- Email: [email protected]
Highly improved electrochemical hydrogen storage performances of the Nd–Cu–added Mg2Ni-type alloys by melt spinning
Journal of Alloys and Compounds 584 (2014) 81–86
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: www.elsevier.com/locate/jalcom
Highly improved electrochemical hydrogen storage performances of the Nd–Cu–added Mg2Ni-type alloys by melt spinning Yanghuan Zhang a,b,⇑, Can Li a,b, Ying Cai a, Feng Hu a, Zhuocheng Liu a, Shihai Guo b a b
Key Laboratory of Integrated Exploitation of Baiyun Obo Multi-Metal Resources, Inner Mongolia University of Science and Technology, Baotou 014010, China Department of Functional Material Research, Central Iron and Steel Research Institute, Beijing 100081, China
a r t i c l e
i n f o
Article history: Received 23 July 2013 Received in revised form 1 September 2013 Accepted 2 September 2013 Available online 12 September 2013 Keywords: Mg2Ni-type alloy Nd addition Melt spinning Nanocrystalline and amorphous Electrochemical hydrogen storage performances
a b s t r a c t In order to improve the electrochemical hydrogen storage performances of the Mg2Ni-type alloy, Nd and Cu were added jointly. The (Mg24Ni10Cu2)100 xNdx (x = 0, 5, 10, 15, 20) alloys with nanocrystalline and amorphous structure were fabricated by melt-spinning technique. The effects of spinning rate and Nd content on the structures and electrochemical hydrogen storage performances of the alloys were investigated. The structure characterizations of XRD, TEM and SEM linked with EDS reveal that the as-cast (x = 15) alloy holds a multiphase structure, containing Mg2Ni major phase as well as some secondary phases Mg6Ni, Nd5Mg41 and NdNi. The as-spun Nd-free alloy displays an entire nanocrystalline structure, whereas the as-spun Nd-added alloys hold a nanocrystalline and amorphous structure, meanwhile, the degree of amorphization visibly increases with both spinning rate and Nd content rising, suggesting that the addition of Nd facilitates the glass forming of the Mg2Ni-type alloys. Furthermore, the addition of Nd and melt spinning dramatically improve the electrochemical hydrogen storage performances of the alloys. The discharge capacity and cycle stability of the alloys conspicuously augment with spinning rate growing. When the spinning rate rises from 0 (The as-cast was defined as the spinning rate of 0 m/s) to 40 m/s, the discharge capacity of the (x = 10) alloy is enhanced from 86.4 to 452.8 mA h/g, and capacity maintain rate (S20) (the capacity maintain rate at 20th cycle) from 53.2% to 89.7%, respectively. Ó 2013 Elsevier B.V. All rights reserved.
1. Introduction Among all the magnesium-based alloys, Mg2Ni-type alloys are looked upon as one of the most promising hydrogen storage materials applied in hydrogen fuel cell vehicle or negative electrode in Ni–MH battery [1] due to their major advantages, such as the higher gaseous hydrogen absorption capacity (3.6 wt%) for Mg2NiH4 and the larger gravimetric capacity (about 1000 mAh/g) [2,3]. However, some inherent shortcomings, such as relatively high H-desorption temperature, sluggish hydriding/dehydriding kinetics and extremely poor electrochemical cycle stability, make the attempt of the practical application of the alloys to be frustrated seriously. In spite of facing huge challenges, the researchers in this field still persist in the firm confidence to improve the properties of the alloys and achieve a series of important progress. It was documented that the hydriding and dehydriding kinetics of the Mg and Mg based alloys are very sensitive to their structures [4]. Especially, the hydrogen storage properties are strongly
⇑ Corresponding author. Address: Department of Functional Material Research, Central Iron and Steel Research Institute, No. 76 Xueyuannan Road, Haidian District, 100081 Beijing, China. Tel.: +86 10 62183115; fax: +86 10 62187102. E-mail address: [email protected] (Y. Zhang). 0925-8388/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jallcom.2013.09.019
affected by nanometer scale structures because of the thermodynamic and kinetic aspects. Teresiak et al. [5] also reported that amorphous and/or nanocrystalline Mg–Ni-based alloys can electrochemically absorb and desorb large amounts of hydrogen even at room temperature. Moreover, it was ascertained that the capacity degradation is associated with the irreversible oxidation of the alloy by the electrolyte (KOH) leading to the formation of Mg(OH)2 layer on the surface of the alloy particles [6,7]. Hence, a serious challenge in the improvement of the cycle stability is how to improve the corrosion resistance of the alloy. It was confirmed that the addition of RE (RE = La, Ce, Y) to Mg–Ni or Mg–Cu alloys facilitated the glass-formation, increasing the corrosion resistance and improving the cycle stability [8]. The high energy ball milling (HEBM) and melt-spinning technique are considered to be very effective methods to obtain amorphous and/or nanocrystalline structure. However, the cycle stabilities of the milled Mg and Mg-based alloys are very poor owing to the vanishment of the metastable structures generated by ball milling during the multiple hydrogen absorbing and desorbing cycles [9]. On the contrary, the microstructure created by melt spinning displays much higher stability during the hydrogen absorbing and desorbing cycle as compared with the microstructure generated by HEBM [10]. Meanwhile, the nanocrystalline and amorphous Mg-based alloys prepared by melt-spinning
82
Y. Zhang et al. / Journal of Alloys and Compounds 584 (2014) 81–86
exhibit excellent hydriding characteristics, similar to the alloys produced by the HEBM [11]. Our previous investigations have found that the substitution of M (M = Co, Cu, Mn) for Ni and La for Mg improved the electrochemical and gaseous hydrogen storage performances of the Mg2Ni-type alloys dramatically [12,13]. Therefore, it is expected that the joint addition of Cu and Nd combining with a proper melt spinning technique may ameliorate the electrochemical hydrogen storage performances of the Mg2Ni-type alloys more pronouncedly. To validate this, a systematical investigation about the effects of melt spinning and Nd content on the structures and electrochemical hydrogen storage performances of the (Mg24Ni10Cu2)100 xNdx (x = 0–20) electrode alloys has been performed, and some experimental results were provided. 2. Experimental The experimental alloys with a composition of (Mg24Ni10Cu2)100 xNdx (x = 0, 5, 10, 15, 20) were prepared by using a vacuum induction furnace in a helium atmosphere at a pressure of 0.04 MPa to prevent Mg from volatilizing. For convenience, the alloys were denoted with Nd content as Nd0, Nd5, Nd10, Nd15 and Nd20, respectively. A part of the as-cast alloys were re-melted and spun by melt spinning with a rotating copper roller cooled by water. The spinning rates used in the experiment were 10, 20, 30 and 40 m/s, respectively, which were approximately expressed by the linear velocity of the copper roller. The phase structures of the as-cast and spun alloys were determined by X-ray diffraction (XRD) (D/max/2400). The diffraction, with the experimental parameters of 160 mA, 40 kV and 10°/min respectively, was performed with CuKa1 radiation filtered by graphite. A Philips SEM (QUANTA 400) linked with an energy dispersive spectrometer (EDS) was used for morphological characterization and chemical composition analysis of the as-cast alloys. The thin film samples of the as-spun alloys prepared by using ion etching technology were observed by high resolution transmission electron microscope (HRTEM) (JEM-2100F, operated at 200 kV) and their crystalline states were ascertained by electron diffraction (ED). The alloys were pulverized by mechanical milling into fine powder with a diameter of 35 lm, and then mixing the alloy powder with carbonyl nickel powder in a weight ratio of 1:4. The mixture was cold pressed under a pressure of 35 MPa into round electrode pellets with 15 mm in diameter whose total weight is 1 g. The electrochemical measurements were performed at 30 °C by a tri-electrode open cell consisting of a working electrode (the metal hydride electrode), a sintered Ni(OH)2/ NiOOH counter electrode as well as a Hg/HgO reference electrode, which were immersed in 6 mol/L KOH electrolyte. The voltage between the negative electrode and the reference one was defined as the discharge voltage. In every cycle, the alloy electrode was first charged with a constant current density of 40 mA/g, after resting for 15 min, it was discharged at the same current density to cut-off voltage of 0.500 V.
Fig. 1. XRD profiles of the as-cast and spun (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys: (a) Nd15 alloy, (b) As-spun (10 m/s).
amorphous structure, suggesting that the addition of Nd facilitates the glass forming in the Mg2Ni-type alloys. This conclusion is also evidenced by TEM detections, as demonstrated in Fig. 2. It is found that the as-spun Nd0 alloy displays an entire nanocrystalline structure with an average crystal size of about 20 nm and its electron diffraction (ED) pattern appears sharp multi-haloes, corresponding to a crystalline structure. Nevertheless, the as-spun Nd5 and Nd15 alloys differing from Nd0 alloy exhibit a clear feature of the nanocrystalline embedded in the amorphous matrix, and their electron diffraction patterns consisted of broad and dull halo, indicating the presence of an amorphous structure, which conforms to the XRD observations as depicted in Fig. 1. Meanwhile, the phase components of the as-cast alloys are analyzed by SEM, as illustrated in Fig. 3. Apparently, the addition of Nd results in the morphologies of the as-cast alloys changing obviously. Specifically, the typical dendritic structure of the as-cast Nd0 alloy completely disappears and the some secondary phases emerge, namely Nd5Mg41 and NdNi phases determined by EDS patterns. It is noted that the amount of the Mg6Ni phase is small and it attaches itself to the Mg2Ni phase in the process of growing, so, it is difficult to observe the morphology of the Mg6Ni phase. 3.2. Electrochemical hydrogen storage performances
3. Results and discussion 3.1. Microstructure characteristics The phase components and structure characteristics of as-cast and spun (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys are subjected to XRD analysis, just as presented in Fig. 1. It is found that the as-cast Nd15 alloy is of multiphase structure, including major phase Mg2Ni along with secondary phases NdNi and Nd5Mg41 which originate from Nd adding. The broad and diffuse diffraction peak of the as-spun (10 m/s) Nd15 alloy exhibits the characteristic of amorphous and nanocrystalline structures. With the spinning rate increasing, the diffraction peaks of the Nd15 alloy evidently broadened, meaning that the amount of amorphous phase grows. Also, the sharp and narrow diffraction peaks of the as-spun (10 m/s) Nd0 alloy show an entire nanocrystalline structure with a grain size of about 20 nm, which is calculated by Scherrer’s equation based on the FWHM values of the major diffraction peak (2 0 3) in Fig. 1 (b). Interestingly, it is found that, even if spun at the same spinning rate, the Nd-added alloys display completely different diffraction peaks from that of the Nd0 alloy. The very broad and flat diffraction peaks indicate that the as-spun Nd-added alloys hold an
3.2.1. Activation capability The activation process of an alloy electrode is the course that the discharge capacity of the alloy gradually reaches to the maximum value through charge/discharge cycles at a constant current density. The activation capability was characterized by the cycle number (n) required for attaining the greatest discharge capacity at a constant current density. Evidently, the smaller the n is, the better the activation capability will be. The activation capability of the alloy is inspected by constituting the relationship between the discharge capacity and the cycle number, as illustrated in Fig. 4. It is found that the as-cast and spun alloys can reach to the maximum discharge capacity at most three charging/discharging cycles, displaying a superior activation capability, which is considered to be ascribed to the multiphase structure in the as-cast alloy or nanocrystalline structure in the as-spun alloy. It is known to all that the activation capability of an alloy electrode is closely related with the change of the internal energy of the hydride system before and after hydrogen absorption. The larger the additive internal energy is, the poorer activation performance of the alloy will be [4]. In general, the internal energy involves both the surface energy and the strain energy. The former originates from oxidation
Y. Zhang et al. / Journal of Alloys and Compounds 584 (2014) 81–86
83
Fig. 2. HRTEM micrographs and ED patterns of the as-spun (10 m/s) Nd0, Nd5 and Nd15 alloys: (a) Nd0, (b) Nd5, (c) Nd15.
Fig. 3. SEM images together with typical EDS spectra of the as-cast Nd0 and Nd15 alloys: (a) Nd0, (b) Nd15.
film forming on the surface of the electrode alloy, and the latter is generated by hydrogen atom entering the tetrahedral or octahedral lattice interstitial positions. For the as-cast alloys, the superior activation performance is deemed to be ascribed to multiphase structure due to the phase boundaries providing good diffusing tunnels for hydrogen atoms, further improving the activation capability [14]. And in respect of the as-spun alloys, it is ascribed to the nanocrystalline/amorphous structure originated from melt spinning because the numerous grain boundaries probably act as buffer areas which can greatly release the lattice distortion and strain energy accumulated during hydrogen absorption process. 3.2.2. Discharge potential and discharge capacity Fig. 5 provides the variations of the discharge potentials of the as-cast and spun (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys with the discharge capacity. It is noted that the discharge potentials of all the alloy electrodes decline in varying degree during the discharge process, meaning that the output power of the battery is not constant. Evidently, the discharge potential characteristic, which is directly related to the stability of the output power, is
an extremely important performance of the alloy electrode and is characterized by the potential plateau of the discharge curve of the alloy. The longer and the more horizontal the discharge potential plateau is, the better the discharge potential characteristics of the alloy will be. Clearly, the discharge curves have different width of the discharge potential plateau based on the oxidation of desorbed hydrogen from the hydride. It can be seen from Fig. 5 that the melt spinning and the addition of Nd considerably modify discharge potential characteristics of the alloys, making the discharge potential rise and the discharge plateau lengthen. Meanwhile, it is found that, besides improving potential characteristic, the melt spinning and the addition of Nd also enable the discharge capacity of alloys to increase dramatically. In order to examine the effect of the Nd content on the discharge capacity of the alloys, the relationship between the discharge capacity of the alloys and the amount of Nd is established, just as demonstrated in Fig. 6. Apparently, the discharge capacities of the alloys augment with the Nd content rising up to x = 10, but the excessive amount of Nd adding (x > 10) results in an undesired reduction in the discharge capacity. But even so,
84
Y. Zhang et al. / Journal of Alloys and Compounds 584 (2014) 81–86
Fig. 6. Evolution of the discharge capacity of the as-cast (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys with Nd content varying.
Fig. 4. Activation capability of the as-cast and spun the (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys: (a) Nd15 alloy, (b) As-spun (10 m/s).
and
spun
the alloy. When the Nd content is x = 10, the as-cast and spun alloys yield the maximum discharge capacities, namely 86.4, 200.5, 266.3, 402.5 and 452.8 mA h/g corresponding to the spinning rate of 0, 10, 20, 30 and 40 m/s, respectively. Furthermore, it must be noticed that, for a fixed Nd content, the discharge capacity of the alloy markedly increases with spinning rate rising. Particularly, enhancing the spinning rate from 0 to 40 m/s enables the discharge capacity to grow from 42.5 to 100.6 m A h/g for the Nd0 alloy and from 86.4 to 452.8 mA h/g for the Nd10 alloy, respectively. Obviously, the positive contribution originated from the melt spinning to the discharge capacity of the Nd-added alloys is much more pronounced than the Nd-free one. The benefaction of Nd adding on the discharge capacity of the alloy is considered most likely to be associated with the following factors. On the one hand, the glass forming facilitated by Nd adding is believed to be a crucial factor because an adequate proportion of amorphous and nanocrystalline in Mg–Ni-based alloy possesses superior discharge property [15]. On the other hand, the secondary phases Mg6Ni and NdNi probably
Fig. 5. Discharge potential curves of the as-cast and spun (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys: (a) Nd15 alloy, (b) As-spun (10 m/s).
the Nd-added alloys still show much higher discharge capacity as compared to the Nd-free alloy, suggesting that the addition of Nd totally engenders beneficial impact on the discharge capacity of
Fig. 7. Evolution of the capacity retaining rates (Sn) of the as-cast and spun alloys with cycle number: (a) Nd15 alloy, (b) As-spun (10 m/s).
Y. Zhang et al. / Journal of Alloys and Compounds 584 (2014) 81–86
play a catalytic role in activating the Mg2Ni phase to reversibly absorb/desorb hydrogen in the alkaline electrolyte. As a matter of fact, the excessive amount of Nd (x >10) gives rise to an undesired reduction in the discharge capacity, for which the sharp decrease of the Mg2Ni phase is mainly responsible. In terms of the discharge capacity improved by melt spinning, it is definitely believed to be related to nanocrystalline and amorphous structure owing to the grain boundaries possessing the distribution of the maximum hydrogen concentrations [16]. Also, the nanocrystalline and amorphous structure generated by melt spinning evidently promotes the hydriding and dehydriding abilities of Mg2Ni-type alloys, as reported by Kumar et al. [2] and Teresiak et al. [5]. 3.2.3. Cycle stability Cycle stability, one of the major performances which are used to evaluate whether a kind of an alloy can be applied as a negative electrode material or not, is signified by the capacity retaining rate (Sn), defined as Sn = Cn/Cmax 100%, where Cmax is the maximum discharge capacity and Cn is the discharge capacity at the nth charge–discharge cycle, respectively. Clearly, it means that the larger the capacity retaining rate (Sn) is, the better the cycle stability of the alloy will be. The variations of the Sn values of the as-cast and spun (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys depending on cycle number are presented in Fig. 7, from which the degradation process of the discharge capacity of the alloys can be seen clearly. The slopes of the curves qualitatively reflect the degradation rate of the discharge capacity during the charging-discharging cycles, namely the smaller the slope of the curve is, the better the cycle life of the alloy will be. It is found from Fig. 7(a) that the curve’s slope of the Nd15 alloy always declines with the spinning rate growing, indicating that the melt spinning makes the positive contribution to the cycle stability of the Nd15 alloy. In order to clearly show the effect of the spinning rate on the cycle stability of the alloy, here we build the relationship between the S20 value of the Nd15 alloy and the spinning rate, as inserted in Fig. 7(a), from which it is derived that the increasing of the spinning rate from
85
0 to 40 m/s renders the S20 value of the Nd15 alloy rising from 59.2% to 85.9%. Also, it is also noted that the slopes of the curves in Fig. 7(b) are changeable with Nd content varying, but no matter how much Nd content is, the curves’ slopes of the Nd-added alloys are all less than that of the Nd-free one, indicating that the addition of Nd plays a beneficial role on the cycle life of the alloy electrode. Likewise, the relationship between the S20 values of the as-spun (10 m/s) alloys and the Nd content is also established, as inserted in Fig. 7(b). It is evident that the S20 value of the as-spun (10 m/ s) alloy first mounts up and then goes down with the Nd content growing. Here, we provide some elucidations as the reasons why the melt spinning and Nd adding ameliorate the cycle stability of the alloy. It has come to light that the sharp degradation of the discharge capacity of Mg2Ni-type alloy during the charge–discharge cycling is caused by forming and thickening of Mg(OH)2 surface layer, which hinders the hydrogen atoms from diffusing in or out, in alkaline solution [17]. Moreover, the hydrogen storage material suffers an inevitable volume change during the charge–discharge process which aggravates the alloy’s cracking and pulverizing and then makes the surface of the material apt to be oxidized, which is also evidenced by our experimental results, as demonstrated in Fig. 8. The alloy particles (denoted as A) with a size of about 30 lm and ultrafine Ni powders appearing an aggregation sate (denoted as B) can be seen clearly. Furthermore, it is found that there are numerous cracks on the surfaces of the alloy particles after electrochemical cycle, but the sizes of the alloy particles have no obvious change, suggesting that the pulverization of alloy particles scarcely take place in the process of the electrochemical cycle. Noticeably, the alloy particles after cycling are covered by a rough and gossypine layer, which is inspected to be Mg(OH)2 by XRD, as illustrated in Fig. 8(d). Also, we note that the thickness of the covering layer on the surface of the Nd20 alloy particle is much less than that of the Nd0 alloy, meaning that the addition of Nd can notably increases the anti-corrosion and antioxidation ability of the Mg2Ni alloy. A similar result was reported by Teresiak et al. [5]. Besides, it can be seen that there are much
Fig. 8. SEM morphologies together with typical XRD patterns of the as-spun (10 m/s) Nd0 and Nd20 alloys before and after electrochemical cycle: (a) and (b) Nd0 alloy before and after cycling, (c) Nd20 alloy after cycling, (d) XRD of the Nd0 alloy after cycling.
86
Y. Zhang et al. / Journal of Alloys and Compounds 584 (2014) 81–86
fewer cracks on the surface of the Nd20 alloy particle after cycling than that of the Nd0 one, reflecting that the addition of Nd can markedly enhance the anti-pulverization ability of the alloy. The above-mentioned functions of Nd adding are most probably ascribed to two aspects. Firstly, adding the rare elements (La, Nd, Sm, Y) engenders the cell volume of the Mg2Ni alloy enlarging clearly [13], decreasing the ratios of expansion/contraction in the process of hydrogen absorption/desorption, so as to enhance the anti-pulverization capability. Secondly, the glass formation facilitated by Nd adding is extremely important because an amorphous phase improves not only anti-pulverization ability but also anticorrosion and anti-oxidation abilities of the alloy electrode in a corrosive electrolyte [12]. Based on above-mentioned causes, it is very easy to understand why the Nd-added alloys exhibit much higher cycle stability than the Nd-free alloy. 4. Conclusions The effects of melt spinning and Nd adding on the electrochemical hydrogen storage performances of the as-cast and spun (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys were investigated, and the major conclusions are summarized as follows: 1. The investigations on the structures of the alloys reveal that the addition of Nd facilitates the glass forming in the Mg2Ni-type alloys. As a result, the as-spun Nd-free alloy holds an entire nanocrystalline structure but an obvious amorphous structure for the as-spun Nd-added alloys. 2. Both the melt spinning and Nd adding make the positive contribution to the electrochemical performances of the alloys, including discharge capacity, potential characteristic and cycle stability. 3. Furthermore, the addition of Nd significantly enhances the anticorrosion and anti-pulverization ability of the alloy, which is
viewed as the root reason of Nd adding improving the cycle stability of the Mg2Ni-type alloys.
Acknowledgements This work is financially supported by National Natural Science Foundations of China (51161015) and Natural Science Foundation of Inner Mongolia, China (2011ZD10). References [1] S. Ruggeri, L. Roué, J. Huot, R. Schulz, L. Aymard, J.-M. Tarascon, J. Power Sources 112 (2002) 547–556. [2] L.H. Kumar, B. Viswanathan, S.S. Murthy, J. Alloys Comp. 461 (2008) 72–76. [3] L. Schlapbach, A. Züttel, Nature 414 (2001) 353–358. [4] M.S. Wu, H.R. Wu, Y.Y. Wang, C.C. Wan, J. Alloys Comp. 302 (2000) 248– 257. [5] A. Teresiak, A. Gebert, M. Savyak, M. Uhlemann, C. Mickel, N. Mattern, J. Alloys Comp. 398 (2005) 156–164. [6] T. Abe, T. Tachikawa, Y. Hatano, K. Watanabe, J. Alloys Comp. 792 (2002) 330– 332. [7] C. Lenain, L. Aymard, J.M. Tarascon, J. Solid State Electrochem. 2 (1998) 285– 290. [8] Z.M. Wang, H.Y. Zhou, Z.F. Gu, G. Cheng, A.B. Yu, J. Alloys Comp. 381 (2004) 234–239. [9] M.Y. Song, S.N. Kwon, J.S. Bae, S.H. Hong, Int. J. Hydrogen Energy 33 (2008) 1711–1718. [10] G.Y. Liang, D.C. Wu, L. Li, L.J. Huang, J. Power Sources 186 (2009) 528–531. [11] S. Todorova, T. Spassov, J. Alloys Comp. 469 (2009) 193–196. [12] Y.H. Zhang, C. Zhao, T. Yang, H.W. Shang, C. Xu, D.L. Zhao, J. Alloys Comp. 555 (2013) 131–137. [13] H.P. Ren, Y.H. Zhang, B.W. Li, D.L. Zhao, S.H. Guo, X.L. Wang, Int. J. Hydrogen Energ. 34 (2009) 1429–1436. [14] D.H. Xie, P. Li, C.X. Zeng, J.W. Sun, X.H. Qu, J. Alloys Comp. 478 (2009) 96– 102. [15] T. Spassov, U. Köster, J. Alloys Comp. 287 (1999) 243–250. [16] S. Orimo, H. Fujii, Appl. Phys. A 72 (2001) 167–186. [17] M. Dornheim, S. Doppiu, G. Barkhordarian, U. Boesenberg, T. Klassen, O. Gutfleisch, R. Bormann, Scripta Mater. 56 (2007) 841–846.
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: www.elsevier.com/locate/jalcom
Highly improved electrochemical hydrogen storage performances of the Nd–Cu–added Mg2Ni-type alloys by melt spinning Yanghuan Zhang a,b,⇑, Can Li a,b, Ying Cai a, Feng Hu a, Zhuocheng Liu a, Shihai Guo b a b
Key Laboratory of Integrated Exploitation of Baiyun Obo Multi-Metal Resources, Inner Mongolia University of Science and Technology, Baotou 014010, China Department of Functional Material Research, Central Iron and Steel Research Institute, Beijing 100081, China
a r t i c l e
i n f o
Article history: Received 23 July 2013 Received in revised form 1 September 2013 Accepted 2 September 2013 Available online 12 September 2013 Keywords: Mg2Ni-type alloy Nd addition Melt spinning Nanocrystalline and amorphous Electrochemical hydrogen storage performances
a b s t r a c t In order to improve the electrochemical hydrogen storage performances of the Mg2Ni-type alloy, Nd and Cu were added jointly. The (Mg24Ni10Cu2)100 xNdx (x = 0, 5, 10, 15, 20) alloys with nanocrystalline and amorphous structure were fabricated by melt-spinning technique. The effects of spinning rate and Nd content on the structures and electrochemical hydrogen storage performances of the alloys were investigated. The structure characterizations of XRD, TEM and SEM linked with EDS reveal that the as-cast (x = 15) alloy holds a multiphase structure, containing Mg2Ni major phase as well as some secondary phases Mg6Ni, Nd5Mg41 and NdNi. The as-spun Nd-free alloy displays an entire nanocrystalline structure, whereas the as-spun Nd-added alloys hold a nanocrystalline and amorphous structure, meanwhile, the degree of amorphization visibly increases with both spinning rate and Nd content rising, suggesting that the addition of Nd facilitates the glass forming of the Mg2Ni-type alloys. Furthermore, the addition of Nd and melt spinning dramatically improve the electrochemical hydrogen storage performances of the alloys. The discharge capacity and cycle stability of the alloys conspicuously augment with spinning rate growing. When the spinning rate rises from 0 (The as-cast was defined as the spinning rate of 0 m/s) to 40 m/s, the discharge capacity of the (x = 10) alloy is enhanced from 86.4 to 452.8 mA h/g, and capacity maintain rate (S20) (the capacity maintain rate at 20th cycle) from 53.2% to 89.7%, respectively. Ó 2013 Elsevier B.V. All rights reserved.
1. Introduction Among all the magnesium-based alloys, Mg2Ni-type alloys are looked upon as one of the most promising hydrogen storage materials applied in hydrogen fuel cell vehicle or negative electrode in Ni–MH battery [1] due to their major advantages, such as the higher gaseous hydrogen absorption capacity (3.6 wt%) for Mg2NiH4 and the larger gravimetric capacity (about 1000 mAh/g) [2,3]. However, some inherent shortcomings, such as relatively high H-desorption temperature, sluggish hydriding/dehydriding kinetics and extremely poor electrochemical cycle stability, make the attempt of the practical application of the alloys to be frustrated seriously. In spite of facing huge challenges, the researchers in this field still persist in the firm confidence to improve the properties of the alloys and achieve a series of important progress. It was documented that the hydriding and dehydriding kinetics of the Mg and Mg based alloys are very sensitive to their structures [4]. Especially, the hydrogen storage properties are strongly
⇑ Corresponding author. Address: Department of Functional Material Research, Central Iron and Steel Research Institute, No. 76 Xueyuannan Road, Haidian District, 100081 Beijing, China. Tel.: +86 10 62183115; fax: +86 10 62187102. E-mail address: [email protected] (Y. Zhang). 0925-8388/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jallcom.2013.09.019
affected by nanometer scale structures because of the thermodynamic and kinetic aspects. Teresiak et al. [5] also reported that amorphous and/or nanocrystalline Mg–Ni-based alloys can electrochemically absorb and desorb large amounts of hydrogen even at room temperature. Moreover, it was ascertained that the capacity degradation is associated with the irreversible oxidation of the alloy by the electrolyte (KOH) leading to the formation of Mg(OH)2 layer on the surface of the alloy particles [6,7]. Hence, a serious challenge in the improvement of the cycle stability is how to improve the corrosion resistance of the alloy. It was confirmed that the addition of RE (RE = La, Ce, Y) to Mg–Ni or Mg–Cu alloys facilitated the glass-formation, increasing the corrosion resistance and improving the cycle stability [8]. The high energy ball milling (HEBM) and melt-spinning technique are considered to be very effective methods to obtain amorphous and/or nanocrystalline structure. However, the cycle stabilities of the milled Mg and Mg-based alloys are very poor owing to the vanishment of the metastable structures generated by ball milling during the multiple hydrogen absorbing and desorbing cycles [9]. On the contrary, the microstructure created by melt spinning displays much higher stability during the hydrogen absorbing and desorbing cycle as compared with the microstructure generated by HEBM [10]. Meanwhile, the nanocrystalline and amorphous Mg-based alloys prepared by melt-spinning
82
Y. Zhang et al. / Journal of Alloys and Compounds 584 (2014) 81–86
exhibit excellent hydriding characteristics, similar to the alloys produced by the HEBM [11]. Our previous investigations have found that the substitution of M (M = Co, Cu, Mn) for Ni and La for Mg improved the electrochemical and gaseous hydrogen storage performances of the Mg2Ni-type alloys dramatically [12,13]. Therefore, it is expected that the joint addition of Cu and Nd combining with a proper melt spinning technique may ameliorate the electrochemical hydrogen storage performances of the Mg2Ni-type alloys more pronouncedly. To validate this, a systematical investigation about the effects of melt spinning and Nd content on the structures and electrochemical hydrogen storage performances of the (Mg24Ni10Cu2)100 xNdx (x = 0–20) electrode alloys has been performed, and some experimental results were provided. 2. Experimental The experimental alloys with a composition of (Mg24Ni10Cu2)100 xNdx (x = 0, 5, 10, 15, 20) were prepared by using a vacuum induction furnace in a helium atmosphere at a pressure of 0.04 MPa to prevent Mg from volatilizing. For convenience, the alloys were denoted with Nd content as Nd0, Nd5, Nd10, Nd15 and Nd20, respectively. A part of the as-cast alloys were re-melted and spun by melt spinning with a rotating copper roller cooled by water. The spinning rates used in the experiment were 10, 20, 30 and 40 m/s, respectively, which were approximately expressed by the linear velocity of the copper roller. The phase structures of the as-cast and spun alloys were determined by X-ray diffraction (XRD) (D/max/2400). The diffraction, with the experimental parameters of 160 mA, 40 kV and 10°/min respectively, was performed with CuKa1 radiation filtered by graphite. A Philips SEM (QUANTA 400) linked with an energy dispersive spectrometer (EDS) was used for morphological characterization and chemical composition analysis of the as-cast alloys. The thin film samples of the as-spun alloys prepared by using ion etching technology were observed by high resolution transmission electron microscope (HRTEM) (JEM-2100F, operated at 200 kV) and their crystalline states were ascertained by electron diffraction (ED). The alloys were pulverized by mechanical milling into fine powder with a diameter of 35 lm, and then mixing the alloy powder with carbonyl nickel powder in a weight ratio of 1:4. The mixture was cold pressed under a pressure of 35 MPa into round electrode pellets with 15 mm in diameter whose total weight is 1 g. The electrochemical measurements were performed at 30 °C by a tri-electrode open cell consisting of a working electrode (the metal hydride electrode), a sintered Ni(OH)2/ NiOOH counter electrode as well as a Hg/HgO reference electrode, which were immersed in 6 mol/L KOH electrolyte. The voltage between the negative electrode and the reference one was defined as the discharge voltage. In every cycle, the alloy electrode was first charged with a constant current density of 40 mA/g, after resting for 15 min, it was discharged at the same current density to cut-off voltage of 0.500 V.
Fig. 1. XRD profiles of the as-cast and spun (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys: (a) Nd15 alloy, (b) As-spun (10 m/s).
amorphous structure, suggesting that the addition of Nd facilitates the glass forming in the Mg2Ni-type alloys. This conclusion is also evidenced by TEM detections, as demonstrated in Fig. 2. It is found that the as-spun Nd0 alloy displays an entire nanocrystalline structure with an average crystal size of about 20 nm and its electron diffraction (ED) pattern appears sharp multi-haloes, corresponding to a crystalline structure. Nevertheless, the as-spun Nd5 and Nd15 alloys differing from Nd0 alloy exhibit a clear feature of the nanocrystalline embedded in the amorphous matrix, and their electron diffraction patterns consisted of broad and dull halo, indicating the presence of an amorphous structure, which conforms to the XRD observations as depicted in Fig. 1. Meanwhile, the phase components of the as-cast alloys are analyzed by SEM, as illustrated in Fig. 3. Apparently, the addition of Nd results in the morphologies of the as-cast alloys changing obviously. Specifically, the typical dendritic structure of the as-cast Nd0 alloy completely disappears and the some secondary phases emerge, namely Nd5Mg41 and NdNi phases determined by EDS patterns. It is noted that the amount of the Mg6Ni phase is small and it attaches itself to the Mg2Ni phase in the process of growing, so, it is difficult to observe the morphology of the Mg6Ni phase. 3.2. Electrochemical hydrogen storage performances
3. Results and discussion 3.1. Microstructure characteristics The phase components and structure characteristics of as-cast and spun (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys are subjected to XRD analysis, just as presented in Fig. 1. It is found that the as-cast Nd15 alloy is of multiphase structure, including major phase Mg2Ni along with secondary phases NdNi and Nd5Mg41 which originate from Nd adding. The broad and diffuse diffraction peak of the as-spun (10 m/s) Nd15 alloy exhibits the characteristic of amorphous and nanocrystalline structures. With the spinning rate increasing, the diffraction peaks of the Nd15 alloy evidently broadened, meaning that the amount of amorphous phase grows. Also, the sharp and narrow diffraction peaks of the as-spun (10 m/s) Nd0 alloy show an entire nanocrystalline structure with a grain size of about 20 nm, which is calculated by Scherrer’s equation based on the FWHM values of the major diffraction peak (2 0 3) in Fig. 1 (b). Interestingly, it is found that, even if spun at the same spinning rate, the Nd-added alloys display completely different diffraction peaks from that of the Nd0 alloy. The very broad and flat diffraction peaks indicate that the as-spun Nd-added alloys hold an
3.2.1. Activation capability The activation process of an alloy electrode is the course that the discharge capacity of the alloy gradually reaches to the maximum value through charge/discharge cycles at a constant current density. The activation capability was characterized by the cycle number (n) required for attaining the greatest discharge capacity at a constant current density. Evidently, the smaller the n is, the better the activation capability will be. The activation capability of the alloy is inspected by constituting the relationship between the discharge capacity and the cycle number, as illustrated in Fig. 4. It is found that the as-cast and spun alloys can reach to the maximum discharge capacity at most three charging/discharging cycles, displaying a superior activation capability, which is considered to be ascribed to the multiphase structure in the as-cast alloy or nanocrystalline structure in the as-spun alloy. It is known to all that the activation capability of an alloy electrode is closely related with the change of the internal energy of the hydride system before and after hydrogen absorption. The larger the additive internal energy is, the poorer activation performance of the alloy will be [4]. In general, the internal energy involves both the surface energy and the strain energy. The former originates from oxidation
Y. Zhang et al. / Journal of Alloys and Compounds 584 (2014) 81–86
83
Fig. 2. HRTEM micrographs and ED patterns of the as-spun (10 m/s) Nd0, Nd5 and Nd15 alloys: (a) Nd0, (b) Nd5, (c) Nd15.
Fig. 3. SEM images together with typical EDS spectra of the as-cast Nd0 and Nd15 alloys: (a) Nd0, (b) Nd15.
film forming on the surface of the electrode alloy, and the latter is generated by hydrogen atom entering the tetrahedral or octahedral lattice interstitial positions. For the as-cast alloys, the superior activation performance is deemed to be ascribed to multiphase structure due to the phase boundaries providing good diffusing tunnels for hydrogen atoms, further improving the activation capability [14]. And in respect of the as-spun alloys, it is ascribed to the nanocrystalline/amorphous structure originated from melt spinning because the numerous grain boundaries probably act as buffer areas which can greatly release the lattice distortion and strain energy accumulated during hydrogen absorption process. 3.2.2. Discharge potential and discharge capacity Fig. 5 provides the variations of the discharge potentials of the as-cast and spun (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys with the discharge capacity. It is noted that the discharge potentials of all the alloy electrodes decline in varying degree during the discharge process, meaning that the output power of the battery is not constant. Evidently, the discharge potential characteristic, which is directly related to the stability of the output power, is
an extremely important performance of the alloy electrode and is characterized by the potential plateau of the discharge curve of the alloy. The longer and the more horizontal the discharge potential plateau is, the better the discharge potential characteristics of the alloy will be. Clearly, the discharge curves have different width of the discharge potential plateau based on the oxidation of desorbed hydrogen from the hydride. It can be seen from Fig. 5 that the melt spinning and the addition of Nd considerably modify discharge potential characteristics of the alloys, making the discharge potential rise and the discharge plateau lengthen. Meanwhile, it is found that, besides improving potential characteristic, the melt spinning and the addition of Nd also enable the discharge capacity of alloys to increase dramatically. In order to examine the effect of the Nd content on the discharge capacity of the alloys, the relationship between the discharge capacity of the alloys and the amount of Nd is established, just as demonstrated in Fig. 6. Apparently, the discharge capacities of the alloys augment with the Nd content rising up to x = 10, but the excessive amount of Nd adding (x > 10) results in an undesired reduction in the discharge capacity. But even so,
84
Y. Zhang et al. / Journal of Alloys and Compounds 584 (2014) 81–86
Fig. 6. Evolution of the discharge capacity of the as-cast (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys with Nd content varying.
Fig. 4. Activation capability of the as-cast and spun the (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys: (a) Nd15 alloy, (b) As-spun (10 m/s).
and
spun
the alloy. When the Nd content is x = 10, the as-cast and spun alloys yield the maximum discharge capacities, namely 86.4, 200.5, 266.3, 402.5 and 452.8 mA h/g corresponding to the spinning rate of 0, 10, 20, 30 and 40 m/s, respectively. Furthermore, it must be noticed that, for a fixed Nd content, the discharge capacity of the alloy markedly increases with spinning rate rising. Particularly, enhancing the spinning rate from 0 to 40 m/s enables the discharge capacity to grow from 42.5 to 100.6 m A h/g for the Nd0 alloy and from 86.4 to 452.8 mA h/g for the Nd10 alloy, respectively. Obviously, the positive contribution originated from the melt spinning to the discharge capacity of the Nd-added alloys is much more pronounced than the Nd-free one. The benefaction of Nd adding on the discharge capacity of the alloy is considered most likely to be associated with the following factors. On the one hand, the glass forming facilitated by Nd adding is believed to be a crucial factor because an adequate proportion of amorphous and nanocrystalline in Mg–Ni-based alloy possesses superior discharge property [15]. On the other hand, the secondary phases Mg6Ni and NdNi probably
Fig. 5. Discharge potential curves of the as-cast and spun (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys: (a) Nd15 alloy, (b) As-spun (10 m/s).
the Nd-added alloys still show much higher discharge capacity as compared to the Nd-free alloy, suggesting that the addition of Nd totally engenders beneficial impact on the discharge capacity of
Fig. 7. Evolution of the capacity retaining rates (Sn) of the as-cast and spun alloys with cycle number: (a) Nd15 alloy, (b) As-spun (10 m/s).
Y. Zhang et al. / Journal of Alloys and Compounds 584 (2014) 81–86
play a catalytic role in activating the Mg2Ni phase to reversibly absorb/desorb hydrogen in the alkaline electrolyte. As a matter of fact, the excessive amount of Nd (x >10) gives rise to an undesired reduction in the discharge capacity, for which the sharp decrease of the Mg2Ni phase is mainly responsible. In terms of the discharge capacity improved by melt spinning, it is definitely believed to be related to nanocrystalline and amorphous structure owing to the grain boundaries possessing the distribution of the maximum hydrogen concentrations [16]. Also, the nanocrystalline and amorphous structure generated by melt spinning evidently promotes the hydriding and dehydriding abilities of Mg2Ni-type alloys, as reported by Kumar et al. [2] and Teresiak et al. [5]. 3.2.3. Cycle stability Cycle stability, one of the major performances which are used to evaluate whether a kind of an alloy can be applied as a negative electrode material or not, is signified by the capacity retaining rate (Sn), defined as Sn = Cn/Cmax 100%, where Cmax is the maximum discharge capacity and Cn is the discharge capacity at the nth charge–discharge cycle, respectively. Clearly, it means that the larger the capacity retaining rate (Sn) is, the better the cycle stability of the alloy will be. The variations of the Sn values of the as-cast and spun (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys depending on cycle number are presented in Fig. 7, from which the degradation process of the discharge capacity of the alloys can be seen clearly. The slopes of the curves qualitatively reflect the degradation rate of the discharge capacity during the charging-discharging cycles, namely the smaller the slope of the curve is, the better the cycle life of the alloy will be. It is found from Fig. 7(a) that the curve’s slope of the Nd15 alloy always declines with the spinning rate growing, indicating that the melt spinning makes the positive contribution to the cycle stability of the Nd15 alloy. In order to clearly show the effect of the spinning rate on the cycle stability of the alloy, here we build the relationship between the S20 value of the Nd15 alloy and the spinning rate, as inserted in Fig. 7(a), from which it is derived that the increasing of the spinning rate from
85
0 to 40 m/s renders the S20 value of the Nd15 alloy rising from 59.2% to 85.9%. Also, it is also noted that the slopes of the curves in Fig. 7(b) are changeable with Nd content varying, but no matter how much Nd content is, the curves’ slopes of the Nd-added alloys are all less than that of the Nd-free one, indicating that the addition of Nd plays a beneficial role on the cycle life of the alloy electrode. Likewise, the relationship between the S20 values of the as-spun (10 m/s) alloys and the Nd content is also established, as inserted in Fig. 7(b). It is evident that the S20 value of the as-spun (10 m/ s) alloy first mounts up and then goes down with the Nd content growing. Here, we provide some elucidations as the reasons why the melt spinning and Nd adding ameliorate the cycle stability of the alloy. It has come to light that the sharp degradation of the discharge capacity of Mg2Ni-type alloy during the charge–discharge cycling is caused by forming and thickening of Mg(OH)2 surface layer, which hinders the hydrogen atoms from diffusing in or out, in alkaline solution [17]. Moreover, the hydrogen storage material suffers an inevitable volume change during the charge–discharge process which aggravates the alloy’s cracking and pulverizing and then makes the surface of the material apt to be oxidized, which is also evidenced by our experimental results, as demonstrated in Fig. 8. The alloy particles (denoted as A) with a size of about 30 lm and ultrafine Ni powders appearing an aggregation sate (denoted as B) can be seen clearly. Furthermore, it is found that there are numerous cracks on the surfaces of the alloy particles after electrochemical cycle, but the sizes of the alloy particles have no obvious change, suggesting that the pulverization of alloy particles scarcely take place in the process of the electrochemical cycle. Noticeably, the alloy particles after cycling are covered by a rough and gossypine layer, which is inspected to be Mg(OH)2 by XRD, as illustrated in Fig. 8(d). Also, we note that the thickness of the covering layer on the surface of the Nd20 alloy particle is much less than that of the Nd0 alloy, meaning that the addition of Nd can notably increases the anti-corrosion and antioxidation ability of the Mg2Ni alloy. A similar result was reported by Teresiak et al. [5]. Besides, it can be seen that there are much
Fig. 8. SEM morphologies together with typical XRD patterns of the as-spun (10 m/s) Nd0 and Nd20 alloys before and after electrochemical cycle: (a) and (b) Nd0 alloy before and after cycling, (c) Nd20 alloy after cycling, (d) XRD of the Nd0 alloy after cycling.
86
Y. Zhang et al. / Journal of Alloys and Compounds 584 (2014) 81–86
fewer cracks on the surface of the Nd20 alloy particle after cycling than that of the Nd0 one, reflecting that the addition of Nd can markedly enhance the anti-pulverization ability of the alloy. The above-mentioned functions of Nd adding are most probably ascribed to two aspects. Firstly, adding the rare elements (La, Nd, Sm, Y) engenders the cell volume of the Mg2Ni alloy enlarging clearly [13], decreasing the ratios of expansion/contraction in the process of hydrogen absorption/desorption, so as to enhance the anti-pulverization capability. Secondly, the glass formation facilitated by Nd adding is extremely important because an amorphous phase improves not only anti-pulverization ability but also anticorrosion and anti-oxidation abilities of the alloy electrode in a corrosive electrolyte [12]. Based on above-mentioned causes, it is very easy to understand why the Nd-added alloys exhibit much higher cycle stability than the Nd-free alloy. 4. Conclusions The effects of melt spinning and Nd adding on the electrochemical hydrogen storage performances of the as-cast and spun (Mg24Ni10Cu2)100 xNdx (x = 0–20) alloys were investigated, and the major conclusions are summarized as follows: 1. The investigations on the structures of the alloys reveal that the addition of Nd facilitates the glass forming in the Mg2Ni-type alloys. As a result, the as-spun Nd-free alloy holds an entire nanocrystalline structure but an obvious amorphous structure for the as-spun Nd-added alloys. 2. Both the melt spinning and Nd adding make the positive contribution to the electrochemical performances of the alloys, including discharge capacity, potential characteristic and cycle stability. 3. Furthermore, the addition of Nd significantly enhances the anticorrosion and anti-pulverization ability of the alloy, which is
viewed as the root reason of Nd adding improving the cycle stability of the Mg2Ni-type alloys.
Acknowledgements This work is financially supported by National Natural Science Foundations of China (51161015) and Natural Science Foundation of Inner Mongolia, China (2011ZD10). References [1] S. Ruggeri, L. Roué, J. Huot, R. Schulz, L. Aymard, J.-M. Tarascon, J. Power Sources 112 (2002) 547–556. [2] L.H. Kumar, B. Viswanathan, S.S. Murthy, J. Alloys Comp. 461 (2008) 72–76. [3] L. Schlapbach, A. Züttel, Nature 414 (2001) 353–358. [4] M.S. Wu, H.R. Wu, Y.Y. Wang, C.C. Wan, J. Alloys Comp. 302 (2000) 248– 257. [5] A. Teresiak, A. Gebert, M. Savyak, M. Uhlemann, C. Mickel, N. Mattern, J. Alloys Comp. 398 (2005) 156–164. [6] T. Abe, T. Tachikawa, Y. Hatano, K. Watanabe, J. Alloys Comp. 792 (2002) 330– 332. [7] C. Lenain, L. Aymard, J.M. Tarascon, J. Solid State Electrochem. 2 (1998) 285– 290. [8] Z.M. Wang, H.Y. Zhou, Z.F. Gu, G. Cheng, A.B. Yu, J. Alloys Comp. 381 (2004) 234–239. [9] M.Y. Song, S.N. Kwon, J.S. Bae, S.H. Hong, Int. J. Hydrogen Energy 33 (2008) 1711–1718. [10] G.Y. Liang, D.C. Wu, L. Li, L.J. Huang, J. Power Sources 186 (2009) 528–531. [11] S. Todorova, T. Spassov, J. Alloys Comp. 469 (2009) 193–196. [12] Y.H. Zhang, C. Zhao, T. Yang, H.W. Shang, C. Xu, D.L. Zhao, J. Alloys Comp. 555 (2013) 131–137. [13] H.P. Ren, Y.H. Zhang, B.W. Li, D.L. Zhao, S.H. Guo, X.L. Wang, Int. J. Hydrogen Energ. 34 (2009) 1429–1436. [14] D.H. Xie, P. Li, C.X. Zeng, J.W. Sun, X.H. Qu, J. Alloys Comp. 478 (2009) 96– 102. [15] T. Spassov, U. Köster, J. Alloys Comp. 287 (1999) 243–250. [16] S. Orimo, H. Fujii, Appl. Phys. A 72 (2001) 167–186. [17] M. Dornheim, S. Doppiu, G. Barkhordarian, U. Boesenberg, T. Klassen, O. Gutfleisch, R. Bormann, Scripta Mater. 56 (2007) 841–846.