Advanced Materials

25th Anniversary Article: Galvanic Replacement: A Simple and Versatile Route to Hollow Nanostructures with Tunable and Well-Controlled Properties


  • Xiaohu Xia,

    1. The Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University, Atlanta, Georgia, USA
    Search for more papers by this author
  • Yi Wang,

    1. The Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University, Atlanta, Georgia, USA
    Search for more papers by this author
  • Aleksey Ruditskiy,

    1. School of Chemistry and Biochemistry, School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Atlanta, Georgia, USA
    Search for more papers by this author
  • Younan Xia

    Corresponding author
    1. The Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University, Atlanta, Georgia, USA
    2. School of Chemistry and Biochemistry, School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Atlanta, Georgia, USA
    Search for more papers by this author


This article provides a progress report on the use of galvanic replacement for generating complex hollow nanostructures with tunable and well-controlled properties. We begin with a brief account of the mechanistic understanding of galvanic replacement, specifically focused on its ability to engineer the properties of metal nanostructures in terms of size, composition, structure, shape, and morphology. We then discuss a number of important concepts involved in galvanic replacement, including the facet selectivity involved in the dissolution and deposition of metals, the impacts of alloying and dealloying on the structure and morphology of the final products, and methods for promoting or preventing a galvanic replacement reaction. We also illustrate how the capability of galvanic replacement can be enhanced to fabricate nanomaterials with complex structures and/or compositions by coupling with other processes such as co-reduction and the Kirkendall effect. Finally, we highlight the use of such novel metal nanostructures fabricated via galvanic replacement for applications ranging from catalysis to plasmonics and biomedical research, and conclude with remarks on prospective future directions.

1 Introduction

Metal nanostructures have received ever increasing interest over the past several decades owing to their remarkable properties and intriguing applications in many areas such as catalysis, sensing, imaging, and biomedicine.[1-13] We can tailor the properties of a metal nanostructure by adjusting a set of physical parameters, including composition, size, shape, and internal structure (e.g., solid vs. hollow).[14-20] For a specific application, the performance of metal nanostructures can often be optimized by controlling any one of these parameters, or a combination of them. Of the many methods that have been demonstrated for achieving such controls, galvanic replacement offers a particularly effective and versatile approach due to its abilities to control the size and shape, and to tune the composition, internal structure, and morphology of the resultant nanostructures.[21-23]

Galvanic replacement is an electrochemical process that involves the oxidation of one metal (which is often referred to as a sacrificial template) by the ions of another metal having a higher reduction potential. Upon contact in a solution phase, the template will be oxidized and dissolved into the solution while the ions of the second metal will be reduced and plated onto the outer surface of the template. This simple reaction can be employed to generate a wide variety of metal nanostructures, and is limited by no more than the requirement of a favorable difference in the reduction potentials of the two metals.[21-23] The power of galvanic replacement in engineering the properties of metal nanostructures can be understood from the following major aspects: i) Composition. Elemental compositions of the final products can be tuned by reacting the template with different amounts of salt precursors (one type or more than one). When more than one type of metal ions are involved in the galvanic replacement, the spatial distributions of different metals in the final product can also be controlled by simply altering the order of addition for the salt precursors.[24] ii) Internal structure. In general, the oxidation and dissolution of atoms from the template tend to occur from specific sites of the template due to the difference in chemical reactivity and/or the shielding of certain regions by a layer of the second metal deposited during galvanic replacement. As a result, the final structure is typically a porous shell, whose thickness can be easily controlled by varying the amount of precursor(s) added into the reaction solution. iii) Morphology. Since the newly formed atoms are deposited on the surface of the template, the final product typically possesses a shape closely resembling that of the original template with a slight increase in dimensions. Overall, galvanic replacement offers a facile and versatile route to a variety of advanced multifunctional nanostructures often characterized by tightly controlled sizes and shapes, hollow interiors, porous walls, and tunable elemental compositions.[21-23]

The first example of galvanic replacement involving nanoscale objects was reported by our group in 2002 for the preparation of Au hollow nanostructures from Ag nanoparticles.[25] Over the past decade or so, we and other research groups have successfully extended this approach to many other combinations of metals and nanoscale templates in the exploration of various new applications.[26-37] The major goal of this article is to provide an updated account of the mechanistic understanding of galvanic replacement, as well as its ability to generate complex, multifunctional porous nanostructures. We start with a brief discussion on the mechanism of galvanic replacement, as exemplified by the reaction between Ag nanoparticles and aqueous AuCl4 to generate Au-Ag alloyed nanoshells and nanocages. We then explicitly discuss the selectivity towards different facets on the template by using Ag cuboctahedrons covered by a mix of {111} and {100} facets as a model system. We also discuss in detail how alloying and dealloying, two key processes involved in galvanic replacement, influence the final morphology and internal structure of the nanostructures prepared using this approach. This discussion is further enhanced by two examples in which galvanic replacement is shown to be promoted or prevented by controlling the experimental conditions. In the following section, we highlight the use of galvanic replacement for generating more complex nanostructures by coupling it with other mechanisms such as co-reduction of multiple precursors and the Kirkendall effect. Finally, we present several examples to highlight some of the unique properties and applications for the nanostructures fabricated using galvanic replacement in areas such as catalysis, plasmonics, and biomedicine.

2 Mechanistic Understanding

Galvanic replacement is a redox process, in which a metal is corroded (sacrificed) by the ions of a second metal when they are in contact in a solution phase. A galvanic replacement reaction can be broken down into two half reactions: the oxidation (loss of electrons) or corrosion of the first metal at the anode and the reduction (gain of electrons) of the ions of the second metal, and subsequent deposition, of this metal at the cathode. The driving force for this reaction comes from the difference in reduction potentials of the two metals involved, with the potential of the second metal necessarily being higher than that of the first metal. Table 1 summarizes the standard reduction potentials of metals commonly used in galvanic replacement studies. It should be pointed out that the potentials listed here are for the standard conditions only. Variations in temperature and concentrations of relevant ions, as well as the involvement of other non-standard conditions can all affect the actual value of reduction potential.[38, 39] These changes may reverse the direction of a replacement reaction and result in the termination or prevention of galvanic replacement (see Section 'Promotion and Prevention of Galvanic Replacement'). Please delete “, Supporting Information” following the “Section 'Promotion and Prevention of Galvanic Replacement'”, which appeared in the PDF proof.

Table 1. Reduction potentials of metals relative to the standard hydrogen electrode (SHE)
Reduction reactionE0 (V vs. SHE)a)
  1. a)For ideal conditions at 25 °C and 1 atm.

Co2+ + 2e→ Co–0.28
Cu2+ + 2e→ Cu0.34
Rh3+ + 3e→ Rh0.76
Ag+ + e→ Ag0.80
Pd2+ + 2e→ Pd0.95
Ir3+ + 3e→ Ir1.16
Pt2+ + 2e→ Pt1.18
Au3+ + 3e→ Au1.50

When the galvanic replacement involves a macroscopic object, the reaction can be directly visualized and monitored with the naked eye, and the pertinent mechanistic details can often be found in an introductory chemistry textbook.[40, 41] For example, when a Cu wire is immersed in an aqueous AgNO3 solution (Figure 1A, left), the Cu atoms from the surface of the wire can quickly react with Ag+ ions to release Cu2+ ions into the solution. Simultaneously, Ag+ ions are reduced into Ag atoms and deposited on the surface of the Cu wire. As a result, the initially colorless AgNO3 solution turns blue due to the appearance of Cu2+ ions, while the original Cu wire is increasingly covered by a grey precipitate of Ag (Figure 1A, right). The equations involved in this galvanic replacement reaction can be summarized as follows:

Figure 1.

(A) Photographs of a glass vial containing AgNO3 solution (left) immediately after the insertion of a Cu wire and (right) after the galvanic replacement reaction had proceeded for 10 min. The insets show atomic models used to illustrate the mechanism involved in the galvanic replacement reaction between Cu and AgNO3. (B) Schematic illustration of the morphological and structural changes at different stages of the galvanic replacement reaction between a Ag nanoparticle and HAuCl4 in an aqueous solution.

Half reactions:

display math(1)
display math(2)

Combined reaction:

display math(3)

Galvanic replacement has found extensive use in our everyday life. For instance, the galvanic replacement between Zn and MnO2 has been applied to fabricate low-cost household “carbon-zinc” batteries.[42] Water tanks and pipes composed of multiple metals (e.g., copper and cast iron) are often protected from corrosion by introducing into the system a sacrificial anode made of a metal (e.g., aluminum) with a higher reduction potential. Failure to regularly replace the sacrificial anode may greatly shorten the service lifetime of the appliance.

When the metal object shrinks from macroscopic to nanoscale, galvanic replacement can be used as a simple, versatile, and robust method for generating novel nanostructures. With a Ag nanoparticle as an example, Figure 1B shows the major steps involved in a galvanic replacement reaction with HAuCl4. When an aqueous HAuCl4 solution is added into an aqueous suspension of Ag nanoparticles, galvanic replacement will be initiated immediately at the site with the highest surface energy (e.g., defect, stacking fault, or step).[43-45] As a result, Ag atoms will be oxidized and dissolved into the solution, generating a small hole on the surface of the nanoparticle. At the same time, the electrons will quickly migrate to the surface of the nanoparticle and be captured by AuCl4 to generate Au atoms via a reduction reaction. The newly formed Au atoms tend to be deposited epitaxially on the surface of the Ag nanoparticle (step 1 in Figure 1B) due to a good match between Ag and Au in terms of both crystal structure (face-centered cubic, or fcc) and lattice constant (4.086 vs. 4.078 Å for Ag and Au, respectively).[25, 46] The deposition will lead to the formation of a thin and incomplete layer of Au on the surface of each nanoparticle, which can prevent the underneath Ag from reacting with AuCl4. As a result, the small hole will serve as the primary site for continuous dissolution of Ag. The opening will also allow all the species involved in the reaction (e.g., Ag+ and AuCl4) to diffuse in and out of the cavity.[46] Accompanying the deposition of Au, alloying will occur with the underlying Ag (step 2 in Figure 1B) because a homogeneous alloy is thermodynamically more stable than a mixture of segregated Au and Ag.[47] In the end, complete dissolution of pure Ag from the template transforms the nanoparticle into a nanostructure characterized by a hollow interior and an alloyed shell (step 3 in Figure 1B). It is worth noting that the small hole should have been sealed at this point, creating a shell characterized by a smooth, hole-free surface. The disappearance of small holes can be attributed to the continuous deposition of Au on the surface and/or a mass-transport process, such as lateral diffusion of Au atoms across the surface.[46, 48] In addition, the shell thickness and void size depend not only on the size of the Ag nanoparticle (sacrificial template), but also on the stoichiometric ratio between Ag and Au3+. In comparison with Au3+, the reaction between a Ag nanoparticle and the same amount of Au+ ions will generate a Au-Ag shell with a thicker wall since every three Ag atoms dissolved will result in the reduction of three Au+ ions rather than one Au3+ ion.[49, 50] If more HAuCl4 is added into the reaction system, the AuCl4 will cause dealloying for the Au-Ag shells by selectively removing Ag atoms from the alloyed shells.[51-54] During the course of dealloying, many lattice vacancies will be generated as the Ag atoms are extracted by AuCl4, leading to an increase in the surface free energy.[55] To make up this energy penalty, the vacancies coalesce, generating small holes in the shells.[56] Further dealloying enlarges the hole size, thus generating hollow nanostructures with porous walls (step 4 in Figure 1B), which are commonly referred to as nanocages.[21] Complete dealloying typically causes the nanocages to collapse into small fragments of pure Au (step 5 in Figure 1B). The driving force for a galvanic replacement reaction tends to drop as the reaction proceeds due to the involvement of alloying and thus the change in reduction potential for atoms in the template. It should be pointed out that the mechanistic details for galvanic replacement are specific to neither particle shape nor material, and should be applicable to all different combinations of sacrificial templates and metal ions as long as there is an appropriate difference in reduction potentials between the two metals involved.[26-37]

3 Facet Selectivity

For a sacrificial template covered by only one type of facet, both dissolution of template and deposition of the newly formed atoms are expected to take place on the same type of facet. Examples include the galvanic replacement between HAuCl4 and Ag nanocubes with sharp corners.[46, 57] Since each Ag nanocube is enclosed by six {100} facets, both dissolution of Ag and deposition of Au should occur on the {100} facets, resulting in the formation of Au-Ag nanocages with pores randomly distributed on the side faces. A similar trend was also observed for Ag octahedrons solely covered by eight {111} facets.[45, 58] When more than one type of facets are present on the surface of a template, galvanic replacement can proceed with facet selectivity,[59, 60] allowing for rational design and engineering of the final nanostructure. In this case, galvanic replacement tends to start from the facet(s) with the highest surface free energy, whereas the deposition of newly formed atoms preferentially occurs on the remaining facets with lower surface free energies.[43-45]

It has been established that the surface free energies of low-index facets of an fcc noble-metal crystal decrease in the order of γ{110}{100}{111}.[61, 62] When a clean polyhedron with no capping agent on the surface is employed as the sacrificial template for a galvanic replacement reaction, the dissolution of atoms should begin from the {110} and {100} facets due to their higher surface free energy, while deposition of the newly formed atoms should occur on the {111} facets regardless of the area and shape of the facets.[60] In the solution-phase synthesis of nanocrystals, however, a facet-specific capping agent is often intentionally introduced into the reaction system to promote the formation of desired facets.[63-68] As such, the surface of a nanocrystal is often covered by the capping agent. The adsorption of a capping agent can alter the surface free energies of various facets and even reverse their order. Therefore, the presence of a capping agent has to be taken into consideration when predicting the facet to be involved in the initiation of a galvanic replacement reaction. For instance, poly(vinyl pyrrolidone) (PVP) can specifically bind to the {100} facets of Ag nanocrystals, resulting in γ{100}{111}.[62, 69] When 42-nm Ag cuboctahedrons, enclosed by six {100} and eight {111} facets, were used as sacrificial templates for the galvanic replacement reaction with HAuCl4, the dissolution of Ag atoms began from the {111} facets, whereas the deposition of Au atoms occurred on the {100} facets covered by PVP (Figure 2A).[23] Unlike the cases when Ag nanocubes and octahedrons were used as the sacrificial templates, the nanocages produced from the Ag cuboctahedrons (Figure 2B) showed uniform pores in terms of both size and shape positioned at the eight corners. Figure 2C shows a typical TEM image. Such well-defined porous structures are attractive for drug delivery applications owning to their well-defined pores (see Section 'Biomedical Research'). In addition to the change in structure, the facet selectivity of a galvanic replacement reaction can also have an impact on the overall morphology of the final product when the size of the template is reduced. For example, when Ag cuboctahedrons of 24 nm in size were used as the templates, the products underwent a shape transformation into octahedrons.[70] This transformation from cuboctahedral to octahedral shape can be attributed to both the selective deposition of Au on {100} facets, and the relatively small size of the templates.

Figure 2.

(A) Schematic illustration of the morphological and structural changes at different stages of the galvanic replacement reaction between a Ag cuboctahedron and HAuCl4 solution. Note that the dissolution of Ag atoms takes place on the eight {111} facets, while the deposition of Au occurs on the {100} facets. (B) SEM and TEM (inset) images of the Ag cuboctahedrons used for the galvanic replacement reaction with HAuCl4 solution. (C) SEM and TEM (inset) images of the Au-Ag nanocages synthesized from the Ag cuboctahedrons in (B). Scale bars in both insets are 50 nm. Reproduced with permission.[23] Copyright 2010 Elsevier.

Despite these interesting results, the facet selectivity of a galvanic replacement reaction is yet to be systematically investigated. The impacts of other capping agents on different metals and their facets, including citrate for Ag{111} and hexadecylamine for Cu{100},[66, 71] are worthy of a thorough investigation in the future. In addition to the capping agents, other factors such as reaction kinetics,[72] the presence of twin defects,[73-75] and involvement of surface diffusion[76] can all affect the facet selectivity of galvanic replacement and thus deserve careful study.

4 Alloying vs. Dealloying

Alloying and dealloying are two important processes involved in a galvanic replacement reaction;[77] both of them have significant impacts on the structure and morphology of the final product.

4.1 Alloying

Alloying is involved in the initial stage of a galvanic replacement reaction when a thin layer of metal A is deposited on the template composed of metal B (as shown by steps 2 and 3 in Figure 1B). The kinetics of alloying obey Fick's second law of diffusion:

display math(4)

where D is the interdiffusion coefficient; C is the atomic concentration of metal A as a function of time (t) and distance (x) from the A-B interface; and Cs is initial concentration of metal A at the A-B interface (x = 0), which is in direct proportion to the number of A atoms deposited on metal B; D is strongly dependent on temperature and the energy barrier to diffusion,[76, 78] and can be described by the Arrhenius equation:

display math(5)

here D0 is the pre-exponential factor; Ediff is the energy barrier to diffusion, which is determined by a set of parameters such as bonding energy and lattice mismatch between the two metals;[76]T is the absolute temperature; and R is the ideal gas constant. Clearly, D has a strong dependence on the temperature at which the galvanic replacement reaction is conducted. For example, the value of D for the interdiffusion of atomic Au through a layer of Ag would increase from 10−24 m2 s−1 to 10−19 m2 s-1 as the temperature increases from room temperature to 100 °C.[79, 80] Collectively, the degree of alloying increases with elevation of reaction temperature, thickening of A layer deposited on the template of metal B, and depression of the energy barrier to diffusion.

In general, alloying is the basis for retaining the morphology of the original template during a galvanic replacement reaction (Figure 3A). If there is no alloying, it will be difficult to retain the morphology of the template. A good example is the galvanic replacement between Ag nanocubes and Pd2+ or Pt2+ ions.[26] Reaction of Ag nanocubes with Na2PdCl4 resulted in the formation of smooth, single-crystal nanoboxes composed of a Pd-Ag alloy (Figure 3B). However, when Na2PdCl4 was replaced by Na2PtCl4, poly-crystalline nanoboxes with bumpy surfaces composed of Pt nanoparticles were formed (Figure 3C). The difference in morphology observed in these two cases arose from the fact that, unlike Au and Pd, Pt does not readily undergo solid-solid interdiffusion over the entire surface of a Ag template to form a conformal Pt-Ag alloy. This difficulty can be attributed to the metal-metal bonding energy. The much higher bonding energy of Pt-Pt (307 kJ mol−1) relative to that of Pt-Ag (218 kJ mol−1) would lead to: i) an island growth pattern for the deposited Pt because the newly formed Pt atoms tend to adhere to each other rather than interacting with the Ag surface;[76] and ii) a relatively larger energy barrier to interdiffusion, i.e., Ediff in Equation (5). As a result, it is difficult to generate a conformal layer of Pt-Ag alloy over the Ag surface. In contrast, the bonding energies of Pd-Pd (100 kJ mol−1) and Au-Au (226 kJ mol−1) are lower than those of Pd-Ag (137 kJ mol−1) and Au-Ag (229 kJ mol−1), respectively,[76, 81, 82] favoring the formation of Pd-Ag and Au-Ag alloys rather than separation into different phases.

Figure 3.

(A) Schematic illustration of the morphological and structural changes involved in the galvanic replacement reaction between a Ag nanocube and Pd2+ or Pt2+ ions. (B) SEM and TEM (inset) images of Pd-Ag nanoboxes prepared from the galvanic replacement reaction between Ag nanocubes and Na2PdCl4 solution. Scale bar in the inset is 40 nm. (C) TEM image of Pt-Ag nanoboxes prepared from the galvanic replacement reaction between Ag nanocubes and Na2PtCl4 solution. (B) and (C) are reproduced with permission.[26] Copyright 2005 American Chemical Society.

4.2 Dealloying

Dealloying plays an important role in the later stages of a galvanic replacement reaction, especially in controlling the porosity of the wall, when the metal with a lower reduction potential is selectively removed from the alloyed wall (as shown by steps 4 and 5 in Figure 1B). In general, dealloying can be accomplished either by adding more of the metal ions involved in the galvanic replacement or by utilizing a conventional wet etchant.[46, 52, 83]

Figure 4 shows a schematic illustration and electron microscopy images of the nanostructures at different stages of the aforementioned dealloying processes. The starting sample was prepared by stopping the galvanic replacement reaction between Ag nanocubes and HAuCl4 at a relatively early stage.[46] The first approach involved the continuous addition of HAuCl4 to the aqueous suspension of Au-Ag nanoboxes.[46] The morphology and structure underwent a number of major changes, including: i) removal of remaining Ag in the interior and closing of pores on the surface, together with corner truncation, to generate hole-free nanoboxes (Figure 4B); ii) generation of nanocages with pores on the surface by extracting Ag atoms from the Au-Ag alloyed walls (Figure 4C); and iii) increase in size of the surface pores and collapse of the Au structure into small particles with irregular shapes (Figure 4D). Notably, in contrast to the case of using Ag cuboctahedrons as templates (Figure 1B), shape reconstruction from cubes to truncated cubes was observed in the early stage of dealloying for templates based on Ag nanocubes. The appearance of truncation at corners can be attributed to a combination of both selective deposition of Au on {100} facets and the inherent driving force to lower the total surface free energy. A detailed discussion about the dealloying mechanism can be found in Section 'Mechanistic Understanding' and Ref. [46].

Figure 4.

(A) Schematic illustration of two different methods for dealloying that involved the use of HAuCl4 and Fe(NO3)3, respectively, as the etchant. (B-D) SEM images of samples obtained by dealloying 120 nm partially hollow Au-Ag nanoboxes with increasing amounts of HAuCl4. The 100 nm scale bar in (D) applies to (B) and (C). (B-D) are reproduced with permission from Ref. [46.] Copyright 2004 American Chemical Society. (E-G) TEM and SEM (insets) images of samples obtained by dealloying 50-nm partially hollow Au-Ag nanoboxes with increasing amounts of Fe(NO3)3. The 50 nm scale bar in (G) applies to (E) and (F). The scale bars in the insets of (E-G) are 50 nm. (E-G) are reproduced with permission.[84] Copyright 2007 American Chemical Society.

In the second approach, Fe(NO3)3 was used as a wet etchant to selectively dissolve Ag from the Au-Ag alloyed nanoboxes.[84] After a small amount of Fe(NO3)3 solution had been introduced, the nanoboxes evolved into nano­cages with porous walls (Figure 4E). As the amount of Fe(NO3)3 increased, the pore size was gradually expanded (Figure 4F). When all of the Ag atoms were removed, the central portion of each nanocage wall disappeared, generating a nanoframe made of pure Au (Figure 4G). Unlike the dealloying with HAuCl4, which involved the concurrent deposition of Au atoms, the reaction between Au-Ag nanoboxes and Fe(NO3)3 was only a simple dealloying process:

display math(6)

In a sense, the use of Fe(NO3)3 can easily separate the dealloying of Ag from the deposition of Au, making the dealloying process, and thus the morphology of the product, better controlled. For example, the degree of porosity of the Au-Ag nano­cages could be easily controlled by simply adjusting the amount of Fe(NO3)3 added (Figure 4, E-G). In contrast, when using HAuCl4, the coupling between Ag dealloying and Au deposition made it difficult to precisely control the wall thickness and porosity. It should be pointed out that, in addition to Fe(NO3)3, other chemicals such as NH4OH and H2O2 can also serve as etchants to dissolve Ag from the alloyed walls.[85, 86] The etching powers of these compounds may differ, leading to variations in the final products even when the same batch of alloyed nanoboxes is used. For instance, when NH4OH, a much weaker etchant relative to Fe(NO3)3,[87] was used to etch Ag from the same batch of Au-Ag nanoboxes, the final product became a mixture of nanoframes and nanocages. This difference can be attributed to the weaker etching power of NH4OH, and thus the incomplete removal of Ag from the Au-Ag alloyed walls.

5 Promotion and Prevention of Galvanic Replacement

5.1 Promotion

In principle, galvanic replacement should occur between any two metals with the favorable reduction potentials. However, the experimental results could be completely different from what we would expect due to the variation in reaction kinetics. An effective means for promoting a galvanic replacement reaction is to introduce an ionic species to facilitate the reaction. In addition to their role to change the reduction potentials of the metal/ion pairs and thus shift the equilibrium towards the direction of galvanic replacement, the ionic species can also increase the reaction kinetics. A notable example can be found in the galvanic replacement reaction between Pd nanocubes and PtCl62− ions as reported in our recent work.[88] On the basis of the reduction potentials (0.74 V vs. SHE for the PtCl62−/Pt pair and 0.62 V vs. SHE for the PdCl42−/Pd pair), galvanic replacement should occur according to the following equation:

display math(7)

However, experimental observations indicated that the reaction could not occur without the introduction of halides into the reaction system. We initially added extra Cl ions to the reaction solution, which were expected to promote the galvanic replacement by shifting the equilibrium to the right according to Equation (7). Our results demonstrated that Cl ions could indeed help initiate the galvanic replacement between Pd and PtCl62−. However, the galvanic replacement induced by Cl showed no selectivity toward different facets of the Pd templates, resulting in the formation of Pd-Pt nanocrystals with an irregular morphology.[88] We then switched to Br ions since they could easily replace the Cl in the Pd and Pt precursors due to much higher stability constants for [PdBr42−] and [PtBr62−] relative to [PdCl42−] and [PtCl62−], respectively.[89, 90] Accordingly, the reduction potentials of PtCl62−/Pt and PdCl42−/Pd were changed significantly by Br ions. The chemical reaction equation became:

display math(8)

where the reduction potential of PtBr62−/Pt (0.61 V vs. SHE) was also more positive than that of PdBr42−/Pd (0.49 V vs. SHE). In addition, the presence of Br ions shifted the equilibrium to the right based on Equation (8), further promoting the galvanic replacement. Owing to the selective adsorption of Br ions on Pd{100} facets,[76, 91] the PtBr62− complex was expected to have a higher concentration in the vicinity of the {100} facets. As a result, Pd atoms on the {100} facets of a nanocube were preferentially oxidized and dissolved by reacting with PtBr62− (Figure 5A), leading to the formation of Pd-Pt concave nanocubes (Figure 5, B-D) and eventually octapods (Figure 5E). In a sense, the binding of Br ions to Pd{100} facets helped induce the galvanic corrosion of Pd by PtBr62− selectively adsorbed on the side faces of the Pd nanocube, with Pt then deposited on the Pd{111} facets located at the corner sites. This result is in contrast with the galvanic replacement reaction between Ag cuboctahedrons and HAuCl4 (Figure 2),[23, 70] in which the reaction started simultaneously from eight {111} facets, while the resultant Au atoms were mainly deposited on {100} facets due to the protection of Ag(100) surface by PVP. It is worth mentioning that this same strategy, based on alteration of reduction potential, was also used to promote the galvanic replacement between Ag and RhCl3 by using I ions.[92] In this case, the reduction potential of Ag was greatly reduced from +0.8 (for Ag+/Ag) to −0.15 V (for AgI/Ag) vs. SHE, thereby creating a significant difference in reduction potential between the Ag template and RhCl3.

Figure 5.

(A) Schematic illustration of the morphological and structural changes involved in the galvanic replacement reaction between PtCl62− ions and a Pd nanocube whose {100} facets were covered by chemisorbed Br ions. (B-E) TEM images of Pd-Pt bimetallic nanocrystals prepared through the galvanic replacement between PtCl62− ions and 18 nm Pd cubes after the reaction had proceeded for different periods of time: (B) 0.5, (C) 4, (D) 9, and (E) 20 h. The insets (scale bars: 10 nm) show TEM images of individual nanocrystals at higher magnifications. Adapted and reproduced with permission.[88] Copyright 2011 American Chemical Society.

5.2 Prevention

Sometimes, it is also necessary to prevent the galvanic replacement reaction from occurring, especially when the replacement reaction is too vigorous to be controlled. In general, reducing the difference in redox potentials is an effective method for the prevention of galvanic replacement. For example, Yin and co-workers demonstrated the synthesis of highly stable Ag@Au core-shell nanoplates (Figure 6A) by depositing a thin and uniform layer of Au on the surfaces of Ag nanoplates.[93] In this case, the galvanic replacement between Ag nanoplates and HAuCl4 was inhibited in order to prevent the formation of hollow Ag/Au alloyed nanoplates. To prevent galvanic replacement, I ions were introduced into the reaction system. In the presence of I, a more stable AuI4 complex was formed due to the displacement of Cl in AuCl4 by I.[94] Accordingly, the reduction potential of the HAuCl4 (Au3+/Au, 1.50 V vs. SHE or AuCl4/Au, 0.93 V vs. SHE) was reduced to a level (AuI4/Au, 0.56 V vs. SHE) below that of the Ag template (0.80 V vs. SHE). As such, reduction of the Au precursor and deposition of Au atoms on the Ag nanoplates was mainly controlled by the reductant (i.e., ascorbic acid) added into the reaction solution rather than by the Ag atoms from the template via galvanic replacement. Taken together, the Ag nanoplates remained essentially intact during the deposition of Au, with the formation of bimetallic core-shell nanostructures as the final products (Figure 6B).

Figure 6.

(A) TEM image and a model (inset) of Ag@Au core-shell nanoplates that were prepared by coating a thin layer of Au on the surface of Ag nanoplates in the presence of I ions. (B) Atomic percentage profile of a Ag@Au nanoplate calculated from the relative counts in the energy-dispersive X-ray (EDX) line scan, as indicated by the arrow in the scanning transmission electron microscopy (STEM) image below. Note that some of the nanoplates contain small pits or cavities due to the involvement of galvanic replacement. Reproduced with permission.[93] Copyright 2012 John Wiley and Sons.

Another route for the prevention of galvanic replacement is the introduction of a strong antioxidant or reductant into the system. In the presence of a strong reductant, the salt precursor will be rapidly reduced into metal atoms by the reductant, leaving little possibility for them to galvanically react with the template. For instance, Xue and co-workers demonstrated that the galvanic replacement reaction between HAuCl4 and Ag nanoplates could be prevented in the presence of hydroxylamine (a strong reductant) and NaOH.[95] Since the reducing power of hydroxylamine was greatly enhanced at high pH,[96] the reduction of HAuCl4 was dominated by hydroxylamine rather than the Ag nanoplates and no galvanic replacement will occur. In addition, once a thin layer of Au was coated on the surface of a Ag nanoplate, it could serve as a barrier to isolate the underlying Ag from the HAuCl4 solution, thereby limiting the galvanic replacement reaction. As a result, the final products were Ag@Au core-shell nanoplates with morphologies similar to those shown in Figure 6A. Other strong reductants, such as ascorbic acid paired with NaOH,[97] can also effectively inhibit the galvanic replacement between a Ag template and HAuCl4.

6 Coupling with Other Mechanisms

As discussed in Sections 'Mechanistic Understanding'-'Promotion and Prevention of Galvanic Replacement', galvanic replacement is a powerful method for generating metal nanostructures with hollow interiors. In a typical galvanic replacement reaction, the deposition of atoms from a precursor and the dissolution of atoms from a template will occur on the outer surface and the interior of the template, respectively. Therefore, the final product obtained through galvanic replacement will generally be a single-walled nanobox or nanocage with morphology closely resembling that of the original template. For instance, Au-Ag nanoboxes or nanocages with cubic shapes were prepared by reacting Ag nanocubes with HAuCl4, while the Au-Ag nanoboxes or nanocages with octahedral shapes were produced when Ag nanocubes were replaced by Ag octahedrons.[46, 58] In a sense, the morphology and structure of the final product is more or less predictable from the choice of the starting template and salt precursor. By coupling galvanic replacement with other chemical/physical processes, however, it is possible to generate more complex structures beyond expectation, which often have unique properties and improved performances in certain applications.

6.1 Coupling with Sequentially Deposited Templates

By coupling galvanic replacement with sequentially deposited templates, we have fabricated hollow metal nanostructures with multiple walls.[98, 99] Figure 7, A-C, summarizes the design and preparation of three different types of such nanostructures: nanorattles with a yolk-shell structure, multi-walled nanoshells, and multi-walled nanotubes. To synthesize the nanorattles, a layer of Ag was first coated on the surface of a Au-Ag alloyed nanoparticle through the reduction of Ag+ ions and subsequent surface growth.[98] The Ag-coated nanoparticles were then reacted with HAuCl4, which galvanically oxidized the Ag layer, generating a Au-Ag shell separated from the Au-Ag solid core by a thin gap (Figure 7D). In this case, the gap between the core and shell could be controlled by varying the thickness of the deposited Ag. This strategy could also be extended to the preparation of multi-walled hollow metal nanostructures. Figure 7E shows a triple-walled nanoshell created by alternating between galvanic replacement with HAuCl4 and deposition of a Ag layer.[98] Specifically, single-walled Au-Ag nanoshells were first prepared via the galvanic replacement between Ag cuboctahedrons and HAuCl4. Subsequently, a thin layer of Ag was coated on the shell through reduction, followed by another round of galvanic replacement with HAuCl4, resulting in the formation of double-walled nanoshells. Repeating this cycle once more yielded the triple-walled Au-Ag nanoshells. In principle, these processes can be further repeated to generate nanoshells with multiple walls. Interestingly, this approach could also be applied to Ag nanowires to fabricate multi-walled metal nanotubes (Figure 7F).[99]

Figure 7.

(A-C) Schematics showing coupling of sequential reduction with galvanic replacement for the fabrication of (A) nanorattles, (B) multiple-walled nanoshells, and (C) multi-walled nanotubes composed of Au-Ag alloys. (A-C) are adapted and reproduced with permission.[23] Copyright 2010 Elsevier. (D,E) TEM images of nanorattles and triple-walled Au-Ag nanoshells, respectively. Reproduced with permission.[98] Copyright 2004 American Chemical Society. (F) SEM images of double-walled Au-Ag nanotubes. Reproduced with permission.[99] Copyright 2004 John Wiley and Sons. Insets in (D-F) show electron microscopy images of individual nanostructures at higher magnifications.

6.2 Coupling with Co-Reduction

The galvanic replacement reaction can be coupled with co-reduction by introducing a reductant into a galvanic replacement reaction. The concurrent reduction of metal ions by both the reductant in solution and by the metal template increases the complexity of the reaction mechanism relative to the simple reduction method. In order to create the desired nanostructure, one has to carefully control both rates of co-reduction and galvanic replacement reactions. Recently, we successfully synthesized Pd-Pt nanocages with greatly enhanced catalytic activities (Figure 8B,C) by coupling the galvanic replacement between Pd nanocubes and K2PtCl4 with a co-reduction process by using citric acid as the reductant.[100] Figure 8A shows the reaction mechanism responsible for the formation of the Pd-Pt nanocages. As discussed in Section 'Promotion', the galvanic replacement reaction was initiated at the {100} facets of a Pd nanocube owing to the adsorption of Br ions on the Pd(100) surfaces. As such, the side faces of a Pd nanocube were slightly excavated, producing concave nanocubes at the very early stage (step 1) of the reaction. Meanwhile, co-reduction of the Pd2+ (from the dissolution of Pd nanocube) and the Pt2+ (from K2PtCl4) by citric acid began to occur, and was accompanied by the galvanic replacement reaction throughout the synthesis. The newly formed Pd and Pt atoms preferentially co-deposited onto the side faces of the concave nanocube, resulting in the formation of a Pd-Pt alloy shell (step 2). As the reaction proceeded, the interior Pd gradually disappeared due to galvanic replacement. At the same time, walls composed of Pd-Pt alloys with exposed {100} facets were formed due to the co-reduction and stabilization of {100} facets by Br ions (step 3). Eventually, a Pd-Pt nanocage with hollow interior and porous walls was formed via this coupled process. Note that the morphology of the final product was sensitive to both the rates of galvanic reaction and co-reduction. For example, the above process resulted in Pd-Pt nanodendrites, rather than nanocages, when citric acid was replaced by ascorbic acid (a stronger reductant) owing to the rapid consumption of PtCl42− and thus the reduced rate of galvanic replacement. In addition to the Pd-Pt system, co-reduction can also be coupled with the galvanic replacement reactions between Ag templates and Au precursors to generate Au-Ag alloy nanostructures with distinct optical and catalytic properties.[45, 74]

Figure 8.

(A) Schematic illustration of the morphological and structural changes in a synthesis that involved coupling of a co-reduction with a galvanic replacement reaction between PtCl42− and a Pd nanocube. Citric acid (CA) was used as a reductant. (B) High-angle annular dark-field STEM (HAADF-STEM) image of the Pd-Pt nanocages prepared using the procedure shown in (A). The inset shows the HAADF-STEM image of an individual Pd-Pt nanocage at a higher magnification. (C) EDX mapping (top trace) and line-scan (bottom trace) profiles of a single Pd-Pt nanocage. Adapted and reproduced with permission.[100] Copyright 2011 American Chemical Society.

6.3 Coupling with the Kirkendall Effect

The Kirkendall effect is a phenomenon created by the unbalanced interdiffusion between two materials, driven by their difference in interdiffusion rates.[101-103] The characteristic consequence of the Kirkendall effect is the formation of voids (also referred to as Kirkendall voids) in the bulk of the component with the faster diffusion rate. Although the first documented study can be traced back to the work by Ernest Kirkendall on the diffusion between zinc and alpha brass more than 70 years ago,[104] only in the last decade has the Kirkendall effect become a powerful approach to the preparation of nanoscale materials with hollow structures,[105-110] as pioneered by Alivisatos and Yin.[102] Recently, Puntes and co-workers developed a route for the production of complex hollow nanostructures by sequential or simultaneous action of galvanic replacement and the Kirkendall effect.[111] Using Ag nanocrystals as templates, and Au or Pd as an oxidizing agent, polymetallic hollow nanocrystals with various morphologies and compositions were obtained at room temperature. Figure 9, A and B illustrates the morphological evolution during the preparation of Au-Ag double-walled nanoboxes (Figure 9C) through the simultaneous action of galvanic replacement and the Kirkendall effect. The galvanic replacement between Ag template and Au precursor began with the deposition of a Au shell on the template surface and the formation of pinholes in the walls of the template (A1-A4, Figure 9A). Once a cavity formed in the interior of the template via continuous galvanic replacement, and both the inner and outer surfaces were covered by Au, the wall could be viewed as a thin film of Ag between two layers of Au. This structure allowed for the formation of Kirkendall voids within the nanocrystal walls (diffusion of Ag in Au is faster than that of Au in Ag[46]), while the Ag continued to be removed by galvanic replacement (A5, Figure 9A). The voids then grew and coalesced to form a continuous cavity between the layers of Au, leading to the formation of Au-Ag double walled nanoboxes (A6, Figure 9A). Interestingly, galvanic replacement and the Kirkendall effect can be coupled not only simultaneously but also sequentially. For example, Au-Pd-Ag multi-layered nanostructures were obtained through the sequential action of galvanic replacement and the Kirkendall effect (Figure 9E). Initially, during the cavity formation in the Pd-Ag structures (Figure 9D), the removal of Ag was dominated by galvanic replacement. After the addition of Au precursor, a gap was formed around the cavity (Figure 9E), primarily driven by the Kirkendall effect.

Figure 9.

(A) Schematic illustration of the morphological and structural changes involved in the preparation of Au-Ag double-walled nanoboxes through simultaneous galvanic replacement and Kirkendall effect. (B) HAADF-STEM images and EDX mapping for the samples obtained at different stages of the reaction shown in (A). (C) TEM images of Au-Ag double-walled nanoboxes prepared using the procedure shown in (A). (D,E) TEM images of nanostructures prepared via sequential action on Ag nanocubes by (D) galvanic replacement with a Pd precursor and (E) Kirkendall effect with a Au precursor. Reproduced with permission from Ref. [111]. Copyright 2011 American Association for the Advancement of Science.

7 Niche Applications

7.1 Catalysis

Metal nanostructures prepared by galvanic replacement have many unique features that make them attractive for various catalytic applications: i) a much larger reactive surface area compared to solid particles of similar sizes owing to the presence of inner surfaces and numerous pores on the walls; ii) porous surfaces and hollow interiors that allow efficient mass transport of both the product and reactant in and out of the catalytic system; iii) alloyed, single-crystal structures with good flexibility and stability, allowing them to survive harsh catalytic reaction environments (e.g., high temperature), thus increasing their catalytic longevity; iv) tunable compositions as controlled by the amount of metal ions added to a galvanic replacement reaction, allowing for the optimization of catalytic activity for specific reactions; and v) large surfaces which can serve as good electrical connections to accommodate both the oxidation and reduction half reactions of a redox system, while the thin wall is still able to provide high catalytic activity comparable to small sized solid particles.

Over the past decade, nanocages composed of different metals were developed and studied as effective catalysts for a variety of reactions such as Suzuki cross-coupling reactions,[112] redox reactions,[113-115] and oxygen reduction reactions.[116, 117] As demonstrated in our previous study, Au-Ag nanocages, prepared via the galvanic replacement between Ag nanocubes and HAuCl4, can be used as highly effective catalysts for redox reactions.[118] The study was motivated by the need to overcome the dilemma encountered with conventional solid catalytic nanoparticles used for redox reactions. It is known that the catalytic activity of a nanoparticle in redox reactions strongly depends on its size. Typically, smaller nanoparticles provide higher catalytic activity due to their much greater surface-to-volume ratio.[119, 120] However, as the nanoparticles become increasingly smaller, the redox reactions may not proceed since the oxidation and reduction half reactions will be forced to take place on different particles. This occurs when the surface area of an individual particle is too small to accommodate both reaction intermediates. We proposed that this problem could be solved by replacing solid nanoparticles with nanocages possessing ultrathin, porous walls with large, electrically conductive surfaces that could accommodate both half reactions. Using a model redox reaction, based on the reduction of p-nitrophenol by NaBH4, we compared the catalytic properties of different Au-based nanostructures including 50 nm Au-Ag nanocages, Au-Ag nanoboxes, partially hollow Au-Ag nanoboxes, and 50- and 5-nm solid Au spheres (Figure 10A).[118] The reaction was monitored in real time by measuring the extinction peaks from p-nitrophenolate ions at 400 nm as they gradually dropped in intensity during the reaction.[121, 122] Figure 10B summarizes the rate constants and activation energies for different types of Au-based catalysts towards the reduction of p-nitrophenol by NaBH4. As expected, the Au-Ag nanocages show the best performance with a rate constant of 2.83 min−1. In comparison, the rate constants were calculated to be 1.12, 0.59, 0.20, and 0.95 min−1 for the Au-Ag nanoboxes, partially hollow Au-Ag nanoboxes, 50- and 5-nm solid Au particles, respectively. An opposite trend was observed in the activation energies, with the lowest value calculated for the Au-Ag nanocages (28.04 kJ mol−1). In addition to the large electrically conductive surface, other factors such as a large accessible surface area may also contribute to the superior performance of the Au/Ag nanocages. Notably, Au-Ag nano­cages also showed the best activity for photocatalytic reactions relative to Au-Ag nanoboxes and nanoframes of the same sizes. The superior performance of the nanocages was mainly attributed to their higher reactive surface area and the presence of an inner cavity which served as a reaction volume for intermediate collisions. More details about this study can be found in Ref. [123].

Figure 10.

(A) Representative TEM images together with schematic drawings of the Au-based nanocages, nanoboxes, partially hollow nanoboxes, and solid nanoparticles (from left to right) used as catalysts for the reduction of p-nitrophenol by NaBH4. (B) A table summarizing the rate constants (k) and activation energies (Ea) for different types of Au-based catalysts shown in (A). Adapted and reproduced with permission.[118] Copyright 2010 American Chemical Society.

Furthermore, these hollow nanostructures can be used as catalysts for the oxygen reduction reaction (ORR) at the cathodes of proton exchange membrane fuel cells (PEMFCs).[124-126] In a recent study, it was demonstrated that Pd-Pt nanocages exhibited considerably enhanced ORR activities compared to solid Pd@Pt core-shell nanocrystals.[127] Catalytic activities were evaluated for two different types of Pd-Pt nanocages ca. 25 nm in size: octahedral and cubic Pd-Pt nanocages prepared through co-reduction coupled with galvanic replacement between Pd octahedrons and cubes, respectively, with K2PtCl4 (Figure 11A). For comparison, another two samples of core-shell Pd@Pt nanoparticles of similar sizes and shapes but with solid interiors were also evaluated: octahedral Pd@Pt particles and cubic Pd@Pt particles. Figure 11B shows the mass activities and specific activities of these four different types of Pd-Pt bimetallic nanostructures. The mass activities of these catalysts were found to decrease in the order of octahedral Pd/Pt nanocages (393.9 mA/mg) > octahedral Pd@Pt particles (286.4 mA mg-1) > cubic Pd/Pt nanocages (178.1 mA mg-1) > cubic Pd@Pt particles (128.5 mA mg-1). The area-specific activities showed a similar trend as the mass activity. These results showed that the porous, hollow Pd-Pt nanocages outperformed the solid core-shell Pd@Pt particles of the same size and shape. These enhanced activities for Pd-Pt nanocages can be attributed to their relatively larger reactive surface areas (40.3 and 42.3 m2 g-1 for octahedral and cubic Pd-Pt nanocages, respectively) when compared to the solid Pd@Pt particles (22.2 and 25.9 m2 g-1 for octahedral and cubic Pd@Pt particles, respectively). Note that the overall ORR activities of the octahedral shaped Pd-Pt nanostructures were higher than that of the cubic nanostructures, mainly due to their {111} facets on the surface that have been shown to be more active than {100} facets for ORR.[128-130] In addition to ORR, the porous Pd-Pt nanocages have been successfully applied to catalyzing the preferential oxidation (PROX) reaction of CO,[131-133] outperforming both pure Pd and pure Pt solid particles.[100] Significantly, the Pd-Pt alloy showed a lower onset temperature than both of the pure metal catalysts, as well as a higher selectivity for CO.

Figure 11.

(A) Representative TEM images together with schematic drawings of the cubic and octahedral Pd-Pt nanocages, and cubic and octahedral Pd@Pt core-shell nanoparticles (from left to right) used as catalysts for the oxygen reduction reaction (ORR). (B) Mass and area-specific catalytic activities at 0.85 V vs. a reversible hydrogen electrode for the catalysts shown in (A). Adapted and reproduced with permission.[127] Copyright 2012 American Chemical Society.

7.2 Plasmonics

One of the most attractive properties of Au nanostructures is their localized surface plasmon resonance (LSPR),[10] arising from the collective oscillations of conduction electrons in response to the incident light under resonant conditions.[15, 134] Galvanic replacement provides a simple, and easy method for precisely tuning the LSPR peaks of the hollow/porous Au nanostructures by adjusting the amount of HAuCl4 relative to Ag templates.[21, 135]Figure 12A shows the UV-vis-near infrared (UV-vis-NIR) extinction spectra obtained by titrating an aqueous suspension of Ag nanocubes (ca. 40 nm in edge length) with different volumes of an aqueous HAuCl4 solution (0.1 mM).[57] The LSPR peaks of the resultant Au-Ag nanocages were constantly tuned from the visible to the NIR region as the volume of HAuCl4 solution was increased. As shown in Figure 12B, the Au-Ag nanocages can have extremely large cross sections for both scattering and absorption components in the NIR region.[136] The latter allows for effective conversion of NIR light into heat, making the nanocages particularly well-suited for a variety of applications such as photoacoustic imaging and targeted cancer therapy.[137-139] It is worth pointing out that the magnitudes of both scattering and absorption components of a Au-Ag nanocage can be tailored by simply controlling the size and porosity of the nanocage, and more details about this capability can be found in our previous publications.[136, 140]

Figure 12.

(A) UV-vis-NIR extinction spectra taken from an aqueous solution of Ag nanocubes (40 nm in edge length) after titration with different volumes of a 0.1 mM HAuCl4 solution (as labeled on each curve). Reproduced with permission.[57] Copyright 2007 Nature Publishing Group. (B) UV-vis spectra calculated using the discrete dipole approximation (DDA) method for a Au-Ag nanocage with an inner edge length of 50 nm, a wall thickness of 5 nm, and 5-nm pores at all corners. Reproduced with permission.[136] Copyright 2007 John Wiley and Sons. (C) Plot of the change in height (Dh) and temperature (DT) as a function of time upon irradiation of aqueous suspensions of Au-Ag nanocages at four different concentrations: 1.0 × 109 (▿), 2.5 × 1019 (▵), 5.0 × 109 (○), and 1.0 × 1010 (◻) particles mL-1 (from bottom to top). Each cycle of irradiation lasted for 40 min (on for 10 min and then off for 30 min), with the cycle being repeated twice. (D) Plot of the energy conversion efficiencies (η) as a function of particle concentration. In (C) and (D), the laser density was set to 0.4 Wcm−2 and the illuminated area was 1.13 cm2. (C) and (D) are reproduced with permission.[141] Copyright 2013 John Wiley and Sons.

In a recent study, we also quantified the absorption cross section (σ) of the Au-Ag nanocages, a fundamental parameter that determines their photothermal conversion efficiency, using a homemade plasmon-assisted optofluidic (PAOF) system.[141] The system contained a diode laser as the energy source, an aqueous suspension of Au-Ag nanocages as the photo-responsive fluidic medium, and a capillary as a means to monitor the expansion in volume as the medium is heated by the photothermal effect. We could quantify the photothermal energy conversion efficiencies (η) of Au-Ag nanocages at different concentrations by simply measuring the change in height (h) for the liquid in the capillary (Figure 12C, detailed experimental designs and methods can be found in Ref. [141]). For comparison, we also quantified the η values for two other types of Au nanostructures (i.e., Au nanorods[142, 143] and hexapods,[144] whose major LSPR peaks were tuned to the same position as the nanocages) using the same PAOF system (Figure 12D). The η values were calculated to be 22.1 ± 1.7% and 29.6 ± 1.9% for the Au nanorods and hexapods, respectively, which were less than half of the value calculated for the Au-Ag nanocages (63.6 ± 4.2%) at the same particle concentration. Based on the values of η and the use of indocyanine green (ICG) as a reference with known absorption cross section, the PAOF system was also used to measure the σ values of the Au-Ag nanocages, Au nanorods, and Au hexapods. Their corresponding values were calculated to be 12.4 × 10−15, 3.5 × 10−15, and 4.9 × 10−15 m2, respectively, at a particle concentration of 1.0 × 1010 particles mL−1.

7.3 Biomedical Research

Of the many interesting nanostructures prepared by galvanic replacement, Au-Ag nanocages are particularly desirable for biomedical applications because of the good biocompatibility of the metals,[135] their strong LSPR peaks in the NIR region ranging from 700 to 900 nm,[57] and their easy surface functionalization via the well-established Au-S linkage.[145, 146] It should be emphasized that LSPR properties of the Au nanostructures in the NIR region are particularly important for biomedical applications since the background interference in the NIR region (including the absorption by blood and water, and the scattering by soft tissues) is significantly reduced.[147, 148] The Au-Ag nanocages also have additional advantages when compared with other nanostructures with tunable LSPR peaks in the NIR region, including large absorption cross section,[137] accessible hollow interiors,[149] precisely controlled LSPR peak position,[57] and availability of a broad range of sizes (20–500 nm).[150] All of these properties make them ideal for applications such as optical imaging, photothermal treatment, and drug delivery. In the following discussion, we only highlight a few studies over the past three years, with the Au-Ag nanocages referred to as “AuNCs” for the purpose of simplicity.

Owing to the presence of hollow interiors and the photothermal effect, AuNCs were utilized as smart vehicles for controlled release and drug delivery. The concept was initially demonstrated by functionalizing the outer surface of AuNCs with thermally responsive polymers to control the release of a pre-loaded drug by NIR laser irradiation or high-intensity focused ultrasound (HIFU).[149] More recently, we demonstrated another delivery strategy by filling the hollow interior of an AuNC with a drug-doped phase-change material (PCM) such as 1-tetradecanol, possessing a melting point of 38–39 °C (Figure 13A).[151] Once the drug was mixed with a PCM, it could be easily loaded into the AuNC via diffusion at a temperature above the melting point of the PCM. Typically, surfactant-like PCMs, such as those containing hydrophilic heads and long hydrophobic tails, are preferred to ensure good miscibility with different types of drug molecules. After loading, the AuNC encapsulated the drugs in the interior until it was heated to a temperature slightly beyond the melting point of the PCM,[152] which resulted in a quick release of the drug through diffusion. Importantly, this heat-induced drug release could be conveniently controlled by either varying the power of HIFU or the duration of exposure to HIFU (Figure 13B).

Figure 13.

(A) Schematic illustration of the process involved in the release of a dye from a Au-Ag nanocage through the aid of a phase-change material (PCM). (B) Release profiles of the dye when the samples were exposed to high-intensity focused ultrasound (HIFU) at different applied powers. Adapted and reproduced with permission.[151] Copyright 2011 American Chemical Society.

Cherenkov luminescence imaging has recently received great interest as an emerging modality for molecular imaging owing to its low cost, easy operation, high throughput, and clean autofluorescence background.[153] Cherenkov luminescence is referred to as the light emitted during the decay of a radionuclide such as 18F, 64Cu, or 198Au.[153-156] In a recent study, we reported a new approach to the synthesis of 98Au-doped AuNCs by adding H98AuCl4 to the precursor solution used for a galvanic replacement reaction with Ag nanocubes (Figure 14A).[157] The direct incorporation of 198Au atoms into the walls of AuNCs ensured good stability for in vivo imaging, as compared to the conventional methods where 198Au was attached to the surface of nanostructures through a ligand or physical encapsulation.[158, 159] Moreover, the specific radioactivity of the 198Au-doped AuNCs could be easily and precisely controlled by varying the amount of H198AuCl4 used for the galvanic replacement reaction, allowing for more accurate analysis. We then evaluated the feasibility of in vivo Cherenkov luminescence imaging by introducing the 198Au-doped AuNCs into small animals. Experimentally, the AuNCs doped by 198Au and then covered by poly(ethylene glycol) were intravenously injected into a mouse bearing EMT-6 tumor xenografts in the right hind legs (60–70 μCi per mouse) and bioluminescence images were captured at different post-injection times (Figure 14B). As confirmed by semiquantitative analysis of the region-of-interest after decay correction, the bioluminescence intensities at the tumor area increased constantly up to 24 hours, indicating the potential use of this new imaging platform for quantitative analysis of tumors in small animals. When combined with the tunable LSPR properties in the NIR region, the radioactive AuNCs may find uses as a new multimodal imaging platform for diagnosing both small animals and human bodies.

Figure 14.

(A) Schematic illustration of the procedure used for the preparation of 198Au-doped nanocages via galvanic replacement between Ag nanocubes and a mixture of 197Au3+ and 198Au3+ ions. (B) Representative luminescence images of a mouse bearing EMT-6 tumor at different time points after the injection of PEGylated 198Au-doped nanocages (64 μCi per mouse) through the tail vain. Reproduced with permission.[157] Copyright 2013 American Chemical Society.

8 Concluding Remarks

Through a large number of examples, we have demonstrated that template-engaged galvanic replacement offers a remarkably simple and versatile route to the fabrication of multifunctional nanostructures with hollow interiors and many other attractive features. This approach relies on the use of two concurrent processes to precisely engineer the properties of nanostructures: oxidation/dissolution of atoms from a template and reduction of a salt precursor, followed by deposition of the metal atoms onto the template. Such a reaction is driven by the difference in reduction potentials between the two metal/ion pairs involved.

Despite its great success, the galvanic replacement approach still deserves further study to achieve a better understanding and more precise control of the process. Firstly, it is a must that the metal derived from the salt precursor has a higher reduction potential than the metal in the template in order to initiate a galvanic replacement reaction. On the other hand, the reaction kinetics is determined by other parameters such as temperature and concentrations of the salt precursors. In principle, the kinetics can be experimentally controlled by varying the reaction temperature or adding ionic species into the system to promote or prevent a galvanic process. Secondly, the site selectivity on the initial template has a strong impact on the morphology of the final product. It is generally acknowledged that the oxidation/dissolution of template will preferentially occur on facets with the highest surface free energy whereas the deposition of new atoms will occur on the remaining low-energy facets. The chemisorption of a capping agent can alter the surface free energies of various facets and even reverse their order. In a sense, it is possible to control the site selectivity of a reaction by introducing appropriate facet-capping agents. Accordingly, screening new facet-capping agents that can selectively bind to a specific set of facets, with varying adsorption energies and degree of surface coverage, is also a promising direction that deserves further exploration. Thirdly, alloying of the two metals through interdiffusion has to be promoted in order to generate a smooth surface for the resultant nanostructures. The rate of interdiffusion, and thus the smoothness of the surface of the final product, can be adjusted by controlling the temperature.

While not emphasized in this article, the stoichiometry of a reaction can greatly impact the porosity and morphology of the final product. We have previously investigated the reactions of AuCl2 and AuCl4 with Ag nanocubes to examine the effect of stoichiometric ratio on the morphology of resultant product.[49, 50] Other metallic precursors with different oxidation states may also result in fundamentally different products and thus need to be thoroughly studied. As discussed in Section 'Coupling with Other Mechanisms', galvanic replacement can be coupled with sequential reduction, co-reduction, and the Kirkendall effect to generate nanostructures with greatly increased complexity. Combinations with other physical/chemical processes or synthetic techniques may enable the fabrication of nanomaterials with unexplored structures and properties. Notably, the sacrificial templates involved in galvanic replacement reactions are not limited to nanostructures made of metals. In a recent report, Hyeon and co-workers demonstrated that nanocrystals of metal oxide (e.g., Mn3O4 and Co3O4) could also serve as templates for galvanic replacement reactions with salt precursors, generating functional hollow nanostructures made of multimetallic oxides.[160] In this new galvanic system, the higher-oxidation-state ions with a higher reduction potential in the metal oxide nanocrystals was replaced by the lower-oxidation-state metal ions with a lower reduction potential from a salt precursor, which is different from what was observed in metallic systems (Section 'Mechanistic Understanding'). For example, Fe2+ ions could reduce and replace Mn3+ ions in the Mn3O4 nanocrystals to produce Mn3O4/γ-Fe2O3 or γ-Fe2O3 hollow nanostructures, owing to the lower reduction potential of Fe3+/Fe2+ (0.77 V vs. SHE) relative to that of Mn3O4/Mn2+ (1.82 V vs. SHE). In contrast, Mn3O4 nanocrystals could not be transformed into hollow nanostructures by reacting with Co2+ ions because the reduction potential of Co3+/Co2+ (1.87 V vs. SHE) is higher than that of Mn3O4/Mn2+.

Although the final product of a galvanic replacement reaction can be readily characterized using a number of techniques such as electron microscopy and EDS elemental scanning/mapping, in situ monitoring and quantitatively analyzing the structural and atomic changes of templates during a galvanic reaction are still challenging. These challenges are expected to be overcome in the near future as new tools are applied to analyze such reactions. For instance, in a recent study, Sun and co-workers have successfully monitored the galvanic replacement reaction between Ag nanowires and HAuCl4 solution in real time by using in situ transmission X-ray microscopy in combination with a flow cell reactor.[161] The in situ observations clearly show the detailed morphological and structural evolution from the solid Ag nanowires to hollow Au-Ag nanotubes, which involved multiple steps: anisotropic etching of the Ag nanowires, uniform deposition of Au atoms onto the surface of nanowires, and reconstruction of the nanotube walls via an Ostwald ripening process.

Finally, it should be pointed out that the hollow nanostructures may find new applications other than those presented in this article. For example, the tunable plasmonic properties of these materials make them attractive for a range of optical applications such as ultrasensitive detection and optical amplification. The hollow interiors of these nanostructures, coupled with the ability to functionalize their surfaces, may find use in applications such as gene delivery and biosensors. Ultimately, we hope the studies presented in this article will provide a strong foundation for inspiring future work.


This article is part of an ongoing series celebrating the 25th anniversary of Advanced Materials. This work was supported in part by a grant from the NIH (R01 CA13852701), an NIH Director's Pioneer Award (DP1 OD00798), and startup funds from Georgia Institute of Technology. As a jointly supervised Ph.D. student from Southwest University, Yi Wang was partially supported by a Fellowship from the China Scholarship Council. Aleksey Ruditskiy was partially supported by an NSF Graduate Research Fellowship.


  • Image of creator

    Xiaohu Xia has been a postdoctoral fellow in the Xia group at Georgia Institute of Technology since January 2012. He received his BS degree in biotechnology (2006) and PhD degree in biochemistry and molecular biology (2011) from Xiamen University, China. He worked as a visiting graduate student at Washington University in St. Louis from October 2009 to December 2011. His current research interests include the design and synthesis of nanomaterials and exploration of their applications in catalysis and biomedicine.

  • Image of creator

    Younan Xia has been the Brock Family Chair and Georgia Research Alliance (GRA) Eminent Scholar in Nanomedicine at Georgia Institute of Technology since January 2012. He holds joint appointments in the Wallace H. Coulter Department of Biomedical Engineering, School of Chemistry and Biochemistry, and School of Chemical and Biomolecular Engineering. Xia received his BS degree from the University of Science and Technology of China (USTC) in 1987, MS degree from the University of Pennsylvania in 1993, and PhD degree in physical chemistry from Harvard University (with Professor George M. Whitesides) in 1996. He started as an Assistant Professor of chemistry at the University of Washington (Seattle) in 1997, and was promoted to Associate Professor and Professor in 2002 and 2004, respectively. From July 2007 to December 2011, he was the James M. McKelvey Professor for Advanced Materials in the Department of Biomedical Engineering at Washington University in St. Louis. His current research interests include nanomaterials, biomaterials, nanomedicine, regenerative medicine, imaging, electrospinning, and colloidal science.