Precisely tuning the longitudinal localized surface plasmon resonance of gold nanorods via additive-regulated overgrowth

Gold nanorods (GNRs) with desired longitudinal localized surface plasmon resonance (LLSPR) and strong scattering intensity are important for extending their practical applications in bioimaging and sensing. Herein, a simple additive (HCl and Na2S)-regulated overgrowth approach has been proposed for preparing GNRs with tunable LLSPR. In this approach, HCl is used to slow down the growth reaction rate by changing chemical equilibrium, while Na2S is utilized to halt the reaction when LLSPR is reaching the expected wavelength under monitoring by a UV-Vis spectrometer. Under optimal conditions, GNRs with an LLSPR range from 850 to 650 nm could be facilely prepared with a high precision of 3 nm deviation. The TEM images reveal that GNRs have high monodispersity, displaying an increase in both length and diameter but a decrease in the aspect ratio. With the increase in size, the produced GNRs show enhanced scattering intensity and are applicable for single nanoparticle imaging due to the enlarged absorption and scattering cross-section and improved matching efficiency toward the CCD response.


Introduction
Gold nanorods (GNRs) with unique optical and surface properties have attracted increasing interests in sensing, phototherapy, gene and/or drug delivery, and bioimaging. [1][2][3][4][5][6] As the most intriguing and important feature of GNRs, the longitudinal localized surface plasmon resonance (LLSPR) originating from the collective oscillations of free electrons and conned to the nanoparticle surface can range from visible to near-infrared regions. 7 Unlike the LSPR of spherical gold nanoparticles, the LLSPR of GNRs is only dependent on the aspect ratio (AR, length/diameter) and is important for their further applications. 8,9 For example, GNRs with LLSPR in the range of 750-1300 nm are more favorable as in vivo imaging probes because of the low background from tissues in this wavelength window. 10,11 Recently, the anisotropic characteristics have made GNRs promising optical tools for understanding the biological events at the single particle level through monitoring both translational and rotational motions. 12,13 Wavelengthdependent photo collection efficiency of optical detectors (e.g., charge-coupled devices (CCDs) and photomultipliers) and a high signal-to-noise ratio can be realized when LLSPR matches the detector's optical response curve. 14 Besides, for photothermal therapeutic applications, shorter or longer LLSPR than the laser wavelength would reduce the light-heat efficiency and decrease the therapeutic effect. 15 Therefore, the development of facile routes to repeatedly synthesize LLSPR-tunable GNRs is of signicance to construct versatile GNR-based optical platforms.
Toward this goal, much efforts have been dedicated to the alteration of AR and geometry with various synthetic strategies. In general, a seed-mediated growth method, with the advantages of simplicity, high yield, size tunability, and monodispersity, has been widely applied in the preparation of GNRs with designable LLSPR properties. 1,16 For example, GNRs with a supreme LLSPR of 981 nm and uniform morphology are synthesized under alkaline conditions by replacing ascorbic acid (a reducing agent) with H 2 O 2 . 17 With the addition of aromatic additives during seed-mediated synthesis, GNRs with tunable LLSPR beyond 1000 nm and improved purity are obtained. 18,19 However, seed-mediated synthesis sometimes shows low reproducibility, limiting their applications. 20 To overcome this concern, post-synthesis, that is, subjecting the presynthesized GNRs to further oxidation or overgrowth has been applied in the production of LLSPR designable GNRs. [21][22][23][24][25] For instance, the length of GNRs can increase/remain with/without adding thiol molecules. 26,27 In comparison to oxidation, the overgrowth strategy is more valuable to yield GNRs with high light absorption and scattering efficiency by increasing their sizes. 26 Despite the tunable LLSPR of GNRs with the overgrowth strategy, tuning precision is still a problem. Thus, the exploration of a simple approach for precisely and reproducibly tuning the LLSPR of GNRs is still appealing.
In this study, a simple and reproducible strategy for the preparation of GNRs with the precise tuning of LLSPR through HCl and Na 2 S-regulated overgrowth was proposed. The addition of HCl can slow the overgrowth reaction rate by altering chemical equilibrium, while the addition of Na 2 S leads to the depletion of the gold element due to the formation of Au 2 S in the solution, which terminates the overgrowth process, as shown in Scheme 1. By simply changing the amount of HCl and Na 2 S, GNRs with tunable LLSPR from 650 to 850 nm could be easily produced with a precision of 3 nm. Single nanoparticle imaging was also conducted with darkeld microscopy, proving that GNRs with LLSPR at 650 nm showed the strongest scattering because of the largest cross-section and the best matching of LLSPR with the CCD response. This work provides a promising strategy to produce desired GNRs with precise LLSPR and strong scattering intensity, which may facilitate the assays at the single-nanoparticle level.

Characterizations
UV-Vis absorption spectroscopy was performed on a UV-1800 Spectrometer (Hitachi, Japan). TEM images were obtained using a JEM 1230 transmission electron microscope operating at 100 kV (JEOL, Japan). Single GNR images were acquired with an NI-U upright microscope (Nikon, Japan), which was equipped with a 100 W halogen tungsten lamp, an oil immersion darkeld condenser (NA 1.20-1.43), a 40Â plan uor objective, and a DP 73 camera (Olympus, Japan).

Synthesis of GNR seeds
GNR seeds for further overgrowth were prepared through a seed-mediated method according to previously reported methods. 28,29 Briey, small spherical gold nanoparticles with a diameter of around 3 nm were initially obtained by reducing Au 3+ with a strong reducing agent. Then, 48 mL of ice-cold 0.010 M NaBH 4 was injected into 8 mL solution containing 0.1 M CTAB and 0.00025 M HAuCl 4 under vigorous stirring. Then, this solution was kept at 28 C for at least 2 hours prior to use. For GNR-seed synthesis, 412 mL of 24.28 mM HAuCl 4 , 0.5 mL of 4 mM AgNO 3 , and 105 mL of 0.1 M AA were added to 20 mL 0.1 M CTAB solution in order. Aer shaking this solution to make the colour change from bright yellow to colourless, 60 ml of the nanoparticle solution was further injected. The mixture was shaken for 20-30 seconds again and then kept at 28 C for 4 h. In order to assess the reproducibility of GNR synthesis, three researchers repeated the experiment with the same recipe.

HCl and Na 2 S on the overgrowth of GNRs
Understanding and controlling the dynamics of GNR overgrowth in solution are essential to attune the LLSPR of GNRs. The effect of pH on the dynamics was examined by adding HCl or NaOH. First, 72 mL of 24.28 mM HAuCl 4 and 26.7 mL of 0.1 M AA were added to 3 mL 0.1 M CTAB solution in a 5 mL cuvette, followed by shaking until the solution became colorless. Then, 0 mL, 30 mL, and 300 mL aliquots of 1 M HCl or 30 mL of 1 M NaOH was injected to adjust pH, and extra distilled water was added to make up the nal volume to 3.3 mL. Aer adding 375 mL GNR-seed solution, the UV-Vis spectra of the mixtures were monitored using a UV-Vis spectrometer at an interval of 30 seconds. To obtain the optimum amount of Na 2 S to halt the GNR growth when LLSPR reaches 800 nm, 17.5 mL, 35 mL, 70 mL, 140 mL, 280 mL, and 560 mL of 0.1 M Na 2 S solutions were injected. The UV-Vis spectra of GNR solutions before and aer the addition of Na 2 S at 1.0 min and 15 min were measured. Finally, GNRs with LLSPR of 650 nm, 700 nm, 750 nm, 800 nm, and 850 nm were prepared by adding 30 mL of HCl to slow the growth and 140 mL of 0.1 M Na 2 S to stop it. Scheme 1 Tuning LLSPR of GNRs through decreasing and halting the overgrowth with HCl and Na 2 S.

Single GNR imaging
To perform single-particle imaging, 10 mL of 1000-fold ultrapure water-diluted GNR sample was dropped on a clean slide. Aer putting a clean cover glass on it, the sample was immediately examined under a microscope and images were acquired with the exposure times of 10 ms, 50 ms, 100 ms, and 500 ms.

HCl and Na 2 S-regulated overgrowth of GNRs
The problem of traditional seed-mediated GNR synthesis is variable LLSPR even with the same protocol. 20,30 At the starting point, GNRs using 4.8 Â 10 À4 M HAuCl 4 , 9.5 Â 10 À5 M AgNO 3 , and 5.0 Â 10 À4 M AA in 20 mL 0.1 M CTAB solution and 60 mL of small gold sphere nanoparticle solution were prepared by three researchers. The UV-Vis absorption spectra of the resulting GNR solution were measured (Fig. 1). Obviously, among these GNRs, the largest wavelength of LLSPR was 854 nm, while the smallest one was 760 nm. The deviation achieved was nearly 94 nm. The absorbances were also apparently different, where the highest absorption intensity of the GNR solution was three times the lowest value. Similar results were also reported previously. 20 The reproducibility of the shape and morphology of GNRs still remains an issue since these are highly dependent on the reaction conditions and the residues of some organics may cause such differences. 20 Another important reason is that NaBH 4 can react with water to produce H 2 and form bubbles, which affect the nal amount of NaBH 4 used for preparing spherical gold nanoparticles. Therefore, the LLSPR of the asprepared GNRs is inevitably variable and cannot be anticipated; correspondingly, it is difficult to synthesize GNRs with particular LLSPR using controlling recipes.
Since the overgrowth of GNRs results in continual blueshis in LLSPR, 26 we herein proposed an extensively applicable routine to repeatedly prepare GNRs with specic LLSPR, that is, halting the GNR overgrowth at a particular LLSPR, and this was monitored with a UV-Vis spectrometer. Controlling the kinetic dynamics of the GNR overgrowth process is promising to obtain GNRs with desired LLSPR. The chemical reaction of GNR overgrowth can be briey described as follows: The reaction rate is dependent on the concentration of both reactants and products. Based on this chemical reaction, the forward reaction rate is simply described as k[  3 . However, aer the introduction of HCl at the beginning, the backward rate was higher initially, causing the production of Au 0 to decline and the overgrowth of GNRs to slow down. The nal equilibrium state would be altered, inducing decrease in the reduction in Au 0 . The UV-Vis spectra of GNRs during the overgrowth in 30 minutes with the addition of different amounts of HCl and NaOH were measured immediately aer the addition of GNR seeds, as shown in Fig. 2. Aer the overgrowth, the LLSPR of GNRs blueshied and the absorbance of their solutions increased, but the dynamic processes were slightly different. Without HCl or NaOH (pH ¼ 3.05), the LLSPR would initially redshi by 24 nm in the rst one min and then gradually blueshi to around 650 nm (Fig. 2C). Probably, the fast deposition of Au onto the end of GNRs leads to a small increase in AR at the start. 31 With the addition of 30 mL of 1 M HCl (pH ¼ 2.72), the LLSPR remained unchanged in the rst 2 min, but the absorbance continually went up (Fig. 2B). Then, the LLSPR blueshied slowly to 650 nm. Interestingly, with the addition of 300 mL of 1 M HCl (pH ¼ 1.98), the LLSPR blueshi went on linearly but more slowly during the overgrowth ( Fig. 2A). The average blueshi rates for the LLSPR of GNR solutions with the addition of 300 mL, 30 mL, and 0 mL of 1 M HCl in the rst 5 min were 4.8 nm min À1 , 14.2 nm min À1 and 40.6 nm min À1 , respectively. In contrast, with the addition of 30 mL of 1 M NaOH (pH ¼ 9.65), the overgrowth was completed within 2 min. The absorbance around 520 nm was higher than LLSPR, indicating the formation of a large amount of spherical gold nanoparticles because of the strong reducing ability of AA under basic conditions. The slow changes in LLSPR suggest the possibility of precisely tuning GNRs with the designed AR. The rate of 40.6 nm min À1 (without the addition of HCl) was too fast to stop the LLSPR of GNRs at a precise wavelength. However, too much HCl (300 mL, 1 M) only triggered a slight increase in absorbance, indicating a minor increase in the GNR size although the shi in LLSPR was slow. As a consequence, 30 mL of 1 M HCl (8.16 mM) was optimal to reduce the growth of GNRs and benecial for producing GNRs with specic LLSPR.
Sulde ions (S 2À ) have strong affinity for metal ions and can react with gold ions to form Au 2 S and Au 2 S 3 . 32-34 Thus, the introduction of S 2À depleted gold ions and stopped the This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 12619-12625 | 12621 overgrowth of GNRs immediately. To understand the effect of S 2À on the overgrowth of GNRs, the UV-Vis spectra of GNR solutions with the addition of different amounts of Na 2 S when LLSPR reached 800 nm were monitored at 1.0 min and 15 min windows. As shown in Fig. 3A and B, the addition of 17.5 and 35 mL of 0.1 M Na 2 S is insufficient to stop the growth, and the LLSPR of GNRs still undergoes blueshis to 676 nm and 708 nm, respectively. A possible reason was that gold ions were not totally exhausted by S 2À . We also noticed that the absorption intensity of transverse surface plasmon resonance (around 520 nm) underwent a huge increase aer 15 min of reaction. The main reason is that with the insufficient addition of S 2À ions, the newly formed Au 2 S and Au 2 S 3 act as the nuclei for further gold deposition and generate spherical gold nanoparticles. Increasing the S 2À concentration would regress such further overgrowth. When 70 mL of Na 2 S was added, the growth was completely terminated and the LLSPR of GNR remained steady (Fig. 3C). Surprisingly, further increasing the S 2À concentration could result in a redshi in LLSPR. For example, aer using 140 mL, 280 mL, and 560 mL of 0.1 M Na 2 S, redshis of 6 nm, 14 nm, and 16 nm, respectively, were observed. Such unusual redshis are attributed to the formed Au 2 S 3 layer on the surface of GNRs with extra S 2À ions. 33,35 The formed Au 2 S 3 coating layer changed the refractive index of GNRs and led to a redshi in LLSPR. Interestingly, the S 2Àterminated GNRs showed high stability. Aer centrifugation and redispersion, the LLSPR of these GNRs had no shi during a 2 month storage period, demonstrating their high stability. 33  Thus, 70 mL of 0.1 M Na 2 S (a nal concentration of 1.83 mM) was adopted to stop GNR growth and obtain LSPR at any wavelength.

Synthesis GNRs with precise LLSPR
To validate our method, GNRs with designed LLSPR were obtained by controlling the use of HCl and sulde. Through regulation, GNRs with LLSPR of 850 nm, 800 nm, 750 nm, 700 nm, and 650 nm were successfully prepared. The normalized UV-Vis absorption spectra and TEM images of these GNRs shown in Fig. 4 demonstrate that the LLSPR of GNRs can be continually, arbitrarily, and exquisitely tuned to any desired wavelength. In addition, the statistical lengths and diameters of these GNRs are displayed in Fig. 4B. These statistical results on lengths and diameters suggest the accurate control of GNRs with designed LLSPR and a small deviation (less than 3 nm), which is more precise than that obtained in some reported methods. 26,36 These GNRs with high yields ($95%) have well-dened morphology without the coexistence of bread-like and multipod-like nanoparticles. The TEM results showed that aer adjusting LLSPR from 850 nm to 650 nm, the length of GNRs increased from 56.9 AE 5.5 nm to 89.9 AE 7.6 nm, while the diameter of GNRs increased from 13.6 AE 0.7 nm to 37.5 AE 4.5 nm. The growth along the longitudinal axis of GNRs (33.0 nm) was faster than that along the transverse axis (13.9 nm). The probable reason is the micelle structure of CTAB on the side of GNRs, which inhibits the growth along the transverse axis. 31 Aer overgrowth, the aspect ratio reduced from 4.2 to 2.3, leading to a decrease in LLSPR. Therefore, our method for ne-tuning the LLSPR of GNRs is highly efficient.

Single GNR imaging
Single nanoparticle imaging has become an important tool to investigate the dynamics of physical, chemical, and biological processes. 13,37 The development of appropriate optical probes has been an important goal of nanomaterial engineering. Coincidentally, GNRs having the characteristics of tunable LLSPR and larger light cross-sections are excellent candidates. 1,5 The darkeld images of GNRs with different LLSPRs were taken under different exposure times from 10 ms to 500 ms, as shown in Fig. 5. We observed only a small amount of GNR 650 with an exposure time of 10 ms, and the brightness and number increased on increasing the exposure time. GNRs exhibit red colour because their LLSPR is in the range of red light (620-760 nm). 38 The intensities of GNRs were different, further proving the unequal dimensions of GNRs. In contrast, GNR 700 can be observed with the exposure times of 100 ms and 500 ms. However, GNR 750 was only seen with a longer exposure time of 500 ms. For GNR 800 and GNR 850 , no recognizable spot was observed even with 500 ms exposure time. GNR 650 was 9.92 times brighter than GNR 700 according to the images obtained for 100 ms exposure time, while this value was calculated to be 6.55 according to Gans' theory. 39 It should be noted that the scattering intensity of a single GNR is related to both the scattering cross-section of GNRs and the overlap between LLSPR and light response of CCD. The scattering cross-section is generally proportional to the size of GNRs. Thus, GNR 650 has a higher scattering cross-section than GNR 700 . On the other This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 12619-12625 | 12623 hand, the used DP73 CCD only has optical responses in the range from 400 to 650 nm. In comparison with GNR 700 , GNR 650 ts the optical response of CCD better. Taken together, GNR 650 is brighter than GNR 700 and other GNRs.
To further conrm the importance of optical matching, we also prepared GNRs with LLSPR larger than 1100 nm, a diameter of 20 nm and a length of 358 nm (the average volume is 1.30 Â 10 3 nm 3 , which is larger than that of GNR 650 with an average value of 9.93 Â 10 4 nm 3 ) according to a previously reported method. 40 As expected, these GNRs with large LLSPR were not observed even with an exposure time of 500 ms, further suggesting the signicant role of the overlap between LLSPR and CCD response. As a result, GNR 650 with high brightness obtained by our methods might be good optical probes for imaging, which can improve the localization accuracy and help monitor biochemical and physical events at a high speed. Moreover, our method is extremely suitable for engineering GNR imaging probes because it can not only adjust the LLSPR of GNRs to t the optical response of detectors but also increase nanoparticle size associated with absorption and scattering cross-section.

Conclusions
In conclusion, we have reported a simple and reliable route to precisely tune the LLSPR of GNRs by HCl and Na 2 S-regulated overgrowth processes. In this assay, HCl decreased the overgrowth reaction rate by affecting chemical equilibrium, while Na 2 S stopped the overgrowth through the formation of Au 2 S to deplete the gold element. Aer a systematic study, 8.16 mM HCl was found to be optimal for suppressing the shi rate of LLSPR, and 1.83 mM Na 2 S was sufficient to stop GNR overgrowth efficiently. With this approach, GNRs with LLSPR at 850, 800, 750, 700, and 650 nm have been successfully prepared with an ultrasmall deviation of 3 nm. In comparison with other GNRs, GNR 650 showed the strongest scattering under darkeld microscopy; it was 9.92 times brighter than GNR 700 and the value was higher than the theoretical calculation value (6.65fold). This is due to the satisfactory matching between the LLSPR of GNR 650 and the response wavelength range of CCD. Through this post-synthesis method, however, only the preparation of blue-shied GNRs compared to GNR seeds is possible. To obtain GNRs with longer LLSPR, GNR seeds with larger LLSPR need to be prepared rst. Although it has some limitations, this method is highly valuable for the preparation of GNRs with tunable LLSPR and strong scattering, which should enhance their practical applications, e.g., imaging and phototherapy.

Conflicts of interest
There are no conicts to declare.