C690

Journal of The Electrochemical Society, 148 10 C690-C694 2001

0013-4651/2001/148 10 /C690/5/$7.00 The Electrochemical Society, Inc.

Effect of Organic Additives on the Electrocrystallization and the Magnetoresistance of Cu-Co Multilayers
Elisabeth Chassaingz
Centre dEtudes de Chimie Metallurgique CNRS, 94407 Vitry-sur-Seine, France The electrodeposition of Cu-Co multilayers was investigated in the presence of two additives, sodium dodecylsulfate SDS and saccharin. Cyclic voltammetry, steady-state polarization, impedance spectroscopy, and potential pulses were carried out using an electrochemical quartz crystal microbalance. Magnetoresistance measurements were performed on Cu-Co multilayers deposited on indium tin oxide glass. Saccharin strongly inhibits the discharge of cobalt whereas SDS has a depolarizing effect. However the reaction path is not markedly changed. The deposition of cobalt starts with a strong evolution of hydrogen due to diffusioncontrolled reduction of protons. The giant magnetoresistance ratio is markedly reduced by these additives. 2001 The Electrochemical Society. DOI: 10.1149/1.1401079 All rights reserved. Manuscript submitted March 12, 2001; revised manuscript received May 25, 2001. Available electronically September 11, 2001.

In many electroplating applications, small amounts of organic compounds are added in order to improve the brightness, the corrosion resistance, etc. Since the pioneering work by Saubestre,1 several studies have been carried out. A classication of these compounds was attempted by Franklin2 and Oniciu and Muresan3 and the effects of the additives were reviewed by Plieth.4 More recently a renewed interest in the effect of the additives was raised for the electrodeposition of alloys of transition metals in order to improve their magnetic properties.5-7 The effects of these additives on the deposition kinetics depend on the nature of the metal or alloy, inhibiting or accelerating effects on the deposition process have been observed.7 During the last decade a large interest has been raised for the electrodeposition of multilayers with nanometric or micrometric thicknesses aiming at preparing structures with enhanced properties i.e., mechanical, magnetic, magnetoresistive properties .8-20 The addition of brighteners and wetting agents improves the smoothness not only of the external surface,13-15 but also the planarity of the interfaces between the layers.15,17,18 As a consequence such additions have a benecial inuence on the corrosion resistance of Cu/Ni layers17,18 and on the antiwearing properties.16 Some additives were shown to decrease the internal stresses.1,17 They may also change the morphology and microstructure of Cu-Co multilayers.15 However according to Lenczowski et al., additions of brighteners such as thiourea and Triton X100 have a catastrophic effect both on the structural quality and on the magnetoresistance of Cu-Co multilayers.19 Except for the recent work by Kelly et al.,14,15 few investigations have been focused in the transient behavior of the effect of these additives, especially using an electrochemical quartz crystal microbalance EQCM . In the present work, we investigate the effect of two different additives on the electrocrystallization and the magnetoresistive properties of Cu/Co multilayers. First sodium dodecylsulfate SDS which is a surfactant, was studied; it is often used as an antipitting agent in Watts bath since it removes hydrogen bubbles formed at the cathode and reduces the number of pits.1 The other additive is saccharin, a rst class brightener for nickel lms, which is also considered to decrease the internal stress.1,17 The additives were used separately and then together, since synergetic effects often occur. An electrochemical quartz crystal microbalance EQCM was used to record both the current response and the mass changes during the process. The magnetoresistance of Cu/Co 20 multilayers electrodeposited on indium tin oxide ITO glass substrates, in the presence of the additives, was measured. Experimental The composition of the basic electrolyte denoted electrolyte 1 is similar to that previously used for the electrodeposition of Cu-Co multilayers.11,12 It contained 0.71 M CoSO4 , 0.19 M CoCl2 ,

6.4 10 3 M CuSO4 , 0.65 M H3BO3, 0.34 M NaCl. Reagent grade chemicals were used. The pH was adjusted to 1.7 0.5 by additions of dilute sulfuric acid. Such a low pH value was shown to improve the magnetoresistive properties of the multilayers.20 Temperature was maintained at 300 K. The solutions were deaerated by nitrogen bubbling before the experiments. During the deposition, the solutions were unstirred to avoid the increase of the copper discharge, which is diffusion controlled. In electrolyte 2, 0.2 g dm 3 of SDS was added; in electrolyte 3, 2 g dm 3 of saccharin was added; electrolyte 4 contained 0.2 g dm 3 SDS and 2 g dm 3 saccharin. A classic three-electrode cell was used. The reference electrode was a saturated calomel one SCE . The counter electrode was platinum gauze. The gold layer of the electrochemical quartz crystal Maxtek PM 740 was used as working electrode AT-cut of 5 MHz resonance frequency . No effect of the oscillations of the quartz crystal on the mass transport of the cupric species, i.e., on the copper current, was observed. The ohmic drop between working and reference electrodes was measured using electrochemical impedance spectroscopy, which also allowed the evaluation of the chargetransfer resistance and the double-layer capacitance value. Cyclic voltammetry was carried out to characterize the electrodeposition process. A potential pulse investigation was performed to determine the partial polarization curves and the current efciency in transient conditions. The magnetoresistance was measured by the four-point method in the presence of an applied eld at 77 K up to 1.5 kOe. Some experiments were also carried out at 4.2 K up to 12 kOe.

Results Cyclic voltammetry.For each solution, two cyclic voltammetry diagrams were carried out with a scanning rate of 10 mV s 1, one in the potential range of the deposition of copper 100 to 600 mV/SCE and the other one in the whole range of deposition 500 to 1200 mV/SCE . The additives, SDS or saccharin, do not have a marked inuence on the discharge of copper. The current efciency for the copper discharge is only slightly decreased in the presence of saccharin or saccharin and SDS. By contrast the additives have a strong effect on the discharge of cobalt. Figures 1a and b clearly show the inhibiting effect of saccharin and the depolarizing effect of SDS; in the case of saccharin, the cathodic polarization curve is shifted toward more negative potentials and the dissolution peak is smaller. The deposited masses in the presence of saccharin (210 10 g cm 2), or SDS saccharin (250 10 g cm 2) are markedly smaller than in electrolytes 1 (320 10 g cm 2) and 2 (380 10 g cm 2). The ratio Qa/Qc of the charge used for the dissolution Qa and for the deposition Qc gives an estimation of the current efciency. It is 0.93 0.05 for the electrolytes 1 or with SDS. It is reduced to

E-mail: Elisabeth.Chassaing@glvt-cnrs.fr

Downloaded 05 Sep 2009 to 130.127.56.12. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Journal of The Electrochemical Society, 148 10 C690-C694 2001

C691

Figure 3. Impedance plots recorded for a current density of 15 mA cm 2 in the electrolytes without additive with SDS saccharin and both additives .

addition of saccharin shifts the polarization curve toward negative potentials whereas the addition of SDS has only a slight depolarizing effect. All the curves show two different behaviors depending on the polarization, indicating a two-step discharge of the Co II species as for the deposition of nickel.21 At low polarization, the slope is close to 25 0.5 V 1 for the basic electrolyte and electrolyte 2 and 17 0.5 V 1 in the presence of saccharin and both additives. For larger polarization, the slope is close to 9 0.5 V 1 for electrolytes 1, 2, and 4 and to 12 0.5 V 1 in the presence of saccharin.23
Figure 1. a Cyclic voltammograms and b mass changes recorded for the electrolytes without additive curve 1 , with SDS curve 2 , saccharin curve 3 , and both additives curve 4 . Initial potential E i 500 mV/SCE.)

0.88 0.05 in electrolyte 3 containing saccharin and 0.85 in electrolyte 4 with both additives.

0.05

Steady-state polarization behavior.The steady-state polarization curves for the deposition of cobalt, corrected for ohmic drop, conrm the results of the voltammetric investigation Fig. 2 . The

Electrochemical impedance investigation.An electrochemical impedance investigation, between 60 kHz and 0.1 Hz, was carried out in the potential range of cobalt deposition. The Nyquist plots show a high-frequency loop corresponding to the relaxation of the double-layer capacitance, C d , in parallel with the charge-transfer resistance, R t Fig. 3 . Table 1 summarizes the results for a current density of 15 mA cm 2. For each solution, the double-layer capacitance decreases with increasing polarization. At potentials greater than 800 mV/SCE, copper is discharged under diffusion control, i.e., with a rough surface area. In this range the double-layer capacitance is large 200300 F cm 2 . As soon as cobalt deposition starts, the double-layer capacitance decreases indicating a smoother surface. In the basic electrolyte or in the presence of SDS, the double-layer capacitance is still rather large 80-100 F cm 2 , indicating a rough morphology. In the presence of saccharin, the capacitance is markedly decreased 40-60 F cm 2 , probably as the result of a blocking adsorption of the additive and/or the smoothing inuence of saccharin.2

Table I. Discharge of cobalt at 15 mA cm2 in the four electrolytes: potential corrected for ohmic drop, double-layer capacitance, C d and R t i product of charge-transfer resistance, R t and the current density, i. Electrolyte Figure 2. Steady-state polarization curves, corrected for ohmic drop, recorded in the potential range for the deposition of cobalt recorded for the electrolytes without additive , with SDS , saccharin , and both additives . 1. 2. 3. 4. No additive SDS Saccharin Saccharin SDS Potential mV/SCE at 15 mA cm 2 800 815 910 890 10 10 10 10 Cd 140 175 50 60 F cm
2

R t i MV 45 41 48 49 5 5 5 5

20 20 5 5

Downloaded 05 Sep 2009 to 130.127.56.12. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

C692

Journal of The Electrochemical Society, 148 10 C690-C694 2001

The R t i product of the charge-transfer resistance and the current density, increases with polarization from 45 to 70 mV. It is slightly larger in the presence of saccharin Table I . For a current density of 15-20 mA cm 2, corresponding to the value used for the deposition of cobalt, the R t i product is close to 45 mV. Figure 3 shows that the impedance plots exhibit, at low polarization, in addition to the high-frequency capacitive feature, a mediumfrequency inductive loop. Its characteristic frequency increases exponentially with polarization. This behavior is quite similar to that observed for the discharge of nickel in a Watts bath.21,22 The medium-frequency loop was attributed to the two-step discharge of nickel via an intermediate adion.21 A similar interpretation can be made for the discharge of cobalt. The cobalt deposition may be considered as the result of the two-step discharge of the Co II species via an intermediate adion, written Co I ads for simplicity, though it may be hydrolyzed or complexed Co II Co I
ads

Co I

ads

1 2

e Co

These reactions are coupled to the hydrogen evolution reaction, written as the global reduction of the protons 2H 2e H2 3

Figure 4. a Potential pulses for the preparation of a Cu-Co trilayer. b Current density response curve 1 and mass changes curve 2 for electrolyte 1 without additive.

With this reaction scheme, we can calculate the polarization curves as already done for nickel discharge.21,22 The characteristic features of the impedance plots (R t i product, characteristic frequencies of the loops, etc. can be calculated and compared with the experimental ones. To account for the increasing value of the R t i product with potential, Reaction 1 is considered as a reversible process.23 In the presence of SDS the kinetics of these reactions are only slightly modied.23 However in the presence of saccharin, the polarization curve is markedly shifted toward the negative potentials. The decrease of the double-layer capacitance may indicate a blocking adsorption of saccharin on the surface area.24 Osaka et al. showed that saccharin was only physically and reversibly adsorbed on the gold electrodes in contrast to other additives such as thiourea.5 It would inhibit the discharge of cobalt, but would not otherwise effect the reaction mechanism. Indeed the characteristic frequency of the inductive loop, which depends on the rate constants of these reactions, has the same dependence with potential.23 Potential pulses.As shown in Fig. 4a, potential pulses have been carried out, with the EQCM, at E 1 400 mV and E 2 1000 mV, values corresponding to those used for the preparation of the multilayers.11 The potential at 1000 mV corresponds to a potential of 900 10 mV when corrected for ohmic drop. Cu-Co trilayers were prepared using predetermined electrical charges: Cu 50 or 100 mC cm 2 /Co 200 mC cm 2 /Cu 50 mC cm 2 . The charges correspond to thicknesses of, respectively, 16 2 and 28 2 nm for copper and 55 5 nm for cobalt, when deposited from the basic electrolyte 1. The frequency shifts recorded with the EQCM were corrected taking into account the mass distribution on the surface area using the model developed by Kelly et al.25 The Wagner number, W a , expresses the uniformity of the current distribution. It is equal to Wa d di r0

where d /di is the slope of the polarization curve i.e., the polarization resistance , the conductivity of the solution, r 0 the radius of the electrode surface area. The conductivity of the electrolytes, measured using a double platinum/platinized electrode, is equal to 0.07 S/cm. For copper deposition, the polarization resistance is large, W a is close to 30, indicating a nearly uniform current distribution. For cobalt deposi-

tion, the polarization resistance is of the order of 45 mV according to the impedance and polarization results Table I and W a is close to 0.4. The correction of the frequency shift was shown to depend not only on the radius of the working area (r 0 0.625 cm) but also on the bottom radius of the electrode here r b r 0 /2). 25 Using the data in Ref. 25, a correction of frequency shift due to inhomogeneous mass distribution on the quartz area of 0.80 to 0.85 has been applied. According to the Sauerbrey equation, for frequency shifts of less than 2% of the resonant frequency, the frequency shift is proportional to the mass increase: f k m, where m is the mass per unit area g cm 2 , k is a constant. The AT quartz was calibrated using 0.5 M CuSO4 , 0.5 M H2SO4 , 1.1 M C2H5OH solution, for which the faradaic efciency is close to unity.25 The coefcient, k was calculated equal to 56.7 0.3 Hz g 1 cm2. At the beginning of each pulse, a current peak is observed corresponding to the charge or discharge of double-layer capacitance Fig. 4b . This leads to a capacitive current of i c E/ R s exp( t/Rs C d), where E is the magnitude of the potential pulse, R s the series resistances 8 to 9 cm2 in our conditions , C d the double-layer capacitance 50-100 F cm 2 . This feature leads to current peaks of the order of 40 mA cm 2, the capacitive current becomes negligible after 0.010 s, which is actually observed. The partial transient curves for cobalt discharge and hydrogen evolution have been calculated from the results of the EQCM Fig. 5 . As previously observed,12 hydrogen evolution is the main reaction at the beginning of cobalt deposition. This was conrmed by in situ magnetization measurements where we observed that the magnetization increased after a certain delay.12 As a result, for pulses of short duration, the efciency for cobalt deposition depends on the length of the pulse. The hydrogen evolution current decreases as a power function of time with an exponent of 0.30 0.05. This behavior is probably related to the diffusion-controlled discharge of the protons. The exponent is larger as compared to the exponent expected from Cottrells law. This is probably because the discharge occurs on the growing area of the cobalt lm. The partial transient curves for hydrogen evolution are only slightly affected by the presence of additives in the bath Fig. 6a . By contrast the partial curves for cobalt deposition depend strongly on the presence and on the nature of the additives Fig. 6b . The addition of SDS increases the deposition rate of cobalt whereas sac-

Downloaded 05 Sep 2009 to 130.127.56.12. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Journal of The Electrochemical Society, 148 10 C690-C694 2001

C693

Figure 5. Total and partial currents for copper, cobalt discharge, and hydrogen evolution during the pulse at 1000 mV vs. SCE in electrolyte 1.

Figure 7. Current density curve 1 and experimental solid line and calculated dashed line frequency shifts curves 2 for the deposition of copper at the reversal of the potential from cobalt to copper discharge in electrolyte 2 containing SDS.

charin inhibits its deposition as compared to the discharge without additives. As a consequence, the current efciency for cobalt discharge at this potential, in steady-state conditions, is markedly smaller, i.e., 0.6, in the presence of saccharin than in the basic electrolyte or in the presence of SDS about 0.8 . This result is in disagreement with those obtained by Kelly et al.,14,15 who observed an increase of current efciency in the presence of saccharin and a decrease in the

presence of SDS. Our electrolyte composition is, however, different. Kellys solution did not contain chloride species but sodium acetate. In addition the pH of their solution was higher 4 instead of 1.7 . As a result, the electroactive species may be different. In our solutions the dominant electroactive species would be CoCl and Co2 , while in Kellys electrolyte these are mainly Co2 and to a lesser extent CoOAc and CoOH species. This may explain the different kinetics of the cobalt discharge. According to Kelly et al.,15 SDS promotes the chemical displacement of cobalt by copper during the pulse-off time: Cu2 Co Co2 Cu. In our case this displacement is probably less pronounced since our experiments are carried out in potentiostatic conditions. In addition, according to Alper et al. a low pH may suppress the dissolution of cobalt.20 The frequency shift for copper deposition was calculated after the potential reversal from E 2 to E 1 , using Sauerbrey equation. The deposited mass for copper discharge, m Cu ( g cm 2), according to Faradays law, is equal to m Cu 63.55 106 Q 2 F

Figure 6. a Partial current for hydrogen evolution recorded for the electrolytes without additive curve 1 and with SDS curve 2 , saccharin curve 3 , and both additives curve 4 . b Partial current for cobalt deposition recorded for the electrolytes without additive curve 1 and with SDS curve 2 , saccharin curve 3 , and both additives curve 4 .

where Q is the electrical charge used C cm 2 , F the Faradays constant, and the Faradaic efciency for copper deposition measured at the end of the rst pulse . Figure 7 shows that, for the bath with SDS addition, the calculated frequency shift curve 2 coincides with the experimental one curve 1 . This was also the case for the other electrolytes. Hence we could not show any cobalt displacement by cupric ions, using this method. Magnetoresistance measurements.The inuence of the additives on the magnetoresistance of Cu-Co multilayers deposited on ITO glass was investigated. Ten to twenty bilayers were prepared, and their resistivity was measured by the four-point method at 77 K in parallel and transverse conditions as previously described.11 The thickness of the individual copper layers was 4.5 nm, and that of cobalt was 0.8 nm. The parallel magnetoresistance ratio, measured for a eld of 1.5 kOe was about 4 1% with SDS addition, 5 1% with saccharin, and 6 1% when both additives are present. At 4.2 K and for a eld of 12 kOe, the magnetoresistance ratio was 9 0.5% for a multilayer deposited from a bath with both additives Fig. 8 . For multilayers deposited from the basic electrolyte without additive, the giant magnetoresistance GMR was 15% at 77 K and 25% at 4.2 K.11,12 This marked decrease of the GMR conrms the results reported by Lenczowski et al. who also observed a drastic reduction of the GMR in the presence of thiourea and Triton 100.19 These authors attribute the decrease to a less perfect crystalline structure.19 According to Osaka et al. the incorporation of saccharin is very weak compared to that of thiourea sulfur inclusions are less than

Downloaded 05 Sep 2009 to 130.127.56.12. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

C694

Journal of The Electrochemical Society, 148 10 C690-C694 2001 the presence of saccharin. No chemical displacement could be observed with the EQCM under our conditions. The calculated frequency shift coincides with the experimental one. The GMR of Cu-Co multilayers is about three times smaller when the electrolyte contains the additives than in the basic electrolyte. Though the planarity of the interfaces has been shown to be better in the presence of saccharin,15 the transport properties are smaller. It is likely that decomposition products of the additives are adsorbed at the interfaces and, even in small amounts, they may have a detrimental effect on the magnetoresistance as already observed in the presence of Triton or thiourea.19 References
1. 2. 3. 4. 5. 6. 7. E. B. Saubestre, Plating, 45, 1219 1958 and references therein. T. C. Franklin, Surf. Coat. Technol., 30, 415 1987 . L. Oniciu and L. Muresan, J. Appl. Electrochem., 21, 565 1991 . W. Plieth, Electrochim. Acta, 37, 2115 1992 . T. Osaka, T. Sawaguchi, F. Mizutani, T. Yokoshima, M. Takai, and Y. Okinaka, J. Electrochem. Soc., 146, 3295 1999 . Y. K. Kim, H. Y. Son, Y. S. Choi, K. S. Moon, and K. H. Sunwoo, J. Appl. Phys., 87, 5413 2000 . I. Tabakovic, S. Riemer, V. Inturi, P. Jallen, and A. Thayer, J. Electrochem. Soc., 147, 219 2000 . W. Schwarzacher, Electrochem. Soc. Interface, 8 1 , 20 1999 and references therein. M. Shima, L. Salamanca Riba, T. P. Moffat, and R. D. McMichael, J. Magn. Magn. Mater., 199, 52 1999 . P. E. Bradley and D. Landolt, Electrochim. Acta, 45, 1077 1999 . E. Chassaing, A. Morrone, and J. E. Schmidt, J. Electrochem. Soc., 146, 1794 1999 . A. Gundel, E. Chassaing, and J. E. Schmidt. in Fundamental Aspects of Electrochemical Deposition and Dissolution Including Modeling, D. L. R. Aogaki, M. Matlosz, Y. Sato, and J. B. Talbot, Editors, PV 99-33, p. 135, The Electrochemical Society Proceedings Series, Pennington, NJ 1999 . G. Wouters, M. Bratoeva, J. P. Celis, and J. R. Roos, Electrochim. Acta, 40, 1439 1995 . J. J. Kelly, P. Kern, and D. Landolt, J. Electrochem. Soc., 147, 3725 2000 . J. J. Kelly, P. Bradley, and D. Landolt, J. Electrochem. Soc., 147, 2975 2000 . Q. J. Xue and W. Zhang, J. Phys. D, 30, 3301 1997 . C. Bertrand, Ph.D. Thesis, Universite of Reims, Champagne-Ardenne 2000 . M. Troyon and L. Wang, Appl. Surf. Sci., 103, 517 1996 . S. K. J. Lenczowski, C. Schonenberger, M. A. M. Gijs, and W. J. M. Dejonge, J. Magn. Magn. Mater., 148, 455 1995 . M. Alper, W. Schwarzacher, and S. J. Lane, J. Electrochem. Soc., 144, 2346 1997 . I. Epelboin, M. Joussellin, and R. Wiart, J. Electroanal. Chem., 119, 61 1981 . E. Chassaing, M. Joussellin, and R. Wiart, J. Electroanal. Chem., 157, 75 1983 . E. Chassaing, To be published. K. M. Yin, J. Electrochem. Soc., 145, 3851 1998 . J. J. Kelly, K. M. A. Rahman, C. J. Durning, and A. C. West, J. Electrochem. Soc., 145, 492 1998 .

Figure 8. Magnetoresistance ratio in parallel conditions for a Cu 4.5 nm /Co 0.8 nm 20 deposited from electrolyte 1 curve 1 and in the presence of saccharin and SDS curve 2 .

0.4 atom % .5 However it has been shown that saccharin affects the microstructure of the deposits.5 According to Kelly et al., in the presence of SDS, the morphology of the multilayers is changed, cobalt-rich columns are surrounded by a copper-rich matrix phase.15 This might lead to pinholes in the layers resulting in a strong decrease of the GMR ratio. Conclusions The additions of SDS and saccharin have only a small inuence on the deposition of copper, but a marked effect on cobalt discharge. SDS slightly depolarized the discharge of cobalt whereas saccharin strongly inhibits the process. The impedance investigation shows that the reaction path for cobalt discharge is not markedly changed, but the addition of saccharin partly blocks the electrode area. This leads to a shift of the polarization curves toward negative potentials and to a decrease of the double-layer capacitance. However no marked change is observed in the Tafel coefcient. The potential pulse investigation shows that at the beginning of cobalt discharge hydrogen evolution is the main reaction, due to diffusion-controlled reduction of protons of the electrolyte. The faradaic efciency increases with pulse time, and it is much smaller in

8. 9. 10. 11. 12.

13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.

Downloaded 05 Sep 2009 to 130.127.56.12. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp