Article pubs.acs.org/JPCA
Experimental Study of Thermal Stability of Thin Nanowires Eugene B. Gordon,*,† Alexander V. Karabulin,‡ Vladimir I. Matyushenko,§ and Igor I. Khodos∥ †
Institute of Problems of Chemical Physics RAS, Academician Semenov Avenue, 1, 142432 Chernogolovka, Moscow Region, Russian Federation ‡ National Research Nuclear University MEPhI (Moscow Engineering Physics Institute), Kashirskoe shosse, 31, 115409 Moscow, Russian Federation § The Branch of Talrose Institute for Energy Problems of Chemical Physics RAS, Academician Semenov Avenue, 1/10, 142432 Chernogolovka, Moscow Region, Russian Federation ∥ Institute of Microelectronics Technology and High Purity Materials RAS, Institutskaya Street, 6, 142432 Chernogolovka, Moscow Region, Russian Federation ABSTRACT: Thin (D < 10 nm) nanowires are in principle promising for their application as catalysts and as elements of nanocomputers and quantum devices. To perform these tasks, their structure and properties must be stable at least at standard conditions. Using our technique based on the capture of small particles to the core of quantized vortices in superfluid helium, we synthesized nanowires made of various metals and alloys and investigated their thermal stability. The indium nanowires (D = 8 nm) were shown to be stable when heated to 100 °C, i.e., almost to the melting point, whereas the silver nanowires (D = 5 nm) disintegrated into traces of individual nanoclusters at 300 K. The gold and platinum nanowires also decomposed at temperatures more than twice as low as the melting point. A model is proposed to explain the premature decay of thin nanowires by unfreezing of the surface-atom mobility in combination with the anomalous dependence of the surface tension on the nanowire radius. Methods for improving the stability limits of thin nanowires by saturation of their surface with immobilized atoms as well as by surface oxidation have been proposed and experimentally tested.
1. INTRODUCTION In their studies started in 1995, Markku Räsänen and his team for the first time successfully used the ability of a lowtemperature matrix to serve as a template that dictates the structure of products of chemical reactions performed in these matrices. Thus, a whole family of noble-gas (Ng) hydrides was produced in noble-gas matrices. These molecules exhibit quite a specific nature of bonding, which leads to unusual chemical and spectroscopic properties (recent reviews on this issue can be found in refs 1 and 2). A natural desire of the authors was to get these unusual compounds in a concentrated form, in particular in the form of noble-gas molecular aggregates (HNgY)n or HNgY crystals. The HXeCCH crystalline compound had been predicted.3 In addition, efforts to achieve true gas-phase identification of the HNgY molecules are now in progress. On the other hand, the rigidity of low-temperature matrices hinders the mobility of most reagents (with the fortunate exception of the H, D, and F atoms) and reaction products. Therefore, the contact reactions predominantly take place. Nevertheless, there is the unique case in which the long thin template providing for a collinear reaction pathway exists at low temperature which allows the free movement of the reactants along its core. Such a template represents the quasi-1D quantized vortices arising under almost any excitation of liquid helium in the superfluid state and existing for a long time. © XXXX American Chemical Society
Initially, this was demonstrated by experiments on embedding molecular hydrogen into the liquid helium.4,5 It was found that hydrogen condensation proceeds completely differently in the normal and superfluid states of the liquid, the latter appearing under cooling below 2.2 K. In the normal liquid helium, the condensation products are only spherical micron-sized particles, whereas in the superfluid helium the condensation proceeds much faster and its products are long filaments, which behave like quantized vortices.4,5 Because any impurity particles have affinity to the vortex core being significant for low temperatures, and because their motion along the vortex core is allowed, this unusual condensation mechanism should always exist. This is why we proposed to use such an approach to grow the nanofilaments from any impurity.5 The quantized vortex is actually a 1D−structure because the diameter of its core is comparable to the size of the atom and the length may be many centimeters. For this reason, we even were afraid that the threads formed in such a way would be too thin to be detected and studied. Fortunately, it turned out that, in the most practically interesting case of metallic nanowires, the filaments produced in the superfluid helium were not so thin: the Special Issue: Markku Räsänen Festschrift Received: August 30, 2014 Revised: November 6, 2014
A
dx.doi.org/10.1021/jp5087834 | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
and mechanical properties. In particular, it was unknown whether such thin metallic nanowires are stable at normal temperatures regardless of the method of their production. Furthermore, it is important to know which metal nanowires are not corroded in ambient air. In our study, we paid particular attention to the preparation and study of the metal nanowebs suitable for use as catalysts of specific chemical reactions. It has been observed for a long time that the mechanical properties of materials at the nanoscale are often different from those at the macroscopic scale. At nanometer scales, because of the increasing surface-to-volume ratio, the surface effects become predominant and can significantly modify the macroscopic properties.13 In this regard, the elucidation of the temperature stability of nanowires is of both applied and fundamental interest. Using our experimental approach, we grow bundles of long, thin nanowires from almost any metal or alloy. All nanowires in their bundle are of almost the same diameter, but this obvious advantage entails quite a significant drawback. Indeed, because the diameter of the nanowire grown by the condensation in vortices in the superfluid helium is governed by the thermal properties of the metal used,6 the nanowires made of a given material are characterized by a certain diameter, which cannot be changed. Therefore, when comparing the properties of nanowires of various metals, we often have to compare nanowires with significantly different diameters. The natural upper limit of the temperature stability of nanoobjects with various dimensions (clusters, films, or wires) is their melting point, which is different from the melting temperature of a bulk sample because of the curvature of the surface. Both calculations and experiments have convincingly shown that the melting size effect, although appreciable, typically diminishes the melting point by no more than 10− 15%. A good estimate is obtained by the formula14
nanowire diameter dependent on the thermal properties of the metal varied from 8 nm for the low melting indium6 to 3 nm for refractory platinum.7 This is due to the fact that even in He(II), which has extremely high thermal conductivity, the coalescence of small metal clusters leads to their self-melting, which resulted in the spherical shape of the products of merging. Starting at only a certain size, the metal clusters begin to coalesce into nanowires. In our experiments, the product of metal embedding into He(II) by laser ablation was a nanoweb wherein the nanowires were interconnected in a metallic way that provided the whole nanoweb electrical conductivity.8 For nonmetals, in principle, the production of superfine threads can be expected in the superfluid helium. There may be chains of atoms or molecules, such as those studied by Räsänen et al. The main difficulty in searching for such structures is the impossibility of studying them by electron microscopy. The optical methods developed by the Helsinki group seem to be more appropriate, and the publication of the paper in this issue can be regarded as our proposal for cooperation in this area. A significant feature of our approach is its versatility: unlike the quasi-0D nanospheres and quasi-2D nanofilms, a universal method of production of the quasi-1D nanowires has not yet been developed. It is also important that the thickness of our nanowires (which is, by the way, in most cases less than that of the nanowires grown by other methods) is suitable for a number of practical applications. For example, their surface curvature was serendipitously close to that optimal for manifestation of the nanocatalytic properties;9 superconductive nanowires made of refractory metals were thin enough to manifest the effect of complete suppression of superconductivity due to phase slip, which makes them promising for use in the logical elements of quantum computers, in 1D superconducting quantum interference device magnetometers, and in other quantum devices.10,11 Our nanowires possess a regular structure and perfect shape because they are grown from the premelted clusters. Furthermore, as an advantage of the experiments in the liquid helium, the study of their properties is in the absence of any contamination both during the growth of the nanowires and because the measurements can be performed inside the cryostat, thereby avoiding exposure to the ambient air, and, if necessary, can be performed immediately at low temperatures. We produced the bundles of nanowires of many metals and alloys in amounts sufficient for studying not only their physical but also their chemical properties.12,7 An important factor for chemical applications is the material resistance to the external factors, first of all to heating. The most interesting application for physical chemistry is the use of a nanoweb as a catalyst of chemical reactions. Catalysis by nanoparticles has developed to date into an independent and promising field of chemistry.9 The use of free-standing nanowires instead of nanoparticles deposited onto a substrate is attractive from several points of view: this makes it possible to eliminate the uncontrolled influence of the support on the process and also allows the optimization of the process by applying an electric field because, already at voltages of several tens of volts, the nanowire surface becomes capable of serving as a cold electron emitter. However, the nanowires were rarely used in nanocatalysis probably because the optimal diameter of the catalyst nanoparticle is less than 5 nm. The production of nanowires with free surface of such a diameter is hardly possible by other methods. Possible practical applications of nanowires have called attention to the problems of stability of their shape, structure,
Tmw = Tmb(1 − αd /D), α = 4/3 for nanowires
(1)
where Tmw and Tmb are the melting points of nanowires and the bulk material, respectively, and d and D are the diameters of atom and wire, respectively. However, mostly nanoparticles and nanofilms have been investigated experimentally, whereas nanowires have been paid much less attention. Our studies of nanowires of low-melting metals, primarily indium,6,12 support the above conclusion. We reinvestigated the bundles of the indium nanowires produced 6 months ago. As can be seen from Figure 1a they remained practically unchanged during the storage in air at room temperature (i.e., at the temperature equal to 70% of its melting temperature), and the indium nanospheres showed a similar stability (Figure 1b). This is why we believed that the nanowires made of much more refractory metals, which are promising for applications, will be stable at least at ambient temperature. However, quite unexpectedly, the experiment showed that the silver nanowires disintegrated at 300 K despite the fact that silver is much more refractory than indium.12 Furthermore, it turned out that the gold nanowires deposited onto a glass surface also decayed into individual clusters during prolonged storage. Moreover, it was observed that the length of any metal nanowires shortens appreciably under heating long before room temperature is reached. On the basis of the literature data and our results, a possible mechanism was proposed for the nanowire decomposition at temperatures several times lower than the melting temperature of the material.12 In the present B
dx.doi.org/10.1021/jp5087834 | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
holes regularly spaced on the surface were used. The simultaneous monitoring of the transformation of the nanowebs both precipitated onto the surface and spanning over the holes made it possible to examine the initial steps of the decay caused by heating without the substrate influence as well as the final stages of the nanowire collapse; however, this examination was complicated in the latter case by the nanowire material wetting of the surface carbon coating.
3. MODEL OF THIN NANOWIRE LOW-TEMPERATURE DECAY For melting of an atomic crystal, the atom bulk mobility must be sufficient to enable atoms to move over distances much longer than the lattice constant. In contrast, most atoms in nanowires are located at or near the surface, where much less energy is required to allow atomic movement. For an object as small as a nanowire, the replacement of one layer of atoms for the distance of few nanometers is sufficient to change the nanowire shape significantly. Energy considerations show that this can occur at temperatures much lower than the bulk melting point. However, the change in shape can occur only if the surface atom mobility is not stochastic; the simplest case is when the process of relaxation to equilibrium changes the shape. Because the surface tension of thin nanowires provides for the major contribution to their energy,13 the equilibrium shape of the wires of a given length, with the coefficient of the surface tension being independent of the wires diameter, is cylindrical. Hence, the surface mobility can result in neither disintegration of the wire to peas nor even formation of a bead-like structure. The surface mobility takes place in nanospheres as well, but it cannot change the spherical shape in principle. However, the nanowires have some peculiarity in their surface tension. It has been recently observed experimentally16 just for the silver nanowires, demonstrating such unusually low stability in our experiments, as well as for the lead nanowires. The measured elastic modulus, proportional to the surface tension coefficient, was independent of nanowire diameter only down to 50 nm, then it began to increase up to a factor of 2 at d = 30 nm (see inset in Figure 2). Unfortunately, thinner wires were not studied. It is more or less clear, however, that the surface tension should become small for very thin wires because the interaction between the surface atoms ceases to be metallic and their binding energy, as well as the surface tension, drops. It seems clear from the results for lead that the surface tension coefficient (ξ) passes over a maximum at small diameter d of the nanowire. Let us assume that the dependence of ξ on d has the form shown in Figure 2. Then, the equilibrium shape corresponding to the energy minimum will depend on the initial diameter of the cylindrical nanowire. In region A, where ξ = const, the cylindrical shape will of course be energetically favorable. However, in region B, where ξ increases with d decreasing, the equilibrium shape is peapod with a period of about d. For a wire with diameter smaller than that corresponding to the maximal surface tension, the decrease of both ξ and the wire perimeter with d decreasing forces the surface atoms to move from the areas of nodes to the antinodes. As a result, the nanowire with a d value belonging to region C should be metastable with respect to the decay into individual nanoclusters. An increase in temperature leads to the removal of metastability. Of course, because the surface tension is the major contributor to the energy of the nanowire, the relaxation of its shape always tends
Figure 1. Indium nanostructures after 6-month storage: (a) nanowires and (b) ball of coagulated (not fused) 6 nm nanospheres.
paper, we describe experiments crucial for testing this hypothetical mechanism as well as propose and experimentally test the ways to improve the stability of the nanowires of interest.
2. EXPERIMENTAL SECTION The experimental setup has been described elsewhere.6,15 It was assembled on the base of an optical helium cryostat. Lowering the temperature of liquid helium was carried out by pumping out its vapor to a pressure of 700 Pa, which corresponded to a temperature of 1.55 K. The atoms and small metal clusters were embedded into the superfluid helium by laser ablation of metal targets immersed into He(II). The experimental cell was mounted at the bottom of the liquid helium bath and represented a stainless steel tray with a vertical rod, to which a metal foil 0.1−0.5 mm thick, serving as a target, was fixed. The solid-state Nd:LSB pulse-repetition laser with diode pumping used for metal ablation had the following characteristics: λ = 1.062 μ, E = 0.1 mJ, τ = 0.5 ns, and f = 500−4000 Hz. The laser beam entered the cryostat through the sapphire windows and was focused on the target surface; the spot size was 50−100 μm. The vertical row of gold-plated electrodes separated by a distance of 3 or 1.4 mm was placed near the target. They represented the commercial electric connectors used in the PC card. The nanowires formed in the quantized vortices of the superfluid helium from the material embedded into the fluid by laser ablation of the targets immersed into helium were either pinned to the tops of the electrodes introduced into the reaction zone or precipitated to the bottom of the cell. In the first case, the bundles of nanowires closed an electrical circuit, and it was possible to measure their resistance to electric current at different temperatures. The standard grids for transmission electron microscopy (TEM) were placed at the bottom of the cell. After the cryostat was heated to room temperature, the grids with deposited nanowires were moved into the vacuum chamber of the TEM instrument. Unfortunately, their contact with air could not be avoided during this action. Carbon-coated copper or gold grids with micron-sized C
dx.doi.org/10.1021/jp5087834 | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
Figure 3. Photo of the bundles of indium nanowires grown in the cryostat after a 5 min laser ablation. Separations between electrodes are 3 mm. Figure 2. Plots regarding the explanation of a possible cause for changes of the equilibrium shape of thin nanowires. Inset: the dependence of elastic modulus on diameter for silver and lead nanowires; data taken from Figure 1 of ref 16.
connections between individual nanowires in the nanoweb. Electrically, the nanoweb is a parallel−series connection of an unknown large number of nanowires. Therefore, the absolute resistivity of individual nanowires could not be determined; we had to use only the two-wire method, which did not enable us to separate the resistances of the conductors and the contacts between them. Nevertheless, the contact resistance can be ignored on the basis of the following experimental fact. In the study of bundles of superconducting nanowires, the resistance drops by several orders of magnitude on transition to the superconducting state, where the measured resistance is only the contact resistance.7,17 The nanowires in the web are quite firmly stuck together. They do not separate when the liquid−gas boundary moves across the specimen during helium evaporation or later under the influence of turbulent flows in the helium gas. In the present study, to determine the effect of oxidation of the platinum nanowires in air on their conductivity, we were able to replace helium in the cryostat at normal pressure and T = 300 K with atmospheric air for 1 month. The pass-through conductivity remained unchanged after this treatment. However, as Figure 4 demonstrates, when the nanoweb moves from the liquid helium (panel a) to the helium gas (panel b), it visually shrinks just as, and for the same reason as, the painter’s brush shrinks when it is pulled out of the water (see Figure 4). Furthermore, as is seen in Figure 5 for gold (panel a) and platinum (panel b) as examples, the electric resistance of the bundle drops by 5−10% after its passage from liquid to gas, presumably because of increasing degree of percolation via extra contacts between the nanowires at bundle shrinking. Both effects, the changes in the visible shape and in the measured resistance, were reversible when the specimen moved back and forth between liquid and gas. It follows that our nanowebs are not rigid cross-linked structures, at least at low temperatures. This is quite natural because the vortex rings generated by cavitation, which is initiated by pulsed heating in the laser focus and further evolved into entangled quantized vortices, are far from being straight lines, and it is difficult to find a reason for the wire straightening during its growth. The temperature dependence of the nanowire resistivity was studied in many papers,18−20 the results of which can be
to reduce its length. If the ends of the nanowire are fixed, it is pulled, and when the threshold of plasticity is overcome, it breaks down. The nature of the break is different in regions A, B, and C. The thick nanowires, like common wires, retain nearly cylindrical shape and tear at the weakest spot. The thin nanowires decay into a set of short units. However, most importantly, the wire in region C disintegrates at temperatures 2−3 times lower than the melting temperature, whereas the wire in region A can survive up to the melting point. Of course, the model based on Figure 2 is oversimplified. The most important shortcoming of the model is that it is based on the experimental measurements of ξ at constant temperature; therefore, it cannot predict what else happens, in addition to the increase in the surface mobility, when the temperature increases. Because the surface tension coefficient in metals increases with temperature and the surface smoothness of the thin nanowire enhancing the surface tension decreases, one can suppose that the curve of Figure 2 will rise and its maximum will shift to higher d with increasing temperature.
4. EVOLUTION OF THE NANOWEB MORPHOLOGY AND STRUCTURE ON HEATING FROM 1.6 TO 300 K The amazing ability of the nanowires grown in the superfluid helium to firmly pin to the tips of the electrodes placed in the condensation zone is of great help in such kinds of studies.5 This pinning ability is due to the affinity of the quantized vortices, inside which the nanowires grow, to the vessel heterogeneities. At long laser exposition, the bundles of nanowires grow so large that they are visible with the naked eye. Figure 3 shows that, despite the fact that the electrodes are parallel to each other, the nanowires prefer to grow mainly at their tips. The observed electrical conductivity between adjacent electrodes is characterized by a linear volt−ampere characteristic independent of the polarity even at low voltage of about 0.1 V, which indicates the metallic nature of the D
dx.doi.org/10.1021/jp5087834 | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
becomes dominant. For the wires a few nanometers thick, the transition to this regime occurs above room temperature.7 However, the effective resistance of the nanowire bundles measured by us for many metals and alloys behaved quite differently. Typical examples of the temperature dependence of the resistance of bundles of the nanowires made of some metals and alloys, all grown in the superfluid helium, obtained by us for a variety of metals and alloys in the present and previous papers are shown in Figure 6. In all cases the slow increase of the effective resistance with temperature turned to a steep growth at a temperature of about 260 K, and is approximately the same for all metals, which often leads to complete circuit disconnection. Because the condensation products surviving at the bottom of the cell after heating to T = 300 K represented strained filaments, this irreversible increase of the resistance was explained in our paper7 by partial rupture of the nanowires in the web due to their increasing tension leading to decreasing degree of percolation and, as a consequence, to increasing the resistance. However, if we began to cool the cryostat slowly by pouring liquid nitrogen into the nitrogen jacket at any T ≥ 250 K, the temperature dependence of the resistance became weak and reversible (under subsequent heating), reflecting the actual dependence of the resistivity of individual nanowires on T.7 When the specimen was heated further, the irreversible increase of resistance was restored just at the same temperature at which the preceding cooling of the specimen had started. Surprisingly, the nanoweb “remembers” exactly, sometimes within a few degrees, the annealing temperature. The processes of heating and cooling were conducted very slowly (more than 10 h per each step). Therefore, this “memory” seems to reflect the existence for a given individual nanowire of well-defined temperature at which the nanowire breaks, possibly dependent upon its thickness, length, shape, and bead inclusions. This fact supports our model according to which the destruction of the nanowires occurs because of changes in their shape accompanied by their shortening. To test the model, we carried out special experiments aimed at clarifying the nature of the break; either it is due to the tension accumulated along the entire length of the bundle or each individual wire breaks independently (see Figure 7). In the same experiment, we compare the temperature dependences of the resistance of the nanowire bundles grown simultaneously between the standard vertically arranged electrodes separated by 1.4 mm and either (Figure 7a) between vertical electrodes with the gap reduced to 70 μm or (Figure 7b) between the electrodes separated by 1.4 mm but deviated from the vertical. In the first case, the length of the bundle was 20-fold reduced; in the second, the bundle was not tensioned by gravity during its growth, rather it became slack, taking the form of “catenary”. All these changes had no visible influence on the losses of percolation under heating, as is seen in Figure 7. The local nature of the rupture, which occurs independently in each nanowire according to our model of thermal evolution of the nanowire shape, is a consequence of the transition from the cylindrical nanowire equilibrium shape to the peapod one with relatively small distances between the peas.
Figure 4. A nanowire bundle shrinking on the liquid helium level passage (a) in liquid and (b) in gas.
Figure 5. Change in resistance of the nanowire bundles after passage of the level of the liquid helium across the specimen: (a) gold and (b) platinum.
summarized as follows. At high temperature, the wire resistance, as that of the thicker wires, is due to the scattering of electrons by phonons, and it increases ∼T. At low temperature, the scattering of electrons by the surface of the wire (or the grain), which is nearly independent of temperature, E
dx.doi.org/10.1021/jp5087834 | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
Figure 6. Temperature dependence of the effective resistance for 1.4 mm bundles of nanowires: (a) tin, (b) indium, (c) gold, (d) platinum, (e) silver, (f) silver−copper (58:42) alloy, (g) indium−lead (88:12) alloy, and (h) niobium.
5. INFLUENCE OF SILVER DOPING BY COPPER ON THE MORPHOLOGY, STRUCTURE, AND STABILITY OF NANOWEBS Unexpectedly low stability of the silver nanoweb at 300 K (ref 12) stimulated more detailed studies of this metal. Indeed, with the obvious exception of mercury, the nanowires deposited on the grid for all metals including such low-melting metals as indium and tin were stable under normal conditions. First of all, it appeared that if we carried out the electron-microscopic studies immediately, a few hours after heating the specimen to room temperature, it was possible to observe the silver nanowires too (see Figure 8). Although the nanowire consisted entirely of beads, it was still a single whole because it also existed in the grid holes, as can be seen in Figure 8a. The period of the peapod structure, though not very constant, amounted to tens of nanometers (see Figure 8b). In addition to the nanowires, chains of clusters, the remains of the nanowires destroyed, were also seen on the carbon surface. If the
specimen was kept at room temperature for 2−3 days, the webs located over the grid holes disappeared and only “fox tracks” of the silver nanoclusters rather than the nanowires were seen on the carbon surface.12 The gold melting point (1064 °C) is higher than that of silver, and the gold nanowires deposited on a carbon-coated films were stable at room temperature for a long time. However, being deposited on glass, they disintegrated into separate clusters, like silver wires, in a few days at standard conditions (Figure 9a). In the left-hand side of Figure 9b, the number of deposited nanowires is so large that they do not adhere tightly to the glass surface; rather, they form layers that are not on the surface. As is clearly seen, the wires preserve their integrity there. A similar picture, while less pronounced, is observed with the fresh silver nanowebs. This observation suggests that wetting the surface (glass in the present case) by the metal stimulates nanowire decomposition. F
dx.doi.org/10.1021/jp5087834 | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
Figure 7. Effects of reducing the spacing between the electrodes and of a deviation of their tips from the vertical arrangement on the temperature dependence of the resistance of nanowire bundles grown between the electrodes. For niobium (a), the spacing between the vertically arranged electrodes is 1.4 mm (trace 1) or 70 μm (trace 2). For the Ag58Cu42 alloy (b), the tip of the lower electrode deviates from vertical (trace 1) by 45° (trace 2). Figure 9. Reflecting electron microscopy (REM) images of golden nanowires deposited onto the glass plate: (a) traces of golden clusters and (b) the devastating effect of the glass surface. The nanowires in the left part did not adhere tightly to the glass and therefore survived.
copper. The choice of copper as dopant was motivated both by the fact that the copper nanowires were completely stable at standard conditions21,12 and by physical-chemistry considerations. According to literature data, the addition of copper to silver even improves the catalytic activity of the wires.22 At the same time, such experiments are of fundamental interest, revealing the mechanism of growth of thin alloy nanowires. There are two scenarios for nanowire growing in the superfluid helium by sticking together preliminarily molten nanoclusters, which can be explained on the silver−copper alloy phase diagram shown in Figure 10. If the elemental composition of the melt fits to the existence of α or β phases, which correspond to small fractions of copper (Ag:Cu ≥ 87:13) or silver (Ag:Cu ≤ 4:96), respectively, then solid solution will be formed after cooling and the wire composition must be uniform. Outside the regions of α and β phases, the original cluster must disintegrate upon solidification to crystallites enriched by silver (α phase) or copper (β phase). The nanoscale adds some specificity to the above considerations. The phase diagram for the surface layers, whose energy is mainly determined by the surface tension, must be different from that for the bulk shown in Figure 10. As a result, the nanowire surface should be enriched by atoms of one component as compared to the bulk. As shown in ref 24, the same effect takes place for the nanoparticles. In addition to saturation of the nanowire surface by atoms of one component, another effect very interesting for applications is expected for the alloys with compositions close to eutectic, namely, the separation into the nanocrystals of α and β phases in the axial direction. Period of alternation of the regions with different
Figure 8. TEM image of silver nanoweb deposited on the surface of carbon-coated grid with 2 μm holes just after its heating to room temperature.
Poor thermal stability severely limits practical applications of thin nanowires. The existence of the low-temperature (without melting) decay channel via the surface migration of atoms appears mortally dangerous for the prospects of using the gold and especially silver nanowires in nanocatalysis, where extremely thin, D ≤ 5 nm, nanowires with a free surface, which are stable at fairly high temperatures, are required. However, because the surface mobility of atoms plays the main role in the nanowire decay, it is possible to try to fill up the surface by atoms of a less mobile metal. To improve the stability of the silver nanowebs, we decided to dope silver with G
dx.doi.org/10.1021/jp5087834 | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
should be followed by separation into the solid solutions rich in silver (α-phase) or copper (β-phase). Therefore, it is natural to interpret the structure of the alloy nanowires shown in Figure 11(f) as alternating crystalline regions rich in silver or copper. A more simple behavior should take place in those rare cases in which the components of the alloy can form solid solutions in arbitrary proportions. We observed such a behavior of the indium−lead alloy, which is known to satisfy this condition25 (Figure 12). The web morphology in this case is typical (Figure
Figure 10. Phase diagram for silver−copper alloy.23 Solid vertical line, eutectics; dashed lines, the alloys investigated in the present work.
composition should be close to the wire diameter. Nanowire heterostructures for various purposes could be created in this way. Our experiments with alloys demonstrate these effects.12 Indeed, the silver nanowires are completely decayed at 300 K, Figure 11a,b, whereas the copper nanowires survived the heating to room temperature, Figure 11c,d. In the meantime, as seen from Figure 11e,f, adding copper to silver improves the thermal stability of the nanowires. Although the numerous breaks of individual nanowires are seen in these micrographs, the web consisting of them still represents a whole and does not sink into the holes on the grid. Even with a small (14%) content of copper in silver nanowires, a few atomic layers on its surface are oxidized. This effect is consistent with the conclusion of ref 24 about strong enrichment of the surface layers of the nanoparticles made of Ag:Cu = 97.5:2.5 alloy by copper atoms and their rapid further oxidation. The highresolution image of Ag:Cu nanowire demonstrates a wellexpressed interference structure evidencing crystalline ordering. Because the 58:42 alloy is quasi-eutectic, its solidification
Figure 12. TEM image of structure and morphology of the bundles of nanowires made of In:Pb = 88:12 superconducting alloy.
12a), and the inner structure of the In:Pb=88:12 nanowire formed in the superfluid helium is uniform, which is also evidenced by the temperature of the nanowire transition to the superconducting state of 5.0 K which is according to our measurements and is close to the 4.5 K measured for the bulk
Figure 11. Morphology and structure of sediments on carbon-coated copper grid under laser ablation of various targets in superfluid helium: (a, b) Ag, (c, d) Cu, and (e, f) Ag:Cu = 58:42. Light halo around the nanowires corresponds to copper oxide formed during the contact with air. Diameter of holes in the grid is 2.5 μm. H
dx.doi.org/10.1021/jp5087834 | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
Figure 13. Indium nanowires: (a) bundle morphology, (b) structure of an individual nanowire, and changes of (c) morphology and (d) structure after annealing at 120 °C. Diameter of holes in the grid is 3 μm.
Figure 14. Gold nanowires: (a) bundle morphology; (b) structure of an individual nanowire; changes of (c) morphology and (d) structure after annealing at 200 °C; (e) and (f), same as (c) and (d) at T = 500 °C. Inset, the result of nanowire disintegration by focused electron beam in TEM chamber. The diameter of holes in the grid is 2.5 μm.
samples.25 It is also interesting that this nanowire was not oxidized in air whereas the lead nanowires were completely oxidized during their transfer into the chamber of the electron microscope. The results of these studies are valid for any thin nanowires regardless of how they have been obtained and modified. In our opinion, they are of great importance. On the one hand, they show how to prepare thin nanowires stable at reasonably high temperature or with respect to surface oxidation. On the other
hand, they pave the way for construction of unique nanoheterostructures for various applications.
6. EVOLUTION OF THE NANOWEB MORPHOLOGY AND STRUCTURE UNDER HEATING ABOVE T = 300 K Unfortunately, we could not know directly which was the structure and morphology of the nanowire bundles at the time I
dx.doi.org/10.1021/jp5087834 | J. Phys. Chem. A XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry A
Article
Figure 15. Platinum nanowires: (a) bundle morphology; (b) structure of individual nanowire; changes of (c) morphology and (d) structure after annealing to 500 °C. Diameter of the holes is 3 μm.
of their formation and how these characteristics changed in the process of heating to room temperature. However, we could observe subsequent transformations of the structure and shape of nanowires when they were heated beyond 300 K. In this way, it was possible to determine the limits of stability for many nanowires, to observe how they decayed, and eventually to try restoring the whole picture of their thermal transformation including extrapolation of the high-temperature data to low temperatures. Because of the distortion of the image of our electron microscope by the current of the sample heater and the uncertainties in the actual specimen temperature, we could not perform thermal studies inside the TEM. Therefore, the study was carried out as follows. Grids with nanowire bundles deposited thereupon were slowly heated in a vacuum furnace to a desired temperature and kept at this temperature for a long period of time. Next, they were slowly cooled to room temperature in vacuum and thereafter were transferred to the TEM vacuum chamber, contacting with the air already being cooled to 300 K. On the basis of the results described above (see Figure 6), we believed that their TEM images corresponded to the nanowires with the structure and the shape stable at the temperature of their maximal heating. The evolution of the structure and the shape of the nanowires of different metals under their annealing is shown in Figures 13−15. The indium nanowires at room temperature, which is 70% of its melting point, are stable for many days (Figure 13 a,b), and notable metamorphoses occurs only when they are annealed at 120 °C, where the absolute temperature is 91% of indium melting point (see Figure 13c,d). The shape of the nanowires is already distorted, but no beads are noticeable yet. Moreover, they are still cross-linked to the web, and some of them remain stretched over the holes in the grids. Interestingly, after heating to 120 °C, the spherical clusters, present in sediments, lose their regular spherical shape because of destruction by large negative internal pressure.26 As already mentioned, the gold nanowires were stable at room temperature for many days but only when deposited on the carbon film (Figure 14a,b). When stored on glass, they decayed in chains of clusters (see Figure 9). The annealing of gold nanoweb deposited on the carbon-coated grid to 200 °C,
i.e., to the temperature constituting only 35% of the absolute melting temperature, already caused significant changes (Figure 14c,d). Numerous breaks appeared at this temperature, and the resulting clusters were similar to those formed when the golden web was destroyed in the TEM vacuum chamber under irradiation by the focused electron beam. The nanowire stretched over the grid holes partially collapsed, although fragments of interconnected nanowires were preserved. Heating to only 500 °C, which is 58% of the melting temperature, led to the complete disappearance of the nanoweb in the holes and to complete destruction of the deposited carbon nanowires into short fragments and “fox tracks” of individual clusters similar to those in the case of silver (Figure 14e,f). As seen in Figure 15, at room temperature the platinum nanoweb is perfectly stable both on the grid surface and over the holes (Figure 15a,b). However, annealing the sample at 500 °C caused nearly full disappearance of the nanowires stretched over the holes and quite a significant decay of the nanowires deposited onto the grid surface (Figure 15c,d). The data obtained on the stability of the nanowebs made of various metals are summarized in Table 1. They can be briefly Table 1. Experimentally Estimated Upper Limits of Temperature Stability (Tdw) of Nanowires from Different Metals Formed in He(II) by Laser Ablation in Comparison with Their Melting Temperatures (Tmw) Calculated by Equation 114 diameter, D (nm) melting temperature, Tmb (K) Tmw/Tmb, from eq 1 Tdw/Tmb, experiment
indium
silver
gold
platinum
8 430 0.95 0.9
5 1235 0.92 0.24
4 1337 0.92
