full papers Nanotube Arrays
Cost-effective Atomic Layer Deposition Synthesis of Pt Nanotube Arrays: Application for High Performance Supercapacitor Liaoyong Wen, Yan Mi, Chengliang Wang, Yaoguo Fang, Fabian Grote, Huaping Zhao, Min Zhou, and Yong Lei*
Due
to the unique advantages of Pt, it plays an important role in fuel cells and microelectronics. Considering the fact that Pt is an expensive metal, a major challenging point nowadays is how to realize efficient utilization of Pt. In this paper, a cost-effective atomic layer deposition (ALD) process with a low N2 filling step is introduced for realizing well-defined Pt nanotube arrays in anodic alumina nanoporous templates. Compared to the conventional ALD growth of Pt, much fewer ALD cycles and a shorter precursor pulsing time are required, which originates from the low N2 filling step. To achieve similar Pt nanotubes, about half cycles and 10% Pt precursor pulsing time is needed using our ALD process. Meanwhile, the Pt nanotube array is explored as a current collector for supercapacitors based on core/ shell Pt/MnO2 nanotubes. This nanotube-based electrode exhibits high gravimetric and areal specific capacitance (810 Fg−1 and 75 mF cm−2 at a scan rate of 5 mV s−1) as well as an excellent rate capability (68% capacitance retention from 2 to 100 Ag−1). Additionally, a negligible capacitance loss is observed after 8000 cycles of random charging-discharging from 2 to 100 Ag−1.
1. Introduction Metal nanostructures with high regularity, complex architectures and large specific surface areas have been extensively utilized in various application fields.[1–5] Due to the unique advantages of Pt including high catalytic ability, chemical stability against oxidation and high work function of 5.6 eV, it plays an especially important role in fuel cells and microelectronics.[6–12] Considering the fact that Pt is a
L. Wen, Y. Mi, Dr. C. Wang, Y. Fang, F. Grote, Dr. H. Zhao, Dr. M. Zhou, Prof. Y. Lei Institut für Physik & IMN MacroNano* (ZIK) Technische Universität Ilmenau Ilmenau, Germany E-mail: [email protected] DOI: 10.1002/smll.201400436
3162
wileyonlinelibrary.com
very expensive metal, a major challenge nowadays is how to realize cost-effective utilization of Pt. To meet this challenge, an approximate fivefold reduction of Pt content is necessary at the present state of industrial development.[4,5] Meanwhile, the trend of device miniaturization requires efficient techniques enabling Pt fabrication at nano-scale range so as to increase the overall surface-to-volume ratio and hence to reduce the Pt loading. Recently, intensive interests have been raised for synthesizing different configurations of Pt nanostructures, especially for nanowires and nanotubes.[13] Anodic alumina nano-porous templates have been widely used for fabricating ordered arrays of diverse nanostructures,[14–18] including Pt nanowires and nanotubes.[19–22] Comparing to dispersed Pt nanoparticles that usually have agglomeration problems, Pt nanowires and nanotubes are highly desirable for applications as catalytic materials and micro-electrodes.[4,12,20] Atomic layer deposition (ALD), which allows conformal and homogenous three-dimensional nano-fabrication with
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2014, 10, No. 15, 3162–3168
Cost-effective Atomic Layer Deposition Synthesis of Pt Nanotube Arrays
high structural controllability at atomic level, becomes a promising alternative approach for realizing Pt nanostructures.[23] To date, Pt has been deposited by ALD process on various substrates.[24–27] Suffering from poor nucleation and inland growth, dispersed Pt nanoparticles usually appeared at the beginning of the ALD process.[23] And a real layer-bylayer atomic growth can only be achieved when the dispersed Pt nanoparticles coalesce together and finally forms a continuous film, which requires a large number of ALD cycles and hence makes the ALD growth of Pt nanostructures high costly. Meanwhile, owing to the reaction limitation and the large molecular mass of Pt precursors (MeCpPtMe3), the diffusion and transportation of the precursors in nano-pores are very slow,[21,22,28,29] which require a sufficient long pulsing time so as to accomplish the precursor diffusion and reaction process within the nano-pores. Just recently, increasing Pt precursor pulsing time up to 30 seconds has been tried to obtain high Pt coverage inside alumina templates,[22] in which a large amount of Pt precursor was consumed and hence further increases the cost of the ALD growth of Pt nanostructures. This is also the reason why most of the Pt ALD growth so far is limited to nanoparticles or low aspect ratio structures. Therefore, how to realize a cost-effective ALD process for high aspect ratio Pt nanostructures presents a timely topic for advanced electronic devices such as energy storage elements, transistors and sensors. Herein, we achieve a highly cost-effective ALD technique for synthesising pre-defined Pt nanotube (NT) arrays within alumina nano-porous templates. This low Pt consuming advantage is attributed to a low N2 filling step of the ALD process, by which much fewer cycles and a shorter Pt precursor pulsing time are required comparing to the conventional ALD growth. It is demonstrated that about half ALD cycles and 10% Pt precursor pulsing time is necessary for obtaining similar Pt NTs using the low N2 filling step, enabling Pt nanostructures more practicable for device utilizations such as superior current collectors. Therefore, starting from these Pt NTs, core/shell Pt/MnO2 NT arrays are fabricated for realizing a high performance supercapacitor. The presented approach shall be applicable for fabricating other noble metal nanostructures, since the ALD processes of these materials are ruled by the similar reaction mechanisms to that of Pt.
in a supercapacitor electrode based on core/shell Pt/MnO2 NT array (Figure 1g). For this core/shell structure, the core Pt NT with high electrical conductivity is used as a current collector to provide paths for charge storage and delivery. Meanwhile, the construction of the NTs can also offer a large surface area to increase the active material (i.e., MnO2) mass loading. The pre-patterned alumina templates are fabricated by a similar process to that in previous report.[30,31] Firstly, a Ni nano-piller stamp, which was copied from a silicon master mold with ordered nanohole arrays, is used to imprint aluminum foils. We have successfully printed a 1 cm × 1 cm defect-free area, and the silicon master mold can be reusable for many times without noticeable damage. Different anodization conditions and wet-chemical etching time are applied to obtain the desirable pre-patterned alumina templates (Figure S1). After that, when a 100-cycle conventional ALD process (the solid line of Figure 2a) is used for the Pt growth, dispersed Pt nanoparticles are formed on the surface of the template (Figure 2b1). After 200 ALD cycles, the dispersed Pt nanoparticles coalesce into a porous film (Figure 2b2), and this behaviour is consistent with the Volmer-Weber island growth model of metal films on oxide surfaces.[23] Instead of using the conventional process, we introduce an innovative ALD process for Pt growth, where a low N2 filling step is used after pulsing (as illustrated in the dot line of Figure 2a). Figure 2b3 shows the template surface after 100 cycles with a 60 s low N2 filling step, where Pt porous film already appeared. This demonstrates that the nucleation delay phenomenon of the Pt ALD growth in our process is largely decreased and about half of the Pt precursor can be saved to obtain a similar Pt film. When the cycle number is increased to 200 (Figure 2b4), the porous Pt film further transforms to a compact film. Importantly, the infiltration depth of the Pt inside the template pores reaches about 5 µm, which is approximately 2.5 times of the infiltration depth in the conventional process (Figures 2c and S2c).
2. Results and Discussion The fabrication procedure is schematically shown in Figure 1. In this process, a Pt NT array is grown on a pre-patterned alumina template, as demonstrated in Figures 1a–1d. After that, a mixed polydimethylsiloxane (PDMS) solution is poured on the surface of the template, followed by dissolving the backside alumina template (Figures 1e and 1f). Finally, a thin MnO2 layer is deposited on the Pt NTs, resulting small 2014, 10, No. 15, 3162–3168
Figure 1. Schematic illustration of the fabrication process of Pt and Pt/MnO2 NT arrays: (a) surface imprinting on an aluminium foil using a Ni nano-piller stamp; (b) anodization for preparing alumina template followed by a chemical etching process; (c) dispersed Pt nanoparticles formed on template after a few ALD growth cycles; (d) continuous Pt NT array obtained after more ALD growth cycles; (e) a mixed PDMS solution poured on alumina template; (f) removing alumina template and resulting in a Pt NT array on PDMS substrate; (g) electrodepositing MnO2 to form Pt/MnO2 NT array for supercapacitor electrode.
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
3163
full papers
L. Wen et al.
Figure 2. (a) Comparison of representative one-cycle ALD growth of Pt between conventional (solid line) and innovative (dot line) process. (b1) and (b2) show the surface of alumina template after 100 and 200 cycles Pt growth in the conventional process, (b3) and (b4) show the surface of alumina template after 100 and 200 cycles Pt growth in the innovative process. (c) Dependence of the Pt infiltration depth on the low N2 filling time, the inset is the relationship of the Pt infiltration depth to the Pt precursor pulsing time.
To investigate the influence of the low N2 filling step on the Pt infiltration depth inside the pores, a series of experiments are conducted in which the low N2 filling time is adjusted in the range of 0 – 90 s. It is revealed in Figures 2c and S2a-S2d that the infiltration depth increases with the increment of the low N2 filling time, where a maximum depth of about 6 µm is obtained when the time is increased to 90 s. Since the pore diameter is about 260 nm, the aspect ratio of Pt NTs increases about 3.5 times (from 7 to 23) with the same Pt precursor pulsing time of 1.3 s. Based on previous reports, the Pt infiltration depth in nanopores vs the Pt precursor pulsing time shows a square root dependence (Pt infiltration depth ∝ Pt pulsing time1/2), which belongs to the diffusionlimited regime of ALD coating on nanopores.[21,28] Different from the above-mentioned relationship, the Pt infiltration depth vs the low N2 filling time shows good square root dependence (Pt infiltration depth ∝ Low N2 filling time1/2) in our ALD process (Figure 2c). In the conventional ALD process, about 12 s of Pt precursor pulsing is required to obtain Pt NTs with an infiltration depth of 6 µm. This assumption is also supported by previous experiment results, where up to
3164 www.small-journal.com
30 s of Pt precursor pulsing time was used for growing about 10 µm Pt NTs inside the alumina template.[22] However, when a 90 s low N2 filling step is introduced in ALD process, only 10% Pt precursor pulsing time is required for the similar Pt infiltration depth. It clearly indicates that the Pt precursor pulsing time can be largely shortened with the additional low N2 filling step in the Pt NT growth. Meanwhile, the influence of Pt precursor pulsing time on Pt infiltration depth is also studied. When the Pt pulsing time increases from 0.5 to 2.1 s with the same N2 low filling step of 30 s, the Pt infiltration depth only extends to about 5.3 µm (as shown in the inset curve of Figures 2c, S2e and S2f). Even so, this value is still less than that of the Pt NTs obtained from a shorter pulsing (1.3 s) but a longer low N2 filling (90 s) process. It is well known that there are two main requirements for the thermal ALD growth of Pt on oxide substrates: (i) adsorption of the Pt precursor (MeCpPtMe3) and (ii) elimination of the precursor ligands. For the Pt growth within a nano-porous template, the Pt coverage is typically determined by the material surface, the applied pressure and the precursor pulsing time.[28,29] Whereas using
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2014, 10, No. 15, 3162–3168
Cost-effective Atomic Layer Deposition Synthesis of Pt Nanotube Arrays
Figure 3. (a) Left SEM image is the cross section of Pt NTs after 200 cycles of ALD growth using the innovative process. The right images are enlarged views of the top and bottom part. (b) XPS data of the Pt NTs in (a). (c) I–V curves of Pt NTs after partially etching the backside template. The inset is the schematic of the conductivity measurement setup with two pairs of electrodes (a and c, b and d), the inter-electrode distance is about 0.7 cm.
the same Pt precursor pulsing time and the same template, the low N2 filling step builds up a higher Pt precursor partial pressure. The higher-pressure condition not only enhances the Pt precursor diffusion and adsorption inside the pores, but also largely extends the nucleation time of the Pt nanoparticles on oxide surfaces. Therefore, instead of extending Pt precursor pulsing time, introducing a low N2 filling step in Pt ALD process is an efficient choice for realizing high aspect ratio Pt NTs with an ultra-less Pt precursor consumption. In addition, a pre-patterned alumina template with a thickness of about 6.0 µm is used to prepare continuous Pt NT arrays with the 90 s low N2 filling step. The left image of Figure 3a presents a cross-sectional view of the sample. The small 2014, 10, No. 15, 3162–3168
whole surface of the template is covered with Pt NTs after 200 cycles of ALD growth. The right two images are the enlarged views of the top and bottom part of the sample, respectively. It is clear that the inside surface of Pt NTs at the top is more compact and smooth than those at the bottom. Even so, we still found that the Pt NT is a continuous film, which is firstly indicated by the XPS measurement. Since the penetrating depth of XPS measurement is about 10 nm, the lack of the Al2O3 XPS signal from the underlying alumina template argues for the continuous characteristic of the Pt NTs (Figure 3b). This point is further confirmed by the smooth and continuous outside surface of Pt NTs after removing the template (Figure S3). A Pt NT sample after partially etching the backside alumina template is used for conductivity measurement. Four silver pastes (denoted as electrode a, b, c and d) are connected to the four edges of the sample, constructing two pairs of measuring electrodes (Figure 3c). The sheet electrical resistance of the continuous Pt NT film is as low as 3.4 Ω/cm2 and there is only small variation between the two different pairs. Pt NTs with about 180 nm tube diameter and 2 µm length (shown in Figure 4a) are used for growing core/shell Pt/ MnO2 NTs for supercapacitor applications. MnO2 shells are electrodeposited on Pt cores. The overall core-shell NT diameter (and hence the thickness of MnO2 shell) can be adjusted by controlling the MnO2 deposition time: the tube diameter and the shell thickness is about 240 and 30 nm when the deposition time is 30 s (Figure 4b), while the diameter and thickness reach up to about 340 and 80 nm respectively for 90 s deposition time (Figure 4c). These pre-defined core-shell Pt/MnO2 NT arrays are used to construct supercapacitor electrodes (a photography of a typical electrode is shown in Figure 4d). In order to study the performance of the core/shell Pt/ MnO2 NT arrays for electrochemical energy storage, cyclic voltammetry (CV) and galvanostatic charge-discharge are measured in a voltage window of 0–0.9 V. The CV tests are performed on four samples at a scan rate of 20 mVs−1 in 1.0 M Na2SO4: bare Pt NT array, 30 s deposition Pt/MnO2 NT array (denoted as 30 s-NT electrode), 90 s deposition Pt/ MnO2 NTs array (denoted as 90 s-NT electrode), MnO2 on a Pt planar film (denoted as PF electrode) with same mass loading of the 30 s-NT sample (Figure 5a). The enclosed areas of the CV curves become much larger after MnO2 loading, which reveals that the Pt electrode contribution to the capacitance can be ignored. The CV curves of the 30 s-NT and 90 s-NT electrodes have a clear quasi-rectangular shape, indicating their ideal electrical double-layer capacitance behaviours. On the contrary, the CV curve of the PF electrode is much worse compared to the NT electrodes. Meanwhile, the rate-dependent CVs of the typical 30 s-NT electrode are investigated from 5 to 100 mV s−1 (Figure 5b). All the curves maintain a nearly ideal capacitive CV shape with only small distortions, showing remarkable rate capabilities of the electrode. After that, the gravimetric specific capacitance (Csp) vs the scan rate is summarized in Figure 5c, which demonstrates that the capacitance of the 30 s-NT electrode reaches to 810 Fg−1 at the scan rate of 5 mV s−1 (Eq S1). And the areal specific capacitance (Casp) also reaches up to 75 mFcm−2
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
3165
full papers
L. Wen et al.
behavior from the current density of 20 to 100 Ag−1 (Figure 5d) and the representative linear voltage-time profiles at 2 Ag−1 (Figure S4) show high symmetric natures in all current densities. The summary plot of Csp versus current density demonstrates that the 30 s-NT electrode reaches up to 793 Fg−1 (73 mFcm−2) at 2 Ag−1 and this value is still at 530 Fg−1 (114 mFcm−2) when the deposition time is increased to 90 s (Figure 5e and Eq S2-S5). And both the NT-electrodes have a much better performance than that of the PF electrode in all current densities. Most importantly, when the current density of the 30 s-NT electrode increases to 100 Ag−1, the Csp still keeps at 542 Fg−1 (50 mFcm−2), which retains at about 68% capacitance of the Csp at 2 Ag−1. For most of the previFigure 4. Tilted and top SEM images of (a) Pt NT array, (b) MnO2 shell on Pt NTs after 30 s ously reported core/shell and hybrid nanodeposition, (c) MnO2 shell on Pt NTs after 90 s deposition. (d) Photograph of a supercapacitor structures (e.g., Au/MnO2 nanopores,[3]Au/ electrode on PDMS substrate using core/shell Pt/MnO2 NT array. MnO2 nanowires[32] and AuPd/MnO2 nanorods,[33] the Csp is very high at low (Eq S5). Even when the deposition time increases to 90 s, the current densities while it decreases dramatically at high curelectrode Csp still keeps at 550 Fg−1 (119 mFcm−2). The Csp of rent densities. The good rate capability of our Pt/MnO2 NT the 30 s- and 90 s-NT electrodes yield much better perfor- arrays electrodes can be attributed to the superior Pt NT mance at all scan rates compared with the PF electrode. For the collector, which provides not only a highly accessible surface 30 s-NT electrode, the typical one-cycle charging-discharging area but also appropriate gaps among the adjacent NTs for
Figure 5. (a) The CVs of bare Pt-NT, 30 s-NT, 90 s-NT and PF electrodes at a scan rate of 20 mV s−1. (b) Typical CVs of the 30 s-NT electrode at different scan rates from 5 to 100 mV s−1. (c) Specific capacitance curves of the 30 s-NT, 90 s-NT and PF electrodes at the different scan rates. (d) Representative linear voltage-time profiles for the charging and discharging of the 30 s-NT electrode at high current densities from 20 to 100 Ag−1. (e) The summary plot of Csp versus current density of the 30 s-NT, 90 s-NT and PF electrodes. (f) Cycling stability of the 30 s-NT electrode at random current densities up to 8000 cycles.
3166 www.small-journal.com
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2014, 10, No. 15, 3162–3168
Cost-effective Atomic Layer Deposition Synthesis of Pt Nanotube Arrays
fast and reversible Faradic reactions and short ion diffusion paths. The long-term cycle stability of the 30 s-NT electrode is also evaluated. The Csp as a function of cycle number is presented in Figure 5f, where a negligible Csp loss at 20 Ag−1 is observed over the first 3000 cycles. After that, random charging and discharging with current densities from 2 to 100 Ag−1 is carried out up to 8000 cycles and the electrode still presents a high cycle stability and a negligible capacitance loss after such a sudden variation. Moreover, the morphology and the elements of the electrode after the cycling test are investigated by EDX mapping over a 20 µm × 20 µm area and by EDX line-scanning across a single Pt/MnO2 NT (Figure S5), where the electrode keeps good core/shell structural feature even after 8000 cycles of random cycling. The Ragone plot (power density vs energy density) of the electrode at various current densities is shown in Figure S6, which delivers a high power density of about 56 kW kg−1 at the energy density of about 61 kWh kg−1. The energy and power density in this study reveals that our Pt/MnO2 NT-based electrodes fulfill the power requirement (15 kW kg−1) for PNGV (Partnership for a New Generation of Vehicles).
3. Conclusion In summary, we successfully establish an innovative ALD process to grow regular arrays of Pt NT arrays on pre-patterned alumina templates, in which a low N2 filling step is introduced. It is demonstrated that with the low N2 filling step, half ALD cycles can be saved for growing similar Pt films. Additionally, Pt NTs with an infiltration depth up to 6 µm are achieved and the aspect ratio of Pt NTs increases about 3.5 times (from 7 to 23) with the same Pt precursor pulsing time of 1.3 s. For obtaining an identical infiltration depth in the conventional process, about ten times of Pt precursor pulsing time is required. The presented approach here should be applicable to obtain other noble metal nanostructures, since they have the similar ALD reaction mechanisms to that of Pt. Meanwhile, Pt/MnO2 core/shell NT array on a PDMS substrate is explored for high-performance supercapacitor applications. This core-shell NT electrode exhibits a high gravimetric and areal specific capacitance (810 Fg−1 and 75 mFcm−2 at 5 mV s−1) as well as an excellent rate capability (68% capacitance retention from 2 to 100 Ag−1). A negligible capacitance loss is observed after 8000 random charging-discharging cycles from 2 to 100 Ag−1. These good performances can be attributed to the well-defined Pt nanotube arrays, which provide a high accessible surface area and inter-tube gaps for fast and reversible Faradic reactions and short ion diffusion paths. Overall, the Pt NT arrays demonstrated here provide desirable candidate structures for high-performance supercapacitors as well as for other Pt-based nano-devices such as energy storage elements, transistors and sensors.
4. Experimental Section Fabrication of pre-patterned alumina templates: Firstly, the surface of a master Si mold (AMO GmbH) is cleaned by small 2014, 10, No. 15, 3162–3168
immersing the sample in a Piranha solution (H2SO4/H2O2; 3:1) for 30 min with ultrasonication. Surface modification is obtained by treating the Si master pattern with 3-aminopropyltriethoxysilane (3-APTES; 1.0 vol% in CH3CH2OH) at 65 °C. After that, Ni is electrodeposited on the Si master mold using a Ni plating solution comprising NiCl2·6H2O (8.41×10−2 M), Ni (H2NSO3)2·4H2O (1.59 M), and H3BO3 (0.33 M). The typical current density for the Ni deposition is 10 mAcm−2. After the electrodeposition process, Ni imprint stamps are obtained by stripping the metal films from the Si master mold, and the Si master mold can be recycling-used for many times for fabricating Ni imprint stamps. The Ni films are used for the imprinting process on Al foils. Typically, the imprinted Al foil is achieved by applying a pressure of about 10 kN cm−2 for 3 min using an oil pressing system. Finally, the anodization of prepatterned Al is conducted under a constant voltage of 160 V in a H3PO4 solution (0.4 M) at 15 °C, where the anodization voltage is chose to satisfy the linear relationship between the interpore distance and the anodization potential (2.5 nm V−1). ALD fabrication of Pt NT arrays: After 1 h anodization under 160 V, the pre-patterned alumina nano-porous template is further etched in a H3PO4 solution (5 wt%) at 30 °C for 90 min. The fabrication of Pt NT arrays is conducted in a Picosun SUNALETM R150 ALD System. The ALD reactor is maintained at a temperature of 300 °C and the chamber pressure varies from 8 hPa to 30 hPa under different growth processes. The temperature of Pt(MeCp)Me3 canister is held at 80 °C. A typical conventional ALD growth cycle consists of four steps: Pt(MeCp)Me3 pulsing (1.3 s) – N2 purging (18 s) – O2 pulsing (1.3 s) – N2 purging (18 s). The N2 carrying gases are kept at 100 sccm and the pressure of the reaction chamber is about 8 hPa during the growth. For our innovative ALD growth cycle, it consists of six steps: Pt(MeCp)Me3 pulsing (1.3 s) – low N2 filling (30 s) -N2 purging (18 s) – O2 pulsing (1.3 s) – low N2 filling (30 s) – N2 purging (18 s). During the low N2 filling step, the N2 carrying gases are decreased to 60 sccm and the pressure of the chamber is gradually raised up to about 21 hPa. For the other steps, the N2 carrying gases are still kept at 100 sccm. When 60 s and 90 s low N2 filling step are used, the chamber pressure will be increased up to about 24 and 30 hPa, respectively. Synthesis of core/shell Pt/MnO2 NT arrays: The pre-patterned alumina template for the core/shell Pt/MnO2 NT arrays is anodized at 160 V for 15 min and then etched in H3PO4 solution (5 wt%) at 30 °C for 30 min. 200 cycles growth of Pt using the innovative ALD process with Pt(MeCp)Me3 pulsing (0.8 s) and low N2 filling (90 s) are used for growing Pt NTs within the templates. Then, Polydimethylsiloxane (PDMS, Sylgard 184 Dow Corning) is poured on the alumina template and is baked at 60 °C for 4 h. The backside aluminum is removed by a mixture solution (3.4 g copper chloride, 100 mL hydrochloric acid and 100 mL deionized water). Pt NTs are released by dissolving the alumina template in a NaOH solution (1.0 M) for 1 h, followed by a rinsing process with deionized water. Subsequently, manganese oxide (MnO2) is deposited at a constant potential of 0.7 V using an aqueous solution of manganese acetate (100 mM) and sodium sulfate (100 mM). The potential is measured versus an Ag/AgCl reference electrode. And a Pt foil is used as the counter electrode. The mass of the MnO2 is determined from the charges passed during electrochemical deposition and assumes 100% efficiency. Structure and electrical characterization: the Pt and Pt/MnO2 NTs are investigated using a field emission scanning electron
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
3167
full papers
L. Wen et al.
microscope (SEM) and energy dispersive X-ray spectrometry (Auriga Zeiss FIB). I–V curves of the Pt NTs are measured by Keithley 2612B. Electrochemical studies are performed on an electrochemical workstation (BioLogic, Inc.) with a 3-electrode system, which consists of an Ag/AgCl reference electrode, a Pt foil counter electrode and a Pt/MnO2 NT arrays working electrode. All the tests are carried out in Na2SO4 aqueous solution (1.0 M). Cyclic voltammetry is carried out at scan rates from 5 to 100 mV s−1. Galvanostatic charge/discharge cycling is measured at different current densities from 2 to 100 Ag−1. A potential window in the range of 0–0.9 V is used in all measurements.
Supporting Information Supporting Information is available from the Wiley Online Library or from the author.
Acknowledgements Financial support from the European Research Council Grant and BMBF (ZIK: 3DNanoDevice) is gratefully acknowledged. Liaoyong Wen and Yan Mi contributed equally to this work.
[1] M. E. Stewart, C. R. Anderton, L. B. Thompson, J. Maria, S. K. Gray, J. A. Rogers, R. G. Nuzzo, Chem. Rev. 2008, 108, 494–521. [2] H. G. Zhang, X. D. Yu, P. V. Braun, Nat. Nanotechnol. 2011, 6, 277–281. [3] X. Y. Lang, A. Hirata, T. Fujita, M. W. Chen, Nat. Nanotechnol. 2011, 6, 232–236. [4] D. F. van der Vliet, C. Wang, D. Tripkovic, D. Strmcnik, X. F. Zhang, M. K. Debe, R. T. Atanasoski, N. M. Markovic, V. R. Stamenkovic, Nat. Mater. 2012, 11, 1051–1058. [5] M. K. Debe, Nature 2012, 486, 43–51. [6] E. P. Lee, Z. M. Peng, D. M. Cate, H. Yang, C. T. Campbell, Y. N. Xia, J. Am. Chem. Soc. 2007, 129, 10634–10635. [7] X. W. Yu, S. Y. Ye, J. Power Sources. 2007, 172, 133–144. [8] E. P. Lee, Z. M. Peng, W. Chen, S. W. Chen, H. Yang, Y. N. Xia, ACS Nano 2008, 2, 2167–2173. [9] S. H. Sun, F. Jaouen, J. P. Dodelet, Adv. Mater. 2008, 20, 3900–3904.
3168 www.small-journal.com
[10] J. L. Shui, J. C. M. Li, Nano Lett. 2009, 9, 1307–1314. [11] H. W. Liang, X. A. Cao, F. Zhou, C. H. Cui, W. J. Zhang, S. H. Yu, Adv. Mater. 2011, 23, 1467–1471. [12] H. Knoops, M. E. Donders, L. Baggetto, M. C. Van de Sanden, P. Notten, W. M. Kessels, ECS Trans. 2009, 25, 333–344. [13] A. C. Chen, P. Holt-Hindle, Chem. Rev. 2010, 110, 3767–3804. [14] Y. Lei, C. H. Liang, Y. C. Wu, L. D. Zhang, Y. Q. Mao, J. Vac. Sci. Technol. B. 2001, 19, 1109–1114. [15] Y. Lei, W. K. Chim, J. Weissmuller, G. Wilde, H. P. Sun, X. Q. Pan, Nanotechnology 2005, 16, 1892–1898. [16] M. H. Wu, L. Y. Wen, Y. Lei, S. Ostendorp, K. Chen, G. Wilde, Small 2010, 6, 695–699. [17] L. Y. Wen, Z. Z. Shao, Y. G. Fang, K. M. Wong, Y. Lei, L. F. Bian, G. Wilde, Appl. Phys. Lett. 2010, 97, 053106–053108. [18] S. R. Gowda, A. L. M. Reddy, X. B. Zhan, H. R. Jafry, P. M. Ajayan, Nano Lett. 2012, 12, 1198–1202. [19] G. Y. Zhao, C. L. Xu, D. J. Guo, H. Li, H. L. Li, J. Power Sources. 2006, 162, 492–496. [20] K. S. Napolskii, P. J. Barczuk, S. Y. Vassiliev, A. G. Veresov, G. A. Tsirlina, P. J. Kulesza, Electrochim. Acta. 2007, 52, 7910–7919. [21] D. J. Comstock, S. T. Christensen, J. W. Elam, M. J. Pellin, M. C. Hersam, Adv. Funct. Mater. 2010, 20, 3099–3105. [22] D. F. Gu, H. Baumgart, K. Tapily, P. Shrestha, G. Namkoong, X. Y. Ao, F. Muller, Nano Res. 2011, 4, 164–170. [23] S. M. George, Chem. Rev. 2010, 110, 111–131. [24] H. B. R. Lee, S. F. Bent, Chem. Mater. 2012, 24, 279–286. [25] H.-B.-R. Lee, S. H. Baeck, T. F. Jaramillo, S. F. Bent, Nano Lett. 2013, 13, 457–463. [26] J. L. Lu, J. W. Elam, P. C. Stair, Acc. Chem. Res. 2013, 46, 1806–1815. [27] A. J. M. Mackus, M. A. Verheijen, N. Leick, A. A. Bol, W. M. M. Kessels, Chem. Mater. 2013, 25, 1905–1911. [28] J. W. Elam, D. Routkevitch, P. P. Mardilovich, S. M. George, Chem. Mater. 2003, 15, 3507–3517. [29] R. G. Gordon, D. Hausmann, E. Kim, J. Shepard, Chem. Vapor Depos. 2003, 9, 73–78. [30] W. Lee, R. Ji, C. A. Ross, U. Gösele, K. Nielsch, Small 2006, 2, 978–982. [31] C. L. Wang, L. Y. Wen, T. Kups, Y. Mi, R. Vellacheri, Y. G. Fang, P. Schaaf, H. P. Zhao, Y. Lei, J. Mater. Chem. C 2013, 1, 8003– 8006. [32] W. B. Yan, J. Y. Kim, W. D. Xing, K. C. Donavan, T. Ayvazian, R. M. Penner, Chem. Mater. 2012, 24, 2382–2390. [33] Z. N. Yu, B. Duong, D. Abbitt, J. Thomas, Adv. Mater. 2013, 25, 3302–3306.
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Received: February 17, 2014 Revised: March 15, 2014 Published online: April 3, 2014
small 2014, 10, No. 15, 3162–3168
Cost-effective Atomic Layer Deposition Synthesis of Pt Nanotube Arrays: Application for High Performance Supercapacitor Liaoyong Wen, Yan Mi, Chengliang Wang, Yaoguo Fang, Fabian Grote, Huaping Zhao, Min Zhou, and Yong Lei*
Due
to the unique advantages of Pt, it plays an important role in fuel cells and microelectronics. Considering the fact that Pt is an expensive metal, a major challenging point nowadays is how to realize efficient utilization of Pt. In this paper, a cost-effective atomic layer deposition (ALD) process with a low N2 filling step is introduced for realizing well-defined Pt nanotube arrays in anodic alumina nanoporous templates. Compared to the conventional ALD growth of Pt, much fewer ALD cycles and a shorter precursor pulsing time are required, which originates from the low N2 filling step. To achieve similar Pt nanotubes, about half cycles and 10% Pt precursor pulsing time is needed using our ALD process. Meanwhile, the Pt nanotube array is explored as a current collector for supercapacitors based on core/ shell Pt/MnO2 nanotubes. This nanotube-based electrode exhibits high gravimetric and areal specific capacitance (810 Fg−1 and 75 mF cm−2 at a scan rate of 5 mV s−1) as well as an excellent rate capability (68% capacitance retention from 2 to 100 Ag−1). Additionally, a negligible capacitance loss is observed after 8000 cycles of random charging-discharging from 2 to 100 Ag−1.
1. Introduction Metal nanostructures with high regularity, complex architectures and large specific surface areas have been extensively utilized in various application fields.[1–5] Due to the unique advantages of Pt including high catalytic ability, chemical stability against oxidation and high work function of 5.6 eV, it plays an especially important role in fuel cells and microelectronics.[6–12] Considering the fact that Pt is a
L. Wen, Y. Mi, Dr. C. Wang, Y. Fang, F. Grote, Dr. H. Zhao, Dr. M. Zhou, Prof. Y. Lei Institut für Physik & IMN MacroNano* (ZIK) Technische Universität Ilmenau Ilmenau, Germany E-mail: [email protected] DOI: 10.1002/smll.201400436
3162
wileyonlinelibrary.com
very expensive metal, a major challenge nowadays is how to realize cost-effective utilization of Pt. To meet this challenge, an approximate fivefold reduction of Pt content is necessary at the present state of industrial development.[4,5] Meanwhile, the trend of device miniaturization requires efficient techniques enabling Pt fabrication at nano-scale range so as to increase the overall surface-to-volume ratio and hence to reduce the Pt loading. Recently, intensive interests have been raised for synthesizing different configurations of Pt nanostructures, especially for nanowires and nanotubes.[13] Anodic alumina nano-porous templates have been widely used for fabricating ordered arrays of diverse nanostructures,[14–18] including Pt nanowires and nanotubes.[19–22] Comparing to dispersed Pt nanoparticles that usually have agglomeration problems, Pt nanowires and nanotubes are highly desirable for applications as catalytic materials and micro-electrodes.[4,12,20] Atomic layer deposition (ALD), which allows conformal and homogenous three-dimensional nano-fabrication with
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2014, 10, No. 15, 3162–3168
Cost-effective Atomic Layer Deposition Synthesis of Pt Nanotube Arrays
high structural controllability at atomic level, becomes a promising alternative approach for realizing Pt nanostructures.[23] To date, Pt has been deposited by ALD process on various substrates.[24–27] Suffering from poor nucleation and inland growth, dispersed Pt nanoparticles usually appeared at the beginning of the ALD process.[23] And a real layer-bylayer atomic growth can only be achieved when the dispersed Pt nanoparticles coalesce together and finally forms a continuous film, which requires a large number of ALD cycles and hence makes the ALD growth of Pt nanostructures high costly. Meanwhile, owing to the reaction limitation and the large molecular mass of Pt precursors (MeCpPtMe3), the diffusion and transportation of the precursors in nano-pores are very slow,[21,22,28,29] which require a sufficient long pulsing time so as to accomplish the precursor diffusion and reaction process within the nano-pores. Just recently, increasing Pt precursor pulsing time up to 30 seconds has been tried to obtain high Pt coverage inside alumina templates,[22] in which a large amount of Pt precursor was consumed and hence further increases the cost of the ALD growth of Pt nanostructures. This is also the reason why most of the Pt ALD growth so far is limited to nanoparticles or low aspect ratio structures. Therefore, how to realize a cost-effective ALD process for high aspect ratio Pt nanostructures presents a timely topic for advanced electronic devices such as energy storage elements, transistors and sensors. Herein, we achieve a highly cost-effective ALD technique for synthesising pre-defined Pt nanotube (NT) arrays within alumina nano-porous templates. This low Pt consuming advantage is attributed to a low N2 filling step of the ALD process, by which much fewer cycles and a shorter Pt precursor pulsing time are required comparing to the conventional ALD growth. It is demonstrated that about half ALD cycles and 10% Pt precursor pulsing time is necessary for obtaining similar Pt NTs using the low N2 filling step, enabling Pt nanostructures more practicable for device utilizations such as superior current collectors. Therefore, starting from these Pt NTs, core/shell Pt/MnO2 NT arrays are fabricated for realizing a high performance supercapacitor. The presented approach shall be applicable for fabricating other noble metal nanostructures, since the ALD processes of these materials are ruled by the similar reaction mechanisms to that of Pt.
in a supercapacitor electrode based on core/shell Pt/MnO2 NT array (Figure 1g). For this core/shell structure, the core Pt NT with high electrical conductivity is used as a current collector to provide paths for charge storage and delivery. Meanwhile, the construction of the NTs can also offer a large surface area to increase the active material (i.e., MnO2) mass loading. The pre-patterned alumina templates are fabricated by a similar process to that in previous report.[30,31] Firstly, a Ni nano-piller stamp, which was copied from a silicon master mold with ordered nanohole arrays, is used to imprint aluminum foils. We have successfully printed a 1 cm × 1 cm defect-free area, and the silicon master mold can be reusable for many times without noticeable damage. Different anodization conditions and wet-chemical etching time are applied to obtain the desirable pre-patterned alumina templates (Figure S1). After that, when a 100-cycle conventional ALD process (the solid line of Figure 2a) is used for the Pt growth, dispersed Pt nanoparticles are formed on the surface of the template (Figure 2b1). After 200 ALD cycles, the dispersed Pt nanoparticles coalesce into a porous film (Figure 2b2), and this behaviour is consistent with the Volmer-Weber island growth model of metal films on oxide surfaces.[23] Instead of using the conventional process, we introduce an innovative ALD process for Pt growth, where a low N2 filling step is used after pulsing (as illustrated in the dot line of Figure 2a). Figure 2b3 shows the template surface after 100 cycles with a 60 s low N2 filling step, where Pt porous film already appeared. This demonstrates that the nucleation delay phenomenon of the Pt ALD growth in our process is largely decreased and about half of the Pt precursor can be saved to obtain a similar Pt film. When the cycle number is increased to 200 (Figure 2b4), the porous Pt film further transforms to a compact film. Importantly, the infiltration depth of the Pt inside the template pores reaches about 5 µm, which is approximately 2.5 times of the infiltration depth in the conventional process (Figures 2c and S2c).
2. Results and Discussion The fabrication procedure is schematically shown in Figure 1. In this process, a Pt NT array is grown on a pre-patterned alumina template, as demonstrated in Figures 1a–1d. After that, a mixed polydimethylsiloxane (PDMS) solution is poured on the surface of the template, followed by dissolving the backside alumina template (Figures 1e and 1f). Finally, a thin MnO2 layer is deposited on the Pt NTs, resulting small 2014, 10, No. 15, 3162–3168
Figure 1. Schematic illustration of the fabrication process of Pt and Pt/MnO2 NT arrays: (a) surface imprinting on an aluminium foil using a Ni nano-piller stamp; (b) anodization for preparing alumina template followed by a chemical etching process; (c) dispersed Pt nanoparticles formed on template after a few ALD growth cycles; (d) continuous Pt NT array obtained after more ALD growth cycles; (e) a mixed PDMS solution poured on alumina template; (f) removing alumina template and resulting in a Pt NT array on PDMS substrate; (g) electrodepositing MnO2 to form Pt/MnO2 NT array for supercapacitor electrode.
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
3163
full papers
L. Wen et al.
Figure 2. (a) Comparison of representative one-cycle ALD growth of Pt between conventional (solid line) and innovative (dot line) process. (b1) and (b2) show the surface of alumina template after 100 and 200 cycles Pt growth in the conventional process, (b3) and (b4) show the surface of alumina template after 100 and 200 cycles Pt growth in the innovative process. (c) Dependence of the Pt infiltration depth on the low N2 filling time, the inset is the relationship of the Pt infiltration depth to the Pt precursor pulsing time.
To investigate the influence of the low N2 filling step on the Pt infiltration depth inside the pores, a series of experiments are conducted in which the low N2 filling time is adjusted in the range of 0 – 90 s. It is revealed in Figures 2c and S2a-S2d that the infiltration depth increases with the increment of the low N2 filling time, where a maximum depth of about 6 µm is obtained when the time is increased to 90 s. Since the pore diameter is about 260 nm, the aspect ratio of Pt NTs increases about 3.5 times (from 7 to 23) with the same Pt precursor pulsing time of 1.3 s. Based on previous reports, the Pt infiltration depth in nanopores vs the Pt precursor pulsing time shows a square root dependence (Pt infiltration depth ∝ Pt pulsing time1/2), which belongs to the diffusionlimited regime of ALD coating on nanopores.[21,28] Different from the above-mentioned relationship, the Pt infiltration depth vs the low N2 filling time shows good square root dependence (Pt infiltration depth ∝ Low N2 filling time1/2) in our ALD process (Figure 2c). In the conventional ALD process, about 12 s of Pt precursor pulsing is required to obtain Pt NTs with an infiltration depth of 6 µm. This assumption is also supported by previous experiment results, where up to
3164 www.small-journal.com
30 s of Pt precursor pulsing time was used for growing about 10 µm Pt NTs inside the alumina template.[22] However, when a 90 s low N2 filling step is introduced in ALD process, only 10% Pt precursor pulsing time is required for the similar Pt infiltration depth. It clearly indicates that the Pt precursor pulsing time can be largely shortened with the additional low N2 filling step in the Pt NT growth. Meanwhile, the influence of Pt precursor pulsing time on Pt infiltration depth is also studied. When the Pt pulsing time increases from 0.5 to 2.1 s with the same N2 low filling step of 30 s, the Pt infiltration depth only extends to about 5.3 µm (as shown in the inset curve of Figures 2c, S2e and S2f). Even so, this value is still less than that of the Pt NTs obtained from a shorter pulsing (1.3 s) but a longer low N2 filling (90 s) process. It is well known that there are two main requirements for the thermal ALD growth of Pt on oxide substrates: (i) adsorption of the Pt precursor (MeCpPtMe3) and (ii) elimination of the precursor ligands. For the Pt growth within a nano-porous template, the Pt coverage is typically determined by the material surface, the applied pressure and the precursor pulsing time.[28,29] Whereas using
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2014, 10, No. 15, 3162–3168
Cost-effective Atomic Layer Deposition Synthesis of Pt Nanotube Arrays
Figure 3. (a) Left SEM image is the cross section of Pt NTs after 200 cycles of ALD growth using the innovative process. The right images are enlarged views of the top and bottom part. (b) XPS data of the Pt NTs in (a). (c) I–V curves of Pt NTs after partially etching the backside template. The inset is the schematic of the conductivity measurement setup with two pairs of electrodes (a and c, b and d), the inter-electrode distance is about 0.7 cm.
the same Pt precursor pulsing time and the same template, the low N2 filling step builds up a higher Pt precursor partial pressure. The higher-pressure condition not only enhances the Pt precursor diffusion and adsorption inside the pores, but also largely extends the nucleation time of the Pt nanoparticles on oxide surfaces. Therefore, instead of extending Pt precursor pulsing time, introducing a low N2 filling step in Pt ALD process is an efficient choice for realizing high aspect ratio Pt NTs with an ultra-less Pt precursor consumption. In addition, a pre-patterned alumina template with a thickness of about 6.0 µm is used to prepare continuous Pt NT arrays with the 90 s low N2 filling step. The left image of Figure 3a presents a cross-sectional view of the sample. The small 2014, 10, No. 15, 3162–3168
whole surface of the template is covered with Pt NTs after 200 cycles of ALD growth. The right two images are the enlarged views of the top and bottom part of the sample, respectively. It is clear that the inside surface of Pt NTs at the top is more compact and smooth than those at the bottom. Even so, we still found that the Pt NT is a continuous film, which is firstly indicated by the XPS measurement. Since the penetrating depth of XPS measurement is about 10 nm, the lack of the Al2O3 XPS signal from the underlying alumina template argues for the continuous characteristic of the Pt NTs (Figure 3b). This point is further confirmed by the smooth and continuous outside surface of Pt NTs after removing the template (Figure S3). A Pt NT sample after partially etching the backside alumina template is used for conductivity measurement. Four silver pastes (denoted as electrode a, b, c and d) are connected to the four edges of the sample, constructing two pairs of measuring electrodes (Figure 3c). The sheet electrical resistance of the continuous Pt NT film is as low as 3.4 Ω/cm2 and there is only small variation between the two different pairs. Pt NTs with about 180 nm tube diameter and 2 µm length (shown in Figure 4a) are used for growing core/shell Pt/ MnO2 NTs for supercapacitor applications. MnO2 shells are electrodeposited on Pt cores. The overall core-shell NT diameter (and hence the thickness of MnO2 shell) can be adjusted by controlling the MnO2 deposition time: the tube diameter and the shell thickness is about 240 and 30 nm when the deposition time is 30 s (Figure 4b), while the diameter and thickness reach up to about 340 and 80 nm respectively for 90 s deposition time (Figure 4c). These pre-defined core-shell Pt/MnO2 NT arrays are used to construct supercapacitor electrodes (a photography of a typical electrode is shown in Figure 4d). In order to study the performance of the core/shell Pt/ MnO2 NT arrays for electrochemical energy storage, cyclic voltammetry (CV) and galvanostatic charge-discharge are measured in a voltage window of 0–0.9 V. The CV tests are performed on four samples at a scan rate of 20 mVs−1 in 1.0 M Na2SO4: bare Pt NT array, 30 s deposition Pt/MnO2 NT array (denoted as 30 s-NT electrode), 90 s deposition Pt/ MnO2 NTs array (denoted as 90 s-NT electrode), MnO2 on a Pt planar film (denoted as PF electrode) with same mass loading of the 30 s-NT sample (Figure 5a). The enclosed areas of the CV curves become much larger after MnO2 loading, which reveals that the Pt electrode contribution to the capacitance can be ignored. The CV curves of the 30 s-NT and 90 s-NT electrodes have a clear quasi-rectangular shape, indicating their ideal electrical double-layer capacitance behaviours. On the contrary, the CV curve of the PF electrode is much worse compared to the NT electrodes. Meanwhile, the rate-dependent CVs of the typical 30 s-NT electrode are investigated from 5 to 100 mV s−1 (Figure 5b). All the curves maintain a nearly ideal capacitive CV shape with only small distortions, showing remarkable rate capabilities of the electrode. After that, the gravimetric specific capacitance (Csp) vs the scan rate is summarized in Figure 5c, which demonstrates that the capacitance of the 30 s-NT electrode reaches to 810 Fg−1 at the scan rate of 5 mV s−1 (Eq S1). And the areal specific capacitance (Casp) also reaches up to 75 mFcm−2
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
3165
full papers
L. Wen et al.
behavior from the current density of 20 to 100 Ag−1 (Figure 5d) and the representative linear voltage-time profiles at 2 Ag−1 (Figure S4) show high symmetric natures in all current densities. The summary plot of Csp versus current density demonstrates that the 30 s-NT electrode reaches up to 793 Fg−1 (73 mFcm−2) at 2 Ag−1 and this value is still at 530 Fg−1 (114 mFcm−2) when the deposition time is increased to 90 s (Figure 5e and Eq S2-S5). And both the NT-electrodes have a much better performance than that of the PF electrode in all current densities. Most importantly, when the current density of the 30 s-NT electrode increases to 100 Ag−1, the Csp still keeps at 542 Fg−1 (50 mFcm−2), which retains at about 68% capacitance of the Csp at 2 Ag−1. For most of the previFigure 4. Tilted and top SEM images of (a) Pt NT array, (b) MnO2 shell on Pt NTs after 30 s ously reported core/shell and hybrid nanodeposition, (c) MnO2 shell on Pt NTs after 90 s deposition. (d) Photograph of a supercapacitor structures (e.g., Au/MnO2 nanopores,[3]Au/ electrode on PDMS substrate using core/shell Pt/MnO2 NT array. MnO2 nanowires[32] and AuPd/MnO2 nanorods,[33] the Csp is very high at low (Eq S5). Even when the deposition time increases to 90 s, the current densities while it decreases dramatically at high curelectrode Csp still keeps at 550 Fg−1 (119 mFcm−2). The Csp of rent densities. The good rate capability of our Pt/MnO2 NT the 30 s- and 90 s-NT electrodes yield much better perfor- arrays electrodes can be attributed to the superior Pt NT mance at all scan rates compared with the PF electrode. For the collector, which provides not only a highly accessible surface 30 s-NT electrode, the typical one-cycle charging-discharging area but also appropriate gaps among the adjacent NTs for
Figure 5. (a) The CVs of bare Pt-NT, 30 s-NT, 90 s-NT and PF electrodes at a scan rate of 20 mV s−1. (b) Typical CVs of the 30 s-NT electrode at different scan rates from 5 to 100 mV s−1. (c) Specific capacitance curves of the 30 s-NT, 90 s-NT and PF electrodes at the different scan rates. (d) Representative linear voltage-time profiles for the charging and discharging of the 30 s-NT electrode at high current densities from 20 to 100 Ag−1. (e) The summary plot of Csp versus current density of the 30 s-NT, 90 s-NT and PF electrodes. (f) Cycling stability of the 30 s-NT electrode at random current densities up to 8000 cycles.
3166 www.small-journal.com
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2014, 10, No. 15, 3162–3168
Cost-effective Atomic Layer Deposition Synthesis of Pt Nanotube Arrays
fast and reversible Faradic reactions and short ion diffusion paths. The long-term cycle stability of the 30 s-NT electrode is also evaluated. The Csp as a function of cycle number is presented in Figure 5f, where a negligible Csp loss at 20 Ag−1 is observed over the first 3000 cycles. After that, random charging and discharging with current densities from 2 to 100 Ag−1 is carried out up to 8000 cycles and the electrode still presents a high cycle stability and a negligible capacitance loss after such a sudden variation. Moreover, the morphology and the elements of the electrode after the cycling test are investigated by EDX mapping over a 20 µm × 20 µm area and by EDX line-scanning across a single Pt/MnO2 NT (Figure S5), where the electrode keeps good core/shell structural feature even after 8000 cycles of random cycling. The Ragone plot (power density vs energy density) of the electrode at various current densities is shown in Figure S6, which delivers a high power density of about 56 kW kg−1 at the energy density of about 61 kWh kg−1. The energy and power density in this study reveals that our Pt/MnO2 NT-based electrodes fulfill the power requirement (15 kW kg−1) for PNGV (Partnership for a New Generation of Vehicles).
3. Conclusion In summary, we successfully establish an innovative ALD process to grow regular arrays of Pt NT arrays on pre-patterned alumina templates, in which a low N2 filling step is introduced. It is demonstrated that with the low N2 filling step, half ALD cycles can be saved for growing similar Pt films. Additionally, Pt NTs with an infiltration depth up to 6 µm are achieved and the aspect ratio of Pt NTs increases about 3.5 times (from 7 to 23) with the same Pt precursor pulsing time of 1.3 s. For obtaining an identical infiltration depth in the conventional process, about ten times of Pt precursor pulsing time is required. The presented approach here should be applicable to obtain other noble metal nanostructures, since they have the similar ALD reaction mechanisms to that of Pt. Meanwhile, Pt/MnO2 core/shell NT array on a PDMS substrate is explored for high-performance supercapacitor applications. This core-shell NT electrode exhibits a high gravimetric and areal specific capacitance (810 Fg−1 and 75 mFcm−2 at 5 mV s−1) as well as an excellent rate capability (68% capacitance retention from 2 to 100 Ag−1). A negligible capacitance loss is observed after 8000 random charging-discharging cycles from 2 to 100 Ag−1. These good performances can be attributed to the well-defined Pt nanotube arrays, which provide a high accessible surface area and inter-tube gaps for fast and reversible Faradic reactions and short ion diffusion paths. Overall, the Pt NT arrays demonstrated here provide desirable candidate structures for high-performance supercapacitors as well as for other Pt-based nano-devices such as energy storage elements, transistors and sensors.
4. Experimental Section Fabrication of pre-patterned alumina templates: Firstly, the surface of a master Si mold (AMO GmbH) is cleaned by small 2014, 10, No. 15, 3162–3168
immersing the sample in a Piranha solution (H2SO4/H2O2; 3:1) for 30 min with ultrasonication. Surface modification is obtained by treating the Si master pattern with 3-aminopropyltriethoxysilane (3-APTES; 1.0 vol% in CH3CH2OH) at 65 °C. After that, Ni is electrodeposited on the Si master mold using a Ni plating solution comprising NiCl2·6H2O (8.41×10−2 M), Ni (H2NSO3)2·4H2O (1.59 M), and H3BO3 (0.33 M). The typical current density for the Ni deposition is 10 mAcm−2. After the electrodeposition process, Ni imprint stamps are obtained by stripping the metal films from the Si master mold, and the Si master mold can be recycling-used for many times for fabricating Ni imprint stamps. The Ni films are used for the imprinting process on Al foils. Typically, the imprinted Al foil is achieved by applying a pressure of about 10 kN cm−2 for 3 min using an oil pressing system. Finally, the anodization of prepatterned Al is conducted under a constant voltage of 160 V in a H3PO4 solution (0.4 M) at 15 °C, where the anodization voltage is chose to satisfy the linear relationship between the interpore distance and the anodization potential (2.5 nm V−1). ALD fabrication of Pt NT arrays: After 1 h anodization under 160 V, the pre-patterned alumina nano-porous template is further etched in a H3PO4 solution (5 wt%) at 30 °C for 90 min. The fabrication of Pt NT arrays is conducted in a Picosun SUNALETM R150 ALD System. The ALD reactor is maintained at a temperature of 300 °C and the chamber pressure varies from 8 hPa to 30 hPa under different growth processes. The temperature of Pt(MeCp)Me3 canister is held at 80 °C. A typical conventional ALD growth cycle consists of four steps: Pt(MeCp)Me3 pulsing (1.3 s) – N2 purging (18 s) – O2 pulsing (1.3 s) – N2 purging (18 s). The N2 carrying gases are kept at 100 sccm and the pressure of the reaction chamber is about 8 hPa during the growth. For our innovative ALD growth cycle, it consists of six steps: Pt(MeCp)Me3 pulsing (1.3 s) – low N2 filling (30 s) -N2 purging (18 s) – O2 pulsing (1.3 s) – low N2 filling (30 s) – N2 purging (18 s). During the low N2 filling step, the N2 carrying gases are decreased to 60 sccm and the pressure of the chamber is gradually raised up to about 21 hPa. For the other steps, the N2 carrying gases are still kept at 100 sccm. When 60 s and 90 s low N2 filling step are used, the chamber pressure will be increased up to about 24 and 30 hPa, respectively. Synthesis of core/shell Pt/MnO2 NT arrays: The pre-patterned alumina template for the core/shell Pt/MnO2 NT arrays is anodized at 160 V for 15 min and then etched in H3PO4 solution (5 wt%) at 30 °C for 30 min. 200 cycles growth of Pt using the innovative ALD process with Pt(MeCp)Me3 pulsing (0.8 s) and low N2 filling (90 s) are used for growing Pt NTs within the templates. Then, Polydimethylsiloxane (PDMS, Sylgard 184 Dow Corning) is poured on the alumina template and is baked at 60 °C for 4 h. The backside aluminum is removed by a mixture solution (3.4 g copper chloride, 100 mL hydrochloric acid and 100 mL deionized water). Pt NTs are released by dissolving the alumina template in a NaOH solution (1.0 M) for 1 h, followed by a rinsing process with deionized water. Subsequently, manganese oxide (MnO2) is deposited at a constant potential of 0.7 V using an aqueous solution of manganese acetate (100 mM) and sodium sulfate (100 mM). The potential is measured versus an Ag/AgCl reference electrode. And a Pt foil is used as the counter electrode. The mass of the MnO2 is determined from the charges passed during electrochemical deposition and assumes 100% efficiency. Structure and electrical characterization: the Pt and Pt/MnO2 NTs are investigated using a field emission scanning electron
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
3167
full papers
L. Wen et al.
microscope (SEM) and energy dispersive X-ray spectrometry (Auriga Zeiss FIB). I–V curves of the Pt NTs are measured by Keithley 2612B. Electrochemical studies are performed on an electrochemical workstation (BioLogic, Inc.) with a 3-electrode system, which consists of an Ag/AgCl reference electrode, a Pt foil counter electrode and a Pt/MnO2 NT arrays working electrode. All the tests are carried out in Na2SO4 aqueous solution (1.0 M). Cyclic voltammetry is carried out at scan rates from 5 to 100 mV s−1. Galvanostatic charge/discharge cycling is measured at different current densities from 2 to 100 Ag−1. A potential window in the range of 0–0.9 V is used in all measurements.
Supporting Information Supporting Information is available from the Wiley Online Library or from the author.
Acknowledgements Financial support from the European Research Council Grant and BMBF (ZIK: 3DNanoDevice) is gratefully acknowledged. Liaoyong Wen and Yan Mi contributed equally to this work.
[1] M. E. Stewart, C. R. Anderton, L. B. Thompson, J. Maria, S. K. Gray, J. A. Rogers, R. G. Nuzzo, Chem. Rev. 2008, 108, 494–521. [2] H. G. Zhang, X. D. Yu, P. V. Braun, Nat. Nanotechnol. 2011, 6, 277–281. [3] X. Y. Lang, A. Hirata, T. Fujita, M. W. Chen, Nat. Nanotechnol. 2011, 6, 232–236. [4] D. F. van der Vliet, C. Wang, D. Tripkovic, D. Strmcnik, X. F. Zhang, M. K. Debe, R. T. Atanasoski, N. M. Markovic, V. R. Stamenkovic, Nat. Mater. 2012, 11, 1051–1058. [5] M. K. Debe, Nature 2012, 486, 43–51. [6] E. P. Lee, Z. M. Peng, D. M. Cate, H. Yang, C. T. Campbell, Y. N. Xia, J. Am. Chem. Soc. 2007, 129, 10634–10635. [7] X. W. Yu, S. Y. Ye, J. Power Sources. 2007, 172, 133–144. [8] E. P. Lee, Z. M. Peng, W. Chen, S. W. Chen, H. Yang, Y. N. Xia, ACS Nano 2008, 2, 2167–2173. [9] S. H. Sun, F. Jaouen, J. P. Dodelet, Adv. Mater. 2008, 20, 3900–3904.
3168 www.small-journal.com
[10] J. L. Shui, J. C. M. Li, Nano Lett. 2009, 9, 1307–1314. [11] H. W. Liang, X. A. Cao, F. Zhou, C. H. Cui, W. J. Zhang, S. H. Yu, Adv. Mater. 2011, 23, 1467–1471. [12] H. Knoops, M. E. Donders, L. Baggetto, M. C. Van de Sanden, P. Notten, W. M. Kessels, ECS Trans. 2009, 25, 333–344. [13] A. C. Chen, P. Holt-Hindle, Chem. Rev. 2010, 110, 3767–3804. [14] Y. Lei, C. H. Liang, Y. C. Wu, L. D. Zhang, Y. Q. Mao, J. Vac. Sci. Technol. B. 2001, 19, 1109–1114. [15] Y. Lei, W. K. Chim, J. Weissmuller, G. Wilde, H. P. Sun, X. Q. Pan, Nanotechnology 2005, 16, 1892–1898. [16] M. H. Wu, L. Y. Wen, Y. Lei, S. Ostendorp, K. Chen, G. Wilde, Small 2010, 6, 695–699. [17] L. Y. Wen, Z. Z. Shao, Y. G. Fang, K. M. Wong, Y. Lei, L. F. Bian, G. Wilde, Appl. Phys. Lett. 2010, 97, 053106–053108. [18] S. R. Gowda, A. L. M. Reddy, X. B. Zhan, H. R. Jafry, P. M. Ajayan, Nano Lett. 2012, 12, 1198–1202. [19] G. Y. Zhao, C. L. Xu, D. J. Guo, H. Li, H. L. Li, J. Power Sources. 2006, 162, 492–496. [20] K. S. Napolskii, P. J. Barczuk, S. Y. Vassiliev, A. G. Veresov, G. A. Tsirlina, P. J. Kulesza, Electrochim. Acta. 2007, 52, 7910–7919. [21] D. J. Comstock, S. T. Christensen, J. W. Elam, M. J. Pellin, M. C. Hersam, Adv. Funct. Mater. 2010, 20, 3099–3105. [22] D. F. Gu, H. Baumgart, K. Tapily, P. Shrestha, G. Namkoong, X. Y. Ao, F. Muller, Nano Res. 2011, 4, 164–170. [23] S. M. George, Chem. Rev. 2010, 110, 111–131. [24] H. B. R. Lee, S. F. Bent, Chem. Mater. 2012, 24, 279–286. [25] H.-B.-R. Lee, S. H. Baeck, T. F. Jaramillo, S. F. Bent, Nano Lett. 2013, 13, 457–463. [26] J. L. Lu, J. W. Elam, P. C. Stair, Acc. Chem. Res. 2013, 46, 1806–1815. [27] A. J. M. Mackus, M. A. Verheijen, N. Leick, A. A. Bol, W. M. M. Kessels, Chem. Mater. 2013, 25, 1905–1911. [28] J. W. Elam, D. Routkevitch, P. P. Mardilovich, S. M. George, Chem. Mater. 2003, 15, 3507–3517. [29] R. G. Gordon, D. Hausmann, E. Kim, J. Shepard, Chem. Vapor Depos. 2003, 9, 73–78. [30] W. Lee, R. Ji, C. A. Ross, U. Gösele, K. Nielsch, Small 2006, 2, 978–982. [31] C. L. Wang, L. Y. Wen, T. Kups, Y. Mi, R. Vellacheri, Y. G. Fang, P. Schaaf, H. P. Zhao, Y. Lei, J. Mater. Chem. C 2013, 1, 8003– 8006. [32] W. B. Yan, J. Y. Kim, W. D. Xing, K. C. Donavan, T. Ayvazian, R. M. Penner, Chem. Mater. 2012, 24, 2382–2390. [33] Z. N. Yu, B. Duong, D. Abbitt, J. Thomas, Adv. Mater. 2013, 25, 3302–3306.
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Received: February 17, 2014 Revised: March 15, 2014 Published online: April 3, 2014
small 2014, 10, No. 15, 3162–3168
