Exploring the coherent interaction in a hybrid system of hollow gold nanoprisms and cyanine dye J-aggregates: role of plasmon-hybridization mediated local electric-field enhancement

Kamalika Das , Bidhan Hazra and Manabendra Chandra *
Department of Chemistry, Indian Institute of Technology, Kanpur, Uttar Pradesh, India. E-mail: mchandra@iitk.ac.in

Received 10th August 2017 , Accepted 25th September 2017

First published on 26th September 2017

In this work, we probed the possibility of observing strong plasmon–exciton interactions in hollow gold nanoprism–J-aggregate nanocomposites. Several different hollow gold nanoprisms (HGNs) with different aspect ratios were synthesized. This allowed us to systematically tune the LSPR energies through the exciton energy of the PIC–J-aggregate, which in turn allowed us to have direct determination of the coupling strength of HGN–J-aggregate composites. Hybrid nanosystems were prepared by adsorbing and assembling 1,1′-diethyl-2,2′-cyanine (pseudoisocyanine or PIC) iodide onto the surface of hollow gold nanoprisms. Plasmon–exciton interactions were studied using extinction spectroscopy. The experimental results were analysed, and complemented by the results obtained from numerical simulations. Our results reveal that the HGN–PIC–J-aggregate hybrid nanosystem shows coherent coupling between the localized surface plasmons of the HGN and excitons of the PIC–J-aggregate, as obvious from the observation of a clear transparency dip and the formation of two new hybrid plexcitonic modes in the plexcitonic spectra. Anti-crossing behaviour of the plexcitonic modes, together with large Rabi splitting and coupling constant, asserts strong coupling between the plasmon and the exciton, overwhelming the decoherence effects, in our hybrid nanosystem. Analysis of the calculated near-field distribution establishes that the plasmon-hybridization mediated large electric-field enhancement holds the key to the strong coupling.


Interaction between plasmons and excitons in hybrid nanosystems comprising metal nanoparticles and molecular or other quantum-sized excitons has garnered a surge of interest in recent years due to the tunable and exciting optical properties of noble metal nanostructures, originating from their localized surface plasmon resonance (LSPR). The LSPR of the metal nanoparticles couples with the excitonic transition displayed by molecular J-aggregates or quantum dots. A thorough understanding of plasmon–exciton coupling not only holds the key for improving our present state of knowledge about various fundamental phenomena like coherent energy transfer, quantum entanglement, cavity quantum electrodynamics,1etc. but also for the practical implementation of a wide variety of applications in the fields of sensing,2 artificial light harvesting,3 nanoscale optical devices,4 quantum information processing,5–7etc. Plasmon–exciton interactions generate hybridized energy states, known as plexcitons, and result in many exotic optical phenomena such as induced transparency, Rabi splitting, Fano resonance, plasmon resonance energy transfer, enhanced absorption, etc.8–12

Among several choices, molecular J-aggregates of cyanine dyes are particularly suitable and are the most used excitonic material in plexciton research. Cyanine dyes have been extensively studied due to their use in photosensitization and optoelectronic materials. Under reduced solubility conditions, the cyanine dyes are known to form so-called J-aggregates. Generally, the absorption spectra of J-aggregates exhibit very narrow linewidths, have high oscillator strength, and are always red shifted with respect to the monomer absorption peak. This spectral shift is attributed to the formation of excited states by coherent coupling of molecular transition dipoles leading to the formation of a Frenkel-type exciton.13,14 Due to the features of these molecular J-aggregates and the ease of their fabrication, cyanine dye-coated plasmon-supporting nanoparticles are ideal and the most used systems for probing plasmon–exciton interactions. While the characteristics of molecular excitons are important, the structure and morphology and hence the optical properties of the plasmon-supporting nanoparticles play the most pivotal role in plexcitonic coupling. Therefore, the ability to tune the structure and optical properties of the plasmonic nanoparticles in a controlled manner is the key to achieving the desired light–matter interaction in plexcitonic systems and their subsequent applications. Fortunately, substantial progress in the synthesis and control over the structure and morphology of plasmonic nanoparticles has been made in recent years and this has resulted in the synthesis of nanoparticles with various shapes and morphologies, e.g., nanorods, nanoprisms, multipods, nanostars, nanocrescents, nanorice, silica core–metal shell particles, hollow nanospheres, etc.15–23 Interestingly, the LSPR tunability and the electric-field enhancing capability are the two most essential prerequisites for a metal nanoparticle in order to exhibit strong plexciton coupling. Because of the need for a large electric-field enhancement, silver nanostructures have always been the prime choice as the plasmonic component although they suffer from oxidation problems.8,12 The number of studies with gold nanostructures is rather low and they are limited to geometries like nanorod or nanosphere (or nanoshell) dimers.9,10 The nanostructure most relevant to the present work is the hollow gold nanoprism (HGN).24,25 HGN is an equilateral gold nanoprism with a cavity at its center of mass. Such a nanoparticle with an anisotropic shape and a hollow interior allows one to exploit the so-called plasmon hybridization to widely tune the LSPR.24–28 Moreover, due to efficient plasmon hybridization, such anisotropic and hollow nanoparticles can localize and amplify the electric field of the exciting light with extreme efficiency.29 Therefore, HGNs can be used as a very good plasmonic component in attaining strong plasmon–exciton coupling when excitons are in close proximity to the surface of the HGN.

In this paper, we report the investigation into the plasmon–exciton interaction in HGN–J-aggregate nanocomposites. The main goal of this study was to probe whether the J-aggregate coated HGNs show the signatures of plasmon–exciton interaction and also to determine whether these hybrid nanosystems can overcome the damping effects of the interband transitions and exhibit strong plasmon–exciton coupling. Several different HGNs with different aspect ratios were synthesized. This allowed us to systematically tune the LSPR energies through the exciton energy of the PIC–J-aggregates, which in turn allowed us to have direct determination of the coupling strength of HGN–J-aggregate composites. Hybrid nanosystems were prepared by adsorbing and assembling 1,1′-diethyl-2,2′-cyanine (pseudoisocyanine or PIC) iodide onto the surface of the hollow gold nanoprism. Plasmon–exciton interaction was studied using extinction spectroscopy. The experimental results were analysed and complemented by the results obtained from numerical simulations. Our results reveal that the HGN–PIC–J-aggregate hybrid nanosystem shows coherent coupling between the localized surface plasmons of the HGN and the excitons of the PIC–J-aggregate, as is obvious from the observation of a clear transparency dip and the formation of two new hybrid plexcitonic modes in the plexcitonic spectra. The anti-crossing behaviour of the plexcitonic modes together with large Rabi splitting and coupling constant assert strong coupling between the plasmon and the exciton, overwhelming the decoherence effects in our hybrid nanosystem. Analysis of the calculated near-field distribution establishes that the plasmon-hybridization mediated large electric-field enhancement holds the key behind strong coupling.

Materials and methods

All chemicals were purchased from Sigma-Aldrich and used as received. HPLC grade water was used as solvent in all the experiments.

Synthesis of hollow gold nanoprisms (HGNs)

Hollow gold nanoprisms were synthesized using a sacrificial galvanic replacement method.25,30 HGN synthesis is a two-step process in which the first step involves the formation of silver nanoseed, which is added to the growth solution in the second step, leading to the formation of a hollow nanoprism. Nine different HGNs were prepared by varying the amount of Ag nanoseeds added to the growth solution. A more detailed synthetic procedure is provided in the ESI. All the HGN samples were aged for 7 days before characterizing them or carrying out any experiment with them.

Optical and structural characterization

Extinction spectra of the samples were recorded using either a fiber-optic spectrophotometer (AvaSpec 3648, Avantes) or a dual-channel spectrophotometer (Jasco V670). Surface Enhanced Raman Scattering (SERS) measurements were carried out using a confocal Raman optical microscope (Olympus BX51) equipped with a triple grating imaging spectrometer (Princeton Instrument Acton Spectra Pro SP-2500). Details of the sample preparation for the SERS measurements are given in the ESI. Transmission electron microscopy (TEM) images were acquired on a ‘FEI Tecnai G2 12 Twin’ transmission electron microscope operated under an acceleration voltage of 200 kV. Zeta-potential measurements for HGN and PIC-coated HGN samples were performed using a Malvern Zetasizer Nano ZS90 equipped with a 633 nm laser. Both PIC-coated and uncoated HGN solutions were half-diluted with water before injecting them through one of the two ports of a folded capillary cell.

Preparation of HGN–PIC composites

A saturated aqueous solution (3.52 × 10−4 M) of PIC was prepared and stored in the dark at room temperature and used as a stock solution. Varying amounts of the stock solutions of the dye were added to the HGN solution to make a total volume of 2 ml, which was mixed well and then kept undisturbed in the dark for 72 hours. Excess cyanine dye molecules were removed by repeated centrifugations at 8000 rpm for 15 minutes followed by washing and the final sediment was redispersed in water.

Numerical simulation methods

The finite difference time domain (FDTD) technique was employed to numerically simulate the optical properties of ‘bare’ HGNs and the HGN–PIC hybrid nanocomposites. For this purpose, we used a commercially available software package “FDTD solutions v8.12” (Lumerical). A detailed discussion of the methods of numerical simulation of HGNs can be found in previous reports.24,25 Simulation was performed on a single HGN (or PIC-coated HGN) suspended in water. Before proceeding for the simulation of plexcitons, the effect of dielectrics of the molecular layer (essentially the dye) on the HGN spectra was checked by running simulations on a single HGN coated with a dye layer of zero oscillator strength. For simulating the optical response of plexcitons, the complex dielectric permittivity of the PIC J-aggregate covering the HGN was modeled using the single Lorentzian function9,31
image file: c7cp05455f-t1.tif
where ε = 1.5 is the high-frequency component of the J-aggregate molecular layer, f = 0.01 is the reduced oscillator strength, the J-aggregate resonance frequency was fixed at ωe = 2.15 eV and the resonance linewidth was set to γe = 25.6 meV. Except for the case of thickness-dependent simulations, the thickness of the adsorbed dye layer was kept fixed at 4 nm throughout. A judicious choice of thickness of the dye layer and the oscillator strength was necessary to achieve a reasonable agreement with the experimental results.

Results and discussion

PIC is one of the most well-studied J-aggregate forming cyanine dyes (its chemical structure is shown in Fig. 1), which exists mostly in its monomeric form in aqueous solution at concentrations of 10−4 M or lower.32 J-Aggregates are formed under reduced solubility conditions or upon adsorption on the surface of a substrate. J-Aggregate formation can be easily induced by high salt concentration at room temperature. Fig. 1 shows the absorption spectra of 1.7 × 10−5 M PIC dye in water (grey) and in 5 M aqueous NaCl (black). The absorption band corresponding to the PIC J-aggregate is red shifted (λmax = 2.16 eV) with respect to the monomer absorption peak and has a very narrow linewidth (FWHM 24 meV). According to the Frenkel exciton model, the reason for this shift is attributed to the formation of excited states by coherent coupling of molecular transition dipoles.13,14
image file: c7cp05455f-f1.tif
Fig. 1 Absorption spectra of 17 μm PIC in water (grey) and in 5 M aqueous NaCl (black). The broad peak at around 2.37 eV is due to dye monomers. Appearance of the new peak (black curve) at around 2.16 eV indicates the formation of J-aggregates induced by the salt. The chemical structure of PIC iodide is also shown.

In order to study plasmon–molecular exciton interactions, the first step is to prepare a stable composite of nanoparticles and J-aggregates. Such composites are normally prepared by mixing a nanoparticle dispersion with a pre-prepared J-aggregate solution, which normally contains a high concentration of salts or bases. HGNs are quite stable in salt solution (e.g., up to 0.3 M NaCl), as shown in a previous report.25 However, a very high (>1 M) salt or base concentration can lead to aggregation of HGNs over a period of time, if not immediately. Since particle aggregation can obscure the exact effect of plasmon–exciton interactions, we decided not to use salt or base to form J-aggregates. Instead, we added a monomeric PIC solution into an HGN dispersion and allowed the dye monomers to self-assemble onto the nanoprism surfaces and thereby form J-aggregates. We mixed different aliquots of the stock PIC solution to the as-prepared HGN solution leading to final PIC concentrations ranging from 1.8 × 10−6 M to 8.8 × 10−5 M. At these concentrations, the aqueous solutions of the PIC dye do not form J-aggregates, as is evident from the absence of any J-band peak in their spectra (see the ESI). The result of the addition of one specific amount (4 × 10−5 M) of PIC to one particular hollow gold nanoprism (HGN-V) is presented in Fig. 2. Fig. 2a and b show the spectra of the HGN-V and aqueous PIC solutions, respectively. The TEM image of HGN-V is also shown as an inset of Fig. 2a. The black curve in Fig. 2c shows the effect of addition of PIC into the HGN-V solution. We have also shown the incoherent sum of the individual spectra of HGN-V and the PIC monomer in Fig. 2d for comparison. It is obvious from Fig. 2c and d that the spectrum of the HGN–PIC composite is not just an incoherent sum of the individual extinction spectra of HGN (Fig. 2a) and the dye monomer (Fig. 2b). The extinction spectrum of the HGN–PIC composite shows a transparency dip at ∼2.12 eV. The formation of the transparency dip is almost instantaneous (within a few seconds). The dip increases with time and after a certain period of time, no change in the dip intensity was observed (Fig. 2e). The position of the observed dip-minimum is almost identical to the absorption-maximum of the J-band of PIC aggregates. The kinetics of plexciton formation observed by us in this work is quite similar to a very recent result on the Ag nanoprism–TDBC system, as published by Balci et al.33 The presence of excess free monomeric PIC molecules obscures the blue side of the dip. We centrifuged and washed the as-prepared HGN–PIC composite solution twice to get rid of the excess free PIC monomers. The supernatant solution contains only PIC monomers, as is evident from its extinction spectrum (not shown). The spectrum of the centrifuged and washed HGN–PIC composites is shown in Fig. 2f. The spectral dip, between two new peaks, remains unperturbed and looks very prominent. The appearance of the dip at the J-band frequency even after centrifugation and washing clearly suggests that (i) PIC molecules are strongly adsorbed and self-assembled onto the metal surface forming J-aggregates, and (ii) the excitons of the thus-formed J-aggregates interact coherently with the localized surface plasmons of the HGNs. Before probing into the details of the plasmon–exciton interactions in the HGN–PIC system, we investigated PIC adsorption and J-aggregate formation on the HGN surface in detail. We probed the dye adsorption on the HGN surface by temperature-dependent extinction spectroscopy, surface enhanced Raman scattering (SERS) spectroscopy and zeta-potential measurements.

image file: c7cp05455f-f2.tif
Fig. 2 Extinction spectrum of HGN-V (a), absorption spectrum of 42.8 μm aqueous solution of PIC (b) and extinction spectrum of a homogeneous solution containing 8 × 10−11 (M) HGN-V and 42.8 μm PIC (c) are shown. A transparency dip at ∼2.12 eV can be clearly seen in (c). (d) Simple addition of the individual extinction spectrum of the as-prepared HGN-V and the 42.8 μm aqueous PIC solution is shown. (e) The time evolution of the transparency dip is shown. (f) Extinction spectrum of the HGN–PIC composite, as shown in (c), after repeated centrifugation and washing. Plexcitonic mode splitting can be clearly seen.

The zeta potentials of the CTAB-capped HGN-V and HGN-V–PIC were found to be 60.7 ± 5.5 mV and 46.9 ± 2.6 mV, respectively. Reduction in zeta potential is indicative of significant replacement of the CTAB double layers at the HGN surfaces by the PIC molecules. It is worth noting here that the adsorption of the PIC molecules onto the HGN surface is not detrimental toward the stability of the colloids, as is obvious from their very high zeta-potential value (46.9 mV). Now, these surface-adsorbed PIC molecules can either remain in a monomeric form or self-assemble through a range of noncovalent interactions including electrostatic interactions as well as π–π interactions.34 Further evidence for the fact that the PIC molecules are adsorbed onto the HGN surface and that they possibly self-assembled there, came from the temperature-dependent extinction measurements. If the PIC molecules are truly adsorbed and self-assembled on the HGN surface via weak noncovalent interactions, they can easily be desorbed from the HGN surface by applying some external energy, e.g. heat. A centrifuged and washed HGN–PIC sample (whose spectrum is shown in Fig. 2f), redispersed in water, was heated at 55 °C in a water bath for an hour. This led to complete disappearance of the dip, which suggests that the ‘J-aggregate’ no longer exists on the HGN surface. When this solution was cooled down to room temperature, no further change was observed in the spectrum and the dip also did not reappear. The observed spectrum resembles the spectrum of pure HGN except the shoulders near the 480–520 nm region, which is qualitatively very similar to Fig. 2d. This means that the system became merely a non-interacting mixture of HGN and free PIC monomer. As a matter of fact, centrifugation of this solution led to the complete separation of the PIC monomers (supernatant) and pure HGN (residue). All these observations are presented in Fig. 3. These results suggest that the PIC dyes are not only adsorbed onto the HGN surface but also self-assembled. The reason behind not seeing a reappearance of the dip, when cooled, is that the number of PIC molecules in the solution (note that these PIC molecules are those that get desorbed from the HGN surfaces at 55 °C) is too small to drive the adsorption equilibrium.

image file: c7cp05455f-f3.tif
Fig. 3 The effect of temperature on HGN–PIC composites is described. The black curve represents the extinction spectrum of centrifuged and washed HGN–PIC composites. The transparency dip completely disappears (blue curve) when the solution is heated for 1 hour at 55 °C and does not reappear even after cooling the solution down to room temperature. Centrifugation of the heat-treated solution gives back the spectra of pure HGN (residue, green curve) and PIC monomer (supernatant, orange curve). This observation suggests that PIC dye molecules are adsorbed and self-assembled onto the HGN surface.

In order to confirm whether the surface-adsorbed PIC molecules did really form J-aggregates, we performed a SERS experiment with a centrifuged and washed HGN–PIC composite. Fig. 4 shows the observed SERS spectrum. It is known that J-aggregate formation leads to the enhancement of relative intensities of some vibrational Raman modes of PIC such as one doublet near 1630 cm−1, one triplet near 1360 cm−1,and also a low frequency band at 605 cm−1.25,34–38 The in-plane stretching vibrations/deformations of phenyl groups or pyridyl groups give rise to these modes. Since the π–π interaction between the aromatic rings is the main force that holds the aggregates together, it is expected that changes in the induced electronic distribution in adjacent aromatic groups will amplify the aromatic group vibrations, which leads to enhanced polarizability and hence enhanced Raman scattering.34–38 Observation of enhancements of these modes in SERS spectra proves the existence of PIC J-aggregates. In the case of our HGN–PIC system, we can easily see that all these bands are present, with enhanced intensities, in its SERS spectrum (Fig. 4). It should be noted here that (i) the amount of PIC dye solution added to prepare the HGN–PIC composite in no way shows any signature of J-aggregates when dissolved in pure water and (ii) the excess monomer has already been removed by centrifugation and separation processes, which was monitored by extinction measurement. So we can infer that the observed SERS from the HGN–PIC nanocomposite originates from the J-aggregates, which were formed at the surface of the HGN. With this understanding of dye adsorption and aggregation on the nanoparticle surface, let us now bring our attention back to the foregoing discussion on the plasmon–exciton interaction in the HGN–PIC system.

image file: c7cp05455f-f4.tif
Fig. 4 Surface enhanced Raman scattering spectrum of the PIC dye-coated HGN. The spectrum clearly shows that PIC dyes exist as J-aggregates on the surface of HGN.

We can see from Fig. 2f that there are two newly formed peaks separated by the dip. One of the peaks is blue-shifted and the other one is red shifted with respect to the dip. Generally speaking, the formation of both blue-shifted and red-shifted bands indicates a coherent coupling between the plasmon and the exciton. This coherent coupling between the localized surface plasmon mode of the nanoparticle and the excitonic energy states of J-aggregates is very much analogous to ‘orbital hybridization’. Such a coupling leads to the formation of two new hybrid energy states, often referred to as plexcitonic states (an upper plexcitonic branch or UPB, and a lower plexcitonic branch or LPB), exactly as we see in the present case (Fig. 2f). Moreover, the LSPR energy of HGN-V and the energy of the surface-bound PIC J-aggregates, whose coupling resulted the spectrum in Fig. 2f, are nearly equal. The energy gap between the UPB and LPB under such a ‘resonant’ coupling condition is known as the Rabi splitting energy (ℏΩR). The strength of plexcitonic coupling is determined by the Rabi splitting energy. In the present case, we measured an energy splitting of 192 meV for the mixture of as-prepared HGN-V and 4 × 10−5 M aqueous PIC (Fig. 2f). Also, the plexcitonic spectrum is broadened compared to the uncoupled spectra. Such a large value of mode splitting along with broadening of the spectrum upon coupling indirectly suggests a strong coupling between plasmons and excitons. To gain detailed insights into the problem, we carried out numerical simulations on the PIC-coated HGNs using a commercially available FDTD package (Lumerical™). The details of the simulation procedure are discussed in the Materials and methods section. The dimensional parameters of the HGN were taken from TEM data. The simulated absorption, scattering and extinction cross-sections of the dye-coated (4 nm thick) HGN as well as the uncoupled systems are shown in Fig. 5a–c. The qualitative agreement between the experimental and simulated extinction spectra is quite good. We can see from Fig. 5a and b that spectral dips at the exciton absorption line appear in both absorption and scattering spectra and this is indicative of a true strong coupling regime.

image file: c7cp05455f-f5.tif
Fig. 5 Simulated absorption (a), scattering (b), and extinction (c) spectra of HGN (grey) and PIC J-aggregate coated HGN (black) are shown. The exciton resonance of PIC–J-aggregates was modeled using a single Lorentzian function (see text) with a reduced oscillator strength of 0.01. The thickness of the exciton layer was kept fixed at 4 nm. The presence of a transparency dip in the absorption as well as scattering spectrum suggests a strong coherent coupling between HGN and PIC J-aggregates.

If the HGN–PIC J-aggregate interaction indeed falls in the strong coupling regime, then the UPB and LPB will not cross each other when the plasmon resonance is detuned with respect to the PIC J-aggregate resonance. To gain further insights into the problem, we firstly prepared a total of 9 different HGNs having different aspect ratios and hence different LSPR frequencies.24,25 This allowed us to tune the plasmon frequency between 1.85 eV and 2.21 eV across the PIC J-aggregate resonance at 2.16 eV. We note here that the LSPRs of HGNs are broad, and therefore it is difficult to pinpoint the center energies with 1 meV accuracy even though our spectrometer has a resolution of ∼1 meV. We used the 1st derivative of the LSPR spectra of the HGNs to extract the center energies. This method allows us to accurately pinpoint the center energies with accuracies of tens of meV. Next, we prepared the PIC J-aggregate composites of all the different HGNs and recorded their spectra after getting rid of excess PIC monomers through centrifugation. The extinction spectra of all the 9 HGN–PIC samples showed mode splitting due to plexcitonic coupling. In order to extract the value of Rabi splitting, we modeled the PIC J-aggregate coated HGN as a two-coupled harmonic oscillator. This model takes into account of all the radiative and nonradiative losses. According to this model, ωUPB and ωLPB values are given by

image file: c7cp05455f-t2.tif(1)
image file: c7cp05455f-t3.tif(2)
where ωp and ωe are the transition frequencies of the individual HGN and the PIC J-aggregate, respectively, ωpωe is the detuning between LSPR and exciton resonance frequencies, Γp and Γe are the linewidths of the LSPR spectrum of the HGN and the exciton resonance, and g is the coupling constant. The difference between ωUPB and ωLPB values is given by,
image file: c7cp05455f-t4.tif(3)
Note that at zero detuning (i.e., when ωpωe = 0), eqn (3) reduces to the Rabi splitting frequency
image file: c7cp05455f-t5.tif
We plotted all the experimentally determined ωUPB and ωLPB values against the plasmon frequencies in Fig. 6a. We fit all the ωUPB and ωLPB data using eqn (1) and (2) in order to extract the Rabi splitting (ℏΩR) and the coupling constant (g) values. Interestingly, we can immediately see from the fitted curves in Fig. 6a that the two plexcitonic modes display a clear anti-crossing (also called avoided crossing) behavior at the position where the uncoupled HGN-LSPR matches with the J-aggregate resonance frequency. Such an anti-crossing behavior is characteristic of strong coupling between two harmonic oscillators. A very similar mode of anti-crossing behavior was obtained in the numerical simulation too (Fig. 6b). These results, therefore, clearly suggest that the interaction between the plasmon and the exciton in our HGN–PIC system falls in the strong coupling regime.31,39 Here we note that it is possible to observe a very similar anti-crossing behavior by using the linewidths, instead of the peak energies, of the plexcitonic branches provided one measures the reflectance spectra of plexcitonic systems as a function of incident angle.40,41 In fact, analyses of plexciton linewidths can bring out much more useful information regarding plasmon–exciton coupling.

image file: c7cp05455f-f6.tif
Fig. 6 (a) The energies of the hybrid plexcitonic states, extracted from experimental data, are plotted as a function of their corresponding LSPR frequencies. The blue and green squares refer to the upper- and lower plexcitonic modes, respectively. The black curves are obtained through fitting the data using a coupled harmonic oscillator model (see Results and discussion). The wine-red line at around 2.16 eV represents the exciton resonance energy and the grey line indicates the uncoupled LSPR frequencies of the HGNs. It is obvious that two plexcitonic modes display an anti-crossing behavior. A similar mode anti-crossing behavior can be seen in (b), which displays the results obtained through FDTD simulations. Observation of mode anti-crossing behavior suggests a strong plasmon–exciton coupling.

We extracted the value of Rabi splitting from the fitting (Fig. 6a), assuming that ωUPB + ωLPB = ωp + ωe and taking the ωe value always fixed at ≈2.16 eV, and the value was found to be 198 meV, which is among the highest values reported for gold nanostructures.9,10,42,43 On the other hand, the value of coupling constant was found to be 91 meV. As mentioned earlier, the coupling constant, g, is the measure of the strength of the coupling. When image file: c7cp05455f-t6.tif, the system under consideration is said to have reached a strong coupling regime, indicating that the coupling has overcome all the decoherence processes.31,39 To find out whether or not the HGN–PIC system satisfies these criteria, the linewidths of the plasmon and exciton were extracted by Lorentzian fitting of the respective extinction spectra. For HGN-V, whose LSPR matches with that of the PIC J-aggregate resonance, we estimated Γp ≈ 408 meV. The exciton linewidth, Γe, for the PIC J-aggregate system was measured to be ∼21 meV. With these Γ values, it is clear that HGN–PIC systems satisfy the criteria image file: c7cp05455f-t7.tif and hence the plasmon–exciton interaction in these systems certainly falls within the strong coupling category.

It is worth mentioning here that very large Rabi splitting values (∼400 meV) have been reported for silver nanoprism–TDBC J-aggregate systems.33,44,45 However, all those studies used J-aggregates of an anionic cyanine dye TDBC as the excitonic system. The dye used in the present case is a cationic dye, PIC. Ag or gold nanoprisms are commonly capped with CTAB molecules, which bear positive charge and therefore surface adsorption of anionic dyes happens more strongly than cationic dyes. Interestingly, a recent study on plexcitonic coupling between silver nanoprisms and cationic PIC–J-aggregates reported a Rabi splitting value (207 meV) very similar to that of ours.8 So, clearly the coupling strength depends not only on the metal but also on the dye. In this regard, the present study shows that HGNs are as good substrates for plexcitonic coupling as silver nanostructures.

The results presented above are to some extent unexpected. The extinction spectra of the HGNs are quite damped and broadened due to an overwhelming presence of the sp ← d interband transitions, very similar to what is experienced in the case of solid gold nanospheres.46 Due to the presence of strong interband transitions, unlike silver nanostructures or gold nanorods, LSPs in solid gold nanospheres are not expected to show strong coupling with excitons.46 Interestingly though, large mode splitting values of ∼200 meV have been reported for Au nanosphere–exciton coupling.43 However, it is necessary to point out here that these large mode splitting values were measured for Au nanospheres whose LSPRs were not in resonance with the exciton resonance and therefore these values do not actually represent ‘Rabi splitting’. Not to mention that in the case of HGNs, due to their plasmon-hybridization mediated spectral tunability, we could access resonance conditions and this resulted in large Rabi splitting and coupling constant. Now, the question that we ask here is: what is the reason behind observing strong coupling in the HGN–PIC system in spite of the HGNs’ spectra being dominated by interband transitions? The parameters that control the coupling constant and hence the coupling strength include the number (N) of excitons adsorbed, the transition dipole moment (μ) of the exciton and the local electric field (E) felt by the excitons such that gN1/2μe|E|.6,47 Now, μe is constant for a given molecule but N and E are strongly dependent on the nanoparticle used. The surface area of a gold nanoprism is much higher than a nanosphere of comparable volume and therefore can host a larger number of surface-adsorbed molecular excitons, which in turn leads to an increase in coupling strength. The last thing that remains to be checked is the effect of the enhancement on the local electric field. To gain better insights, we simulated the local electric-field distribution of HGN-V using the FDTD technique. The results are shown in Fig. 7a. For comparison, we have also simulated the local electric fields of solid gold nanoprisms, and solid gold nanospheres of comparable surface area and volume (Fig. 7b and c). Furthermore, the result obtained for a solid gold nanoprism (SGN) is shown in Fig. 7d. It is obvious from the electric-field scales (logarithmic) in Fig. 7 that the maximum local electric-field enhancement for HGN is higher than the solid gold nanospheres and the SGN. Most importantly, we see that apart from the tips, the cavity of the HGN hosts a large amount of electric field. Therefore, the overall electric-field enhancement is much higher in HGN compared to the solid gold nanospheres. This large enhancement of the local electric field in HGN is due to an extremely efficient ‘plasmon hybridization’, as we have already explained in our previous works.24,25 Now, a larger field enhancement means that the molecular excitons will feel much more local electric field when they are adsorbed on the surface of the HGN compared to solid gold nanosphere of comparable volume and surface area or even solid gold nanoprism. So, both in terms of the number of adsorption sites for molecular excitons and the local electric field, HGNs are much more capable of increasing the plasmon–exciton coupling strength. Clearly, the larger surface area and most importantly, the plasmon-hybridization mediated large electric-field enhancement help to overwhelm the damping effects of interband transitions.

image file: c7cp05455f-f7.tif
Fig. 7 Spatial electric-field color maps for HGN (a), solid gold nanospheres of radii 16 nm and 13 nm (b and c), and solid gold nanoprism (d) at their respective LSPR maxima. The scale that correlates color to the electric-field intensity (|E|2/|E0|2, in the logarithmic scale) is shown to the right of each respective plot. The nanosphere with a radius of 16 nm has the same surface area and the nanosphere with a radius of 13 nm has the same volume as that of the HGN. Large electric-field enhancement within the cavity of the HGN can be clearly seen. Also, it is very obvious that the maximum local electric-field enhancement for HGNs is higher than other solid gold nanostructures of comparable volume.

We have shown in our previous studies that hollow gold nanoprisms can provide an excellent RI-sensing platform due to their spectral tunability and larger sensing volume owing to plasmon-hybridization mediated electric-field enhancement.24,25 Now, plexcitonic nanoparticles are potentially better suited for sensing applications than plasmonic nanoparticles as plexcitonic resonances are more sensitive toward the local environment.48 Therefore, molecular J-aggregate coated HGNs can possibly act as better transducers in molecular sensing applications. Such possibilities are currently being explored in our laboratory.


We have explored the plasmon–exciton interaction in a hybrid system of hollow gold nanoprisms and one of the most commonly used cyanine dyes, PIC. Preparation of the HGN–PIC–J-aggregate hybrid system is very simple and does not require any salt or base and also, no pretreatment of HGNs is necessary. The PIC molecules are adsorbed and self-assembled onto the HGN surface leading to the formation of J-aggregates as evidenced from surface enhanced Raman scattering (SERS) spectroscopy, temperature-dependent extinction spectroscopy, and zeta-potential measurements. The extinction spectra of the J-aggregate-coated HGNs show two new energy bands, separated by a transparency dip, which signifies a coherent coupling between the localized surface plasmons of HGNs and the exciton resonances of the PIC J-aggregates. An anti-crossing behavior of the two plexcitonic modes was observed with a Rabi splitting energy (Ω) of 198 meV and a coupling constant (g) of 91 meV when the LSPR frequencies of the HGNs were varied across the exciton frequency. Analysis of the spectral linewidths reveals that image file: c7cp05455f-t8.tif. This fact together with the observation of a transparency dip and mode anti-crossing behavior in the plexcitonic spectra confirms a ‘strong plasmon–exciton coupling’ in our HGN–PIC hybrid nanosystem. The agreement between our experimental results and the theoretical results, obtained through FDTD simulations, is excellent. Analysis of the electric-field distribution, obtained through FDTD simulations, suggests that the observation of such a strong coupling even in the presence of pronounced plasmon damping caused by interband transitions can be attributed to the plasmon-hybridization mediated large local electric-field enhancement in HGNs.

Conflicts of interest

There are no conflicts of interest to declare.


We thank Dr Anindita Gayen for the TOC artwork. We also thank Dr Sri Shivakumar for allowing us to use his zeta-potential analyser to perform the zeta-potential measurements. M. C. gratefully acknowledges the Science and Engineering Research Board, Govt. of India, for financial support (Grant No. EMR/2016/001605).

Notes and references

  1. M. S. Tame, K. R. McEnery, Ş. K. Özdemir, J. Lee, S. A. Maier and M. S. Kim, Nat. Phys., 2013, 9, 329–340 CrossRef CAS.
  2. J. Lee, P. Hernandez, J. Lee, A. O. Govorov and N. A. Kotov, Nat. Mater., 2007, 6, 291–295 CrossRef CAS PubMed.
  3. C. Gonzalez-Ballestero, J. Feist, E. Moreno and F. J. Garcia-Vidal, Phys. Rev. B: Condens. Matter Mater. Phys., 2015, 92, 121402 CrossRef.
  4. S. W. Bishnoi, C. J. Rozell, C. S. Levin, M. K. Gheith, B. R. Johnson, D. H. Johnson and N. J. Halas, Nano Lett., 2006, 6, 1687–1692 CrossRef CAS PubMed.
  5. A. Imamoglu, D. D. Awschalom, G. Burkard, D. P. DiVincenzo, D. Loss, M. Sherwin and A. Small, Phys. Rev. Lett., 1999, 83, 4204 CrossRef CAS.
  6. G. Khitrova, M. Gibbs, M. Kira, S. W. Koch and A. Scherer, Nat. Phys., 2006, 2, 81 CrossRef CAS.
  7. D. E. Chang, V. Vuletic and M. D. Lukin, Nat. Photonics, 2014, 8, 685 CrossRef CAS.
  8. B. G. DeLacy, O. D. Miller, C. W. Hsu, Z. Zander, S. Lacey, R. Yagloski, A. W. Fountain, E. Valdes, E. Anquillare, M. Soljačić, S. G. Johnson and J. D. Joannopoulos, Nano Lett., 2015, 15, 2588–2593 CrossRef CAS PubMed.
  9. A. E. Schlather, N. Large, A. S. Urban, P. Nordlander and N. J. Halas, Nano Lett., 2013, 13, 3281–3286 CrossRef CAS PubMed.
  10. A. M. Fales, S. J. Norton, B. M. Crawford, B. G. DeLacy and T. Vo-Dinh, Phys. Chem. Chem. Phys., 2015, 17, 24931–24936 RSC.
  11. F. Tam, G. P. Goodrich, B. R. Johnson and N. J. Halas, Nano Lett., 2007, 7, 496 CrossRef CAS PubMed.
  12. T. J. Antosiewicz, S. P. Apell and T. Shegai, ACS Photonics, 2014, 1, 454–463 CrossRef CAS.
  13. A. S. Davydov, Theory of Molecular Excitons, Plenum Press, New York, 1971 Search PubMed.
  14. M. van Burgel, D. A. Wiersma and K. Duppen, J. Chem. Phys., 1995, 102, 20–33 CrossRef CAS.
  15. T. K. Sau and C. J. Murphy, Langmuir, 2004, 20, 6414–6420 CrossRef CAS PubMed.
  16. J. E. Millstone, S. Park, K. L. Shuford, L. Qin, G. C. Schatz and C. A. Mirkin, J. Am. Chem. Soc., 2005, 127, 5312–5313 CrossRef CAS PubMed.
  17. E. Hao, G. C. Schatz and J. T. Hupp, J. Fluoresc., 2004, 14, 331–341 CrossRef CAS PubMed.
  18. P. S. Kumar, I. Pastoriza-Santos, B. Rodríguez-González, F. J. G. de Abajo and L. M. Liz-Marzán, Nanotechnology, 2008, 19, 015606 CrossRef PubMed.
  19. R. Bukasov, T. A. Ali, P. Nordlander and S. Shumaker-Parry, ACS Nano, 2010, 4, 6639–6650 CrossRef CAS PubMed.
  20. R. Bardhan, O. Neumann, N. Mirin, H. Wang and N. J. Halas, ACS Nano, 2009, 3, 266–272 CrossRef CAS PubMed.
  21. C. L. Nehl, N. K. Grady, G. P. Goodrich, F. Tam, N. J. Halas and J. H. Hafner, Nano Lett., 2004, 4, 2355–2359 CrossRef CAS.
  22. A. M. Schwartzberg, T. Y. Olson, C. E. Talley and J. Z. Zhang, J. Phys. Chem. B, 2006, 110, 19935–19944 CrossRef CAS PubMed.
  23. M. Chandra, A. Dowgiallo and K. L. Knappenberger Jr., J. Am. Chem. Soc., 2010, 132, 15782–15789 CrossRef CAS PubMed.
  24. B. Hazra and M. Chandra, ACS Sens., 2016, 1, 536–542 CrossRef CAS.
  25. B. Hazra, K. Das, S. Das Chakraborty, M. S. Verma, M. M. Devi, N. K. Katiyar, K. Biswas, D. Senapati and M. Chandra, J. Phys. Chem. C, 2016, 120, 25548–25556 CAS.
  26. E. Prodan, C. Radloff, N. J. Halas and P. Nordlander, Science, 2003, 302, 419–422 CrossRef CAS PubMed.
  27. E. Prodan and P. Nordlander, J. Chem. Phys., 2004, 120, 5444–5454 CrossRef CAS PubMed.
  28. P. Nordlander, C. Oubre, E. Prodan, K. Li and M. I. Stockman, Nano Lett., 2004, 4, 899–903 CrossRef CAS.
  29. B. Hazra, K. Das and M. Chandra, Phys. Chem. Chem. Phys., 2017, 19, 18394–18399 RSC.
  30. D. Senapati, T. Senapati, P. S. Wate, R. Kanchanapally, Z. Fan, A. K. Singh and P. C. Ray, Chem. Commun., 2012, 48, 6034–6036 RSC.
  31. G. Zengin, G. Johansson, P. Johansson, T. J. Antosiewicz, M. Käll and T. Shegai, Sci. Rep., 2013, 3, 3074 CrossRef PubMed.
  32. I. A. Struganova, M. Hazell, J. Gaitor, D. McNally-Carr and S. Zivanovic, J. Phys. Chem. A, 2003, 107, 2650–2656 CrossRef CAS.
  33. S. Balci, B. Kucukoz, O. Balci, A. Karatay, C. Kocabas and G. Yaglioglu, ACS Photonics, 2016, 3, 2010–2016 CrossRef CAS.
  34. D. L. Akins, Nanomater. Nanotechnol., 2014, 4, 1–4 CrossRef.
  35. L. J. Pace and E. L. Pace, Spectrochim. Acta, 1980, 36A, 557–561 CrossRef CAS.
  36. K. Kneipp, H. Kneipp and M. Rentsch, J. Mol. Struct., 1987, 156, 331–340 CrossRef CAS.
  37. C. Guo, M. Aydin, H. R. Zhu and D. L. Akins, J. Phys. Chem. B, 2002, 106, 5447–5454 CrossRef CAS.
  38. B. Gu and D. L. Akins, Chem. Phys. Lett., 1985, 113, 558–562 CrossRef CAS.
  39. G. Zengin, M. Wersäll, S. Nilsson, T. J. Antosiewicz, M. Käll and T. Shegai, Phys. Rev. Lett., 2015, 114, 157401 CrossRef PubMed.
  40. P. Vasa, W. Wang, R. Pomraenke, M. Lammers, M. Maiuri, C. Manzoni, G. Cerullo and C. Lienau, Nat. Photonics, 2013, 7, 128–132 CAS.
  41. W. Wang, P. Vasa, R. Pomraenke, R. Vogelgesang, A. D. Sio, E. Sommer, M. Maiuri, C. Manzoni, G. Cerullo and C. Lienau, ACS Nano, 2014, 8, 1056–1064 CrossRef CAS PubMed.
  42. G. A. Wurtz, P. R. Evans, W. Hendren, R. Atkinson, W. Dickson, R. J. Pollard and A. V. Zayats, Nano Lett., 2007, 7, 1297–1303 CrossRef CAS PubMed.
  43. D. D. Lekeufack, A. Brioude, A. W. Coleman, P. Miele, J. Bellessa, L. D. Zeng and P. Stadelmann, Appl. Phys. Lett., 2010, 96, 253107 CrossRef.
  44. S. Balci, Opt. Lett., 2013, 38, 4498–4501 CrossRef CAS PubMed.
  45. M. Wersall, J. Cuadra, T. Antosiewicz, S. Balci and T. Shegai, Nano Lett., 2017, 17, 551–558 CrossRef PubMed.
  46. J. A. Faucheaux, J. Fu and P. K. Jain, J. Phys. Chem. C, 2014, 118, 2710–2717 CAS.
  47. T. Yoshie, A. Scherer, J. Hendrickson, G. Khitrova, H. M. Gibbs, G. Rupper, C. Ell, O. B. Shchekin and D. G. Deppe, Nature, 2004, 432, 200 CrossRef CAS PubMed.
  48. O. Pérez-González, N. Zabala and J. Aizpurua, Nanotechnology, 2014, 25, 035201 CrossRef PubMed.


Electronic supplementary information (ESI) available: Consists of synthetic protocols, sample preparation methods, and spectra of hollow gold nanoprisms. See DOI: 10.1039/c7cp05455f

This journal is © the Owner Societies 2017