Introduction

Noble-metal nanocrystals have received considerable attention in recent years for their size- and shape-dependent plasmonic properties, and they have been used in various application areas including photonics [1], electronics [2], catalysis [3], sensing [4], and biomedicine [5]. Highly uniform silver (Ag) nanostructures of different shapes (spheres [6], rods [7], bars [8], belts [9], wires [10], prisms [11], disks [12]) have been widely reported in the literatures. It is interesting to note that plasmonic applications associated with the longer wavelength region (e.g., 500 nm and above) of the visible spectrum is largely covered by gold (Au) nanoparticles (NP)—instead of silver nanoparticles—despite that gold offers relatively lower field enhancement factors because of higher losses. Until now, reported cases of AgNPs primarily cover the UV-blue region. This impedes widespread applications of AgNPs. With the aim of extending the wavelength of localized surface plasmon resonance (LSPR) in AgNPs, Xia et al. successfully synthesized silver nanocubes that have LSPR peaks in blue-green region (420–500 nm) [13, 14]. On the other hand, silver nanodecahedrons (AgNDs) should hold some promise on expanding LSPR wavelength further in to the red region. Because of their sharp corners/edges, AgND solutions demonstrate remarkable optical properties including narrow scattering peaks and highly reproducible transmission spectra [15, 16]. Despite such special optical properties, literature reporting the growth of AgNDs is quite limited. Earlier works on the synthesis of AgNDs using photochemical [15, 17] or N, N-dimethylformamide (DMF) reduction [18] methods have been reported by several groups. Kitaev et al. reported that growth of nanodecahedrons in a mixture of NP seeds and precursor solution can be activated by white light illumination [16]. While our experimental results indicate that a broadband light source will lead to the formation of poly-dispersive Ag nanoparticles (see Supplementary Information), it is the aim of this work to investigate the possibility of controlling the growth uniformity by using varying the wavelength of the light source. Our experiments reveal that when wavelength of the LED illumination matches with the LSPR wavelength of the AgND seeds, a seed-mediated plasmon-driven regrowth mechanism will occur, and the resultant product is highly mono-dispersive AgNDs with LSPR anywhere in the range of 489–590 nm.

Experimental Section

Chemicals

Silver nitrate (ACS reagent, ≥99.0%), Poly(vinyl pyrrolidone) (PVP, Mw = 40,000 g/mol), l-Arginine (BioUltra, ≥99.5%), and 4-Methylbenzenethiol (98%) were purchased from Sigma-Aldrich. Sodium citrate dehydrate and sodium borohydride (99%) were purchased from Fisher Scientific. All chemicals were used with no further purification. Water used in the synthesis experiment was purified by Spectra-Teknik ultrapure water purification system. High-brightness light-emitting diodes (LEDs) with emission peaks at 465, 500, 520, and 578 nm were purchased from RS Components.

Synthesis of Ag Nanodecahedron Seeds

Silver nanodecahedron (ND) seeds were prepared using the method reported in ref. [16]. A 0.5 mL trisodium citrate (0.05 M), 0.015 mL poly(vinyl pyrrolidone) (PVP, Mw = 40,000 g/mol, 0.05 M), 0.05 mL l-arginine (0.01 M), and 0.15 mL silver nitrate (0.005 M) were mixed with 7 mL deionized water in a clear 10 ml beaker. An 80-μL ice-cold sodium borohydride (0.1 M) was rapidly injected into the mixed solution with continuous stirring. The solution was incubated for 2 h in a dark environment at room temperature. The final bright yellow solution, which was also used for precursor in the following regrowth steps, was exposed to 465 nm radiation from a blue LED for 6 h.

Seed-mediated Regrowth of Ag Nanodecahedrons

In a typical synthesis experiment, 2 mL AgNDs seed solution was mixed with 2–4 mL precursor solution. The final mixture was then exposed to LED illumination for 6–9 h depending on the light intensity. The spectral peak of the LED was chosen such that it was close to the LSPR peak of the ND seeds. All experiments were conducted under room temperature.

SERS Assessment of Ag Nanodecahedrons

In a standard procedure, 100 μL AgNDs colloid after two times centrifugation wash was mixed with 900 μL 0.1 mM 4-methylbenzenethiol (4-MBT) solution in ethanol for 1 h. Then, 10 μL of the mixture was transferred to the surface of a silicon substrate and left to dry. The substrate was then rinsed by ethanol thoroughly to remove any unbounded 4-MBT. SERS measurement of single ND was conducted immediately after sample preparation. Normal Raman spectroscopy was measured with solution of 14 mM 4-MBT in NaOH (~6 M) for the calculation of average enhancement factor (EF) of the single AgND.

Characterization

Field-emission scanning electron microscopy (FESEM) was performed using a Carl Zeiss Ultra Plus FESEM and the operating voltage was 4.96 kV. Transmission electron microscopy (TEM) images were acquired using a FEI CM120 microscope at 120 kV. The average size was determined from SEM and TEM images by averaging the edge length measurements obtained from at least 100 AgNDs. Extinction spectra were measured using a Shimadzu UV-3101 UV/Vis/NIR scanning spectrophotometer. SERS spectra were recorded using a Renishaw inVia confocal Raman spectrometer coupled to a Leica microscope. Backscattered photons were collected using a × 100 objective (NA = 0.85). The 514-nm excitation source was obtained from an argon laser. The scattering spectra were recorded in the range of 550–2,000 cm−1. Except for the normal Raman spectrum of 4-MBT, which required 20 mW and 30 s, typical accumulation time was 10 s per data point, and laser power was 1 mW.

Results and Discussions

Variation of LSPR Peak with Size of Silver Nanodecahedrons

Table 1 shows the extinction peak location of the final products produced from various illumination wavelengths and seed-precursor volume ratios. Extinction spectra of the products are shown in Fig. 1a, and their corresponding visual color in reflection and transmission modes are respectively shown in Fig. 1b, c. The full-width-half-maximum (FWHM) of the extinction spectra of sample 1 is about 45 nm, which indicates that the colloidal silver NDs are highly symmetric and of small size variation. Their narrow FWHM also signifies that the silver NDs may offer attractive figure-of-merit (\( {\text{FOM}} = \frac{1}{\text{FWHM}} \bullet \frac{{\Delta \omega }}{{\Delta n}} \)) [19], for sensing applications. The gradual widening of FWHM with increasing red shift is an intrinsic effect characteristic to increasing the particle size of the NDs [20]. The weak peaks nearer to the blue region can be attributed to the silver NDs' transverse resonance (see Fig. S2 in Supplementary Information) and dipole resonance of a small amount of unconsumed nanoparticles. In addition, the bi-color character (i.e., different sample colors in reflection and transmission modes) of the colloid solutions shown in Fig. 1b, c indicates that the silver NDs have low size variation and possess strong scattering coefficient in the visible region, i.e., light reflected from the colloid mainly composed of the wavelength components scattered by the silver NDs while scattered light has very limited transmission.

Table 1 Summary of final products obtained from exposing seed and precursor solution to LED illumination at various wavelengths
Full size table
Fig. 1
figure 1

a UV-vis extinction spectra and photographs of final silver ND colloids taken in b reflected and c transmitted modes

Full size image

Figures 2a–d show the FESEM and TEM images of silver NDs with extinction peaks at 516, 522, 541, and 550 nm, respectively. Apart from a small amount of nanoprisms and nanoplates, highly uniform silver NDs make up over 90% of the products (see Fig. S3 in Supplementary Information). As shown in Fig. 3a, side view images of the silver NDs, reveal that the distance between two vertices of the two back-to-back penta-prisms is almost the same as their ridge lengths, thus indicating that the enlarged silver NDs as grown from their seeds have preserved the original shape. This is quite different from the case of silver nanodecahedrons evolving into silver pentagonal nanorods as reported in the literature [7, 21]. In such situations, strong binding of PVP to the {100} face results in preferential deposition of silver to the poorly passivated {111} face. In our regrowth experiments, with both PVP and citrate acting as surfactants, no nanorod has been found in the final solutions. It is thought that the incident optical radiation induces strong electromagnetic field at the edges, which will in turn reduce the selective binding of PVP to the {100} face, hence allowing the growth of NDs from their seeds. TEM and FESEM images of at least 100 nanoparticles have been used for estimating the average dimensions of the silver NDs, and the relationship between the major extinction peaks and the average edge lengths is shown in Fig. 3b. Quantitatively the LSPR peaks of the silver NDs, i.e., 489, 516, 522, 550, 572, and 590 nm, are related to their edge lengths (standard deviation) in the order of 28.3 (±3.6), 38.9 (±4.0), 41.7 (±4.5), 53.5 (±5.1), 61.4 (±7.6), 67.4 (±4.7) nm, respectively. (Fig. S4 in Supplementary Information shows the size distribution histograms of the silver NDs and their respective extinction peaks.) A linear relation has also been found: \( l = 0.39 \times {\lambda_{{\max }}} - 163.33({R^2} = 0.998) \), where l and \( {\lambda_{{\max }}} \) are the edge length and position of major LSPR peak, respectively.

Fig. 2
figure 2

FESEM images of silver nanodecahedrons having extinction peaks at a 516 nm, b 522 nm, and TEM images of silver nanodecahedrons having extinction peaks at c 541 nm, d 550 nm. The scale bars in all images are 100 nm

Full size image
Fig. 3
figure 3

a TEM side view images of silver nanodecahedrons at different sizes. All red triangles are equilateral and the scale bars are 50 nm. b Calibration curve showing linear dependence between major localized surface plasmon resonance (LSPR) peak position and edge lengths of silver nanodecahedrons

Full size image

Regrowth Mechanism

The regrowth mechanism of our silver NDs is thought to be based on silver redox cycles, which involves reduction of Ag+ on the silver ND surface under LSP excitation with the assistance of citrate and oxidative dissolution of small silver particles by O2 [22]. During dipole excitation under LED illumination, a large number of hot electrons and hot holes are generated on the silver ND surface. Citrate bonding on the silver surface will be oxidized and an electron is released to neutralize the hot hole. At the same time, Ag+ in the solution will receive the hot electron to become Ag0, which then binds to the surface and finally results in the enlargement of silver ND. On the other hand, small silver nanoparticles introduced as precursor will be oxidized by O2 to release Ag+ ions, which serve as a material source for silver ND regrowth. As shown in Fig. 4, the major LSPR peak (~500 nm) of the silver NDs red shifts slowly with increasing LED illumination duration, thus suggesting the involvement of a single regrowth process during the enlargement of the silver NDs. In addition, the peak at ~400 nm decreases in intensity continuously, thus indicating that small silver nanoparticles are being consumed and gradually oxidized to form silver cations. For the small peak in the long-wavelength region (550–600 nm) during the early stage of the regrowth, we attribute this to the LSPR of small amount larger silver NDs. The seeds are quickly transformed into enlarged NDs in the presence of high concentrations of Ag+ shortly after mixing with the precursor solution. As growth continues, Ag+ concentration in the mixture very soon returns to a steady-state level. It should be noted that the growth of silver NDs is effectively driven by a dynamic equilibrium process, which allows the co-existence of material etching and regrowth, while the silver cation concentration in the solution is maintained. In the presence of LED irradiation, LSPR will result in the promotion of material growth. In our case, irradiation at ~500 nm will affect both small NDs with LSPR at ~516 nm as well as large NDs with LSPR in the range of 550–600 nm. Since the ~500 nm incident irradiation is close to the small NDs' LSPR peak of 516 nm, small NDs will experience much higher growth rate than that of large NDs [23]. While rapid growth of small NDs depletes silver cation concentration, the only pathway for maintaining equilibrium is to generate silver cations through oxidation of other nanoparticle species including large NDs. This phenomenon is particularly true during 5–6 h after the addition of precursor as most of the small silver nanoparticles have been consumed, and the deficiency of silver cations favors the etching of large NDs. This process will continue until the large NDs have become small enough so that their LSPR peaks are similar to that of the small NDs. This explains the observed growth in small NDs as indicated by a LSPR red shift, while large NDs undergo etching as indicated by a LSPR blue shift. In our experiment, we stop the growth process when we see that the FWHM of the extinction spectrum has reached its minimum. This typically takes 7 h.

Fig. 4
figure 4

Monitoring of regrowth after exposing a reaction mixture (2 mL of silver ND seed (LSPR ∼ 489 nm) solution and 4 mL of precursor solution) to 500 nm LED illumination

Full size image

It has been proposed in the literatures that silver precursor containing small silver nanoparticles and citrate will produce silver nanoprisms and nanoplates under the irradiation of 500 nm or longer wavelengths [15, 17, 24]. The small amount of silver nanoprisms and nanoplates found in the final product also confirms that the same reaction has taken place in our experiment. However, nanoprisms and nanoplates only amount to less than 5%. In the present case, LSP excitation has drastically increased the speed of reduction from Ag+ to Ag0 at the surface of the silver NDs. This leads to a much faster regrowth rate of silver NDs than the formation self-transforming seeds into nanoplates, as supported by ref. [23]. These two competitive processes, as highlighted in Scheme 1, can be controlled by the use of a light source. When LED emission peak matches with the LSPR peak of the silver ND seeds, formation of silver nanoprisms and nanoplates in the solution will be suppressed.

Scheme 1
scheme 1

With the use of radiation close to the LSPR peak of the silver NDs seeds, regrowth of silver ND is much more favored then the formation of nanoprisms and nanoplates

Full size image

LED Irradiation at Wavelength away from LSPR Peak of Silver ND Seed

Figure 5a, b are respectively the UV-vis extinction spectra of the final products obtained from exposing the reaction mixture (2 mL of silver ND seed solution (LSPR ∼ 556 nm) mixed with 4 mL of precursor solution) to LED irradiation at 465 nm and 578 nm for 9 h. The 465-nm case has resulted in slight growth of the seeds and the addition of another strong peak at 496 nm. This peak belongs to newly formed silver NDs having smaller sizes, which makes the final product more poly-dispersive. For the 578-nm case as shown in Fig. 5b, the peak due to the silver ND seeds remains nearly unchanged, signifying the absence of any regrowth in the solution, while a new peak at 660 nm associated with silver nanoprisms and nanoplates has emerged. FESEM images of the product also support the presence of these nanoparticles (see Fig. S5 in Supplementary Information). The peak at around 400 nm is due to unconsumed silver nanoparticles introduced from the precursor. The reason for the observed enlargement of silver NDs seeds under 465-nm irradiation and nearly no regrowth in the 578-nm case could be attributed that photon energy of 578 nm is not sufficient to excite hot electrons and holes on the silver ND surface. Thus, little or no reduction can happen on the silver ND surface, while the only reaction to take place will be for the small silver nanoparticles to self-transform into nanoprisms and nanoplates.

Fig. 5
figure 5

Extinction spectra of silver ND seeds (LSPR ∼ 556 nm) and precursor after LED irradiation at a 465 nm and b 578 nm. Dashed lines are for the extinction of ND seed solutions, and solid lines for the final products

Full size image

SERS Characterization of Silver Nanodecahedrons

Figure 6 shows the SERS spectra measured separately from individual samples of silver NDs whose extinction peaks are 489, 522, 550, and 590 nm. Each spectrum is obtained by averaging the spectra of 20 different sites. Majority of the Raman peaks are associated with different vibration modes of 4-MBT [25, 26]. The silicon substrate also contributes a broad peak in the 900–1,000 cm−1 region. In order to establish the correlation between the size of the silver NDs and their SERS EF, the amplitudes of two characteristic peaks of 4-MBT at 1,077 and 1,581 cm−1 are monitored. The method for calculating the average EF of single silver ND we adopted the equation: average EF = (I sers × N bulk)/(I bulk × N sers), where I sers and I bulk are respectively the average area of the Raman peak of interest in the 20 SERS and the same Raman peak area in bulk spectra. N sers and N bulk refer to the number of molecules on a single silver ND and in the focal volume of the 14 mM 4-MBT solution, where we assume the footprint of a single 4-MBT molecule on the silver surface is 0.19 nm2 [27], and the focal volume of our Raman system is measured to be 10 fL according to ref. [28]. The average EFs our silver NDs possessing different extinction peaks are shown in Table 2. Silver NDs with extinction peak at 522 nm provides the largest enhancement for both Raman bands, while ND with 580 nm peak provides comparatively lower enhancement. This is due to a mismatch between the extinction peak and the Raman excitation laser wavelength (514 nm). When the irradiation wavelength from the source coincides with the resonance peak, the electric field around the silver nanoparticle under is much stronger. Since the tunability of the present approach that can provide Ag NDs with extinction peaks anywhere in the range of 489–590 nm, one can always achieve an optimal EF situation for any given Raman excitation wavelength.

Fig. 6
figure 6

SERS spectra of individual single silver ND samples whose extinction peak are at 489, 522, 550, and 580 nm. Dotted rectangle region is the enlarged region of spectra within 1,537–1,606 cm−1

Full size image
Table 2 Summary of calculated average EFs of different sizes silver NDs at 1,077 cm−1 and 1,581 cm−1 band
Full size table

Conclusion

By controlling and tuning the mixing ratio of seed and precursor concentrations, uniform silver nanodecahedrons (NDs) with major localized surface plasmon resonance (LSPR) ranging from 489 to 590 nm have been synthesized by a seed-mediated plasmon-driven method. It is found that two processes are happening in the synthesis. The choice of LED illumination wavelength that matches with the LSP peak of the silver ND seeds can drastically suppress the formation of nanoprims and nanoplates, thus improving the uniformity of the desired nanostructures. Due to their sharp tips and edges, silver NDs provide up to 106 electric field enhancement, with the possibility of further increase if one optimize the matching between the LSP peak and the irradiation laser wavelength. The reported silver NDs should find applications in biosensing and bioimaging.