Inhomogeneous electrodeposition of copper in a magnetic field

Electrochemistry Communications 11 (2009) 379–382 Contents lists available at ScienceDirect Electrochemistry Communications journal homepage: www.elsevier.com/locate/elecom Inhomogeneous electrodeposition of copper in a magnetic field Dámaris Fernández a,b, J.M.D. Coey a,* a b School of Physics and CRANN, Trinity College, Dublin 2, Ireland Departamento de Ciencia de Materiales, Universidad de Santiago, Chile a r t i c l e i n f o a b s t r a c t Article history: Received 10 September 2008 Received in revised form 14 October 2008 Accepted 31 October 2008 Available online 27 November 2008 Keywords: Magnetoelectrochemistry Electrodeposition of copper Lorentz force Magnetohydrodynamics Plating profiles Normally, magnetoconvection driven by the Lorentz force increases the limiting current in the masstransport limited regime, roughly as the one-third power of the applied magnetic field. Here we show that an applied field can actually diminish the rate of copper electrodeposition at low overpotentials. The effect is related to the formation of a vortex at the leading edge of the flow. Similar, but weaker effects are due to gravity. Ó 2008 Elsevier B.V. All rights reserved. Many of the effects of a magnetic field on processes occurring in an electrochemical cell can be traced to the Lorentz force. The magnetic field B interacts with the current density j to produce a body force FL ¼ j  B ð1Þ 3 2 The force density FL is in N m when j is in A m and B is in tesla. The force creates a convective flow on some scale, which has consequences for the rate of electrodeposition [1–3], the morphology of the electrodeposits [2,4], rates of corrosion [5,6] and the rest potential [7]. The first of these effects to be understood quantitatively was electrodeposition in the mass-transport limited regime. There, a quasi-equilibrium is set up at the cathode surface, where the mass flow of cations to the electrode is driven by the concentration gradient, giving a limiting current density jL ¼ nFDrc ð2Þ where n is the number of electrons transferred, F is Faraday’s constant (96,506 C mol1) and D is the diffusion coefficient. The concentration gradient rc is approximated as c0/d where c0 is the ionic concentration far from the electrode and d is the diffusion layer thickness, which is a distance of order 100 lm in typical electrochemical cells. The diffusion layer thickness is governed by the conditions of natural or forced convection in the cell. It is deliberately reduced by agitation in industrial plating baths, in order to speed up the * Corresponding author. E-mail address: [email protected] (J.M.D. Coey). 1388-2481/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.elecom.2008.10.057 plating rate. Where there is a component of applied magnetic field, transverse to the plating current, a flow of electrolyte is established across the cathode surface which is somehow equivalent to gentle stirring [2]. For example, plating rates may be doubled by applied fields of order 1 T. More recently, Weier et al. have discussed the flow structure in the cell in terms of the interplay of gravity-driven and magnetically-driven convection [8,9]. The phenomenon of field-induced stirring, known as the magnetohydrodynamic effect, was investigated and analysed by Aogaki [1], Fahidy [10] and others [11], deducing and adding contributions to the following basic expression for the increase of limiting current in a well-defined geometry 4=3 jL ðBÞ ¼ jL ð0Þ þ aB1=3 c0 ð3Þ where a is a positive constant depending on the properties of the electrolyte and the cell geometry. There is good evidence for the B1/3 variation in the electrochemical literature [1,2,10–12]. Nevertheless, there are circumstances where an applied magnetic field seems to reduce, rather than increase, the current in an electrochemical cell. These include plating of copper, cobalt [13] and other metals [14]. Here we investigate these circumstances for copper electrodeposition. Our main result is that it is possible for the Lorentz force to interfere with natural convection and create inhomogeneous plating currents across the surface of small electrodes, thereby reducing or even suppressing electrodeposition. The system we have studied is a CuSO4 + 1.2 M H2SO4 + 0.25 mM KCl bath with copper concentrations of 0.03–0.6 M. The working electrode was usually sputtered copper on silicon, which had a root mean square roughness of less than 1 nm. Coupons were D. Fernández, J.M.D. Coey / Electrochemistry Communications 11 (2009) 379–382 0.1 M CuSO4, B vertical, V a 2 0 Current density, A/m 10  10 mm2 and the cathode area was usually defined by a mask of 70 lm thick Kapton tape to be a circle of diameter 5 mm, but smaller areas and Teflon-embedded microelectrodes with diameters of 1000, 130 and 35 lm were also used. The counter-electrode was a ball of flattened Pt wire with an area of at least 40 times the working electrode, and a pseudo-reference electrode was made from Cu wire. Magnetic fields of up to 5 T were generated in the 110 mm vertical bore of a cryogen-free superconducting magnet where the field was parallel to the axis of the bore, and the electrodes were arranged vertically (V) or horizontally (H) with either the anode or cathode on top (a/c or c/a). The latter arrangement minimises natural convection. The vertical arrangement maximises magnetically-driven convection. Alternatively, an electromagnet with 200 mm pole faces was used to generate horizontal fields of up to 1.5 T. Most data were collected potentiostatically, with a Solatron model SI 1280B potentiostat at an overpotential of 40, 200 or 550 mV. A series of voltammograms in different applied fields are shown in Figs. 1 and 2a. The cell develops instabilities in high fields and at high overpotentials, especially at the higher copper concentrations. When the overpotential is sufficient to place the system in the mass-transport limited regime, and the working electrode is parallel to the field, an approximately B1/3 increase is observed, as expected (Figs. 1, 2a and b). This is also often the case at 200 mV. However, at the smallest overpotential of 40 mV, the behaviour is different. Here the field can clearly diminish the measured current density, as seen in Fig. 2b, which shows data for 0.1 M CuSO4. The effect of the field depends critically on the electrode configuration, as shown in Fig. 2c. When there is a gravitational contribution, it combines with magnetically-driven convection to increase or decrease the current. The influence of electrode area is illustrated in Fig. 3. Even in the mass-transport limited region, the field enhancement effect is reduced for the smaller electrode areas. In order to reveal what is going on, optical and scanning electron microscope images were compared with thickness profiles of the deposits, measured with a Dektak stylus profilometer. Results for the vertical electrode case at two different overpotentials are compared in Fig. 4. In the field-enhanced deposition conditions of Fig. 4a, the deposit covers all the exposed area and its thickness increases with applied field, as expected. However, in the conditions of Fig. 4b, the deposit has a quite different appearance; it does -100 η = 40mV -200 0.0 T 0.5 T 1.0 T 2.0 T 3.0 T 4.0 T η = 200 mV -300 η = 550 mV -0.6 -0.4 -0.2 0.0 Overpotential, V b 0.1 M CuSO4, B vertical, H c/a 3.5 3.0 Normalized shift in jMEAS 380 2.5 2.0 1.5 - 40 mV - 200 mV - 550 mV 1.0 0.5 0.0 0 1 2 3 4 5 B, Tesla 0.1 M CuSO4, η = - 40 mV c 40 30 /-j/, A/m 2 0.3 M CuSO4, B vertical, V Current density, A/m 2 0 0.0 T 0.5 T 1.0 T 1.5 T 2.0 T 2.5 T 3.0 T 3.5 T 4.0 T 5.0 T -500 -1000 -1500 -0.6 -0.4 -0.2 20 10 H - c/a V H - a/c 0 0 2 4 6 B, Tesla Fig. 2. (a) Linear sweep voltammograms for 0.1 M CuSO4 pH 1 electrolyte, at different applied fields with electrode in the vertical configurations. (b) Measured current versus applied magnetic field at different overpotentials for in the horizontal c/a configuration. (c) Measured current versus applied field for different electrode configurations are illustrated for an overpotential of 40 mV. 0.0 Overpotential, V Fig. 1. voltammograms of cathodic copper electrodeposition measured from open circuit potential, OCP, to a cathodic overpotential of 600 mV with a vertical electrode configuration and vertical magnetic field 0 6 B 6 5 T. [Cu2+] = 0.3 M, pH 1. not cover the whole exposed area, as can be seen from optical microscopy images. The profile is clearly non-uniform across the width of the substrate. Around the edges upstream of the flow, the deposit at 40 mV appears bald, and the bald area increases with D. Fernández, J.M.D. Coey / Electrochemistry Communications 11 (2009) 379–382 Normalized increment in jMEAS 3.5 0.3 M CuSO4, B horizontal, S vertical 3.0 2.5 2.0 1.5 35 µm 130 µm 1000 µm 5000 µm 1.0 0.0 0.5 1.0 1.5 B, Tesla Fig. 3. Increase in measured current, expressed as a normalized value j(B)/j(0) in 0.3 M CuSO4 pH 1 electrolyte, for electrodes of different areas at an overpotential of 550 mV. Magnetic field is applied horizontally, parallel to the vertical electrode surface. applied field. Furthermore the amount of copper deposited on the substrate decreases with field, whereas the current density increases slightly. The images show that field-induced convective flow is able to inhibit electrodeposition at low overpotentials, while the opposite is seen at high overpotentials, where a buildup of material at the leading edge of the flow occurs. Close examination of the deposit obtained at 40 mV without a field, Fig. 4b, shows traces of baldness there too, along the bottom edge. This is the leading edge for gravitationally-induced convec- 381 tion. The depleted copper solution is less dense than the surroundings, and buoyancy forces drive it upwards. The buoyancy force, for a density of 1015 kg m3 is 150 N m3, which is of the same magnitude as the Lorentz force (1) for a current density of 30 A m2 in a field of 5 T. The two forces are of similar magnitude in the low overpotential conditions. A similar case has been described by Uhlemann et al. [12] where reductions in measured current density are observed. As regards scanning electron microscope images, it is apparent that in those areas where the deposit is clearly formed, no difference in morphology can be attributed to the applied field. Moreover, we have observed that growth morphology of the bald zones produced either by gravitationally or magnetically-induced convection is similar. A plausible explanation of the non-uniform deposition is the formation of a back-streaming vortex or eddy localized near the leading edge of the deposit. Magnetohydrodynamic simulations for an ideal geometry where two parallel rectangular plates are immersed in a rectangular cell were carried out using the FLUENT code [15]. Three dynamic equations governing the system are solved simultaneously, namely the Navier–Stokes equation including FL, the continuity equation and the convective–diffusion equation. The concentration gradient at each point on the electrode surface determines j from Eq. (2), which in turn determines FL. At low Bj values, of order 10 T A m2, a back-streaming vortex appears in the simulations, as illustrated in Fig. 5 [15]. The Lorentz force density Bj needed to form a vortex is similar to that used in our experiments at low overpotential. This vortex will have a relatively greater effect on small electrodes. The step due to the thickness of the Kapton tape mask is not an essential feature, since we observe similar behaviour in a series of measurements with a polished copper rod sheathed in Teflon, where there is no such step. Fig. 4. Comparison of copper electrodeposits obtained in conditions where the field enhances (a) or diminishes (b) the deposition rate. The direction of the convective flow is indicated by the black arrows, where FNC and FL refer to natural convection and Lorentz forces, respectively. In (a) the typical appearance of deposits obtained after 300 s of potentiostatic deposition at an overpotential of 200 mV, very similar either by optical or scanning electron microscopy. The profiles, however, show that field enhances deposition rate and produce a build-up of material on the leading edge of the flow. In (b) results obtained after 300 s of potentiostatic deposition applying 40 mV of cathodic overpotential. By comparison of optical microscopy and profilometry it is clear that, in these conditions, the field inhibits growth by means of magnetically-induced flow. SEM images show that the deposition pattern is different only at the leading edge of the flow, while it remains similar over the rest of the sample. 382 D. Fernández, J.M.D. Coey / Electrochemistry Communications 11 (2009) 379–382 Fig. 5. Hydrodynamic simulations which show the formation of a vortex at the leading edge of Lorentz-force-induced flow of electrolyte past a planar cathode [15]. Its extension increases with the applied field. Velocities are of order 0.01 m s1. The reduction in thickness of the deposit in 4 T in Fig 4b indicates that in these conditions, the flow inhibits the build-up of reduced copper crystallites. It is in the conditions of a downwardfacing horizontal electrode, where the current and field are nominally parallel, but a strong circular flow nevertheless is established by the Lorentz force around the circular cathode [16], that both current density and deposit thickness are sharply reduced by the applied field. In conclusion, we have shown how the Lorentz force in the vicinity of a relatively small electrode can increase or decrease the plating current, as a result of the forced convective flow. Increases are observed with larger electrodes, and current densities in excess of 100 A m2 in the present system, when the Lorentz forces exceed those due to buoyancy. Strong decreases in measured current density and deposition rate are observed at low overpotentials, especially in a horizontal electrode configuration where gravitational convection is minimised. Acknowledgement This work was supported by Science Foundation Ireland, as part of the MANSE Project. References [1] R. Aogaki, K. Fueki, T. Mukaibo, Denki Kagaku 43 (1975) 504. [2] G. Hinds, F.E. Spada, J.M.D. Coey, T.R. Ní Mhíocháin, M.E.G. Lyons, J. Phys. Chem. B 105 (2001) 9487. [3] P. Fricoteaux, B. Jonvel, J.P. Chopart, J. Phys. Chem. B 107 (2003) 9459. [4] H. Matsushima, A. Bund, W. Plieth, S. Kikuchi, Y. Fukunaka, Electrochim. Acta 53 (2007) 161. [5] K. Shinohara, K. Hashimoto, R. Aogaki, Chem. Lett. 31 (2002) 738. [6] F.M.F. Rhen, G. Hinds, J.M.D. Coey, Electrochem. Commun. 6 (2004) 413. [7] F.M.F. Rhen, D. Fernández, G. Hinds, J.M.D. Coey, J. Electrochem. Soc. 153 (2006) J1. [8] T. Weier, J. Hüller, G. Gerbeth, F. Weiss, Chem. Eng. Sci. 60 (2005) 293. [9] T. Weier, K. Eckert, S. Mühlenhoff, C. Cierpka, A. Bund, M. Uhlemann, Electrochem. Commun. 9 (2007) 2479. [10] T.Z. Fahidy, J. Appl. Electrochem. 13 (1983) 553. [11] S. Legeai, M. Chatelut, O. Vittori, J.P. Chopart, O. Aaboubi, Electrochim. Acta 50 (2004) 51. [12] M. Uhlemann, H. Schlörb, K. Msellak, J.P. Chopart, J. Electrochem. Soc. 151 (2004) C598. [13] A. Krause, J. Kosa, A. Ispas, M. Uhlemann, A. Gebert, A. Bund, Electrochim. Acta 52 (2007) 6338. [14] C. O’Reilly, G. Hinds, J.M.D. Coey, J. Electrochem. Soc. 148 (2001) C674. [15] T.R. Ni Mhiochain, PhD Thesis, Trinity College, Dublin, 2001. [16] C. Cierpka, T. Weier, G. Gerbeth, M. Uhlemann, K. Eckert, J. Solid State Electrochem. 11 (2007) 687.