Electrochemical growth of Co nanowires in ultra-high aspect ratio InP membranes: FFT-impedance spectroscopy of the growth process and magnetic properties
© Gerngross et al.; licensee Springer. 2014
Received: 25 April 2014
Accepted: 27 May 2014
Published: 25 June 2014
The electrochemical growth of Co nanowires in ultra-high aspect ratio InP membranes has been investigated by fast Fourier transform-impedance spectroscopy (FFT-IS) in the frequency range from 75 Hz to 18.5 kHz. The impedance data could be fitted very well using an electric circuit equivalent model with a series resistance connected in series to a simple resistor-capacitor (RC) element and a Maxwell element. Based on the impedance data, the Co deposition in ultra-high aspect ratio InP membranes can be divided into two different Co deposition processes. The corresponding share of each process on the overall Co deposition can be determined directly from the transfer resistances of the two processes. The impedance data clearly show the beneficial impact of boric acid on the Co deposition and also indicate a diffusion limitation of boric acid in ultra-high aspect ratio InP membranes. The grown Co nanowires are polycrystalline with a very small grain size. They show a narrow hysteresis loop with a preferential orientation of the easy magnetization direction along the long nanowire axis due to the arising shape anisotropy of the Co nanowires.
KeywordsCo nanowires single-crystalline InP membranes FFT-impedance spectroscopy Maxwell element RC element
In this paper, the galvanic filling of InP membranes will be discussed which is an essential step for special magnetic field sensors based on magnetoelectric composites. Sensing biomagnetic signals either from the heart or the brain of a human have become more and more important in modern medical diagnostics, e.g. to detect malfunctions of the heart by magnetocardiography (MCG) [1, 2] or to find the origin for seizures in the brain by magnetoencephalography (MEG) [3, 4]. These biomagnetic signals to be detected lie in the order of 10−12 to 10−15 T. Up to now, this requires rather huge and expensive superconducting quantum interference device (SQUID)-based systems that limit the application to university hospitals or hospital centers. As an additional disadvantage, the SQUID-based systems cannot be applied directly to the patient because of the need for thermal insulation due to liquid helium respectively liquid nitrogen cooling of the SQUIDs. This gives rise to the potential replacement by magnetoelectric composite sensors. In principle, different composite geometries are possible. Magnetoelectric 1–3 composites - one-dimensional magnetostrictive structures in a three-dimensional piezoelectric matrix - have the potential advantage of millions of magnetoelectric elements in parallel and also the very high contact area between the magnetostrictive and piezoelectric component.
The galvanic deposition of magnetic and nonmagnetic metals into porous materials is a challenging field especially for ignoble metals, mainly in terms of conformal filling from the bottom of the pore [5–7]. Most of the deposition research has been done in porous alumina membranes [8–10]. It was recently shown in  that it is possible to galvanically grow dense Ni nanowires in ultra-high aspect ratio porous InP membranes when coating the pore walls with a very thin dielectric interlayer prior to the galvanic deposition. The dielectric layer electrically passivates the pore walls so that a nucleation of metal clusters on the pore walls is prevented. It is well known that the magnetic properties of galvanically grown nanowires strongly depend on the growth conditions. The galvanic deposition parameters have been widely exploited and optimized for thin films [12–18], but not for the application in high and ultra-high aspect ratio structures. The huge difference between thin films and high aspect ratio structures is the mass transport of the species taking part in the deposition reaction. In case of high aspect ratio structures, this mass transport is highly diffusion-limited. The present work makes use of the fast Fourier transform-impedance spectroscopy (FFT-IS) to characterize the growth process of Co nanowires directly at the metal electrolyte interface deep in the pore under specific deposition conditions. The obtained results are then correlated to the results of the structural and magnetic investigation of the Co nanowires/InP membrane composite.
The templates for the growth of Co nanowires are porous InP membranes. These membranes are fabricated in an electrochemical multistep process. The porous InP membranes are fabricated from single-crystalline InP wafers sulfur-doped at a doping concentration of 1.1·1017 cm−3 and a resistivity of 0.019 Ωcm. The surface of the InP wafers is double-side polished and epi-ready. The wafer thickness is 400 ± 10 μm, and the sample size is A = 0.25 cm2. All electrochemical process steps are carried out in electrochemical double cell as described elsewhere .
The first step in the membrane formation is the electrochemical etching of the current-line-oriented pore (curro-pore) array. This is done in an aqueous 6 wt% HCl electrolyte at 20°C. To ensure a homogenous nucleation of the curro-pores, a voltage pulse of 17 V for 1 s is applied that is followed by a constant anodic potential of 10 V for 36 min for the growth of the curro-pores.
In the second step, the membrane is formed. This is done in a combined photoelectrochemical and photochemical process. At first, a layer consisting of crystallographically-oriented pores (crysto-pores) is grown in the bulk wafer back side that is subsequently dissolved photochemically. The etching is carried out in the same electrochemical cell in a 6 wt% aqueous HCl electrolyte at 20°C. More details on the fabrication process are given elsewhere .
In the third step, the membrane structure is post-etched in an HF/HNO3/EtOH/HAc (3:8:15:24) electrolyte at 20°C under a bias potential of −0.8 V for 48 h to obtain an overlapping of the space charge region (SCR) around each pore with SCRs around neighboring pores and therefore semi-insulating properties. Besides this effect, the post-etching also results in perfectly rectangular pores with pore walls exhibiting an equal thickness.
The final step of the template fabrication is the electric passivation of the pore walls by an 8-nm-thick layer of Al2O3 deposited by atomic layer deposition (ALD) to avoid unfavorable current flow through the pore walls during galvanic deposition. This is done in 80 cycles of trimethylaluminum (TMA) and H2O with extended diffusion time at 300°C in a Picosun Sunale R200 ALD tool (Espoo, Finland).
Prior to the galvanic Co deposition, a Au layer with a thickness of about 400 nm is deposited on the InP membrane back side serving as a plating base ensuring a complete coverage of the membrane back side.
The growth of Co nanowires in the InP membrane template is carried out in an electrochemical cell with a three-electrode configuration: a Pt counter electrode and a Pt pseudo-reference electrode, which are both immersed in the Co electrolyte. The Au plating base of the InP membrane template serves as the working electrode.
The Co electrolyte is an aqueous electrolyte with 60 g/l CoSO4 and 45 g/l H3BO3 adjusted to a pH value of 3 by HCl. The electrolyte is kept constantly at a temperature of 35°C. The Co nanowires are grown at a constant current density of 12 mA/cm2 for 20 min. During the entire deposition process, FFT-IS is performed, i.e. every 2 s, a spectrum of 26 frequencies from 75 Hz to 18.5 kHz is applied simultaneously and the corresponding impedance data is recorded as well as the deposition voltage. The impedance data are analyzed ex situ.
The InP membrane/Co nanowires composite structure was investigated using a ZEISS Supra 55 VP scanning electron microscope (SEM) (Oberkochen, Germany) and a Seifert X-ray diffraction (XRD) 3000 TT (Olympia, WA, USA) (Cu Kα = 0.154 nm). The magnetic properties were investigated by a Lake Shore 7300 vibrating sample magnetometer (VSM; Westerville, OH, USA).
Results and discussion
Impedance analysis of the galvanic Co nanowire growth
It is a rather complex model consisting of a series resistor Rs that is connected in series with a resistor-capacitor (RC) element and in series with a Maxwell element. The RC element is a parallel arrangement of the resistor Rp and Cp. The capacitor Cp itself does not occur as a separate fit parameter but is integrated in the time constant τp. The Maxwell element is built up of a parallel arrangement of the resistor Ra and the capacitor Ca and the series connection of the resistor Rb and the capacitor Cb. It is well known that the same impedance data can be described by several corresponding equivalent circuits. All possible arrangements starting from various Voigt model arrangements over Maxwell model arrangements have been evaluated. The chosen Maxwell model was the best-suited model to describe and explain the recorded impedance data most consistently for two reasons. The first reason is, it is shown in literature [15, 17] that the Co deposition can occur via at least two reaction pathways. The second reason is that the decoupling of the seven fit parameters vs. time is best for the chosen Maxwell model in comparison to other investigated equivalent circuit models as will be discussed in the following.
The Co deposition voltage U decreases exponentially with time starting from a value of about −1.25 V and reaches a constant deposition voltage of about −1 V after approximately 10.5 min. The series resistance Rs increases linearly with the time starting from about 90 Ω going up to about 130 Ω with slight oscillations towards the end.
The transfer resistance Rp is negative over the entire deposition time. It linearly increases starting from about −25 Ω up to about −35 Ω, reaching a constant level after about 16 min. Similar to the series resistance, also Rp shows oscillations towards the end but significantly more pronounced in amplitude. Unlike the Rp, the associated process time constant τp remains constant over the entire deposition time. It also shows higher oscillations towards the end.
In the first three minutes, the Maxwell resistance Ra decreases linearly from about 18 Ω to about 16 Ω before Ra linearly increases to 18 Ω and saturates after 16 min with pronounced oscillations during the entire time. The associated capacity Ca does not exhibit the change in slope after three minutes as observed for Ra. It decreases constantly from about 21 μF down to about 15 μF after 15 min before it saturates like Ra.
The Maxwell resistance Rb increases linearly from about 10 Ω up to about 25 Ω. Compared to Ra, the oscillations in Rb are extremely reduced. The corresponding capacity Cb decreases linearly from about 100 μF down to about 50 μF after 10.5 min and decreases further down to about 25 μF with a drastically reduced slope. Similar to Ca, Cb only shows slight oscillations over the complete deposition time.
Interpretation of the impedance fit parameters
As evident from the time dependence of the deposition voltage and the seven fit parameters shown in Figure 2, the galvanic Co deposition can be subdivided into two major sections. Section I is characterized by the exponential decline in the deposition voltage, section II by the constant deposition voltage.
The linear increase of Rs could be understood in terms of the Co nanowire growth. With proceeding deposition time, the Co nanowires increase their length contributing to the series resistence as well as, e.g. ohmic losses in the electrolyte.
A negative resistance can be understood as a process that is acting similar as a catalyst supporting the reaction. Hoare  found for Ni that boric acid in the deposition electrolyte acts in such a way that it is supporting the Ni deposition by forming complexes that can be reduced at lower overpotential compared to the boric acid-free electrolyte. Thus, the transfer resistance Rp and the process time constant τp could describe the influence of boric acid on the Co deposition in ultra-high aspect ratio InP pore arrays. The increase of Rp towards more negative values could be due to an increase in the concentration of boric acid in the pores with increasing deposition time as a result of a reduced diffusion limitation, since the Co nanowires grow towards the pore openings reducing the effective pore depth. The stronger oscillations in Rp might be due to a competition for adsorbing sites on the Co nanowire surface between boric acid-complexed Co ions and other adsorbed species.
The Maxwell resistance Ra could be related to the charge transfer resistance of the direct Co deposition. The decline in the first three minutes could be due to the diffusion limitation of the boric acid that forms complexes with Co2+ ions for an easier deposition. The following linear rise might be attributed to an increased surface coverage of the growing Co nanowires by adsorbed ions impeding the Co deposition. The constant level in Ra after 16 min coincides with the constant level in Rp suggesting that these adsorbed ions might be related to boric acid, such as e.g. B(OH)4−. The ending of the diffusion limitation for the boric acid might be the reason for the constant level in Ra after 16 min.
The Maxwell capacity Ca could be attributed to the corresponding double layer capacity of the direct Co deposition. The decline in Ca correlates with the concentration increase of boric acid species due to a reduced diffusion limitation (see time dependence of Rp) and mirrors also the constant level after 16 min.
The Maxwell resistance Rb and the capacity Cb describe the slowest process during the Co deposition. It could be related to the indirect Co deposition via Co(OH)2 as experimentally observed by Santos et al. . This process takes place in parallel to the direct Co deposition process. Therefore, Rb is assigned to the charge transfer resistance of the Co deposition process via Co(OH)2. The constant gain in Rb indicates that the Co deposition via this process becomes more and more unfavorable compared to the direct Co deposition process represented by Ra. After around 10.5 min, the charge transfer resistances of Ra and Rb exhibit the same value. This allows splitting the entire Co deposition process into two sections. In section I, Rb is lower than Ra. This means that the Co deposition occurs primarily via the indirect mechanism (via Co(OH)2). In section II, the situation is vice versa. The Co deposition occurs primarily via the direct mechanism. The share of the direct Co deposition out of the overall process is determined by 1 − Ra / (Ra + Rb). Consequently, the share of the Co deposition via Co(OH)2 is given by 1 − Rb / (Ra + Rb). The absence of strong oscillations in Rb also indicates that this process appears to be independent from the ending of the diffusion limitation of boric acid.
The capacitance Cb is assigned to the corresponding double layer capacity of the indirect Co deposition. The decline in Cb could be explained in the same way as for Ca. The change in the slope of Ca after about 10.5 min is most probably related to the now preferential Co deposition via the direct deposition process.
As an additional side reaction of the Co deposition, hydrogen can form , but a process related to this hydrogen evolution during the Co deposition could not be identified in the recorded FFT-IS data within the investigated frequency range, most probably because it is a very slow process that is outside the investigated frequency range as it is found for the Ni deposition .
In general, it is decisive that the magnetization in the magnetic material is aligned perpendicular to the applied magnetic field for an optimal magnetostrictive effect, e.g., if the magnetization in the magnetic material is parallel to applied field, the magnetostrictive effect is zero. Another important factor for the application as magnetoelectric sensor is a small hysteresis loop, since magnetic AC fields shall be measured.
For α = 0°, the hysteresis losses of the 0.5 and 1 kOe minor loops are significantly higher compared to the corresponding minor loops for α = 90°. The same behavior is found for the maximum normalized magnetization. This behavior suggests that the easy magnetization direction of the Co nanowires lies along the long nanowire axis z (α = 0°) due to the high aspect ratio of the Co nanowires giving rise to a pronounced shape anisotropy that exceeds the magnetocrystalline anisotropy of Co . The remanence squareness of 0.07 found for the easy magnetization direction is very low compared to a single nanowire with the magnetization also along the long nanowire axis z. One could understand this behavior by taking into account the nucleation of domains with inverse magnetization at the bottom or at the top of the Co nanowires. These domains with inverse magnetization could efficiently reduce stray fields and might be also the reason for the reduced the remanence squareness. The magnetostatic interactions between neighboring Co nanowires might also play an important role, since the interwire distance is far smaller compared to the diameter of the Co nanowires.
Another interesting effect is that for external magnetic fields Ha larger than 500 Oe, the minor loops show a distinct hysteresis that disappears completely for very small Ha (20 and 100 Oe). These minor loops show a reversible linear magnetic field dependence with a higher slope observed for α = 0°. The reversible linear magnetic field dependence means that the magnetization reversal at very small fields Ha occurs by domain rotation and reversible domain wall motion and not by irreversible domain wall motion as observed for higher external fields.
The angular dependence of the coercivity is presented in Figure 4b. The coercivity shows a completely different angular behavior. It is smallest for α = 0° (around 150 Oe) and increases constantly to about 210 Oe for α = 60°, where it peaks for α = 60° and α = 75° before it slightly decreases to around 205 Oe for α = 90°. The magnified view on the differential normalized susceptibility χnorm around H = 0 Oe - depicted in Figure 4c - shows an inverse angular behavior with respect to the maximum χnorm. With increasing angle α, the maximum χnorm decreases steadily from about 0.43/kOe for α = 0° reaching a plateau at about 0.3/kOe for α = 75° and α = 90°. In addition to that, two characteristic peak positions are observed represented by the two solid lines at around 160 Oe and by the two dashed lines at around 280 Oe.
These unusual experimental results in the angular dependence of HC and χnorm could be consistently described taking into account the magnetization reversal that does not occur by a single magnetization reversal mechanism but by two different magnetization reversal mechanisms depending on the incidence angle α of the external magnetic field. The peak positions of χnorm suggest that magnetization reversal mechanism I is predominant for α = 0° and becomes less dominant with increasing α, while the dominance of mechanism II increases with increasing α. Therefore, the maximum in HC for α = 60° and α = 75° could be understood as the result of an interplay between the two magnetization reversal modes. The exact type of these magnetization reversal mechanisms could not be identified by the conducted hysteresis loop measurements. Nevertheless, one might speculate that these reversal modes are most probably the transversal and vortex magnetization reversal mode as found by micromagnetic simulations for Ni nanowires by Han et al. .
Correlating these magnetic results with the structural characterization, one could understand the comparatively high coercivity of the Co nanowires as a direct consequence of the small grain size accompanied by the high amount of grain boundaries that hinder the domain wall movement. The small grain size itself is most probably a consequence of the deposition via the two simultaneously occurring Co deposition processes, as already discussed in the first part of this paper.
The electrochemical growth of Co nanowires in ultra-high aspect ratio InP membranes could be successfully characterized by the analysis of the FFT-IS data. The corresponding fit model is represented by a rather complex electric equivalent circuit containing a series resistance and three RC elements. This fit model is not limited to the Co deposition but has also been successfully applied for the deposition of Ni in ultra-high aspect ratio InP membranes.
Based on the impedance data, the Co nanowire growth could be divided into two separate processes, most possibly the direct Co deposition and the indirect Co deposition via Co(OH)2. The share of each Co deposition process on the overall Co deposition can be determined directly from the transfer resistances of the two processes obtained from the fitted impedance data. These also indicate a beneficial effect of boric acid on the Co deposition. This characterization of the Co deposition process by FFT-IS will help in optimizing the deposition parameters such as temperature, deposition current, electrolyte composition, etc. with respect to the crystal orientation and thus also of the magnetic properties necessary for the application in magnetoelectric 1– 3 composites.
This work was funded by the DFG as part of the special research field 855 ‘Magneto-electric composite materials - biomagnetic interfaces of the future.’
- Wakai RT, Leuthold AC, Martin CB: Atrial and ventricular fetal heart rate patterns in isolated congenital complete heart block detected by magnetocardiography. Am J Obstet Gynecol 1998, 179(1):258. 10.1016/S0002-9378(98)70282-0View Article
- Park J-W, Hill PM, Chung N, Hugenholtz PG, Jung F: Magnetocardiography predicts coronary artery disease in patients with acute chest pain. Ann Noninvas Electro 2005, 10(3):312. 10.1111/j.1542-474X.2005.00634.xView Article
- Sutherling WW, Crandall PH, Engel J Jr, Darcey TM, Cahan LD, Barth DS: The magnetic field of complex partial seizures agrees with intracranial localizations. Ann Neurol 1987, 21(6):548. 10.1002/ana.410210605View Article
- Paulini A, Fischer M, Rampp S, Scheler G, Hopfengärtner R, Kaltenhäuser M, Dörfler A, Buchfelder M, Stefan H: Lobar localization information in epilepsy patients: MEG – A useful tool in routine presurgical diagnosis. Epilepsy Res 2007, 76(2–3):124.View Article
- Granitzer P, Rumpf K, Pölt P, Simic S, Krenn H: Three-dimensional quasi-regular arrays of Ni nanostructures grown within the pores of a porous silicon layer - magnetic characteristics. Phys Status Solidi C 2008, 5(12):3580. 10.1002/pssc.200780133View Article
- Tiginyanu I, Monaico E, Monaico E: Ordered arrays of metal nanotubes in semiconductor envelope. Electrochem Comm 2008, 10(5):731. 10.1016/j.elecom.2008.02.029View Article
- Fang C, Foca E, Xu S, Carstensen J, Föll H: Deep silicon macropores filled with copper by electrodeposition. J Electrochem Soc 2007, 154(1):D45. 10.1149/1.2393090View Article
- Nielsch K, Wehrspohn RB, Barthel J, Kirschner J, Fischer SF, Kronmüller H, Schweinböck T, Weiss D, Gösele U: High density hexagonal nickel nanowire array. J Magn Magn Mater 2002, 249: 234. 10.1016/S0304-8853(02)00536-XView Article
- Göring P, Pippel E, Hofmeister H, Wehrspohn RB, Steinhart M, Gösele U: Gold/Carbon Composite Tubes and Gold Nanowires by Impregnating Templates with Hydrogen Tetrachloroaurate/Acetone Solutions. Nano Lett 2004, 4(6):1121. 10.1021/nl049542vView Article
- McGary PD, Tan L, Zou J, Stadler BJH, Downey PR, Flatau AB: Magnetic nanowires for acoustic sensors (invited). J Appl Phys 2006, 99: 08B310.View Article
- Gerngross M-D, Chemnitz S, Wagner B, Carstensen J, Föll H: Ultra-high aspect ratio Ni nanowires in single-crystalline InP membranes as multiferroic composite. Phys Status Solidi RRL 2013, 7(5):352. 10.1002/pssr.201307026View Article
- Jiang SP, Tseung ACC: Reactive deposition of cobalt electrodes II. role of bubbling oxygen. J Electrochem Soc 1990, 137(11):3381. 10.1149/1.2086225View Article
- Jiang SP, Tseung ACC: Reactive deposition of cobalt electrodes III. role of anions. J Electrochem Soc 1990, 137(11):3387. 10.1149/1.2086226View Article
- Cui CQ, Jiang SP, Tseung ACC: Electrodeposition of cobalt from aqueous chloride solutions. J Electrochem Soc 1990, 137(11):3418. 10.1149/1.2086232View Article
- Pradhan N, Subbaiah T, Das SC, Dash UN: Effect of zinc on the electrocrystallization of cobalt. Electrochim Acta 1997, 53: 644.
- Jeffrey MI, Choo WL, Breuer PL: The effect of additives and impurities on the cobalt electrowinning process. Miner Eng 2000, 13(12):1231–1241. 10.1016/S0892-6875(00)00107-2View Article
- Matsushima JT, Trivinho-Strixino F, Pereira EC: Investigation of cobalt deposition using the electrochemical quartz crystal microbalance. Electrochim Acta 2005, 51: 1960–1966.View Article
- Santos JS, Matos R, Trivinho-Strixino F, Pereira EC: Effect of temperature on Co electrodeposition in the presence of boric acid. Electrochim Acta 2007, 53: 644. 10.1016/j.electacta.2007.07.025View Article
- Langa S, Tiginyanu IM, Carstensen J, Christophersen M, Föll H: Formation of porous layers with different morphologies during anodic etching of n-InP. Electrochem Solid-State Lett 2000, 3(11):514.View Article
- Gerngross M-D, Carstensen J, Föll H: Single-Crystalline membranes in indium phosphide: fabrication process and characterization using FFT impedance Analysis. J Electrochem Soc 2012, 159(11):H857. 10.1149/2.041211jesView Article
- Hoare JP: On the role of boric acid in the Watts bath. J Electrochem Soc 1986, 133(12):2491. 10.1149/1.2108456View Article
- Davalos CE, Lopez JR, Ruiz H, Mendez A, Antano-Lopez R, Trejo G: Study of the role of boric acid during the electrochemical deposition of Ni in a sulfamate bath. Int J Electrochem Sci 2013, 8: 9785.
- Li F, Wang T, Ren L, Sun J: Structure and magnetic properties of Co nanowires in self-assembled arrays. J Phys Condens Matter 2004, 16: 8053. 10.1088/0953-8984/16/45/027View Article
- Sun D-L, Gao J-H, Zhang X-Q, Zhan Q-F, He W, Sun Y, Cheng ZH: Contribution of magnetostatic interaction to magnetization reversal of Fe3Pt nanowires arrays: a micromagnetic simulation. J Magn Magn Mater 2009, 321(18):2737. 10.1016/j.jmmm.2009.03.079View Article
- Han N, Guo G, Zhang L, Zhang G, Song W: Magnetization reversal for Ni nanowires studied by micromagnetic simulations. J Mater Sci Technol 2009, 25(2):151.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.