- Email: [email protected]
Pulsed-laser fabrication of gas-filled hollow Co–Pt nanospheres
Available online at www.sciencedirect.com
ScienceDirect Acta Materialia 61 (2013) 7924–7930 www.elsevier.com/locate/actamat
Pulsed-laser fabrication of gas-filled hollow Co–Pt nanospheres S. Sturm a,⇑, K.Z. Rozman a, B. Markoli b, N.S. Antonakakis c, E. Sarantopoulou c, Z. Kollia c, A.C. Cefalas c, S. Kobe a a Department for Nanostructured Materials, Jozef Stefan Institute, 1000 Ljubljana, Slovenia Faculty of Natural Sciences and Engineering, University of Ljubljana, 1000 Ljubljana, Slovenia c Theoretical and Physical Chemistry Institute, National Hellenic Research Foundation, 11635 Athens, Greece b
Received 22 May 2013; received in revised form 16 September 2013; accepted 18 September 2013 Available online 8 October 2013
Abstract We report on nitrogen-filled hollow Co–Pt nanospheres produced via pulsed-laser ablation in ambient nitrogen gas. The resulting nanospheres are characterized by a single-crystalline face-centred cubic Co55±3Pt45±3 shell and a void filled with molecular nitrogen, typically occupying the sphere’s central region. The average diameter of the spheres and the voids is 35 ± 8 and 16 ± 2 nm, respectively. The calculated number density of nitrogen atoms, measured within these voids, is 1.58 ± 0.4 nm3. The resulting pressure in the voids near ambient temperature (300 K) and at the boiling temperature for the Co–Pt alloy (3000 K) is estimated to be 1.9 ± 0.3 and 34.3 ± 9 MPa, respectively. The gas-filled Co–Pt hollow spheres are formed in only one step involving two physical processes. First, after each laser pulse, the vaporized, supersaturated Co–Pt ablated species are condensated in the plume under high pressure and temperature, resulting in nitrogen gas trapping. Between two laser pulses, the pressure and temperature in the plume drop rapidly, the nitrogen-rich liquid nanospheres become thermodynamically unstable and the nitrogen gas bubble starts to expand until the solidification of the nanospheres. The fast solidification of the solid shell prevents further outward diffusion of nitrogen and thus an amount of nitrogen gas is preserved in the void. These nanospheres have the potential in biomedical, magnetic and catalytic applications. Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Pulsed-laser ablation; Hollow-nanospheres; Nitrogen; TEM; EELS
1. Introduction Structures with hollow interiors, such as hollow nanospheres, have recently received considerable scientific attention owing to their unique properties, which could facilitate breakthrough applications in various fields of nanoscience [1–5]. In view of this, various synthesis methods for the fabrication of hollow nanostructures have been proposed that, to some extent, enable control over the properties of the shell. A large number of reported synthesis procedures are based on using prefabricated nanostructured materials as templates, which are subsequently followed by various chemical transformations, such as diffusion, oxidation, ion exchange and galvanic replacement, resulting in the ⇑ Corresponding author. Tel.: +386 1477 3418; fax: +386 1477 3221.
E-mail address: [email protected] (S. Sturm).
formation of hollow structures [6–9]. As an example, the synthesis of hollow nanostructures from their solid counterparts can be achieved using the Kirkendall effect, where the voids are created by the accumulation of vacancies resulting from different ion mobilities. Based on the Kirkendall effect, different binary metal–oxide or metal–sulfide nanospheres can be formed via a two-step synthesis that involves the formation of a metal particle, which is then followed by subsequent oxidation processing [10,11]. A different template-directed synthesis approach was proposed recently, where hollow nanoparticles were successfully fabricated by a so-called bubble synthesis, which uses an electrochemically evolved hydrogen bubble as a template and a reducing agent for electroless deposition [12]. Although these methods have proved to be successful in the synthesis of hollow nanostructures, their major drawback is the restriction related to the choice of materials,
1359-6454/$36.00 Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.actamat.2013.09.033
S. Sturm et al. / Acta Materialia 61 (2013) 7924–7930
which is often limited to simple solids. For this reason, only a few reports are available on hollow, binary alloy nanoparticles. Given these challenges, physical processing routes like pulsed-laser ablation (PLA) in the presence of a background gas have tremendous potential for applications because they offer flexibility in terms of the choice of materials to be ablated, and at the same time the ability to produce various complex nanostructures [13]. It was shown previously that Sm–Fe–Ta-based hollow nanospheres can be produced via a single-step procedure by the ablation of a Sm–Fe–Ta-based alloy target into an ambient nitrogen gas. The formation of hollow nanospheres, typically with molecular nitrogen occupying the central void of the sphere, was explained by a nanoscale melt-solidification phenomenon [14]. The Co–Pt system is a promising system covering areas of magnetism, and catalysis, with the possibility to fine tune the physical properties by adjusting both the composition and the morphology of nanoparticles [15–17]. In this study we have demonstrated that by applying PLA in an ambient nitrogen gas the gas-filled hollow Co–Pt nanospheres can be successfully produced, where the composition of the particles is controlled by the Co–Pt target composition. By means of various techniques of transmission electron microscopy we aimed to characterize both structure and composition of synthesised nanospheres as well as to determine the nitrogen pressure inside the individual voids. The data were further employed for reconstruction of the formation mechanism of Co–Pt gas-filled nanospheres, and suggestions towards the general formation mechanism for gas-filled nanospheres in other metallic systems have been made. 2. Materials and methods 2.1. Materials The PLA apparatus consisted of a molecular fluorine laser at 157 nm and a stainless steel vacuum chamber, placed in a computerized X–Y–Z–h micro-translation stage. A freshly polished Co–Pt alloy target with the composition Co48Pt52 was positioned perpendicular to the laser beam and focused with a high-quality CaF2 lens through a 5 mm aperture in a compact geometry. The distance between the target and the substrate was fixed at 1 cm. The deposition parameters were 30 mJ, a 3–5 J cm2 fluence (600 lm laser spot size on the target) at 15 Hz. The nanoparticles were deposited in high-purity nitrogen at 0.1 MPa. The nanoparticles were deposited directly on a lacy-carbon copper grid. The latter deposition procedure allowed a direct investigation of the obtained product by the use of transmission electron microscopy (TEM). 2.2. Methods For the detailed structural and compositional investigations of hollow nanospheres a field-emission (scanning)
7925
transmission electron microscope TEM/STEM (JEM2010F) operated at 200 keV (Cs = 0.48 mm) equipped with a high-angle annular dark-field (HAADF) detector, energy-dispersive X-ray spectroscopy (EDXS), (LINK ISIS EDS 300) and electron energy-loss spectroscopy (EELS) (Gatan PEELS 766) was applied. The probe size in the STEM was 2 nm, while the HAADF collection angle was set between 60 and 160 mrad. The EELS spectra were background-subtracted, and corrected for the dark current and the channel-to-channel gain variation of the photodiode detector. The full width at half maximum (FWHM) of the zero-loss line was 1.2 eV. Prior to the quantification of the EELS spectra the effect of the plural scattering was removed by applying the Fourier-ratio deconvolution method [18]. The number density of the nitrogen atoms in the voids of the nanospheres was determined by a modified procedure initially proposed for measuring the density of He atoms in nanometre-sized bubbles in steels using a combined STEM–EELS technique [19,20]. The following equation gives the number density of nitrogen atoms (nN) in the void: nN ¼ I N =ðrN I ZL dÞ
ð1Þ
IN and IZL are the intensities measured from the N–K ionization edge and the zero-loss peak, respectively. The corresponding intensities were integrated over an energy window (DE) of 8 eV. The intensity of the N–K edge was measured starting from the ionization edge onset. rN is the angle-integrated hydrogenic cross-section for the nitrogen K-shell ionization, calculated for the experimental collection angles. d represents the measured diameter of the void. This equation assumes that the void diameter is significantly larger than the electron beam and the measurement was performed in the central region of the void, providing localized information on an individual void. The error values related to the accuracy in the determination of the cross-section, the N–K signal and the void diameter were estimated to be 10%. The total error of the calculated density, applying standard error-propagation relations, amounts to 17%. 3. Results 3.1. Crystal structure and composition analysis Nanoparticles with a high spherical morphology, hereafter referred to as nanospheres, were produced by the direct ablation of the CoPt target on a lacy-carbon copper grid, as shown in a bright field (BF) TEM image (Fig. 1). The selected-area electron-diffraction (SAED) pattern presented in the inset picture in Fig. 1a, which was taken from the observed specimen area, confirms that the spheres are crystalline. Their crystal structure was determined by comparing the experimental and calculated SAED patterns. The experimental and corresponding rotational averaged SAED pattern matches with the calculated diffraction pattern, based on the Co50Pt50 face-centred cubic (fcc) phase
7926
S. Sturm et al. / Acta Materialia 61 (2013) 7924–7930
Fig. 1. BF-TEM image of PLA products using a CoPt target in a nitrogen atmosphere is characterized by (a) nanospheres often organized in a necklacelike structure. The corresponding experimental and calculated SAED patterns indicate the fcc structure. (b) Spheres were found to be single crystalline, as shown by a superimposed NBED pattern obtained from the marked circular region. Some spheres contained planar defects (marked by an arrow).
Fig. 2. (a) HRTEM image of representative spheres characterized by a diffuse spherical contrast located in their central regions. The corresponding image detail marked by a rectangle shows (b) lattice fringes combined with a diffuse round contrast. No structural difference could be observed between the central and outer regions of the sphere, as indicated from the corresponding FFTs.
with a cell parameter a = 0.384 nm [21]. A representative nanobeam electron-diffraction (NBED) pattern, acquired from individual spheres, proves that they are single crystalline (Fig. 1b). In addition, the NBED pattern matches with the Co50Pt50 fcc phase viewed in the [0 0 1] zone axis. Some spheres contained planar defects (marked by an arrow in Fig. 1b), i.e. stacking faults or twins. They are most probably caused by internal stresses due to the high cooling rates and are typical for metals with the fcc structure [22]. A detailed image analysis revealed one morphologically distinctive group of spheres, which were characterized by a strong spherical contrast located in their central regions. A large number of such spheres were regularly observed in the analysed samples, as shown in a high-resolution TEM (HRTEM) image (Fig. 2). The enlarged central section of such a sphere (marked by the rectangle in Fig. 2a), besides crystal fringes, which extend over the whole analysed area, also reveals a background diffuse spherical contrast (Fig. 2b). The corresponding fast Fourier transforms (FFTs), shown as insets in Fig. 2b, were calculated from the central (top) and border (bottom) regions of the image, respectively. Identical reflection points are observed in both the FFT images, indicating that there are no significant changes in the crystal structure and/or crystal orientation
between the inner and outer regions of the sphere. In order to unambiguously interpret the origin of the observed contrast, HAADF-STEM combined with EDXS and EELS analyses were employed. BF-TEM and HAADF-STEM images were acquired from an identical string of spheres and are shown in Fig. 3. A large number of the spheres observed in the BF-TEM image show a unique contrast feature located in their central regions, which corresponds perfectly with the dark spherical contrast observed in the HAADF-STEM image. The darker contrast in the HAADF-STEM image could signify regions of large compositional differences. However, the compositional fluctuations that were measured from these spheres could not explain such a dramatic drop in the measured intensity. Fig. 4a shows typical spheres with a dark, spherical contrast. The corresponding normalized intensity profile, obtained from the corresponding HAADF-STEM image, is shown in Fig. 4b. The compositional difference between the central region and the adjacent shell region, i.e. CoCENTRAL/CoSHELL at.% ratio, was 1.05. The exact compositions of the central and the shell regions were Co57Pt43 and Co54Pt46, respectively. However, such small compositional differences cannot explain the almost 30% drop in the intensity measured from the intensity profile.
S. Sturm et al. / Acta Materialia 61 (2013) 7924–7930
7927
Fig. 3. (a) BF-TEM image obtained from a string of spheres. The majority of spheres have a round, bright contrast in their central regions. (b) HAADFSTEM image of the same specimen detail. The round, dark contrast in the sphere’s central regions coincides with the previously described bright contrast found in the TEM images.
Fig. 4. (a) HAADF-STEM image of hollow spheres and (b) the normalized intensity line profile obtained from the HAADF-STEM image (x–x0 line) showing an almost 30% drop in intensity in the sphere’s central region, which indicates the presence of a void.
According to the HAADF-STEM image analysis and EDXS results, the dark, round contrast is most probably due to the presence of voids within the nanospheres. The overall average composition deduced from the eight spheres was Co55±3Pt45±3. The analysed sphere’s diameter ranged between 24 and 41 nm, with a mean value and standard deviation of 34 ± 5 nm. No significant correlation could be found between their sizes and the composition. 3.2. Characterization of nitrogen in voids by EELS The light-element concentrations, N and O in this case, were measured using EELS. Fig. 5a shows a typical EELS spectrum (I) acquired in the spectrum region of both the N–K and O–K ionization edges. The representative spheres, which were analysed, are shown as the inset in Fig. 5a. The spatially resolved EELS analyses performed in the void and in the shell region consistently show the presence of nitrogen only in the voids. The backgroundsubtracted N–K ionization edge is shown in (II). In order to unequivocally identify the presence of the edge signals
a second energy-difference method was applied, which is especially useful for exposing a weak signal superimposed on a slowly varying background [18]. The corresponding second-difference spectra from the void and the shell region are shown in (III) and (IV), respectively. This analysis additionally confirms that the N signal is associated only with the voids, while no signal can be measured in the shell region. Moreover, the presence of oxygen could not be detected, which indicates that the nanospheres remained in the un-oxidized state. To additionally increase the signal-to-noise ratio, the N–K edges acquired from the voids of the analysed spheres were normalized, subsequently realigned and summed together. Fig. 5b shows the resulting N–K edge, which is presented together with the N–K edges obtained from the previously studied Sm–Fe(Ta) –N system [14] and the reference standard spectrum of N2 gas obtained from air [23]. The fine structure for all these N–K edges is distinctive for molecular nitrogen and is characterized by a sharply peaked edge at 401 eV, followed by a broad continuum. Following these results, the nanospheres with the indicated spherical dark contrast, which
7928
S. Sturm et al. / Acta Materialia 61 (2013) 7924–7930
Fig. 5. (a) Raw and processed EEL spectra obtained from the representative hollow spheres (shown in the inset). (I) Raw EEL spectrum acquired from the central region of the sphere shows the presence of a weak N signal. (II) Background-subtracted EEL spectrum obtained from spectrum I, clearly showing the N–K ionization edge. Second difference spectrum confirms that nitrogen is always related to the sphere’s central region (III). In contrast, spectrum IV, which is characterized only by noise, confirms the absence of nitrogen in the outer regions of the sphere, i.e. below the detection limits. The grey area in (II) represents the integrated signal used for calculating the number density of nitrogen atoms in the voids. (b) Fine structure in the N–K edges obtained from the Co–Pt and Sm–Fe–Ta systems and air are distinctive for molecular nitrogen. The EEL spectra have been translated vertically for visual clarity.
was regularly observed in HAADF-STEM images, can be defined as hollow spheres filled with gas or as nanospheres with gas bubbles. 3.3. Determination of nitrogen pressure in voids The number density of nitrogen atoms (n) within the void was calculated by applying Eq. (1) (see Section 2.2). The void diameter was measured from the HAADF-STEM images, by drawing the intensity line profile. The diameters of the measured voids ranged between 15 nm and 19 nm, with a mean value and relatively narrow standard deviation of 16 ± 2 nm. The diameters of the corresponding spheres ranged between 24 nm and 45 nm, with a mean value and a higher standard deviation of 35 ± 8 nm. The resulting calculated number density of nitrogen atoms obtained from six measured spheres was between 1 and 2 nm3, with an average value of n = 1.58 ± 0.4 nm3. The pressure in the void was estimated from the values of the nitrogen density and the related void volume by using a suitable equation of state. To calculate the pressure in the void accurately, with respect to the given density and the temperature range, a standard correction of an ideal gas law using a virial expansion was applied, as follows: 2
P ¼ nkT ð1 þ nB=N A þ n2 C=ðN A Þ þ . . .Þ
ð2Þ
The symbol k represents the Boltzmann constant, T is the absolute temperature and NA is the Avogadro constant, while B and C are the second and third virial coefficients, respectively. The values of the B and C coefficients for nitrogen gas can be found in the work of Sevast’yanov and Chernyavskaya [24] for temperatures from 75 K up to 2500 K. The resulting calculated average pressure in the voids near the ambient temperature (300 K) was 1.9 ± 0.3 MPa. Assuming that the voids filled with gas were formed while the spheres were still in the liquid phase the equilibrium pressure needs to be recalculated for a
given temperature range. According to the literature data [25], the boiling temperature for the Co–Pt alloy at ambient pressure is 3000 K. The corresponding pressure of the nitrogen in the voids at this temperature would increase to a value of 34.3 ± 9 MPa. 4. Discussion The gas-filled hollow spheres presented in this study were most probably formed via a nucleation-and-growth process. The condensation of such particles starts already within the ablated vapour, with vapour species undergoing enough collisions with the background gas for them to nucleate and grow. Consequently, these particles would exhibit a homogenous composition enriched with the more volatile component, due to preferential ablation, which would be size-independent. On the other hand, the particles that are expelled from the melt would be enriched with the less volatile component, since they experience a loss of the more volatile component on their path towards the substrate [26]. This consequently leads to deposits in which the composition depends upon the particle size [27]. For the Co–Pt system, the corresponding vaporization enthalpies (DHvap) of the Co and Pt are 377 and 469 kJ mol1, respectively. According to this, we can expect that the plume will be enriched with Co, which results in Co-rich nanoparticles, as was already observed by Trelenberg et al. [28]. This is also in agreement with our findings where Co-rich nanospheres, with an average composition of Co55±3Pt45±3, were deposited from the Co-deficient target (Co48Pt52), confirming that they have been formed directly from the vapour phase, via the already described evaporation–condensation mechanism. It was reported that the background gas influences the plasma kinetics and, consequently, the formation of the nanoparticles. Observations like plume splitting and plasma deceleration were related to the appearance of the
S. Sturm et al. / Acta Materialia 61 (2013) 7924–7930
shock wave and a contact front, where the shock wave is caused by the sudden impact of the expanding plasma with the background gas, resulting in a propagating perturbation with the contact front establishing the border between this perturbed gas and the plasma [29,30]. The heterogeneous nucleation and growth of the nanoparticles within the plume is bound to this narrow shock-wave region of the highly compressed, supersaturated vapour mixture, which is composed of the ablated vapour and the background gas, with large temperature and density gradients as well as pressure differences [31]. The vapour pressure and the temperature just behind the plasma front can be estimated using the general laws of the conservation of mass, momentum and energy [32]. The estimated vapour pressures of the Co–Pt plasma expanding into the nitrogen gas at 0.1 MPa would be in the range of 100 MPa to several GPa, while the vapour temperatures would range between 103 and 104 K [31,32]. Immediately after the condensed particles are ejected from the high-pressure and high-temperature plasma region, the particle nucleation and growth are heavily suppressed as a result of the rapid drop in pressure and temperature [31]. As a result the dissolved nitrogen in the molten sphere will start to condense and form an initial nitrogen bubble. The amount of dissolved nitrogen in the Co–Pt at the boiling point and ambient pressure was estimated using Sievert’s law for liquid metals. The calculated amount of dissolved nitrogen in a liquid Co–Pt sphere for given temperature and pressure conditions is 14 at.%. It is worth mentioning that the solubility of nitrogen in a molten sphere at a higher pressure and temperature would be even greater. For comparison, the nitrogen solubility in solid Co at 1873 K and ambient pressure is only 0.3 at.%, while in Pt it can be considered as negligible [25]. This is not a surprising result if we consider that prior to the condensation of the Co–Pt–N particles, the temperature and pressure in the plume can briefly reach well over 104 K and several GPa, respectively, which is enough to supersaturate the Co–Pt with nitrogen. The properties of the condensed nanoparticles, such as the composition, the crystal structure and the morphology, are largely dependent on
7929
the interplay between the thermodynamics of the two-phase region, i.e. the liquid–vapour, the rules of solidification for the multi-component system and the kinetic constraints during rapid cooling. Namely, the time scale of these processes is in the range of tenths of a second, which is based on the fact that all the examined nanospheres had a crystalline structure. This implies that the cooling rates must have been in the range from 102 to 103 K s1 or less. Had they been higher than this, the spheres would be in an amorphous state [33]. The nitrogen bubble pressure (P) within the molten Co–Pt sphere is equilibrated over the surface tension (!) of the Co–Pt liquid following the relation P = 2!/r, where r stands for the radius of the nitrogen bubble. The calculated nitrogen pressure in bubbles with an experimentally measured diameter of 32 ± 1.6 nm would yield a value of 2 102 MPa. The fact that the measured pressure in the voids was one order of magnitude lower suggests that the cooling rates were such that it gave the nitrogen sufficient time to partially diffuse from the nanosphere. As the temperature of the molten sphere is dropping due to cooling (primarily through conduction), the permeability of Co–Pt to nitrogen is also dropping exponentially, while the interdiffusion of Co and Pt is severely hindered. For instance, the interdiffusion coefficient for Co and Pt is D1537K ffi 4 1011 cm2 s1, while the diffusion coefficient for nitrogen at the same temperature is much higher, with D1537K ffi 0.9 cm2 s1. Moreover, the preferential diffusion of nitrogen towards the centre of the sphere is promoted by the fact that the solubility of nitrogen in the solid (0.3 at.%) is much lower than in the liquid (14 at.%). This is in accordance with our results, as the nitrogen signal was only measured in the void, while no nitrogen was detected in the rim of the sphere. The arrangement of the Co–Pt spheres in string-type structures is due to the magnetic dipole–dipole interaction between the particles. Depending on the particle volume, the magnetization can be fixed within an individual particle if the volume exceeds a certain critical volume [34]. For Co–Pt with a fcc crystal structure the calculated critical diameter equals 15 nm, which means that the particles
Fig. 6. The formation mechanism for gas-filled hollow metallic nanospheres produced by PLA in a nitrogen-rich atmosphere in sequences indicating: (I) condensation of particles in the high P–T plasma region, (II) the formation of nitrogen bubble in liquid sphere, (III) solidification and (IV) the final solid sphere with nitrogen-filled void.
7930
S. Sturm et al. / Acta Materialia 61 (2013) 7924–7930
produced in our study are ferromagnetic. They are most probably in a single-domain state, since the calculated critical diameter of the single-domain particle for the Co–Pt fcc crystal structure equals 30 nm [35]. Moreover, spontaneous magnetic fields with a magnetic flux density in the range of kilogauss were already observed in laser-produced plasma due to the large temperature gradients, which generate thermoelectric currents [36]. Such magnetically induced self-assembly, which originates from the magnetostatic interactions, can yield unique structures from strings to necklaces, as observed in our study. According to the obtained results we propose the formation mechanism for gas-filled hollow metallic nanospheres, which can be generalized for other metallic systems (Fig. 6). At first, the condensation of the spheres starts from the supersaturated mixture of ablated species and ambient gas at a relatively high pressure and temperature. The melt needs to be acceptable for gas intake, as this provides a molten sphere with the highest concentration of dissolved nitrogen. Then, the metallic system in relation to the background nitrogen gas should be selected so that the high solubility differences between the melt and the corresponding solid are achieved. Finally, the outwards diffusion of nitrogen through the sphere’s solid rim, for example, by providing high cooling rates, should be suppressed. 5. Conclusions The Co–Pt system was chosen as an example to verify and explain the formation mechanism for gas-filled hollow metallic spheres produced by PLA in the presence of an ambient nitrogen gas. The presented results confirm that nitrogen-filled Co–Pt spheres can be fabricated successfully. The spheres were formed by condensation from the plume in a region of high pressure and temperature. We believe that the nitrogen bubble within the liquid nanosphere was initially formed due to the abrupt drop of pressure and temperature in the nitrogen-saturated liquid sphere. Subsequently, the nitrogen could diffuse from the interior of the sphere outwards. However, due to the fast solidification and the formation of the solid rim the outward diffusion of nitrogen is heavily suppressed, preventing the N2 gas from escaping the sphere. The resulting solid sphere is characterized by a void filled with nitrogen gas. The obtained results support the idea that gas-filled hollow spheres could be fabricated in various complex metallic systems by applying PLA in the presence of a background gas, taking into consideration that in relation to the background gas high solubility differences between the melt and corresponding solids are achieved. Acknowledgements This work was financially supported by the Ministry of Higher Education, Science and Technology of the Republic
of Slovenia and by the European Union as part of the Framework 7 program under a contract for an Integrated Infrastructure Initiative, Reference ESTEEM2. References [1] Chen HM, Liu RS. J Phys Chem 2011;115:3513. [2] Shankar SS, Deka S. Sci Adv Mater 2011;3:169. [3] Shan ZW, Adesso G, Cabot A, Sherburne MP, Syed Asif SA, Warren OL, et al. Nat Mater 2008;7:947. [4] Zhang JZ, Noguez C. Plasmonics 2008;3:127. [5] Botella P, Ortega I, Quesada M, Madrigal RF, Muniesa C, Fimia A, et al. Dalton Trans 2012;41:9286. [6] Vasquez Y, Henkes AE, Chris Bauer J, Schaak RE. J Solid State Chem 2008;181:1509. [7] Zhang Q, Wang W, Goebl J, Yin Y. Nano Today 2009;4:494. [8] Jones MR, Osberg KD, MacFarlane RJ, Langille MR, Mirkin CA. Chem Rev 2011;111:3736. [9] Moon GD, Ko S, Min Y, Zeng J, Xia Y, Jeong U. Nano Today 2011;6:186. [10] Yin Y, Rioux RM, Erdonmez CK, Hughes S, Somorjal GA, Alivisatos AP. Science 2004;304:711. [11] Iba´n˜ez M, Fan J, Li W, Cadavid D, Nafria R, Carrete A, et al. Chem Mater 2011;23:3095. [12] Huang C, Li YJ, Muangphat C, Hao Y. Electrochim Acta 2011;56:8319. [13] Swihart MT. Curr Opin Colloid Interface Sci 2003;8:127. [14] Sturm S, Rozman KZ, Markoli B, Sarantopoulou E, Kollia Z, Cefalas AC, et al. Nanotechnology 2010;21:485603. [15] Chen G, Xia D, Nie Z, Wang Z, Wang L, Zhang L, et al. Chem Mater 2007;19:1840. [16] Vasquez Y, Sra AK, Schaak RE. J Am Chem Soc 2005;127:12504. [17] Sun Q, Wang S, Wang R. J Phys Chem C 2012;116:5352. [18] Egerton RF. Electron energy loss spectroscopy. 2nd ed. New York: Plenum; 1996. [19] Walsh CA, Yuan J, Brown LM. Philos Mag A 2000;80:1507. [20] Fre´chard S, Walls M, Kociak M, Chevalier JP, Henry J, Gorse D. J Nucl Mater 2009;393:102. [21] Leroux C, Cadeville MC, Pierron-Bohnes V, Inden G, Hinz F. J Phys F: Met Phys 1988;18:2033. [22] Smallman RE, Bishop RJ. Metals and materials, science, processes, applications. Oxford: Butterworth-Heinemann; 1995. [23] Ahn CC, Krivanek OL, Burgner RP, Disco MM, Swann PR, EELS Atlas. Arizona State University HREM Facility/Gatan Inc.; 1983. [24] Sevast’yanov RM, Chernyavskaya RA. Inzh-Fiz Zh 1986;51:124. [25] Gale WF, Totemeir TC, editors. Smithells metals reference book. Burlington, MA: Butterworth-Heinemann; 2004. [26] Hergenro¨der R. J Anal At Spectrom 2006;21:1016. [27] Chen LC. In: Chrisey DB, Hubler GK, editors. Pulsed laser deposition of thin films. New York: Wiley; 1994. [28] Trelenberg TW, Dinh LN, Stuart BC, Balooch M. Appl Surf Sci 2004;229:268. [29] Horn I, Gu¨nther D. Appl Surf Sci 2003;207:144. [30] De Posada E, Arronte MA, Ponce L, Rodrı´guez E, Flores T, Lunney JG. J Phys: Conf Ser 2011;274:012078. [31] Sharma AK, Thareja RK. Appl Surf Sci 2005;243:68. [32] Zel’dovich YB, Raizer YP. In: Hayes WD, Probstein RF, editors. Physics of shock waves and high-temperature hydrodynamic phenomena. New York: Dover; 2002. [33] Gaskell DR. Introduction to the thermodynamics of materials. 4th ed. New York: Taylor & Francis; 2003. [34] Yu M, Liu Y, Sellmyer DJ. J Appl Phys 2000;87:6959. [35] Cullity BD, Graham CD. Introduction to magnetic materials. 2nd ed. Hoboken, NJ: Wiley-IEEE Press; 2008. [36] Li CK, Se´guin FH, Frenje JA, Rygg JR, Petrasso RD, Town RPJ, et al. Phys Rev Lett 2006;97:135003.
ScienceDirect Acta Materialia 61 (2013) 7924–7930 www.elsevier.com/locate/actamat
Pulsed-laser fabrication of gas-filled hollow Co–Pt nanospheres S. Sturm a,⇑, K.Z. Rozman a, B. Markoli b, N.S. Antonakakis c, E. Sarantopoulou c, Z. Kollia c, A.C. Cefalas c, S. Kobe a a Department for Nanostructured Materials, Jozef Stefan Institute, 1000 Ljubljana, Slovenia Faculty of Natural Sciences and Engineering, University of Ljubljana, 1000 Ljubljana, Slovenia c Theoretical and Physical Chemistry Institute, National Hellenic Research Foundation, 11635 Athens, Greece b
Received 22 May 2013; received in revised form 16 September 2013; accepted 18 September 2013 Available online 8 October 2013
Abstract We report on nitrogen-filled hollow Co–Pt nanospheres produced via pulsed-laser ablation in ambient nitrogen gas. The resulting nanospheres are characterized by a single-crystalline face-centred cubic Co55±3Pt45±3 shell and a void filled with molecular nitrogen, typically occupying the sphere’s central region. The average diameter of the spheres and the voids is 35 ± 8 and 16 ± 2 nm, respectively. The calculated number density of nitrogen atoms, measured within these voids, is 1.58 ± 0.4 nm3. The resulting pressure in the voids near ambient temperature (300 K) and at the boiling temperature for the Co–Pt alloy (3000 K) is estimated to be 1.9 ± 0.3 and 34.3 ± 9 MPa, respectively. The gas-filled Co–Pt hollow spheres are formed in only one step involving two physical processes. First, after each laser pulse, the vaporized, supersaturated Co–Pt ablated species are condensated in the plume under high pressure and temperature, resulting in nitrogen gas trapping. Between two laser pulses, the pressure and temperature in the plume drop rapidly, the nitrogen-rich liquid nanospheres become thermodynamically unstable and the nitrogen gas bubble starts to expand until the solidification of the nanospheres. The fast solidification of the solid shell prevents further outward diffusion of nitrogen and thus an amount of nitrogen gas is preserved in the void. These nanospheres have the potential in biomedical, magnetic and catalytic applications. Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Pulsed-laser ablation; Hollow-nanospheres; Nitrogen; TEM; EELS
1. Introduction Structures with hollow interiors, such as hollow nanospheres, have recently received considerable scientific attention owing to their unique properties, which could facilitate breakthrough applications in various fields of nanoscience [1–5]. In view of this, various synthesis methods for the fabrication of hollow nanostructures have been proposed that, to some extent, enable control over the properties of the shell. A large number of reported synthesis procedures are based on using prefabricated nanostructured materials as templates, which are subsequently followed by various chemical transformations, such as diffusion, oxidation, ion exchange and galvanic replacement, resulting in the ⇑ Corresponding author. Tel.: +386 1477 3418; fax: +386 1477 3221.
E-mail address: [email protected] (S. Sturm).
formation of hollow structures [6–9]. As an example, the synthesis of hollow nanostructures from their solid counterparts can be achieved using the Kirkendall effect, where the voids are created by the accumulation of vacancies resulting from different ion mobilities. Based on the Kirkendall effect, different binary metal–oxide or metal–sulfide nanospheres can be formed via a two-step synthesis that involves the formation of a metal particle, which is then followed by subsequent oxidation processing [10,11]. A different template-directed synthesis approach was proposed recently, where hollow nanoparticles were successfully fabricated by a so-called bubble synthesis, which uses an electrochemically evolved hydrogen bubble as a template and a reducing agent for electroless deposition [12]. Although these methods have proved to be successful in the synthesis of hollow nanostructures, their major drawback is the restriction related to the choice of materials,
1359-6454/$36.00 Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.actamat.2013.09.033
S. Sturm et al. / Acta Materialia 61 (2013) 7924–7930
which is often limited to simple solids. For this reason, only a few reports are available on hollow, binary alloy nanoparticles. Given these challenges, physical processing routes like pulsed-laser ablation (PLA) in the presence of a background gas have tremendous potential for applications because they offer flexibility in terms of the choice of materials to be ablated, and at the same time the ability to produce various complex nanostructures [13]. It was shown previously that Sm–Fe–Ta-based hollow nanospheres can be produced via a single-step procedure by the ablation of a Sm–Fe–Ta-based alloy target into an ambient nitrogen gas. The formation of hollow nanospheres, typically with molecular nitrogen occupying the central void of the sphere, was explained by a nanoscale melt-solidification phenomenon [14]. The Co–Pt system is a promising system covering areas of magnetism, and catalysis, with the possibility to fine tune the physical properties by adjusting both the composition and the morphology of nanoparticles [15–17]. In this study we have demonstrated that by applying PLA in an ambient nitrogen gas the gas-filled hollow Co–Pt nanospheres can be successfully produced, where the composition of the particles is controlled by the Co–Pt target composition. By means of various techniques of transmission electron microscopy we aimed to characterize both structure and composition of synthesised nanospheres as well as to determine the nitrogen pressure inside the individual voids. The data were further employed for reconstruction of the formation mechanism of Co–Pt gas-filled nanospheres, and suggestions towards the general formation mechanism for gas-filled nanospheres in other metallic systems have been made. 2. Materials and methods 2.1. Materials The PLA apparatus consisted of a molecular fluorine laser at 157 nm and a stainless steel vacuum chamber, placed in a computerized X–Y–Z–h micro-translation stage. A freshly polished Co–Pt alloy target with the composition Co48Pt52 was positioned perpendicular to the laser beam and focused with a high-quality CaF2 lens through a 5 mm aperture in a compact geometry. The distance between the target and the substrate was fixed at 1 cm. The deposition parameters were 30 mJ, a 3–5 J cm2 fluence (600 lm laser spot size on the target) at 15 Hz. The nanoparticles were deposited in high-purity nitrogen at 0.1 MPa. The nanoparticles were deposited directly on a lacy-carbon copper grid. The latter deposition procedure allowed a direct investigation of the obtained product by the use of transmission electron microscopy (TEM). 2.2. Methods For the detailed structural and compositional investigations of hollow nanospheres a field-emission (scanning)
7925
transmission electron microscope TEM/STEM (JEM2010F) operated at 200 keV (Cs = 0.48 mm) equipped with a high-angle annular dark-field (HAADF) detector, energy-dispersive X-ray spectroscopy (EDXS), (LINK ISIS EDS 300) and electron energy-loss spectroscopy (EELS) (Gatan PEELS 766) was applied. The probe size in the STEM was 2 nm, while the HAADF collection angle was set between 60 and 160 mrad. The EELS spectra were background-subtracted, and corrected for the dark current and the channel-to-channel gain variation of the photodiode detector. The full width at half maximum (FWHM) of the zero-loss line was 1.2 eV. Prior to the quantification of the EELS spectra the effect of the plural scattering was removed by applying the Fourier-ratio deconvolution method [18]. The number density of the nitrogen atoms in the voids of the nanospheres was determined by a modified procedure initially proposed for measuring the density of He atoms in nanometre-sized bubbles in steels using a combined STEM–EELS technique [19,20]. The following equation gives the number density of nitrogen atoms (nN) in the void: nN ¼ I N =ðrN I ZL dÞ
ð1Þ
IN and IZL are the intensities measured from the N–K ionization edge and the zero-loss peak, respectively. The corresponding intensities were integrated over an energy window (DE) of 8 eV. The intensity of the N–K edge was measured starting from the ionization edge onset. rN is the angle-integrated hydrogenic cross-section for the nitrogen K-shell ionization, calculated for the experimental collection angles. d represents the measured diameter of the void. This equation assumes that the void diameter is significantly larger than the electron beam and the measurement was performed in the central region of the void, providing localized information on an individual void. The error values related to the accuracy in the determination of the cross-section, the N–K signal and the void diameter were estimated to be 10%. The total error of the calculated density, applying standard error-propagation relations, amounts to 17%. 3. Results 3.1. Crystal structure and composition analysis Nanoparticles with a high spherical morphology, hereafter referred to as nanospheres, were produced by the direct ablation of the CoPt target on a lacy-carbon copper grid, as shown in a bright field (BF) TEM image (Fig. 1). The selected-area electron-diffraction (SAED) pattern presented in the inset picture in Fig. 1a, which was taken from the observed specimen area, confirms that the spheres are crystalline. Their crystal structure was determined by comparing the experimental and calculated SAED patterns. The experimental and corresponding rotational averaged SAED pattern matches with the calculated diffraction pattern, based on the Co50Pt50 face-centred cubic (fcc) phase
7926
S. Sturm et al. / Acta Materialia 61 (2013) 7924–7930
Fig. 1. BF-TEM image of PLA products using a CoPt target in a nitrogen atmosphere is characterized by (a) nanospheres often organized in a necklacelike structure. The corresponding experimental and calculated SAED patterns indicate the fcc structure. (b) Spheres were found to be single crystalline, as shown by a superimposed NBED pattern obtained from the marked circular region. Some spheres contained planar defects (marked by an arrow).
Fig. 2. (a) HRTEM image of representative spheres characterized by a diffuse spherical contrast located in their central regions. The corresponding image detail marked by a rectangle shows (b) lattice fringes combined with a diffuse round contrast. No structural difference could be observed between the central and outer regions of the sphere, as indicated from the corresponding FFTs.
with a cell parameter a = 0.384 nm [21]. A representative nanobeam electron-diffraction (NBED) pattern, acquired from individual spheres, proves that they are single crystalline (Fig. 1b). In addition, the NBED pattern matches with the Co50Pt50 fcc phase viewed in the [0 0 1] zone axis. Some spheres contained planar defects (marked by an arrow in Fig. 1b), i.e. stacking faults or twins. They are most probably caused by internal stresses due to the high cooling rates and are typical for metals with the fcc structure [22]. A detailed image analysis revealed one morphologically distinctive group of spheres, which were characterized by a strong spherical contrast located in their central regions. A large number of such spheres were regularly observed in the analysed samples, as shown in a high-resolution TEM (HRTEM) image (Fig. 2). The enlarged central section of such a sphere (marked by the rectangle in Fig. 2a), besides crystal fringes, which extend over the whole analysed area, also reveals a background diffuse spherical contrast (Fig. 2b). The corresponding fast Fourier transforms (FFTs), shown as insets in Fig. 2b, were calculated from the central (top) and border (bottom) regions of the image, respectively. Identical reflection points are observed in both the FFT images, indicating that there are no significant changes in the crystal structure and/or crystal orientation
between the inner and outer regions of the sphere. In order to unambiguously interpret the origin of the observed contrast, HAADF-STEM combined with EDXS and EELS analyses were employed. BF-TEM and HAADF-STEM images were acquired from an identical string of spheres and are shown in Fig. 3. A large number of the spheres observed in the BF-TEM image show a unique contrast feature located in their central regions, which corresponds perfectly with the dark spherical contrast observed in the HAADF-STEM image. The darker contrast in the HAADF-STEM image could signify regions of large compositional differences. However, the compositional fluctuations that were measured from these spheres could not explain such a dramatic drop in the measured intensity. Fig. 4a shows typical spheres with a dark, spherical contrast. The corresponding normalized intensity profile, obtained from the corresponding HAADF-STEM image, is shown in Fig. 4b. The compositional difference between the central region and the adjacent shell region, i.e. CoCENTRAL/CoSHELL at.% ratio, was 1.05. The exact compositions of the central and the shell regions were Co57Pt43 and Co54Pt46, respectively. However, such small compositional differences cannot explain the almost 30% drop in the intensity measured from the intensity profile.
S. Sturm et al. / Acta Materialia 61 (2013) 7924–7930
7927
Fig. 3. (a) BF-TEM image obtained from a string of spheres. The majority of spheres have a round, bright contrast in their central regions. (b) HAADFSTEM image of the same specimen detail. The round, dark contrast in the sphere’s central regions coincides with the previously described bright contrast found in the TEM images.
Fig. 4. (a) HAADF-STEM image of hollow spheres and (b) the normalized intensity line profile obtained from the HAADF-STEM image (x–x0 line) showing an almost 30% drop in intensity in the sphere’s central region, which indicates the presence of a void.
According to the HAADF-STEM image analysis and EDXS results, the dark, round contrast is most probably due to the presence of voids within the nanospheres. The overall average composition deduced from the eight spheres was Co55±3Pt45±3. The analysed sphere’s diameter ranged between 24 and 41 nm, with a mean value and standard deviation of 34 ± 5 nm. No significant correlation could be found between their sizes and the composition. 3.2. Characterization of nitrogen in voids by EELS The light-element concentrations, N and O in this case, were measured using EELS. Fig. 5a shows a typical EELS spectrum (I) acquired in the spectrum region of both the N–K and O–K ionization edges. The representative spheres, which were analysed, are shown as the inset in Fig. 5a. The spatially resolved EELS analyses performed in the void and in the shell region consistently show the presence of nitrogen only in the voids. The backgroundsubtracted N–K ionization edge is shown in (II). In order to unequivocally identify the presence of the edge signals
a second energy-difference method was applied, which is especially useful for exposing a weak signal superimposed on a slowly varying background [18]. The corresponding second-difference spectra from the void and the shell region are shown in (III) and (IV), respectively. This analysis additionally confirms that the N signal is associated only with the voids, while no signal can be measured in the shell region. Moreover, the presence of oxygen could not be detected, which indicates that the nanospheres remained in the un-oxidized state. To additionally increase the signal-to-noise ratio, the N–K edges acquired from the voids of the analysed spheres were normalized, subsequently realigned and summed together. Fig. 5b shows the resulting N–K edge, which is presented together with the N–K edges obtained from the previously studied Sm–Fe(Ta) –N system [14] and the reference standard spectrum of N2 gas obtained from air [23]. The fine structure for all these N–K edges is distinctive for molecular nitrogen and is characterized by a sharply peaked edge at 401 eV, followed by a broad continuum. Following these results, the nanospheres with the indicated spherical dark contrast, which
7928
S. Sturm et al. / Acta Materialia 61 (2013) 7924–7930
Fig. 5. (a) Raw and processed EEL spectra obtained from the representative hollow spheres (shown in the inset). (I) Raw EEL spectrum acquired from the central region of the sphere shows the presence of a weak N signal. (II) Background-subtracted EEL spectrum obtained from spectrum I, clearly showing the N–K ionization edge. Second difference spectrum confirms that nitrogen is always related to the sphere’s central region (III). In contrast, spectrum IV, which is characterized only by noise, confirms the absence of nitrogen in the outer regions of the sphere, i.e. below the detection limits. The grey area in (II) represents the integrated signal used for calculating the number density of nitrogen atoms in the voids. (b) Fine structure in the N–K edges obtained from the Co–Pt and Sm–Fe–Ta systems and air are distinctive for molecular nitrogen. The EEL spectra have been translated vertically for visual clarity.
was regularly observed in HAADF-STEM images, can be defined as hollow spheres filled with gas or as nanospheres with gas bubbles. 3.3. Determination of nitrogen pressure in voids The number density of nitrogen atoms (n) within the void was calculated by applying Eq. (1) (see Section 2.2). The void diameter was measured from the HAADF-STEM images, by drawing the intensity line profile. The diameters of the measured voids ranged between 15 nm and 19 nm, with a mean value and relatively narrow standard deviation of 16 ± 2 nm. The diameters of the corresponding spheres ranged between 24 nm and 45 nm, with a mean value and a higher standard deviation of 35 ± 8 nm. The resulting calculated number density of nitrogen atoms obtained from six measured spheres was between 1 and 2 nm3, with an average value of n = 1.58 ± 0.4 nm3. The pressure in the void was estimated from the values of the nitrogen density and the related void volume by using a suitable equation of state. To calculate the pressure in the void accurately, with respect to the given density and the temperature range, a standard correction of an ideal gas law using a virial expansion was applied, as follows: 2
P ¼ nkT ð1 þ nB=N A þ n2 C=ðN A Þ þ . . .Þ
ð2Þ
The symbol k represents the Boltzmann constant, T is the absolute temperature and NA is the Avogadro constant, while B and C are the second and third virial coefficients, respectively. The values of the B and C coefficients for nitrogen gas can be found in the work of Sevast’yanov and Chernyavskaya [24] for temperatures from 75 K up to 2500 K. The resulting calculated average pressure in the voids near the ambient temperature (300 K) was 1.9 ± 0.3 MPa. Assuming that the voids filled with gas were formed while the spheres were still in the liquid phase the equilibrium pressure needs to be recalculated for a
given temperature range. According to the literature data [25], the boiling temperature for the Co–Pt alloy at ambient pressure is 3000 K. The corresponding pressure of the nitrogen in the voids at this temperature would increase to a value of 34.3 ± 9 MPa. 4. Discussion The gas-filled hollow spheres presented in this study were most probably formed via a nucleation-and-growth process. The condensation of such particles starts already within the ablated vapour, with vapour species undergoing enough collisions with the background gas for them to nucleate and grow. Consequently, these particles would exhibit a homogenous composition enriched with the more volatile component, due to preferential ablation, which would be size-independent. On the other hand, the particles that are expelled from the melt would be enriched with the less volatile component, since they experience a loss of the more volatile component on their path towards the substrate [26]. This consequently leads to deposits in which the composition depends upon the particle size [27]. For the Co–Pt system, the corresponding vaporization enthalpies (DHvap) of the Co and Pt are 377 and 469 kJ mol1, respectively. According to this, we can expect that the plume will be enriched with Co, which results in Co-rich nanoparticles, as was already observed by Trelenberg et al. [28]. This is also in agreement with our findings where Co-rich nanospheres, with an average composition of Co55±3Pt45±3, were deposited from the Co-deficient target (Co48Pt52), confirming that they have been formed directly from the vapour phase, via the already described evaporation–condensation mechanism. It was reported that the background gas influences the plasma kinetics and, consequently, the formation of the nanoparticles. Observations like plume splitting and plasma deceleration were related to the appearance of the
S. Sturm et al. / Acta Materialia 61 (2013) 7924–7930
shock wave and a contact front, where the shock wave is caused by the sudden impact of the expanding plasma with the background gas, resulting in a propagating perturbation with the contact front establishing the border between this perturbed gas and the plasma [29,30]. The heterogeneous nucleation and growth of the nanoparticles within the plume is bound to this narrow shock-wave region of the highly compressed, supersaturated vapour mixture, which is composed of the ablated vapour and the background gas, with large temperature and density gradients as well as pressure differences [31]. The vapour pressure and the temperature just behind the plasma front can be estimated using the general laws of the conservation of mass, momentum and energy [32]. The estimated vapour pressures of the Co–Pt plasma expanding into the nitrogen gas at 0.1 MPa would be in the range of 100 MPa to several GPa, while the vapour temperatures would range between 103 and 104 K [31,32]. Immediately after the condensed particles are ejected from the high-pressure and high-temperature plasma region, the particle nucleation and growth are heavily suppressed as a result of the rapid drop in pressure and temperature [31]. As a result the dissolved nitrogen in the molten sphere will start to condense and form an initial nitrogen bubble. The amount of dissolved nitrogen in the Co–Pt at the boiling point and ambient pressure was estimated using Sievert’s law for liquid metals. The calculated amount of dissolved nitrogen in a liquid Co–Pt sphere for given temperature and pressure conditions is 14 at.%. It is worth mentioning that the solubility of nitrogen in a molten sphere at a higher pressure and temperature would be even greater. For comparison, the nitrogen solubility in solid Co at 1873 K and ambient pressure is only 0.3 at.%, while in Pt it can be considered as negligible [25]. This is not a surprising result if we consider that prior to the condensation of the Co–Pt–N particles, the temperature and pressure in the plume can briefly reach well over 104 K and several GPa, respectively, which is enough to supersaturate the Co–Pt with nitrogen. The properties of the condensed nanoparticles, such as the composition, the crystal structure and the morphology, are largely dependent on
7929
the interplay between the thermodynamics of the two-phase region, i.e. the liquid–vapour, the rules of solidification for the multi-component system and the kinetic constraints during rapid cooling. Namely, the time scale of these processes is in the range of tenths of a second, which is based on the fact that all the examined nanospheres had a crystalline structure. This implies that the cooling rates must have been in the range from 102 to 103 K s1 or less. Had they been higher than this, the spheres would be in an amorphous state [33]. The nitrogen bubble pressure (P) within the molten Co–Pt sphere is equilibrated over the surface tension (!) of the Co–Pt liquid following the relation P = 2!/r, where r stands for the radius of the nitrogen bubble. The calculated nitrogen pressure in bubbles with an experimentally measured diameter of 32 ± 1.6 nm would yield a value of 2 102 MPa. The fact that the measured pressure in the voids was one order of magnitude lower suggests that the cooling rates were such that it gave the nitrogen sufficient time to partially diffuse from the nanosphere. As the temperature of the molten sphere is dropping due to cooling (primarily through conduction), the permeability of Co–Pt to nitrogen is also dropping exponentially, while the interdiffusion of Co and Pt is severely hindered. For instance, the interdiffusion coefficient for Co and Pt is D1537K ffi 4 1011 cm2 s1, while the diffusion coefficient for nitrogen at the same temperature is much higher, with D1537K ffi 0.9 cm2 s1. Moreover, the preferential diffusion of nitrogen towards the centre of the sphere is promoted by the fact that the solubility of nitrogen in the solid (0.3 at.%) is much lower than in the liquid (14 at.%). This is in accordance with our results, as the nitrogen signal was only measured in the void, while no nitrogen was detected in the rim of the sphere. The arrangement of the Co–Pt spheres in string-type structures is due to the magnetic dipole–dipole interaction between the particles. Depending on the particle volume, the magnetization can be fixed within an individual particle if the volume exceeds a certain critical volume [34]. For Co–Pt with a fcc crystal structure the calculated critical diameter equals 15 nm, which means that the particles
Fig. 6. The formation mechanism for gas-filled hollow metallic nanospheres produced by PLA in a nitrogen-rich atmosphere in sequences indicating: (I) condensation of particles in the high P–T plasma region, (II) the formation of nitrogen bubble in liquid sphere, (III) solidification and (IV) the final solid sphere with nitrogen-filled void.
7930
S. Sturm et al. / Acta Materialia 61 (2013) 7924–7930
produced in our study are ferromagnetic. They are most probably in a single-domain state, since the calculated critical diameter of the single-domain particle for the Co–Pt fcc crystal structure equals 30 nm [35]. Moreover, spontaneous magnetic fields with a magnetic flux density in the range of kilogauss were already observed in laser-produced plasma due to the large temperature gradients, which generate thermoelectric currents [36]. Such magnetically induced self-assembly, which originates from the magnetostatic interactions, can yield unique structures from strings to necklaces, as observed in our study. According to the obtained results we propose the formation mechanism for gas-filled hollow metallic nanospheres, which can be generalized for other metallic systems (Fig. 6). At first, the condensation of the spheres starts from the supersaturated mixture of ablated species and ambient gas at a relatively high pressure and temperature. The melt needs to be acceptable for gas intake, as this provides a molten sphere with the highest concentration of dissolved nitrogen. Then, the metallic system in relation to the background nitrogen gas should be selected so that the high solubility differences between the melt and the corresponding solid are achieved. Finally, the outwards diffusion of nitrogen through the sphere’s solid rim, for example, by providing high cooling rates, should be suppressed. 5. Conclusions The Co–Pt system was chosen as an example to verify and explain the formation mechanism for gas-filled hollow metallic spheres produced by PLA in the presence of an ambient nitrogen gas. The presented results confirm that nitrogen-filled Co–Pt spheres can be fabricated successfully. The spheres were formed by condensation from the plume in a region of high pressure and temperature. We believe that the nitrogen bubble within the liquid nanosphere was initially formed due to the abrupt drop of pressure and temperature in the nitrogen-saturated liquid sphere. Subsequently, the nitrogen could diffuse from the interior of the sphere outwards. However, due to the fast solidification and the formation of the solid rim the outward diffusion of nitrogen is heavily suppressed, preventing the N2 gas from escaping the sphere. The resulting solid sphere is characterized by a void filled with nitrogen gas. The obtained results support the idea that gas-filled hollow spheres could be fabricated in various complex metallic systems by applying PLA in the presence of a background gas, taking into consideration that in relation to the background gas high solubility differences between the melt and corresponding solids are achieved. Acknowledgements This work was financially supported by the Ministry of Higher Education, Science and Technology of the Republic
of Slovenia and by the European Union as part of the Framework 7 program under a contract for an Integrated Infrastructure Initiative, Reference ESTEEM2. References [1] Chen HM, Liu RS. J Phys Chem 2011;115:3513. [2] Shankar SS, Deka S. Sci Adv Mater 2011;3:169. [3] Shan ZW, Adesso G, Cabot A, Sherburne MP, Syed Asif SA, Warren OL, et al. Nat Mater 2008;7:947. [4] Zhang JZ, Noguez C. Plasmonics 2008;3:127. [5] Botella P, Ortega I, Quesada M, Madrigal RF, Muniesa C, Fimia A, et al. Dalton Trans 2012;41:9286. [6] Vasquez Y, Henkes AE, Chris Bauer J, Schaak RE. J Solid State Chem 2008;181:1509. [7] Zhang Q, Wang W, Goebl J, Yin Y. Nano Today 2009;4:494. [8] Jones MR, Osberg KD, MacFarlane RJ, Langille MR, Mirkin CA. Chem Rev 2011;111:3736. [9] Moon GD, Ko S, Min Y, Zeng J, Xia Y, Jeong U. Nano Today 2011;6:186. [10] Yin Y, Rioux RM, Erdonmez CK, Hughes S, Somorjal GA, Alivisatos AP. Science 2004;304:711. [11] Iba´n˜ez M, Fan J, Li W, Cadavid D, Nafria R, Carrete A, et al. Chem Mater 2011;23:3095. [12] Huang C, Li YJ, Muangphat C, Hao Y. Electrochim Acta 2011;56:8319. [13] Swihart MT. Curr Opin Colloid Interface Sci 2003;8:127. [14] Sturm S, Rozman KZ, Markoli B, Sarantopoulou E, Kollia Z, Cefalas AC, et al. Nanotechnology 2010;21:485603. [15] Chen G, Xia D, Nie Z, Wang Z, Wang L, Zhang L, et al. Chem Mater 2007;19:1840. [16] Vasquez Y, Sra AK, Schaak RE. J Am Chem Soc 2005;127:12504. [17] Sun Q, Wang S, Wang R. J Phys Chem C 2012;116:5352. [18] Egerton RF. Electron energy loss spectroscopy. 2nd ed. New York: Plenum; 1996. [19] Walsh CA, Yuan J, Brown LM. Philos Mag A 2000;80:1507. [20] Fre´chard S, Walls M, Kociak M, Chevalier JP, Henry J, Gorse D. J Nucl Mater 2009;393:102. [21] Leroux C, Cadeville MC, Pierron-Bohnes V, Inden G, Hinz F. J Phys F: Met Phys 1988;18:2033. [22] Smallman RE, Bishop RJ. Metals and materials, science, processes, applications. Oxford: Butterworth-Heinemann; 1995. [23] Ahn CC, Krivanek OL, Burgner RP, Disco MM, Swann PR, EELS Atlas. Arizona State University HREM Facility/Gatan Inc.; 1983. [24] Sevast’yanov RM, Chernyavskaya RA. Inzh-Fiz Zh 1986;51:124. [25] Gale WF, Totemeir TC, editors. Smithells metals reference book. Burlington, MA: Butterworth-Heinemann; 2004. [26] Hergenro¨der R. J Anal At Spectrom 2006;21:1016. [27] Chen LC. In: Chrisey DB, Hubler GK, editors. Pulsed laser deposition of thin films. New York: Wiley; 1994. [28] Trelenberg TW, Dinh LN, Stuart BC, Balooch M. Appl Surf Sci 2004;229:268. [29] Horn I, Gu¨nther D. Appl Surf Sci 2003;207:144. [30] De Posada E, Arronte MA, Ponce L, Rodrı´guez E, Flores T, Lunney JG. J Phys: Conf Ser 2011;274:012078. [31] Sharma AK, Thareja RK. Appl Surf Sci 2005;243:68. [32] Zel’dovich YB, Raizer YP. In: Hayes WD, Probstein RF, editors. Physics of shock waves and high-temperature hydrodynamic phenomena. New York: Dover; 2002. [33] Gaskell DR. Introduction to the thermodynamics of materials. 4th ed. New York: Taylor & Francis; 2003. [34] Yu M, Liu Y, Sellmyer DJ. J Appl Phys 2000;87:6959. [35] Cullity BD, Graham CD. Introduction to magnetic materials. 2nd ed. Hoboken, NJ: Wiley-IEEE Press; 2008. [36] Li CK, Se´guin FH, Frenje JA, Rygg JR, Petrasso RD, Town RPJ, et al. Phys Rev Lett 2006;97:135003.