitaxial twinned deposits seems to be similarly related to interface regions where two of the twins meet as they grow bigger.Menzer 16* l7 has discussed the fitting together of two such differently oriented twinsS . K . VERMA AND H . WILMAN 193 of a f.c.c. lattice such as nickel, and he concluded that certain superlattices should be possible at the interface region. However, there seems scope for an appreciable misregister of the lattices to occur, with consequent introduction of dislocations and stresses, which might cause such disorientations to develop and to increase during further growth of the deposit. The effects of stirring and of increase in bath temperature in extending region I to higher c.d. (cf. fig. 1, 14, 15) are similar in type to those discussed for Ni on Cu We thank Mr. E. H. Reynolds, Dr. R. M. Hinde and Mr. R. E. Davies of the Central Research and Engineering Division, British Insulated Callender’s Cables Ltd., for initiating the research and providing the materials. We also thank BICC Ltd for the bursary which enabled one of us (S. K, V.) to carry out the experiments. D. M. Evans and H. Wilman, Proc. Phys. SOC. A., 1950,63,298. A. P. Goswami, Ph.D. Thesis (University of London, 1950). D. N. Kumar, 1955, D.I.C. Thesis (Imperial College, London) ; see also H. Wilman, Acta Cryst., 1957, 10, 842. A. K. N. Reddy and H. Wilman, Trans. Inst. Metal Finishing, 1958-9, 36, 97. S. K. Verma and H. Wilman, J. Phys. D.: Appl. Phys., 1971, 4, 1167. S. K. Verma and H. Wilman, J. Phys. D.: Appi. Phys., 1971, 4, 2051. G. I. Finch and D. N. Layton, J. Eiectrodepositors’ Tech. Suc., 1951, 27, 215. H. P. Murbach and H. Wilman, Pruc. Phys. Soc. B, 1953, 66,905. * D. M. Evans and H. Wilman, Acta Cryst., 1952, 5, 731. lo H. Wilman, Pruc. Phys. SOC. B, 1955, 68,474. l 1 H. Wilman, Nature, 1950, 165, 321. l 2 H. Wilman, Proc. Phys. SOC. A, 1951, 64, 329. l 3 A. D. Whapham, J. Inst. Metals, 1956, 84, 109. l4 A. D. Whapham and H. Wilman, Nature, Lond., 1955, 176,460. l5 A. D. Whapham and H. Wilman, Proc. Roy. SOC. A, 1956,237, 513. l6 G. Menzer, Naturwiss., 1938. 26, 385. l7 G. Menzer, 2. Krist., 1938, 99, 378, 410.