Introduction

In addition to its well-known applications in ornaments, electrical conductors, mirrors, photographic films and antimicrobial1,2,3,4, nanostructured silver (Ag) has undergone extensive investigation in recent years due to many promising applications of this noble metal in the fields of catalysis, electronics, optical labeling and biosensing5,6,7,8. In this regard, a great number of methods have been developed to generate nanoscale Ag crystals with different morphologies such as rod, wire, cube, platelet, decahedron and bipyramid etc9,10,11,12,13,14,15,16,17,18,19,20,21,22,23. Among them, triangular Ag platelets (or “prisms”) have received tremendous research attention over the past decade because they possess outstanding plasmonic features across both visible and IR regions; they also show significant surface-enhanced Raman signals and can also be facilely functionalized with sulfur-containing adsorbates17,18,19,20. Concerning the synthesis of Ag prisms, two major approaches have been developed: photoinduced growth17,18,19,20,21 and seed-mediated growth (or thermal process) in solution22,23. Using the photoinduced route, for example, the conversion from Ag sphere to triangular prism would not take place in the darkroom even for a period of more than 2 months17, revealing that presence of photons is indispensable for the formation of Ag prisms.

The geometry and crystal orientation of such important Ag prisms have been well established18,22. As depicted in Figure 1a, for example, a rotation axis with 3-fold symmetry (C3) is perpendicular to the largest surface planes of {111} of silver crystal. If they are attached with a secondary phase (X, for example, a semiconductor) on one of their three edges or corners, their overall symmetry will be significantly lowered (i.e., the point group is changed from D3h to C2v). By imposing this desymmetrization, one actually turns the Ag prisms into X/Ag heterodimers or binary nanocomposites. Furthermore, the newly added material phase X would likely alter the overall electronic structure and thus surface plasmonic resonance (SPR) feature of Ag prisms to some extent, resulting in novel properties that symmetric Ag prisms do not possess.

Figure 1
figure 1

Synthesis and characterization of Ag2S/Ag heterodimers.

(a). Illustration of desymmetrization of Ag prism (in blue): (i) symmetric triangle, (ii) edge attachment, (iii) corner attachment and (iv) planar attachment with an Ag2S sphere (yellow), (b). a panoramic view (TEM image) of typical Ag2S/Ag heterodimers (Experimental Section) and (c). TEM image, elemental line-profiles and chemical mappings of an Ag2S/Ag heterodimer.

Regarding the preparation of general semiconductor/noble-metal heterodimers, significant advancements have been made in recent years26,27,28,29,30,31,32,33,34,35,36,37. Investigations on this type of materials include CdSe/Au26,27,28, ZnO/Ag29, ZnO/Au30, CdS/Pt31,32, PbS/Au33,34, Ag2S/Au35 and Ag2S/Ag as well36,37,38. Nevertheless, it should be mentioned that these semiconductor/noble-metal heterodimers all involve the growth of metals onto their semiconductor counterparts and rather surprisingly, apart from sphere-like particles or polyhedrons, the metal phases have never been prepared into more anisotropic shape such as a triangular platelet structure. Furthermore, the size of metal phase seems always smaller than that of primary semiconductors, namely, noble metals in such reported heterodimers only serve as a secondary phase. It is our belief that by coupling a smaller semiconductor particle to a larger anisotropic metal crystal, new semiconductor/noble-metal heterodimers can be attained, that is, the metal now serves as a dominant material while the semiconductor as a secondary counterpart. Through this reverse structural transformation, additional functionalities of the metal crystals can be attained.

Very recently, we have synthesized Ag2S/Ag heterodimers using a photoassisted solution-based approach37. Because the Ag crystal shape is sphere-like, the findings in that work could not be correlated directly to the rich literature reports for prismatic triangular silver. It should be mentioned that metallic or ionic Ag, which will be investigated in the present work, is a toxic element to microorganisms and shows prominent bactericidal activity against as many as 12 species of bacteria including Escherichia coli24,25. On the other hand, Ag2S is both a direct narrow-band-gap semiconducting metal sulfide and an effective ionic semiconductor in which Ag+ ions behave just like free electrons in metals resulting in cationic vacancy rich Ag2S phase39,40. Such unique chemical and structural properties could endow it as an excellent host mediator for preparation of Ag2S-based heterodimers with improved properties36,37,38,42,43,44,45. In addition to the connectivity to Ag, Ag2S nanocrystals have also been demonstrated recently to act as a mediator/catalyst for preparation of semiconductor/semiconductor heterodimers such as Ag2S/ZnS and Ag2S/CdS in solution41.

In this article, we have devised a convenient room temperature method for preparation of a new type of Ag2S/Ag heterodimers without photon assistance. The following are the objectives of this work: (i) changing the silver phase to shape-defined Ag prisms, (ii) making the silver become a dominant phase in this binary heterodimers and (iii) desymmetrizing Ag prisms in order to alter its surface plasmonic properties for new applications. While all these objectives have been achieved in the current study, our synthetic protocol also greatly simplifies the general preparation of triangular Ag prisms at room temperature, because it does not need light irradiation, it does not require relatively strong reducing agent (e.g., L-ascorbic acid, NaBH4, etc.) and it does not demand process heating (e.g., ≥ 60°C) or use of seed in the previously reported methods17,18,22. These new Ag2S/Ag heterodimers also exhibit significant enhancement in bactericidal ability for E. coli K-12 cells, compared to the Ag2S/Ag in our earlier work37. The enhanced performance can be attributed largely to the realization of desymmetrizing triangular Ag prisms.

Results

We first synthesized monodisperse CdS colloidal nanospheres through a modified polyol process (Supplementary Information, SI-1). Transmission electron microscopy (TEM) images of the as-formed CdS nanospheres are displayed in Figure S1, which are at close to 100% morphological yield with an average diameter of 120 nm. Furthermore, our high-resolution TEM (HRTEM) investigation indicates that the nanospheres were formed through an oriented attachment mechanism with even smaller crystallites although there are structural defects due to rapid aggregation at high temperatures (Figure S2). The CdS nanospheres were used as a solid precursor for subsequent chemical transformation to Ag2S nanospheres via cation-exchange and as a template for metallic Ag deposition. The resultant Ag2S/Ag heterodimers from this process are depicted in Figure 1a.

In Figure 1b, the panoramic view on a product Ag2S/Ag reveals that each Ag2S nanosphere was grown with a triangular prism of Ag crystal, leading to formation of a novel type of asymmetric metal−semiconductor heterodimers. In this regard, the reported symmetrical nanoprisms of Ag have been desymmetrized for the first time due to attachment of Ag2S phase. In general, the resultant Ag2S/Ag heterodimers are also very uniform at a high morphological yield using this synthetic route (Figure 1b). It should be noted that the average diameter of the Ag2S spheres in the dimers is around 120–130 nm, which is quite similar to that of the CdS precursor. From the high-magnification TEM image (Figure S3), we can find that the edge length of the Ag nanoprisms is about 200 nm, which is larger than the average diameter of Ag2S. This type of heterodimers is structurally different from all other semiconductor−mental heterostructures reported in the literature26,27,28,29,30,31,32,33,34,35,36,37. Detailed characterization of these heterodimers is shown in Figure S3. Interestingly, the Ag platelets connect to Ag2S nanospheres through a number of modes such as corner, edge and face attachments (edge-attachment is dominant). Truncated Ag platelets could also be observed, though their population is low (Figure S3). In fact, all colloidal triangular nanoprisms/plates of silver reported in the literature always include certain percentages of non-triangular morphologies such as hexagons etc19. To confirm the resulted heterodimers, EDX line analysis and chemical mapping study were further conducted (Figure 1c). The Kα1 elemental line profiles distinctly show that sulfur element is distributed evenly in the Ag2S solid spheres. A similar trend for the Ag Lα1 signal profile can also be observed for these nanospheres, noting that the drastic rise in signal at the juncture is a superimposition of Ag Lα1 signals arising from both Ag2S and Ag phases. Therefore, the results revealed by the EDX analysis are in excellent agreement with those by the TEM. In addition, chemical mapping images of the Ag and S on the heterodimers further demonstrate the same elemental distributions (Figures S3 and S4). Ag nanoprisms could also be enlarged using the thus-grown Ag2S/Ag as starting seeds. Such an epitaxial overgrowth is mainly an edge expansion of the Ag nanoprisms while their thickness is essentially unaltered (Figure S5). The image of Figure 2a shows a triangular Ag nanoprism prepared from the overgrowth of Ag. It exhibits sharp edges and corners and smooth basal surfaces. The Ag2S phase in this asymmetric heterodimer is a solid sphere. The electron diffraction (ED) pattern taken from the Ag nanoprism is composed of diffraction spots with a 6-fold rotational symmetry (Figure 2b), revealing that the top and bottom faces of the Ag prism/plate are terminated by the {111} facets. Two sets of diffraction spots marked by square and circle with lattice facet spacing of 2.53 Å and 1.25 Å are corresponding to 1/3{-422} and 2/3{-422} respectively in accordance to previous reports23 and the outer spot set (framed by a rhombus) is the diffractions of {-422} with an interplanar spacing of 0.835 Å. Another set of weak spots with a d-spacing of 1.46 Å can be identified as {220} diffraction normally allowed in fcc lattice. Figure 2c–d gives two HRTEM images taken along the [111] direction. The well-recognized lattice fringes further confirm the single crystal nature of the as-obtained Ag triangle plates, indexed as the forbidden diffractions 1/3{-422} of fcc-lattice of Ag (also verified by Figure 2b). HRTEM images detected from different parts of a Ag2S nanosphere have the clear lattice fringes with an interplanar spacing of approximately 0.26 nm, which is consistent well with the {-121} planes of the monoclinic Ag2S, elucidating that these nanoparticles are single-crystalline (Figure S6).

Figure 2
figure 2

TEM, SAED & HRTEM investigation of Ag2S/Ag heterodimers.

(a). TEM image of a representative Ag2S/Ag heterodimer, (b). SAED pattern of (a)., where squared spots correspond to the formally forbidden 1/3{442} diffractions, triangle-framed spots to {202} diffractions, circled spots to 2/3{442} diffractions and rhombic-shaped spots to {422} diffractions and (c. & d.) HRTEM images and related FFT-images and ED patterns for the areas marked in (a).; details on the area (e of a.) can be found in Figure S6. TEM images (lower panel) and corresponding SAED patterns at various tilting angles.

The stereo-configuration of Ag2S/Ag heterodimers is also examined in the TEM/SAED images of Figure 2 taken at different tilting angles. As reported in Figure S7, the thickness of nanoprisms increased only from ca 16 to 21 nm for the reaction times from 1 h to 5 days, while the edge-length of these nanoprisms increased by more than 19 times from around 10 to 190 nm. This result shows that in the presence of polyvinylpyrrolidone (PVP), the Ag atoms were preferentially deposited on the side edges (i.e., {100}) of the triangular Ag instead of the {111} planes. To understand the role of this capping agent, we further carried out a molecular dynamic simulation (SI-2) for the adsorption of PVP on three common low Miller-indexed Ag surfaces. More detailed information on this simulation can be found in SI-2 and Figures S8 to S11. On the basis of this study, it is validated that PVP can bind more favorably to the {111} planes than to the {100} and {110} facets of Ag nanoprisms under our reaction conditions, consistent with the resultant Ag prism morphology reported in Figure 1 and Figure S316,22.

In order to understand the structural evolution of pristine CdS nanospheres to final Ag2S/Ag heterodimers, X-ray diffraction (XRD) technique was also employed in this work. In Figure 3, a series of XRD patterns clearly confirms a gradual conversion of phase-pure hexagonal CdS (JCPDS card no. 41-1049; space group P63mc; lattice constants ao = 4.140 Å and co = 6.719 Å) to the final Ag2S/Ag composite (the monoclinic phase of Ag2S: JCPDS card no. 14-0072, space group P21/n, lattice constants ao = 4.229 Å, bo = 6.931 Å, co = 7.862 Å and β = 99.61°; and the fcc phase of Ag: JCPDS card no. 04-0783, space group , lattice constant ao = 4.086 Å). In excellent agreement with the observed thin prism morphology, the (111) diffraction of the Ag phase is predominant. Accompanied with the phase evolution, EDX and XPS analyses further demonstrate that the solid spheres of CdS were converted essentially into Ag2S/Ag heterodimers when the molar ratio of AgNO3:CdS was set at 4:1 (Figures S12 & S13). The as-prepared Ag2S/Ag heterodimers were also investigated with FTIR spectroscopy. As shown in Figure S14, Fourier transform infrared spectroscopy (FTIR) spectrum of the Ag2S/Ag heterodimers is almost the same as that of pure PVP, showing several main absorption peaks located at 3444, 2956 and 1656 cm−1, which can be assigned to the O–H, C–H and C = O stretching modes respectively. It should be noted that the absorption peak of C = O bond at 1663 cm−1 for the pure PVP is shifted to 1656 cm−1 for the PVP on Ag2S/Ag sample, which suggests a weak coordinative bonding of C = O (of PVP) to the surfaces of Ag nanoprisms. All these IR absorptions validate that the PVP macromolecules are anchored on the Ag2S/Ag heterodimers. To further confirm this kind of interaction, a surface analysis with X-ray photoelectron spectroscopy (XPS) technique was also performed (Figure S15). Indeed, the study reveals electronic interactions between the Ag2S/Ag and capping PVP, supporting the above FTIR findings.

Figure 3
figure 3

Structural characterization.

XRD patterns of as-prepared CdS solid spheres (a). and Ag2S/Ag heterodimers prepared with different mole ratios of AgNO3:CdS [(b). 1:1, (c). 2:1, (d). 3:1 and (e). 4:1].

As stated earlier, the molar ratio between AgNO3 and CdS is a crucial parameter for the formation of Ag2S/Ag heterodimers. When the AgNO3:CdS ratio was smaller than 2:1, the Ag phase was largely in the form of sphere-like particles (Figure S16). If the AgNO3:CdS was increased from 2:1 to 4:1, triangular Ag prisms became predominant in the product and the size of Ag prisms increased from smaller than 50 to near 200 nm (Figure S16). This is understandable, because the newly formed Ag phase has located on the surface of Ag2S upon the reduction by PVP; the prisms being formed can serve as favorable sites for continuous growth of silver. It is important to mention that only some irregular Ag particles could be generated in the absence of CdS solid precursor in the same reaction environment (EtOH-PVP), even with additional irradiation of UV-light (Figure S17). This indicates that the presence of Ag2S is essential for the formation of Ag phase under this reaction setting. Our synthetic experiments show that the molar ratio of AgNO3:CdS = 4:1 is an optimal condition for forming Ag2S/Ag heterodimers with large Ag prisms yet without generating unattended Ag crystals. The presence of PVP is indispensable for forming Ag prisms (Figures S18 to S23). In this regard, we had also varied the quantity of PVP used in synthesis and found that addition of 0.6 g of PVP is another optimal parameter to generate large Ag prisms (Figure S23).

To confirm the above reaction course, we further examined photon effect on the growth of Ag phase. Interestingly, similar Ag2S/Ag products could be produced in the presence of either UV or visible light (Figures S24 to S27). Nevertheless, the conversion from CdS spheres to the Ag2S/Ag heterodimers should be considered to be unrelated to the both kinds of light. In fact, the present formation reaction (AgNO3 + CdS-EtOH-PVP) can be easily carried out at room temperature (25°C) without any light irradiation (i.e., under dark condition (SI-1), Figures S28 & S29), or strong reducing agent (e.g., L-ascorbic acid, NaBH4, etc.), or addition of metal seeds. It is well-known that metallic Ag nanoparticles could be produced from AgNO3 in the presence of ethanol and PVP or other nonionic surfactants37,46,47,48. In addition, our work indicates that AgNO3 (dissolved in aqueous solution) can also react with CdS-PVP suspension in either aqueous or organic media (methanol, ethanol, 2-propanol, or ethylene-glycol) to generate Ag2S/Ag heterodimers (Figures S30 to S31). The Ag2S/Ag products can be easily stored as well. For example, the Ag2S/Ag heterodimers are extremely stable in ethanol solvent at room temperature for a period of more than one year (Figure S32). As they are decorated with PVP, the heterodimers can be easily dispersed in polar solvents such as water or alcoholic organics.

Discussion

In our synthesis, the morphology of the Ag2S/Ag was found to be dependent strongly on reaction parameters such as the process time, the molar ratio between AgNO3 and CdS and the amount of PVP added. For instance, our time-dependent experiments in Figure 4 show that the addition of AgNO3 into the starting CdS-EtOH-PVP suspension caused an immediate color change from yellow to gray and then to black, indicating that such Ag2S/Ag heterodimers were formed instantaneously. Due to an extremely small Ksp of Ag2S (1.0 × 10−49, 18°C), compared to that of CdS (8.0 × 10−27, 18°C), the irreversible cation-exchange reaction between Cd2+ and Ag+ takes place rapidly upon the addition of AgNO3. Apart from the color change, the morphology of Ag phase evolves significantly upon the reaction time; it transforms from sphere-like nanoparticles to plate-like nanoparticles and finally to triangular nanoprisms. The edge length of the attached Ag nanoprisms could be adjusted through controlling growth time (Figure 4). This product evolution is also reflected in the UV-visible-NIR absorption spectra displayed in Figure 5a. Similar to the previously reported results42, all the spectra show a broad absorption band of Ag2S centering around 497 nm. With extension of reaction time, these spectra gradually display three explicit peaks located at 342, 932 and 1040–1090 nm, respectively. Based on the previous studies17,49,50, in particular, the first two peaks at shorter and mid wavelengths can be attributed to the out-of-plane and in-plane quadrupole excitations23; the red-shift in the latter could also arise from the existence of some truncated triangular prisms in the product50. The gradually increased peak in around 1040–1090 nm in NIR region could result from the in-plane dipolar excitation mode related with the prisms whose edge lengths exceeded 100 nm18,23, which agrees well with the TEM observation of Figure 4. Importantly, the sharp peak at 1073 nm with a very narrow peak width observed in these Ag2S/Ag heterodimers is an unprecedented phenomenon to the SPR investigations of phase-pure silver nanostructures, which may suggest an additional well-defined excitation resonance resulting from effective coupling of this biphasic system at this wavelength.

Figure 4
figure 4

TEM images of Ag2S/Ag heterodimers.

The sample were prepared in the darkroom (Figures S28 & S29) at room temperature for different reaction times: (a). 1 h, (b). 5 h, (c). 10 h, (d). 18 h, (e). 2 days, (f). 3 days, (g). 4 d, (h). 5 days and (i). 7 days.

Figure 5
figure 5

Optical absorption properties.

(a). Time-dependent UV-visible-NIR absorption spectra showing the evolution of Ag2S/Ag in the dark at room temperature (i.e., the samples of Figure 4) and (b). Simulated absorption spectra of the samples in (a).

To provide an in-depth insight to the change of SPR properties, the UV-visible-NIR absorption spectra have been simulated in this work using the dimensional data of the time-dependent TEM results (Figure 4). The average diameter of Ag2S nanospheres used in the simulation was set at 125 nm and the thickness of triangular Ag prisms was changed from 16 to 20 nm and the edge length was allowed to vary from 10 to 210 nm. In addition, truncation of the Ag prisms grown from 3 to 7 days was also taken into account. Figure 5b shows the simulated absorption spectra at different reaction times (Note: The time thereafter is actually correlated to dimensional data of the heterodimers). Initially at 1 h, a broad absorption peak at 340 nm (i.e., Ag2S absorption) is observed. With increase in reaction time, three additional peaks appear at 500, 930 and 1070 nm, respectively. Quite encouragingly, these predicted peak positions are in accordance to those from the experimental measurements; even the sharp peak detected at 1073 nm (Figure 5a) is closely matched. The discrepancy in the peak shape and intensity at the NIR region can be attributed to multiple modes of attachment in the real samples, since we only performed the simulations for edge-attached heterodimers. Figure 6 gives the near-field images for the samples prepared at early reactions (1, 5, 10 and 18 h) plotted as intensity |E|. The incident light is perpendicular to the image plane, propagating in the x-axis and polarized in the z-axis. The wavelength of the incident light is based on the highest peak from the corresponding extinction spectra (Figure S33). From 1 to 18 h (i.e., when Ag prisms are small), the Ag2S sphere dominates the image and the lowest intensity is observed within the small triangular Ag prism. The image at 1 h is quite similar to that of a sphere49, in which the two edges possess higher intensity. With increase in reaction time, the Ag prism starts to grow with increasingly stronger intensity on its tip and significant near-field enhancement is observed on the extruding tip of the prism. The near-field enhancement for the larger Ag prisms (i.e., 2, 3, 5 and 7 days) is also very interesting. As shown in these images, the direction of incident light is the same for all the images, but the incident wavelength differs, since the highest peak of extinction curve shifts from 340 to 1070 nm (Figure S33). The observed red shift is primarily caused by the larger size of Ag prism in the latter samples. Comparing to the near-field image of 18 h sample, an enhancement for the sample grown after 2 days is observed on the two side-tips near the Ag2S sphere (Figure 6)50, noting that different incident wavelengths used in these simulations. Regarding the maximum intensity, there is also a substantial jump for the Ag prisms grown at 18 h to 2 days, increasing from 3.14 to 75.3. Such a near-field enhancement can also be augmented at the junction between two closely spaced particles. Therefore, the interface exhibits a higher intensity compared to normal triangular prism49. For all the samples grown during 2 to 7 days, there is no discernible near-field enhancement within the Ag2S sphere. This further confirms that the small Ag2S sphere has negligible contribution to the enhancement at long wavelengths, but indeed it causes significant degeneracy in SPR properties due to desymmetrization of the Ag prisms. On the basis of these trial simulations, furthermore, one can now understand that the SPR properties depending on the size and location of Ag prisms when the size of Ag2S sphere is fixed and the prepared heterodimers can be activated by the incident light across a wide span of photon wavelengths. Because the triangular Ag prisms in Ag2S/Ag heterodimers have distinctive plasmonic enhancement modes for their two sets of tips (Figure 6) with different incident photons, this unique feature may enable these Ag2S/Ag heterodimers for new applications.

Figure 6
figure 6

Investigation on surface plasmonic resonance.

Near-field images simulated for different edge-attached Ag2S/Ag heterodimers at different growth times (refer to Figure 4, h = hour and d = day).

To demonstrate this point further, we have used these samples for catalytic antibacterial application under either visible light exposure or darkroom condition (Experimental Section). In this set of tests, the photocatalytic ability of Ag2S/Ag heterodimers for deactivation of E. coli was evaluated with a cell-concentration of about 1.0 × 108 CFU/mL, as reported in Figure 7. The reaction mixture containing E. coli cell suspension and the Ag2S/Ag catalyst was placed under visible light for 0–40 min. As a reference, E. coli without Ag2S/Ag was also irradiated for 50 min by UV light. Without catalysts, UV-light irradiation did not cause obvious bactericidal effects on E. coli (case a). However, the inactivation of E. coli was notably increased in the presence of our Ag2S/Ag heterodimers under visible light, as compared to UV irradiation alone. The detected inactivation efficiencies were improved more than 5.0 log10 units within 10 min and the E. coli cells could be largely deactivated within 20 min (case b) in the presence of only 0.01 mg/mL of Ag2S/Ag heterodimers. The same amount of E. coli was completely killed within 5 min when the concentration of Ag2S/Ag heterodimers was increased to 0.03 mg/mL under the same visible light exposure (case c); the curves within 5 min are detailed in Figure S34. The deactivation of E. coli was also evaluated in dark with the presence of Ag2S/Ag heterodimers. Without light illumination (Figure 7d), however, the deactivation of E. coli is one or two orders in magnitude lower than that using the same amount of Ag2S/Ag (0.01 mg/mL) but under the visible light. Nevertheless, the bacteria could also be completely inactivated at a longer reaction time (30 min, Figure 7d). In general, the inactivation efficiency is higher for larger Ag prisms, in comparison to that with smaller prisms (case e) (SI-1 & Figure S34). Under UV-light assistance, the synergetic effects of Ag2S/Ag on bactericidal application have been explored in our previous work, where the Ag2S is in a hollow sphere structure and the Ag is in an uncontrolled spherical shape in the studied heterodimers37. Without any light-assistance, in the present work, in contrast, it is surprising to see this new Ag2S/Ag system with triangular Ag prisms can function much better in the same bactericidal application (Figure 7d). More importantly, it is clearly evidenced in Figure 7 that under the visible light, remarkable improvement in antimicrobial performance of Ag2S/Ag heterodimers can be further attained.

Figure 7
figure 7

Inactivation efficiency against E. coli K-12.

(a). E. coli suspension without Ag2S/Ag irradiated by UV light, (b). E. coli suspension with 0.01 mg/mL of Ag2S/Ag heterodimers (prepared with standard conditions, Methods) under visible light, (c). E. coli suspension with 0.03 mg/mL of Ag2S/Ag heterodimers (prepared with standard conditions, Methods) under visible light, (d). E. coli suspension with 0.01 mg/mL of Ag2S/Ag in the dark condition and (e). E. coli suspension with 0.01 mg/mL of Ag2S/Ag with smaller triangular Ag prisms (edge-length 60–90 nm, Figure S30) under visible light.

To gain more understandings on the roles of the heterodimers in the above photocatalytic processes, the morphology of E. coli at different stages of bactericidal experiments in the presence of 0.01 mg/mL of Ag2S/Ag was investigated by TEM and FESEM methods (Figures S35 & S36). Before light exposure (Figures S35 & S37), the ellipse-like E. coli bacteria held a well-defined cell wall and evenly distributed interior content. However, changes had taken place to the E. coli cells, when they were exposed to visible light for only 5 min (Figure S35); parts of cell membranes became disintegrated, suggesting that the decomposition of cells started from their membranes. Interestingly, it should be noted that Ag2S/Ag heterodimers, especially the corners of Ag nanoprisms were bound to the outer membranes of the cell walls. Structurally, the tips of Ag prisms could behave like a “harpoon”: once they contacted the surfaces of the cells, degradation of cell membrane would likely occur. Besides, the reaction system was well stirred and the frequency of the collision between the corners of Ag prisms and the cell walls must be high. In a certain sense, the Ag2S/Ag heterodimers can function as effective antennas for different incident photons in view of their outstanding photo-absorbing ability (Figure 5) across UV-visible to NIR region. Furthermore, different sets of plasmonic “hot spots” at the corners of Ag prisms could be generated in accordance to the wavelengths of the absorbed photons (Figure 6). It is our belief that such intense localized electromagnetic fields are responsible for the observed cell membrane degradation. This finding is indicated more explicitly in Figure 8 for the E. coli cells with 10 min of light exposure; the outer walls largely disappeared and rumples/holes generated (Figure S35). With the time to 20 min, the outer cell walls disappeared completely and the amount of holes increased (Figure S35). On the contrary, the used Ag2S/Ag heterodimers remained integral showing high structural stability under light irradiation. In order to exclude possible dissolution of Ag+ ions from the above Ag2S/Ag catalysts, we further used ICP-MS to determine the concentration of Ag+ ions during the bactericidal processes. From Table S1 (SI-1), it can be seen that the Ag+ concentration remained at an extremely low level of less than 2 ppm and did not change even after overnight reactions. EDX technique was also employed to determine the molar ratio of Ag:S for the spent catalysts. The measured molar ratio of Ag:S of the Ag2S/Ag heterodimers after exposed by visible light for 30 min and placed with the E. coli cells overnight was 2.535, which was almost identical to the value (2.540) before their use (Table S1). On the basis of both ICP-MS/EDX results, it can be thus concluded that no net Ag+ ions were released from the Ag2S/Ag catalyst into the solution phase during the bactericidal processes.

Figure 8
figure 8

TEM images of an E. coli cell together with two Ag2S/Ag heterodimers.

The test was carried out under visible light illumination for 10 min. Insets indicate the detailed views of the heterodimers. Note that the cell surface is decorated with sodium phosphotungstate (i.e., tiny crystallites scattered around the cell at lighter image contrast).

In summary, this work is significant in the following aspects. First, it adds a new structural dimension and a semiconducting phase to the well-established triangular Ag nanoprisms, offering new research opportunity for investigating Ag-prism-containing nanocomposites. Second, it serves as a model example to bridge two important families of inorganic nanostructures (i.e., Ag nanoprisms and semiconductor/metal asymmetric heterodimers). And third, it demonstrates new possibility to employ the chemistry of cation-exchange, in combination with redox reactions and with assistance of stabilizing agents, to prepare noble-metals into complex shapes for asymmetric heterostructures. As investigated above, this new type of photosensitive heterodimers may find future applications, including bacterial deactivation, bio-imaging and sensing, as well as heterogeneous catalysis under ambient conditions, by utilizing their synergistic properties of the biphasic architecture.

Methods

Chemicals

Cadmium nitrate (Cd(, 99.5%, Merck), thiourea (TU, 99%, Merck) and silver nitrate (AgNO3, AR@(ACS), Mallinckrodt Chemical), di(ethylene glycol) (DEG, 99%, Sigma-aldrich), polyvinylpyrrolidone (PVP K30, MW = 40,000), Fluka), sodium citrate dehydrate (, 99+%, Aldrich), ethanol (C2H5OH, absolute for analysis, ACS, Merck), acetone (C3H6O, HPLC, TEDIA). All reagents were used as received without further purification.

Synthesis of CdS colloidal nanospheres

The CdS colloidal nanospheres precursor was prepared through high-temperature polyol-mediated reaction. Briefly, 1.0 mmol of Cd(NO3)2·4H2O, 0.83 g of PVP and 10 mL of DEG were poured into a three-necked flask. When the mixed solution was heated from room temperature to 160°C, 1.0 mL of thiourea (TU, 99%)/DEG (2.0 mmol of TU) stock solution was injected rapidly into the hot solution. The solution turned to yellow when it reached 160 to 185°C (i.e., formation of CdS). The CdS suspension was kept at 185°C for 1 h. The product was rinsed 3 times by precipitation with a mixture of acetone and ethanol followed by centrifugation at 5900 rpm for 10 min and finally dispersed in ethanol or deionized water for further use.

Seed-free preparation of triangular Ag2S/Ag heterodimers

In a typical synthesis, 0.2 mL of CdS suspension (0.05 M in ethanol, prepared above), 0.6 g of PVP and 2 mL of ethanol were poured into a clean vial. The vial was then sealed by rubber septum and vigorously stirred for 20 min. After that, a given amount of AgNO3 aqueous solution (0.4 mL, 0.10 M) was added dropwise into the above mixture and then stirred continually for 5 days in darkroom condition at 25°C; we used black plastic film to wrap the sample vials and all the experiment steps were handled in the darkroom. After reaction, the black precipitate was harvested by centrifugation and washed with the mixture of acetone and ethanol more than 3 times and the sample was named as normal Ag2S/Ag heterodimers. Some comparative experiments were also performed with identical synthetic procedures but under the irradiation of common white lamps or UV light. Unless otherwise specified, the molar ratio between AgNO3 and CdS was set at 4:1 and the amount of PVP was 0.6 g.

Overgrowth of triangular Ag prisms

Based on the above description, initial Ag2S/Ag heterodimers were prepared by reaction in dark condition for 4 days. The Ag2S/Ag heterodimers then served as seeds, to which about 0.2 mL of 0.10 M AgNO3 solution was added and stirred for another 2 days. The product was collected and washed using the same procedures described above.

Antibacterial tests with Ag2S/Ag heterodimers

E. coli K-12 was cultured in a nutrient broth at 37°C overnight (i.e., 16–18 h) at 200 rpm in a rotary shaker to obtain the first generation cells. Then 1 mL of the first generation cultured suspension was taken out and transferred to another 30 mL of nutrient broth for another 4 h culture at 37°C at 200 rpm in the same rotary shaker until reaching the log phase. Afterward, the E. coli cells were collected by centrifugation at 3500 rpm for 10 min and the bacterial pellets were washed twice with sterilized Milli-Q water to remove residual culture media components. The as-prepared cells were then re-suspended and diluted to the required cell density of around 108 colony-forming units per milliliter (CFU/mL) with sterilized Milli-Q water. 50 mL of E. coli suspension and a certain amount of the normal Ag2S/Ag heterodimers (0.01 and 0.03 mg/mL, respectively) were added to a glass bottle. To ensure good mixing, the resultant suspension was magnetically stirred (set at 700 rpm) under light irradiation; a visible light lamp (30 W) at the ambient condition was used as a light source. The distance between the lamp and the reaction bottle was ca. 50 cm. Before and during the light exposure, a 0.50 mL aliquot of the reaction mixture was withdrawn at given time intervals and diluted serially with sterilized Milli-Q water to adjust the bacterial concentration to ensure the growing bacterial colonies could be counted accurately and easily. In this connection, 0.10 mL of the diluted mixture was spread on a nutrient agar medium and the colonies were counted to determine the viable bacterial numbers after being incubated at 37°C for 18–24 h. All of the bactericidal experiments were carried out at room temperature and repeated no less than 5 to 10 times in order to ensure experimental accuracy; the measured data for each set of experiments were expressed with the mean and standard deviation. The diluted E. coli suspension without addition of the Ag2S/Ag heterodimers was irradiated with a Philips HPR UV light source (a high pressure mercury lamp (125 W, the main emission wavelength: 365 nm (UV-A) and luminous efficacy: 23.2 lm/W) and a UV cutoff filter (λ > 290 nm), with other experimental conditions unchanged and was used as a control reference. In addition, culture experiments on the reaction mixture (i.e., E. coli together with normal Ag2S/Ag heterodimers) in darkroom condition were also performed for comparison.

Details on the instrumental analysis and materials characterization for all the investigated samples in this work can be found in Supplementary Information (SI-1).

Simulation models and methods

Vinylpyrrolidone monomer was optimized using density functional theory (DFT)51. The atoms of PVP and Ag were mimicked by the COMPASS force field52. Molecular dynamics (MD) simulations were carried out in a canonical ensemble for three different types of systems, namely PVP chain, Ag surface and the complex with PVP adsorbed on Ag surface. The absorption spectra, extinction spectra and surface plasmonic properties were simulated using discrete dipole approximation (DDA) as implemented in open source software namely DDSCAT53. The refractive index data of Ag and Ag2S were adopted from the literature54,55,56. Further information on the simulations for PVP and Ag interactions can be found in Supplementary Information (SI-2).