Understanding blue shift of the longitudinal surface plasmon resonance during growth of gold nanorods

We have investigated in detail the growth dynamics of gold nanorods with various aspect ratios in different surrounding environments. Surprisingly, a blue shift in the temporal evolution of colloidal gold nanorods in aqueous medium has been observed during the growth of nanorods by UV visible absorption spectroscopy. The longitudinal surface plasmon resonance peak evolves as soon as the nanorods start to grow from spheres, and the system undergoes a blue shift in the absorption spectra. Although a red-shift is expected as a natural phenomenon during the growth process of all nanosystems, our blue shift observation is regarded as a consequence of competition between the parameters of growth solution and actual growth of nanorods. The growth of nanorods contributes to the red-shift which is hidden under the dominating contribution of the growth solution responsible for the observed massive blue shift.


Introduction
Metal nanoparticles have generated huge attraction due to their different distinct properties as compared to their bulk counterparts. The metal nanoparticles show outstanding optical properties associated with the tunability of their localized surface plasmon resonance (SPR) in the visible part of the spectrum [1]. SPR of noble metal nanoparticles is the collective oscillation of electrons induced by the electromagnetic radiation of light. The SPR peak wavelength (λ max ) of nanoparticles depends upon the size, shape, composition, and also the dielectric environment around them [2][3][4][5][6]. Many interesting electronic, optical, catalytic properties have cropped up because of surface plasmon oscillations and have led to exciting applications in several branches of science and technology including information storage, optoelectronics, biological imaging, cancer therapy etc. [7][8][9][10][11][12]. The most challenging task in the synthesis of metal nanoparticles is to keep control over the shape and size of the particles.
For example, spherical gold nanoparticles show a strong single absorption band in the visible region at about 520 nm whereas gold nanorods (NRs) show two absorption bands belonging to transverse and longitudinal LSPR [13].
To explain the existing experimental results a few remarkable theoretical works have been proposed previously [2,[14][15][16][17] which calculate the extinction spectra (derived from absorbance or transmittance) of metal nanoparticles. From several metal nanoparticles, gold nanorods have drawn significant research interest in the last decade due to their multiple plasmon absorption bands, the transverse surface plasmon resonance (TSPR), and longitudinal surface plasmon resonance (LSPR) [14,17]. The LSPR is highly sensitive to the rod aspect ratio (length to width ratio) and dielectric function of particle and that of the surrounding medium, allowing it to tune across a broad wavelength range [17]. Absorption and scattering phenomena become dominant at different aspect ratios and gold nanorods can 3 be targeted to optical applications depending on their dimensions. In a colloidal solution containing a mixture of spherical and rod-shaped nanocrystals, a UV-Visible absorption measurement will reflect an ensemble average for their contributions in TSPR and LSPR peaks in terms of its width, intensity, and position. The longitudinal peak in gold NRs is much more intense than the transverse peak. Therefore, robust understanding and good synthesis control over nanorod growth are essential for effective material applications.
Synthesis of highly monodispersed metallic nanorods in aqueous solution has been well reported for the last few decades [18][19][20]. In one of the very initial reports on gold nanorods in aqueous solutions, Yu et al. had shown the necessity of adding a small amount of silver to the growth solution [21]. Removing silver could lead to the formation of longer nanorods (aspect ratio ~ 20) but with low yield ~ 55% [19]. However, after all these years of continuous investigations, the exact growth mechanism of gold nanorods using the colloidal route is elusive, which could be due to the constraint of proper in-situ experimental facilities which do not alter the physicochemical properties of the system. We used silver assisted, seeded growth mechanism using the quaternary ammonium surfactant, cetyltrimethylammonium bromide (CTAB) for the growth of stable monodisperse gold nanorods, which has received the most favorable route as far as cost-effectiveness and mass production. Various groups have adopted different methods to formulate fundamental correlations between aspect ratio, volume, LSPR peak position, FWHM of LSPR peak, etc. [22][23][24]. This has been achieved by making a fitting of experimental data with theoretical simulations such as Gans approximation [22], Discrete Dipole Approximation (DDA) [23], Finite Difference Time Domain (FDTD) [24] and other numerical approaches.
Here, we report results on the synthesis of monodisperse NRs by seed-mediated growth method. The synthesis process was followed to ensure all the results were reproducible several times.We have discussed our finding of the blue-shift of LSPR and its extinction 4 wavelength and intensity. We have explicitly used non-destructive optical measurement such as UV-visible absorption spectroscopy, and extract all the information by fitting the respective data to understand the correlation between LSPR position and other parameters in the solution [24]. However, the rod solution ensemble itself is a complex system containing different reactants with numerous factors such as environment di-electric properties, [25] physical/chemical interaction at the surface, [26] surface charge, [27] nanorod assembly, [28] interparticle separations, [29] etc. affecting the highly sensitive plasmon peak position. Also, damping at the metal-ligand (capping molecules) interface has broadened the plasmon resonance width [26]. Recent reports have shown that apart from aspect ratio, the net volume of the particle in a solution can substantially alter the extinction spectrum [30,31]. The LSPR can be tuned by adding different additives such as HCl, Na 2 S which regulate the growth reaction by changing the chemical equilibrium [32]. Progressive developments in the quest for better understanding the dominant factors have resulted in several reports [17,22,24,30,33], yet the results are ambiguous and the dynamics of LSPR peak is elusive. As a step forward, here we demonstrate the temporal evolution of LSPR positions and intensity and its correlation with the growth of gold nanorods in the medium.

Synthesis of Gold Nanorods
A two-step seed-mediated process was used to prepare gold nanorods solution as reported in several literatures [18][19][20]. All solutions were prepared in aqueous solutions, and the entire reaction was carried out in distilled water at ambient conditions. For the preparation of NRs, seed and growth solutions were made as described below.

Characterization
To determine the extinction spectrum of the gold nanorods in solution and to discuss the temporal evolution of the growth of gold nanostructures, we have measured UV-visible spectra using the HITACHI U-4100 spectrometer. All the spectra have been measured in constant 25 ºC room temperature and care has been taken to maintain the room temperature for all the measurements. A Gaussian distribution function is used to extract the respective data as defined in the following section to study the UV visible spectrum data. The field effect scanning electron microscope (FESEM) image of Au nanorods was obtained with Carl Zeiss Field Emission Scanning Electron Microscope. For imaging, the gold nanorods were extracted from the solution by centrifuging the samples at 5000 rpm for 10 minutes. The solution was re-dispersed in distilled water to remove the excess surfactants and then centrifuged again at 5000 rpm for 10 minutes. Then the surfactants were discarded and the rods were deposited onto ultrasonically cleaned Si (100) substrates using a spin coater. For calculating particles size distribution more than hundred particles were selected from different parts of the images and measured using ImageJ software. For each sample, the sizes of different nanorods from SEM images were measured to obtain the average size and the size distribution of nanorods. Transmission electron microscopic (TEM) images have been collected on JEOL-JEM 2100 F model using a 200 kV electron source.

Results and Discussion
The The TSPR peak has a contribution mostly from the width of the nanorod which varies depending upon the reaction conditions and also from the spherical particles which are present in the final rod solution [19]. When the seed solution is added to the growth solution, there is some amount of excess NaBH 4 in the seed solution which gets transferred to the growth solution thereby leading to the formation of additional spherical nanoparticles. With the progression of the reaction, the width of the rod also increases by some amount [19].
These two factors contribute to the position of the TSPR peak and the intensity (the number of individual oscillators) increases with time as the number of particles increases in the solution [22] In the present investigation, the TSPR positions are nearly fixed within a trivial range between 520 to 540 nm. The saturation of the TSPR peak position is also an indication of the stability of the width of the nanorods in the solution. 8 We have analyzed the growth dynamics of the nanorod solution for seed solution (12 μl) and AgNO 3 (100 μl) and derived the information, which is shown in Fig. 2. Fig. 2 (a) shows a typical UV-visible spectrum of gold nanorod with TSPR and LSPR peaks. The longitudinal plasmon peak begins to appear after 3 to 4 minutes of mixing of solution and it has been found that the peak position has blue shifts from 752 nm to 635 nm with an average standard deviation of 0.7 nm as rod growth with time [ Fig. 2 (b)]. We can see that LSPR peak position decreases (decrease of wavelength means blue shift) exponentially with time.
Eustis et al. [22] reported that the intensity of the peak is dependent on the net induced dipole moment of a particle. We found that with the progress of the reaction, the length of the rods in solution increases and the increase of length leads to an increase in the number of atoms in the rod. Thus the increase in individual oscillators in the rod leads to an increase in the intensity of LSPR peak initially. However, after reaching a maximum, there is a small decrease in intensity and after which, it becomes stable [ Fig. 2 (c)]. The decrease could be due to the fact that during the seeded growth process, the rods first undergo an increase in length, after which they contract by a small value [19] and finally settle down in the most stable configuration depending on the thermodynamics of the environment in which they are growing.
The full width half maximum (FWHM) of the LSPR peak describes the poly-dispersity of the rod solution. The value of FWHM reflects the lifetime of plasmon and from uncertainty relation (∆ω×∆t ~ 1); the increase in width of spectra indicates a lesser lifetime of excited plasmon. Furthermore, local field enhancement is dependent on the lifetime of plasmon.
Increasing plasmon damping can be seen in increasing FWHM and leads to shorter plasmon lifetime. Fig. 2 (c) shows the FWHM value obtained at a different time of growth of solution confirms the formation of uniform size distribution. We find that peak shapes are symmetric and corresponding FWHM value ranges from 170 to 120 nm with an average standard 9 deviation of 2.5 nm. It has been found that the size distribution becomes narrower and becomes constant after 25 min of growth reaction. The extinction ratio of the longitudinal surface plasmon resonance peak to transverse peak is decreasing with time and becomes stable after 1 hour of growth reaction [ Fig. 2 (d)].
Although all the ingredients are essential for Au nanorod growth, the amount of silver nitrate and seed solution are the most effective way to tune the plasmon peaks [18]. We have Based on our observation, we have tuned the LSPR peak wavelength from 600-820 nm by changing the volume of seed solution and AgNO 3 solution which effectively change the concentration of reagent per unit volume of the mixed solution in the synthesis process of 10 gold nanorods. We can also tune the LSPR peak wavelength from 600 to 900 nm by changing the concentration of the synthesis parameters [12,18,21].
It has been found that during the growth of Au nanorods, LSPR peak position shifted from 780 nm to 650 nm over the period of 60 minutes (reaction time). It undergoes a constant blue shift as observed in Fig. 1. This behavior is inconsistent with the basic understanding of the nanorod growth mechanism. As the rods grow from spheres, the LSPR peak position should, in fact, show a red-shift rather than blue-shift (when we assume other parameters remain unchanged) [17]. The reason for such behavior is that with the increase of the length of the nanorod, the confinement of electrons along the length (long axis) of the rod decreases. As a result, the longitudinal peak should show red-shifting behavior [37]. However, in our case, LSPR peak position undergoes continuous blue-shift. Our results can be compared with that of Sau et al., [19] where  There is a blue shift of 100 nm when the aspect ratio value changes by 0.28, which is within the standard deviation limit. We can also observe that for B5 and B9 almost the same aspect ratio shows plasmon peaks of 734.4 nm and 697.3 nm, which is blue 13 shifted despite the same aspect ratio. Thus variation of aspect ratio approach does not perfectly replicate the entire growth process corresponding to the LSPR. For all particles, those fall within or outside the quasi-static limit; the aspect ratio always shows a linear relationship with LSPR maximum position [17,30]. Recent works by various groups [30,33] also show the strong dependence of particle volume on the optical extinction spectrum. It has been further established that the formula for relating the various parameters of the rods and spectral position of the LSPR can be done on the basis of aspect ratio, the dielectric constant of surrounding environment and volume in a combination thereof [17,22,24]. Interestingly, the linear curve fittings do not reveal any significant variation in dimension and aspect ratio although the blue shift is quite substantial (~ 100 nm variation between B1 and B21 samples).
This indicates that the LSPR peak is no longer explicitly dependent on the rod size; rather the medium plays an important role in tuning the LSPR peak. Based on various experimental and theoretical interpretations [17,19,22,25,26,43,44] our analysis suggests that the blue shift trend of LSPR peak in the present investigation does indeed depend on the temporal evolution of dielectric constant of the surrounding medium. Our results can be compared to those obtained by Near et al. [30] where they have reported that additional parameters are needed to correlate the LSPR position and aspect ratio.

Conclusion
We have synthesized and grown nanorods from spherical gold nanoparticle seeds. We observed characterized two plasmon peaks in the absorbance spectra of gold nanorods. The first peak observed near ~ 520 nm corresponds to the transverse surface plasmon resonance, which corresponds to the short axis of the ellipsoid of nanorods. It stays around ~ 520 nm for all growth conditions. The second peak was observed in the range of ~ 550 nm and above on the wavelength scale in the near-infrared region. This is the longitudinal surface plasmon resonance, localized to the long axis of the ellipsoid of nanorods. We observed that the 14 longitudinal plasmon peak is blue-shifted during its growth process. The blue shift is not due to anisotropic or tailored nanorods size and shape, but this can be explained by the change of dielectric permittivity of the surrounding nanorods media. To understand it clearly, we carried out water dilution experiments and found that the blue shift is imminent even without any substantial change in rod size and mostly depends on the physicochemical properties of the surrounding medium.       We have taken 60 µl seed solutions and different volumes of AgNO 3 to get different nanorods. Other parameters have been kept the same for all the synthesis conditions.

Table S3
The amount of DI water added to the growth solution to control the LSPR peak position. The dimensional parameters corresponding to LSPR position are also mentioned for the nanorods.   Table -S3). Each spectrum in EDS shows the presence of Au along with other elements used in the synthesis of Au nanorods.