*Review* **Applications of Shell-Isolated Nanoparticle-Enhanced Raman Spectroscopy**

**Grégory Barbillon**

EPF-Ecole d'Ingénieurs, 92330 Sceaux, France; gregory.barbillon@epf.fr

**Abstract:** The surface-enhanced Raman scattering (SERS) is mainly used as an analysis or detection tool of biological and chemical molecules. Since the last decade, an alternative branch of the SERS effect has been explored, and named shell-isolated nanoparticle Raman spectroscopy (SHINERS) which was discovered in 2010. In SHINERS, plasmonic cores are used for enhancing the Raman signal of molecules, and a very thin shell of silica is generally employed for improving the thermal and chemical stability of plasmonic cores that is of great interest in the specific case of catalytic reactions under difficult conditions. Moreover, thanks to its great surface sensitivity, SHINERS can enable the investigation at liquid–solid interfaces. In last two years (2019–2020), recent insights in this alternative SERS field were reported. Thus, this mini-review is centered on the applications of shell-isolated nanoparticle Raman spectroscopy to the reactions with CO molecules, other surface catalytic reactions, and the detection of molecules and ions.

**Keywords:** SHINERS; SERS; core–shell nanoparticles; catalysis; electrochemistry; plasmonics

#### **1. Introduction**

Over the past ten years, the realization of plasmonic structures with a very high sensitivity of detection has significantly increased for application to surface-enhanced spectroscopies [1–10]. Among these enhanced spectroscopies, we find the surface-enhanced Raman scattering (SERS), which uses the plasmonic nanostructures or nanoparticles for amplifying the Raman signal of various molecules. For this amplification, a huge number of geometries has been examined as plasmonic nanodimers [11–15], nanorods [16–20], nanotriangles [21–25] and nanostars [26–30]. Furthermore, plasmonic nanopores have been explored in order to improve the SERS enhancement [31–34]. In addition, another type of SERS substrates has been investigated consisting of a metallic mirror on which plasmonic nanoparticles or nanostructures were deposited or fabricated allowing an enhancement of 1 or 2 magnitude orders due to hybridized modes or a coupling between nanoparticles or nanostructures via surface plasmon polaritons on the metallic mirror (film) [35–39]. Another way to improve the Raman signal was the use of hybrid nanomaterials based on zinc oxide or silicon coupled to a plasmonic layer or plasmonic nanoparticles [40–49], and also based on bimetallic nanoparticles [50–55] or other materials as metal oxides [56–61]. Another branches of the SERS field have been also explored, such as the photo-induced enhanced Raman spectroscopy [62–64], the SERS effect generated by high pressure [65,66], and the shell-isolated nanoparticle-enhanced Raman spectroscopy (SHINERS) [67–69]. Concerning SHINERS, this technique has been discovered in 2010 in order to overcome the limitations of SERS regarding the accurate characterization of different surface morphologies, materials, and biological samples [67]. The base concept of SHINERS consists of plasmonic cores that are employed for enhancing the Raman signal of molecules, and a very thin shell of silica improving the thermal and chemical stability of the plasmonic cores, being of significant interest in the specific case of catalytic reactions under difficult conditions [70,71]. By using SHINERS, several groups have already studied catalytic reactions [72–74], applications in electrochemistry [75], and also reported the detection of different chemical molecules [76–81].

**Citation:** Barbillon, G. Applications of Shell-Isolated Nanoparticle-Enhanced Raman Spectroscopy. *Photonics* **2021**, *8*, 46. https:// doi.org/10.3390/photonics8020046

Received: 1 January 2021 Accepted: 9 February 2021 Published: 12 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional clai-ms in published maps and institutio-nal affiliations.

**Copyright:** © 2021 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

The goal of this mini-review is to present the recent advances on the most used applications of SHINERS, such as the catalytic reaction-monitoring processes and the detection of molecules, over the period 2019–2020. Firstly, we will explore the SHINERS applications to the reactions with CO molecules, which are well-known model reactions, then other surface catalytic reactions, and finally the detection of molecules and ions in order to examine the potential of this SHINERS technique. In the final section, we will discuss points to be improved and advantages of the SHINERS technique, and we will address the future directions of this latter.

#### **2. What Is Shell-Isolated Nanoparticle-Enhanced Raman Spectroscopy?**

The shell-isolated nanoparticle-enhanced Raman spectroscopy belongs to the SERS field, and is based on the enhancement of the Raman signal obtained with strong electric fields coming from plasmonic core–shell nanoparticles. In SHINERS, each core-shell nanoparticle plays the role of a metallic tip as for the tip-enhanced Raman spectroscopy (TERS), and this technique allows to obtain a couple of thousand of "TERS tips" on the substrate surface to be analyzed. Thus, the enhanced Raman signal can be jointly obtained from all these plasmonic core-shell nanoparticles ("TERS tips"), allowing a gain of two to three magnitude orders compared to a single TERS tip. Furthermore, the use of the metallic nanoparticles coated with a chemically inert shell can enable the protection of the plasmonic core (SERS-active part) from the substrate surface to be analyzed and the environment. This inert shell can conform to different morphologies of substrates, and also prevent the agglomeration of these core-shell nanoparticles and the oxidation of their plasmonic core. The principal merits of such a technique are a more significant detection sensitivity and a great number of practical applications in life and materials sciences, as well as in food science and environmental pollution.

#### **3. Applications of Shell-Isolated Nanoparticle-Enhanced Raman Spectroscopy**

#### *3.1. SHINERS Application to the Reactions with CO Molecules*

In this section, we present a couple of investigations on the reactions with CO molecules (see Table 1) [82–86]. These reactions with CO molecules are well-known model reactions.

**Table 1.** Shell-isolated nanoparticle Raman spectroscopy (SHINERS) application to the reactions with CO molecules (NPs = nanoparticles; G = Graphene).


Wang et al., have reported on the CO oxidation probed SHINERS technique based on the use of Au@SiO2 nanoparticles, on which are deposited Pt or PtFe nanocatalysts (see Figure 1a) [82]. The authors demonstrated that PtFe catalysts were more active and stable than Pt ones in the CO oxidation. In the CO oxidation, two Raman peaks have been observed at 397 and 485 cm−<sup>1</sup> for Pt catalysts and at 389 and 480 cm−<sup>1</sup> for PtFe catalysts, both corresponding to the adsorption of Pt-CO (see Figure 1b). The redshift of these two Raman peaks recorded for the case with PtFe catalysts has indicated that the CO adsorption was lower on PtFe than on Pt (see the inset of Figure 1b). Moreover, three Raman peaks of oxygen species were detected with PtFe nanocatalysts, indicating that the Pt-C binding

was weakened by the presence of the ferrous center thus producing oxygen species (see Figure 1b) [82].

**Figure 1.** (**a**) Principle scheme of a SHINERS catalyst (PtFe/Au@SiO2) for CO oxidation. (**b**) SHINERS spectra for CO oxidation on Pt (blue line) and PtFe (red line) catalysts. The inset displays a zoom of the two SHINERS spectra (blue and red lines) in the range 350–550 cm−<sup>1</sup> in order to better observe the redshift of Raman peaks. All the figures are reprinted (adapted) with permission from [82], Copyright 2019 American Chemical Society.

Su et al., have reported on the detection of OH and COOH species during the CO electrooxidation process on three Pt surfaces (Pt(100), Pt(110) and Pt(111)) with the SHIN-ERS technique by using Au@SiO2 nanoparticles [83]. The authors have observed that the activity of CO electrooxidation was higher for Pt(111)/Pt(100) surfaces than Pt (110) surface. This increased activity of CO electrooxidation for Pt(111)/Pt(100) surfaces was due to the presence of OH and COOH species. For Pt(110) surface, this activity was weaker due to its high adsorption and coverage of CO on this surface [83]. Furthermore, Wondergem et al. have demonstrated the CO adsorption on Pt nanoparticles which are themselves deposited on Au@SiO2 nanoparticles (see Figure 2a,b) [84]. This investigation was realized by employing the SHINERS technique. From the SHINERS spectra, two characteristic Raman peaks of CO adsorption were recorded in the two ranges of wavenumbers: 350–600 cm−<sup>1</sup> and 1900–2150 cm−1. The first Raman peak corresponds to the CO adsorbed on Pt in a bridge configuration located at 430 and 2010 cm−<sup>1</sup> in these two ranges of wavenumbers (see Figure 2c,d), then the second one corresponds to the CO adsorbed on Pt in a linear configuration located at 505 and 2070 cm−<sup>1</sup> in these two wavenumber ranges (see Figure 2c,d) [84]. Next, the same group has studied the effect of the fabrication method of nickel (Ni) catalysts on Au@SiO2 nanoparticles for SHINERS investigations [85]. Three methods have been tested: spark ablation (SA), colloidal deposition (Col) and precusor (Pr) method. The authors have studied the CO adsorption on these three types of Ni catalysts, and concluded that Pr and Col methods were not suitable for the SHINERS technique due to the use of a high-temperature treatment of reduction. Finally, the SA technique is the most efficient for direct deposition of the nickel catalyst on Au@SiO2 nanoparticles [85].

**Figure 2.** (**a**) Fabrication process of a SHINERS catalyst (Pt/Au@SiO2). (**b**) TEM picture of a SHINERS catalyst (Pt/Au@SiO2). SHINERS spectra for CO adsorption on Pt for (**c**) the low range of wavenumbers and (**d**) the high range of wavenumbers. The grey zones correspond to the Raman peaks of Pt-CO bridge at 430 and 2010 cm<sup>−</sup>1, and Pt-CO linear at 505 and 2070 cm−1. All the figures are reprinted (adapted) with permission from [84] (https://pubs.acs.org/doi/10.1021/acscatal.9b03010 (accessed on 1 January 2021)), Copyright 2019 American Chemical Society (for all further reuses related to the excerpted material, further permissions should be directed to the American Chemical Society).

Hartman et al. have investigated the support effect on the interaction of rhodium (Rh) with CO molecules probed by SHINERS technique [86]. Two types of extrudate support have been tested by introducing alternatively CO and H2: the first one was Rh/SiO2 on which were deposited Au@SiO2 nanoparticles and the second one was Rh/TiO2 on which were also deposited Au@TiO2 nanoparticles. Under the same conditions of CO then H2, the shifts of a Raman peak (named "unknown" by the authors) for the CO hydrogenation were of 70 cm−<sup>1</sup> and 25 cm−<sup>1</sup> for Rh/TiO2 and Rh/SiO2 extrudate supports, respectively. From these Raman shifts for the CO hydrogenation, the authors have deduced that the Rh/TiO2 extrudate support had the strongest interaction with CO molecules during the catalytic process compared to the Rh/SiO2 extrudate support. Thus, this powerful interaction has resulted in a catalyst with higher efficiency for the CO dissociation [86]. Zhang et al., have reported on the adsorption of CO molecules on Pt nanocatalysts, which were deposited on novel core-shell nanoparticles composed of a gold core covered by graphene layers (Au@G nanoparticles). The authors have demonstrated via SHINERS measurements that the adsorption of CO molecules on these Pt/Au@G nanoparticles has occurred in a linear configuration [87]. To finish this section, two groups have reported on the CO reduction catalysis on Cu foil by using the SHINERS technique [88,89]. The authors have employed Au@SiO2 nanoparticles, and allowed them to deduce that the CuO*x*/(OH)*<sup>y</sup>* species were detected on the Cu foil during the CO reduction. Thus, the authors have concluded that the oxygenated species of Cu were unlikely to be the active sites easing the formation of C2+ oxygenates during the process of CO reduction [88,89].

#### *3.2. SHINERS Application to Other Surface Catalytic Reactions*

Here, we address a couple of studies on other surface catalytic reactions (see Table 2) [90–100].

**Table 2.** SHINERS application to other surface catalytic reactions (NPs = nanoparticles; IrO*x* = Iridium oxide; MBT = 2-mercaptobenzothiazole; SnO2 = Tin oxide; pNTP = para-nitrothiophenol; RhB = Rhodamine B; CNNDs = g-C3N4 nanodots; EGLs = Electrochemical exfoliated graphene layers; ITO = Indium tin oxide).


Guan et al. have demonstrated that the adsorption of propargyl alcohol (PA) on Pt(*hkl*) surfaces by using the SHINERS technique [90]. The authors have employed Au@SiO2 nanoparticles for SHINERS experiments, and they obtained adsorption of PA-privileged on Pt(100) and Pt(110) surfaces than on Pt(111) surface. The better surface reactivity for Pt(100) compared to two other Pt surfaces (Pt(100) > Pt(110) > Pt(111)) was due to the more important presence of the primary alcohol group [90]. In the next two examples, the studies of the water oxidation and the configuration of the interfacial water are addressed. Firstly, Saeed et al., have reported on the employment of the SHINERS technique to analyze in realtime the mechanisms of water oxidation with iridium oxides (IrO*x*) as electrocatalyst [91]. To do that, Au@SiO2 nanoparticles were used and deposited on IrO*<sup>x</sup>* surface for the Raman characterization (SHINERS). Thus, the authors demonstrated that SHINERS allowed to observe in real-time the chemical changes on the IrO*<sup>x</sup>* surface during the oxidation of water [91]. Secondly, Li et al., have reported on the configuration of the interfacial water at Au(111) surface, probed by the SHINERS technique using Au@SiO2 nanoparticles for this study [92]. The authors have observed redshifts of the Raman peak corresponding to the *O* − *H* stretching mode of the interfacial water when the potential went towards more negative values, indicating a configuration variation of interfacial water. Thus, the authors have shown three configurations of the interfacial water named parallel, one-H-down and two-H-down, respectively, when the potential shifted to more negative values [92]. In addition, Guo et al., have studied the adsorption of 2-mercaptobenzothiazole (MBT) on pyrite by SHINERS [93]. Au@SiO2 nanoparticles were employed and deposited on pyrite for SHINERS experiments. From the SHINERS spectra recorded with an MBT concentration of 0.01 mM, a Raman peak at 1406 cm−<sup>1</sup> was observed and corresponded to NCS ring stretch mode (see Figure 3a). This Raman peak suggested that double "minerophilic" groups of MBT were bound to pyrite surfaces in the configuration displayed in Figure 3b on the left. From the SHINERS spectra recorded with an MBT concentration of 0.1 mM, two Raman peaks at 1389 and 1409 cm−<sup>1</sup> were observed and also corresponded to NCS ring stretch mode (see Figure 3c). The Raman peak at 1409 cm−<sup>1</sup> has indicated that double "minerophilic" groups of MBT were bound to pyrite surface. The Raman peak at 1389 cm−<sup>1</sup> (starting to appear at −200 mV, see Figure 3c) has also indicated that the MBT molecule was bound to pyrite with the exocyclic sulfur atom without the presence of any nitrogen–metal bond in the configuration displayed in Figure 3b on the right. In summary, the authors

have concluded that the configuration of MBT molecules was preferentially the one in Figure 3b on the left for weaker concentrations of MBT and negative potentials and the one in Figure 3b on the right for higher concentrations of MBT and positive potentials [93].

**Figure 3.** (**a**) SHINERS spectra of a MBT solution (0.01 mM) recorded at pH 9.3 for different potentials. The black dotted line indicates the Raman peak at 1406 cm−1. (**b**) Potential configurations of the MBT adsorption on pyrite. (**c**) SHINERS spectra of a MBT solution (0.1 mM) recorded at pH 4.6 for different potentials. The black dotted lines indicate the Raman peaks at 1389 and 1409 cm<sup>−</sup>1. All the figures are reprinted from [93], Copyright 2020, with permission from Elsevier.

Bodappa et al., have investigated the electrochemical oxidation of Cu(111) and polycrystalline Cu (Cu(poly)) surfaces by using the SHINERS technique [94]. Au@SiO2 nanoparticles were employed for studying the oxidation mechanism of Cu(111) and Cu(poly) surfaces (see Figure 4a). From the SHINERS spectra for Cu(111) oxidation, intermediate species Cu-OH, Cu-O*ad*, and (Cu2O)*sur f* were observed during the oxidation process (i.e., when the potential increased, see Figure 4b). For Cu(poly) oxidation, only Cu-OH and (Cu2O)*sur f* were spotted during the oxidation on the SHINERS spectra (see Figure 4c). Thus, the authors have remarked a difference in the presence of the intermediate species during the oxidation process [94].

**Figure 4.** (**a**) SEM picture of Au@SiO2 nanoparticles on Cu surface (scale bar = 1 μm). The inset displays a TEM picture of a Au@SiO2 nanoparticle. (**b**) SHINERS spectra for Cu(111) oxidation in a 0.01 M KOH solution. (**c**) SHINERS spectra for Cu(poly) oxidation in a 0.01 M KOH solution. All the figures are reprinted (adapted) with permission from [94], Copyright 2019 American Chemical Society.

The next three works have addressed the topic of oxygen reduction reaction (ORR) on Pt(*hkl*) surfaces by employing SHINERS spectroscopy. At first, Galloway et al., have demonstrated the surface specificity of the ORR on Pt(*hkl*) surfaces by SHINERS in sodium– oxygen electrochemistry [95]. The reduction of NaO2 to Na2O2 was favored on Pt(111) and Pt(110) surfaces in 0.1 M NaClO4 dissolved in dimethyl sulfoxide (DMSO), whereas this reduction was not detected on Pt(100) and Pt(poly) surfaces (no characteristic Raman peak of Na2O2 recorded) due to the restricted interactions with adsorbed oxygens [95]. Next, the second work realized by Dong et al. dealt with the observation of intermediate species for ORR on different Pt(*hkl*) surfaces examined by SHINERS. The authors have

spectroscopically evidenced the fact that ORRs on Pt(111) surface was obtained by the generation of OOH species, while for Pt(110) and Pt(100) surfaces by the formation of OH species [96]. Finally, the same group has demonstrated the presence of intermediate species during ORR on high-index Pt(*hkl*) surfaces by SHINERS spectroscopy [97]. Au@SiO2 nanoparticles were used for the study of the ORR activity of these Pt surfaces (see Figure 5a). The authors have observed intermediate species for the two Pt surfaces (Pt(311) and Pt(211)) studied for ORR. From SHINERS spectra recorded for different values of potential, two characteristic Raman peaks at ∼765 and ∼1041 cm−<sup>1</sup> were observed and corresponded to OOH and OH species, respectively (see Figure 5b,c). Moreover, they concluded that the Pt(211) surface had a better reactivity than the Pt(311) surface due to the greater adsorption energy for OOH species with the Pt(311) surface [97].

**Figure 5.** (**a**) Principle scheme of a SHINERS (Au@SiO2) measurement for oxygen reduction reaction on Pt(*hkl*) surfaces. For different values of potential, SHINERS spectra recorded in HClO4 solution (0.1 M) saturated in O2 are shown for the surfaces of (**b**) Pt(311) and (**c**) Pt(211). All the figures are reprinted (adapted) with permission from [97], Copyright 2020 American Chemical Society.

Barlow et al. have investigated the corrosion of 304 stainless steel by using the SHINERS technique. Au@SnO2 nanoparticles were employed for examining this possible corrosion [98]. For the 304 stainless steel, the authors have evidenced a characteristic Raman (SHINERS) peak corresponding to amorphous Fe(OH)2, and also another Raman peak attributed to Cr(VI)–O bindings from a blended oxide based on Cr(VI). When KCl is present in the electrolyte, a Raman peak attributed to *γ*-FeOOH was observed. Finally, the authors have evidenced no green rust, i.e., no intermediate species during the conversion from Fe(OH)2 to *γ*-FeOOH [98]. Besides, Wang et al., have reported on the effects of the size and the nature of nanocatalysts on the hydrogenation of para-nitrothiophenol (pNTP) by employing the SHINERS spectroscopy [99]. Au@SiO2 nanoparticles were used for SHINERS experiments on which Pt, PtCu, PtNi nanocatalysts have been assembled via electrostatic interactions. At first, the authors have studied the size effect of Pt nanocatalysts on the pNTP hydrogenation, and have reported on an optimal size of 6.8 nm for Pt nanoparticles. Then, the authors have studied the kinetics of reaction for Pt, PtCu and PtNi nanocatalysts. They observed that PtCu and PtNi nanocatalysts have shown a quicker and quasi-complete conversion of pNTP than for Pt ones [99]. To finish this section on SHINERS applications to other surface catalytic reactions, Qiu et al. have investigated the effect of the presence of g-C3N4 nanodots (CNNDs) and electrochemical exfoliated graphene layers (EGLs) on the photodegradation of Rhodamine B (RhB) molecules probed by SHINERS technique [100]. For SHINERS experiments, Ag@SiO2 nanoparticles and an illumination wavelength of 632.8 nm were employed for examining this photocatalytic process of degradation of RhB molecules (see Figure 6a). In the absence of CNNDs and EGLs on the ITO substrate, the authors have noted no significant degradation of the intensity of the Raman peaks of RhB molecules. In contrast, the authors have observed a complete degradation of the intensity of these Raman peaks for an illumination time of 20 min with the presence of CNNDs and EGLs (see Figure 6b) [100].

**Figure 6.** (**a**) Principle scheme for the photocatalytic process of degradation of RhB molecules (RhB = red shapes) probed by SHINERS (Ag@SiO2 NPs = blue spheres with a grey shell). The yellow shapes correspond to the g-C3N4 nanodots (CNNDs), and the EGLs/ITO substrate is represented by the white rectangle on which the hexagonal lattice of graphene (EGLs) is depicted. (**b**) SHINERS spectra of RhB molecules recorded for various illumination times in the range 0–22 min. All the figures are reprinted from [100], Copyright 2019, with permission from Elsevier.

#### *3.3. SHINERS Application for the Detection of Molecules and Ions*

In this final section for the SHINERS applications, we report on a couple of works on the detection of molecules and ions (see Table 3) [101–107].


**Table 3.** SHINERS application for the detection of molecules and ions (NPs = nanoparticles; cc-Au = concave cubic gold; PPy = Polypyrrole).

Wondergem et al. have reported the detection of Rhodamine 6G (R6G) molecules by SHINERS spectroscopy [101]. For SHINERS experiments, (Au/SiO2)@SiO2 nanoparticles were used in order to avoid contact between gold nanoparticles and the liquid medium. The authors have obtained a detection limit of 10−<sup>12</sup> M for R6G molecules with these (Au/SiO2)@SiO2 plasmonic superstructures. Moreover, these plasmonic superstructures can enable the study of catalytic reactions in liquids by using SHINERS [101]. In addition, Zhang et al. have investigated the photoinduced behavior of dyes (N719) molecules on three rutile TiO2(*hkl*) surfaces [102]. Au@SiO2 nanoparticles were employed for SHINERS experiments. The authors have remarked that the SCN group of N719 molecules was the group that primarily adsorbed on these three rutile TiO2(*hkl*) surfaces. The authors have evidenced a shift of the Raman peak corresponding to the SCN group after an illumination time of 36 min on TiO2(110) and TiO2(001) surfaces, whereas no shift of this Raman peak was observed for the TiO2(111) surface. They concluded that the N719 molecules adsorbed on TiO2(111) surface were very stable in the time, whereas the N719 molecules had desorbed on TiO2(110) and TiO2(001) surfaces caused by the cleavage of theS=C binding [102]. Furthermore, Sun et al., have reported a detection limit of 10−<sup>9</sup> M for thiram

molecules (pesticides) probed by the SHINERS spectroscopy [103]. Ag@SiO2 nanoparticles on filter paper were employed as well as a miniaturized portable Raman analyzer based on smart-phone for SHINERS experiments. The authors have recorded SHINERS spectra for each concentration of thiram molecules (concentration range = 10<sup>−</sup>9–10−<sup>3</sup> M), where four characteristic Raman peaks of thiram molecules located at 440, 559, 1145 and 1379 cm−<sup>1</sup> are displayed (see Figure 7). By using the Raman peak at 1379 cm−1, the authors have deduced the detection limit (see Figure 7).

**Figure 7.** SHINERS spectra of thiram recorded for various concentrations in the range 10<sup>−</sup>9–10−<sup>3</sup> M, where are indicated the four characteristic Raman peaks as well as the two peaks associated to the filter paper. The figure is reprinted from [103], Copyright 2019, with permission from Elsevier.

In the next two works realized by the same group [104,105], the detection of copper ions and their oxidation states by SHINERS is addressed. Firstly, Forato et al., have investigated the detection of Cu(II) ions by SHINERS spectroscopy. Ag@TiO2 nanoparticles and three excitation wavelengths (514, 633, and 785 nm) were used for SHINERS experiments. The authors have demonstrated an optimal efficiency for the detection of the Raman peak of the Cu–N binding at an excitation wavelength of 633 nm [104]. Then, Quéffelec et al., have reported on the distinctness of Cu(I) and Cu(II) ions by SHINERS measurements [105]. The authors have used the same Ag@TiO2 nanoparticles functionalized with 2,2 -bipyridine phosphonate (bpy-PA) and an excitation wavelength of 633 nm where the efficiency was optimal [104]. From the SHINERS spectra recorded for Ag@TiO2@bpy-PA, Ag@TiO2@bpy-PA-Cu(I) and Ag@TiO2@bpy-PA-Cu(II), the authors have detected the characteristic Raman peaks of N–Cu(II) and N-Cu(I) vibrational modes at 230 and 290 cm−1, respectively (see Figure 8).

**Figure 8.** SHINERS spectra for Ag@TiO2@bpy-PA (in black), Ag@TiO2@bpy-PA-Cu(I) (in red) and Ag@TiO2@bpy-PA-Cu(II) (in blue), where the Raman peaks of N-Cu(II) and N-Cu(I) vibrational modes are depicted. The figure is reproduced from [105] with permission from the Royal Society of Chemistry, Copyright 2020.

Krajczewski et al., have reported the detection of four-mercaptobenzoic acid (p-MBA) by SHINERS measurements [106]. Au@Ag concave cubic nanoparticles (noted: cc-Au@Ag NPs) were employed for the p-MBA detection probed by SHINERS measurements. The authors have demonstrated that these cc-Au@Ag NPs have improved by 35% the enhancement factor of the Raman signal of p-MBA molecules compared to the cc-Au NPs without the thin Ag shell. Moreover, the authors have added a thin layer of SiO2 (<5 nm) on cc-Au@Ag NPs, and they recorded a reduction by 50% of the Raman signal of p-MBA molecules [106]. El-Said et al., have demonstrated the detection of neurotransmitters, such as *γ*-aminobutyric acid (GABA), by using the SHINERS technique [108]. Au@PPy nanobipyramids were used for this investigation, and allowed the detection of GABA with high sensitivity (detection limit of 116 nM). Moreover, these Au@PPy nanobipyramids have also enabled the detection of GABA in the presence of human serum, representing a real sample [108]. Zdaniauskiene et al., have used the SHINERS technique in order to ˙ study *Metschnikowia pulcherrima* yeast cells [109]. Au@SiO2 nanoparticles were employed for this investigation, and allowed to obtain SHINERS spectra more enhanced than SERS spectra. Moreover, the Au@SiO2 nanoparticles also allowed to suppress the appearance of supplementary bands due to potential interactions with the gold nanoparticles, and to identify the wall of yeast cells and their functional elements [109]. To conclude this section and also this review, two works related to cancer research, focusing on the identification of the atypical hyperplasia (AH) of the breast and the detection of tumor cells, are addressed. Zheng et al., have explored the identification of the breast AH by employing the SHINERS technique, which can provide a non-invasive diagnosis and study the cancer mechanisms at a molecular level [110]. The authors have remarked via changes in the Raman band intensities from SHINERS spectra that DNA strands have begun to snap in breast AH, and the presence of amino acid residues was more important than in normal breast tissues [110]. Nicinski et al., have reported on the improvement of the detection sensitivity of tumor cells such as renal cell carcinoma, and blood cells. This improvement was achieved by using Ag@SiO2 nanoparticles via SHINERS measurements. The authors have observed variations in the intensities of Raman bands between cancer and healthy cells due to changes in the structure and quantity of molecules present during the formation of cancer cells [107].

#### **4. Discussions and Future Directions**

The shell-isolated nanoparticle-enhanced Raman spectroscopy has been generally employed with core-shell nanoparticles composed of gold cores (or silver) and silica shells, because the gold or silver cores presented a strong SERS activity. It would be interesting to use other well-known plasmonic materials, such as aluminum, copper, pal-

ladium, and other alternative plasmonic materials (e.g., transparent conductive oxides and transition-metal nitrides) in order to study the influence of the nature of plasmonic material on the efficiency of the SHINERS technique to be analyzed different surface reactions at normal and high temperatures. Other materials for the shell fabrication, such as polymers, can be used to investigate the effect of shell material on this same efficiency cited previously. Through the three sections presented above, we have observed that the shell-isolated nanoparticle-enhanced Raman spectroscopy was non-invasive thanks to the catalytically inactive dielectric shell, and also had other advantages, such as the study of different surface catalytic reactions or adsorption of reactants on several surfaces of different natures and morphologies, and the detection of different molecules. Both these studies [84] were realized under normal conditions of temperature (T = 20–150 ◦C) and pressure. However, a great number of catalytic reactions are produced at high temperatures (T = 300–1000 ◦C) typically in industry. Thus, the thermal stability of SHINERS substrates should be improved for industrial applications. Moreover, the influence of high pressures on the stability of SHINERS substrates is still a research issue to be solved. Another challenging improvement to be addressed is to reduce the shell thickness (<1 nm; without pinhole in shell) in order to achieve a better Raman enhancement. Besides, another advantage would be to use the shell-isolated nanoparticle-enhanced Raman spectroscopy in catalysis as local nanosensors of molecules during the catalytic reactions in order to have a deep understanding of these catalytic reactions at the subnanometer scale [111]. Furthermore, the shell-isolated nanoparticle-enhanced Raman spectroscopy can be extended to other enhanced spectroscopies as tip-enhanced Raman spectroscopy (named shell-isolated TERS) [112] and sum-frequency generation spectroscopy (named SHINE-SFG) [113]. The shell-isolated TERS can allow the exclusion of interferences which are due to the presence of contaminants, and also the investigation of catalytic reactions at the level of a solid–liquid interface [112]. To finish, the SHINE-SFG spectroscopy can enable a novel type of enhancement coming from the non-linear coupling of SHINE-SFG with difference frequency generation. Thus, alternative substrates based on this type of coupling can be designed in order to enhance different signals [113].

#### **5. Conclusions**

In this mini-review, we addressed the applications of the shell-isolated nanoparticleenhanced Raman spectroscopy. We started with the SHINERS application to the reactions with CO molecules. The reactions of oxidation, hydrogenation, and adsorption of CO molecules with various catalysts have been presented. Next, we explored SHINERS studies on other surface catalytic reactions. Among these reactions, we presented a couple of works on oxygen reduction reactions realized on Pt(*hkl*) surfaces. Then, oxidation reactions of water and Cu surfaces have been exposed. Hydrogenation and photodegradation reactions, molecule adsorption, and corrosion have been also addressed. Finally, we reported on the detection of molecules and ions by SHINERS spectroscopy. The SHINERS experiments have enabled to improve the detection sensitivity of pesticides (thiram), tumor cells, and to distinguish the copper oxidation states. In conclusion, the shell-isolated nanoparticle-enhanced Raman spectroscopy can be very useful for obtaining various information on surface catalytic reactions, such as their mechanism and the intermediate species present during these reactions. Moreover, the SHINERS substrates based on core– shell nanoparticles can be employed as very sensitive nanosensors of molecules and ions.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


### *Review* **Nanoplasmonics in High Pressure Environment**

#### **Grégory Barbillon**

EPF-Ecole d'Ingénieurs, 3 bis rue Lakanal, 92330 Sceaux, France; gregory.barbillon@epf.fr Received: 7 July 2020; Accepted: 25 July 2020; Published: 28 July 2020

**Abstract:** An explosion in the interest for nanoplasmonics has occurred in order to realize optical devices, biosensors, and photovoltaic devices. The plasmonic nanostructures are used for enhancing and confining the electric field. In the specific case of biosensing, this electric field confinement can induce the enhancement of the Raman signal of different molecules, or the localized surface plasmon resonance shift after the detection of analytes on plasmonic nanostructures. A major part of studies concerning to plasmonic modes and their application to sensing of analytes is realized in ambient environment. However, over the past decade, an emerging subject of nanoplasmonics has appeared, which is nanoplasmonics in high pressure environment. In last five years (2015–2020), the latest advances in this emerging field and its application to sensing were carried out. This short review is focused on the pressure effect on localized surface plasmon resonance of gold nanosystems, the supercrystal formation of plasmonic nanoparticles stimulated by high pressure, and the detection of molecules and phase transitions with plasmonic nanostructures in high pressure environment.

**Keywords:** plasmonics; localized surface plasmon resonance; high pressure; sensing; SERS

#### **1. Introduction**

During the past decade, nanoplasmonics was employed for the production of photovoltaic devices [1–6], optical devices [7–15], and biosensors [16–20]. Additionally, nanoplasmonics enabled the enhancement of photocatalysis [21–23], the luminescence upconversion enhancement [24,25], and the optical tuning of luminescence and upconversion luminescence [26,27]. In addition, nanoplasmonics can also enhance the sum-frequency generation signal [28–32] and the Förster resonance energy transfer (FRET) [33–37]. Gold and silver were largely used for the production of plasmonic nanostructures, and other alternative plasmonic materials were also employed, such as aluminum [38,39], copper [40,41], palladium [42,43], transition-metal nitrides [44,45], and transparent conductive oxides [46,47]. The plasmonic nanostructures allowed confining the electromagnetic (EM) field into subwavelength-size zones. Concerning to the application to plasmonic biosensing, this EM field confinement allowed inducing an enhancement of Raman signal of analytes named surface enhanced Raman scattering (SERS) [48–53] or the localized surface plasmon resonance (LSPR) shift after detection of analytes on plasmonic nanostructures [54–58]. The confinement of the EM field can be controlled by adjusting the geometry and spatial organization of plasmonic nanosystems, for instance, which can be realized with various techniques of lithography [45,59–66]. In addition, various plasmonic modes can be used for biosensing based on SERS effect or LSPR shifting as dipolar and multipolar resonances [67,68], surface lattice resonances [69,70], and hybridized resonances [71,72]. In the majority of studies cited previously concerning the plasmonic sensing of analytes, the LSPR shifting and SERS measurements were realized in ambient environment (e.g., pressure). However, a relevant subject of nanoplasmonics has emerged over the past decade. This latter concerns nanoplasmonics in high pressure environment [73,74], and its potential application to sensing of molecules. In the nanoplasmonics in high pressure environment, the mechanisms for the LSPR shifts of metallic nanoparticles induced by high pressure are generally based on variations of the refractive index or the phase transitions of the surrounding medium, or deformation of metallic nanoparticles [73,74]. For instance, a study reporting on the effect of high pressure on the LSPR shift of colloidal gold nanoparticles demonstrated that the LSPR redshift of Au nanoparticles in water was due to the linear increasing of the refractive index of the water with pressure [73].

The aim of this short review is to discuss the latest advances on nanoplasmonics in high pressure environment over the period 2015–2020. Firstly, we will present the pressure effect on localized surface plasmon resonance of gold nanosystems, then the use of high pressures for the supercrystal formation of gold nanoparticles, and finish the detection of molecules and phase transitions with plasmonic nanostructures in high pressure environment.

#### **2. Nanoplasmonics in High Pressure Environment**

#### *2.1. Effect of High Pressure on Localized Surface Plasmon Resonance of Metallic Nanoparticles*

In this first section, we discuss the effect of high pressure on LSPR modes of gold nanoparticles with different shapes (see Table 1).

**Table 1.** Effect of high pressure on localized surface plasmon resonance (LSPR) of metallic nanoparticles.


Bao et al. reported the effects of high pressure and the thickness of the gasket in a diamond anvil cell (DAC) on the localized surface plasmon resonance of a colloidal solution of Au spheroidal nanoparticles (AuSNPs). Authors have measured the LSPR of AuSNPs (size = 80 nm) by varying the pressure from 2 to 12 GPa for two gaskets pre-indented to 140 μm (called GPI140) and 317 μm (called GPI317). For the GPI140, the authors have recorded the absorption spectra of AuSNPs for pressures from 2.24 GPa to 11.8 GPa (see Figure 1a).

**Figure 1.** At top, principle scheme of a DAC with a TEM picture of sample (AuSNP size = 80 nm; scale bar = 100 nm). The ruby sphere is employed in order to measure the pressure in chamber by fluorescence. At bottom, absorption spectra of AuSNPs are displayed as function of the wavelength and the pressure with a gasket pre-indented to (**a**) 140 μm and (**b**) 317 μm. Black circles correspond to experimental measurements of the absorption maximum. Red arrows represent the starting of the broadening of absorption peak. Black arrows represent the brutal change in the LSPR shift magnitude for AuSNPs. All of the figures are reprinted from [75], with the permission of AIP Publishing.

A broadening of the absorption peak of AuSNPs has occurred from the pressure of 4.13 GPa (indicated by the red arrow in Figure 1a). Subsequently, a sudden variation in the magnitude of LSPR shift for AuSNPs has occurred at 8.24 GPa (indicated by the black arrow in Figure 1a). This sudden variation is attributed to the deformation of the AuSNP shape. For the second gasket GPI317, they have recorded the absorption spectra of AuSNPs for pressures from 2.75 GPa to 10.47 GPa (see Figure 1b). The broadening of the absorption spectrum and the sudden variation in the LSPR shift magnitude for AuSNPs have occurred at the same pressure of 5.5 GPa (indicated by the red and black arrow, respectively, in Figure 1b). Furthermore, the authors remarked that the sudden variation in the LSPR shift magnitude for AuSNPs was achieved at a higher pressure for the thinnest gasket (GPI140). This was due to a better support of the part of the thinnest gasket located outside the culets in order to do a sharp expansion or contraction of the chamber where the sample is located, emerging at a higher pressure [75]. Gu et al. investigated the effect of quasihydrostatic and non-hydrostatic high pressures on the LSPR of gold nanocrystals (size = 3.9 nm) in a DAC [76]. The used quasihydrostatic and non-hydrostatic pressure media were ethylcyclohexane [78] and toluene [79], respectively. The authors have recorded no variation in the LSPR wavelength of Au nanocrystals for quasihydrostatic high pressures. On contrary, for non-hydrostatic high pressures, they observed a redshift of the LSPR of Au nanocrystals achieving 68 nm, and this latter was reversible when the pressure was decreased. This redshift was due to the deformation of Au nanocrystals (deformed shape with an aspect ratio of ∼2). When the non-hydrostatic pressure was decreased down to ambient pressure, the shape of Au nanocrystal came back its original shape [76]. Martin-Sanchez et al. demonstrated the effects of the hydrostatic pressure on LSPR of gold nanospheres and nanorods [77]. Firstly, the authors reported on the changes in the absorbance spectra of gold nanospheres (AuNS; diameter = 20 nm) in parrafin with pressure. Paraffin was used as solvent due to its easibility of stabilizing gold nanospheres in non-polar media. Authors observed a redshift of the localized surface plasmon resonance of AuNS when the pressure was increased from 0 to 17 GPa, and the redshift magnitude was around 3% of the LSPR wavelength for AuNS (see Figure 2a).

**Figure 2.** (**a**) Absorbance (optical density) spectra of Au nanospheres in paraffin at different pressures. (**b**) Localized surface plasmon resonance wavelength versus pressure. The experimental data are displayed as green points. All of the lines correspond to a fit with the Mie–Gans model where the parameters vary according the considered case. The red dashed line corresponds to the case of an imcompressible particle. The dark and light gray lines correspond to the case where medium and particle are compressible with K0 = 190 GPa (called Nano) and K0 = 167 GPa (called Bulk), respectively. The brown lines correspond to the case of an incompressible medium for a gold nanoparticle (called Nano; in dark brown) and for bulk gold material (called Bulk; in light brown). All of the figures are reprinted (adapted) with permission from [77], Copyright 2019 American Chemical Society.

These weaker LSPR variations with pressure were caused by the higher shape factor (L = 1/3), which decreased the solvent impact. Furthermore, the authors have determined from the experimental measurements the bulk modulus (K0) of the gold nanoparticles (called Nano) by using the Mie–Gans

model [77]. They found a value of K0 equal to 190 GPa, which is bigger as compared to this obtained in the work of Heinz et al. [80] which is equal to 167 GPa for bulk gold (called bulk, see Figure 2b). The Mie–Gans model enables to express the wavelength of the localized surface plasmon resonance at a given pressure (*P*) as follows (for more details, see reference [77]):

$$
\lambda\_{LSPR}(P) = \lambda\_{\mathcal{P}}(0)\sqrt{\frac{V(P)}{V\_0}}\sqrt{\epsilon(0) + \frac{1-L}{L}\epsilon\_{\mathcal{W}}(P)}\tag{1}
$$

where *λp*(0) corresponds to the bulk plasma wavelength at the ambient pressure (corresponding to the pressure *P* = 0), *V*(*P*) and *V*<sup>0</sup> are the particle volume at the pressure *P* and at the ambient pressure, respectively. The ratio *V*(*P*)/*V*<sup>0</sup> depends on the bulk modulus of gold (*K*0) and the first derivative *K* 0. This ratio and *K* <sup>0</sup> express, as follows:

$$\frac{V(P)}{V\_0} = \left(\frac{PK\_0'}{K\_0} + 1\right)^{-1/K\_0'}, K\_0' = \left(\frac{\partial K}{\partial P}\right)\_{P=0} \tag{2}$$

where *K* = *K*<sup>0</sup> + *K* <sup>0</sup>*P* is the bulk modulus of a material (here gold) at a given pressure *P*. *K*<sup>0</sup> is the bulk modulus of a material (here gold) at the ambient pressure. The value of *K* <sup>0</sup> is fixed at 6 [80,81] for all of the studies presented here. *m*(*P*) and (0) are the solvent dielectric function at the pressure *P* and the dielectric constant of gold in the short wavelength limit *λ* → 0 (or *ω* → ∞, which is commonly noted ∞), respectively. *m*(*P*) depends on the ratio *V*0/*V*(*P*). *L* is the shape factor of the nanoparticle.

Secondly, the authors investigated the pressure effect on the LSPR of gold nanorods in hydrostatic regime and beyond this latter for two mixtures of a methanol–ethanol solution. For the first methanol–ethanol (1:4) solution with gold nanorods whose the aspect ratio (AR) is 3.7 (dimensions: 21.7 nm × 5.6 nm, see Figure 3a), they have experimentally observed a redshift of longitudinal plasmonic mode in hydrostatic (*P* = 1–4 GPa) and non-hydrostatic (*P* > 4 GPa) regimes (see Figure 3b). However, in the non-hydrostatic regime (after solution solidification), the optical density decreased abruptly (see Figure 3b).

**Figure 3.** (**a**) TEM picture of the Au nanorods with AR = 3.7 (dimensions: 21.7 nm × 5.6 nm). (**b**) Absorbance (optical density) spectra of Au nanorods (AR = 3.7) in methanol–ethanol (1:4) solution at different pressures. All of the figures are reprinted (adapted) with permission from [77], Copyright 2019 American Chemical Society.

Finally, the authors studied gold nanorods (AR = 3.7, see Figure 3a) in a methanol–ethanol (4:1) solution. They have experimentally observed a redshift of longitudinal plasmonic mode in hydrostatic (1–10 GPa) regime, i.e., up to solution solidification (see Figure 4a). Then, a blueshift in the LSPR wavelength of the longitudinal mode was observed after switching from hydrostatic to non-hydrostatic

regime (see Figure 4). In this non-hydrostatic (*P* > 10 GPa) regime, the LSPR wavelength of the longitudinal mode was again redshifted when the pressure was increased (see Figure 4). However, the optical density decreased more abruptly than in the case of the methanol–ethanol (1:4) solution (see Figure 4a). Besides, weaker LSPR blueshifts for the transversal mode were also observed (see Figure 4b). These blueshifts of the transversal mode were due to the compression of Au nanorods and a higher electron density [77]. The Mie–Gans theory was in agreement with the experimental results for the measurement of the position of the plasmon peak in the hydrostatic regime for both longitudinal and transversal modes. For the non-hydrostatic regime, a difference between experiments and the Mie–Gans theory was observed.

**Figure 4.** (**a**) Absorbance (optical density) spectra of Au nanorods (AR = 3.7) in methanol–ethanol (4:1) solution recorded at different pressures. (**b**) LSPR wavelength versus hydrostatic pressure for the longitudinal and transversal modes of Au nanorods (AR = 3.7) in methanol–ethanol (4:1) solution. Orange points correspond to experimental data. All of the lines correspond to a fit with the Mie–Gans model where the parameters vary according the considered case. For both plasmonic modes, the green line corresponds to the case of an incompressible particle. The gray lines correspond to the case where particle and solvent are compressible (in dark gray = the bulk modulus of gold called Nano; in light gray = the bulk modulus of gold called Bulk). The brown lines correspond to the case of an incompressible medium (in dark brown = the bulk modulus of gold called Nano; in light brown = the bulk modulus of gold called Bulk). The vertical dashed line corresponds to the solution solidification. All of the figures are reprinted (adapted) with permission from [77], Copyright 2019 American Chemical Society.

To conclude this section, a dramatic decrease of the optical density at the LSPR peak was recorded after the solution solidification for both methanol–ethanol solutions with gold nanorods (AR = 3.7). When the pressure of the solution solidification was higher, the optical density decay was more significant (see Figures 3b and 4a, and reference [77]).

#### *2.2. Use of High Pressures for the Supercrystal Formation of Gold Nanoparticles*

In this section, we present studies regarding the pressure effect on the supercrystal formation with gold nanoparticles (see Table 2).


**Table 2.** Studies for the supercrystal formation of metallic nanoparticles stimulated by high pressure.

Schroer et al. investigated the pressure effect on reversibility of the supercrystal formation with a gold nanoparticle suspension. The authors have used Au nanoparticles (AuNPs) functionalized with a shell of poly(ethyleneglycol) (PEG). The radius of AuNPs is around 6 nm, and two lengths of PEG were used (2 and 5 kDa), and these Au nanoparticles coated with PEG were called AuNP@PEG2k and AuNP@PEG5k, respectively. First, the authors have recorded patterns of small-angle X-ray scattering (SAXS) for Au@PEG5k in a CsCl solution of 2 M for two pressures: 1 bar and 4000 bar. For the SAXS pattern obtained for the pressure of 1 bar, they observed a strong forward scattering corresponding to a liquid state of the AuNP@PEG5k solution (see at left in Figure 5a,b). Then, for SAXS pattern recorded for the pressure of 4000 bar, Debye–Scherrer rings were observed, indicating the formation of supercrystals under the form of a face-centered cubic (*f cc*) superlattice (see at middle in Figure 5a,b). Finally, they observed a reversibility of the state of the AuNP@PEG5k solution after the pressure reduction down to 1 bar (see at right in Figure 5a,b) [82].

**Figure 5.** (**a**) Small-angle X-ray scattering (SAXS) patterns of the AuNP@PEG5k in a CsCl solution of 2 M recorded for a pressure of 1 bar (at left), 4000 bar (at middle), again 1 bar (at right). (**b**) Corresponding scheme of the structural assembly of AuNP@PEG5k: at left, liquid state; at middle, face-centered cubic crystallites; at right, return to liquid state. All of the figures are reprinted (adapted) with permission from [82], Copyright 2018 American Chemical Society.

Subsequently, the authors studied the pressure effect on the supercrystal formation with the two types of AuNPs (AuNP@PEG2k and AuNP@PEG5k) in four chloride salts (CsCl, KCl, NaCl, RbCl) at a concentration of 2 M. They remarked that the constant *a* of the *f cc* superlattice of the AuNP@PEG2k and AuNP@PEG5k had decreased when the pressure had increased (see Figure 6a,b). This decreasing was dependent on the cation of the chloride salt solution. Moreover, this pressure effect on the constant *a* of the *f cc* lattice enabled the authors to calculate the effective compressibility *κeff* of the superlattice at the pressure of 4000 bar and at the fixed concentration of 2 M for each chloride salt solution. They observed higher values of *κeff* for the KCl solution for two types of AuNPs: 17.4 × <sup>10</sup>−<sup>5</sup> bar−<sup>1</sup> for AuNP@PEG2k and 39.5 × <sup>10</sup>−<sup>5</sup> bar−<sup>1</sup> for AuNP@PEG5k. They concluded that

the decreasing of the lattice constant is primarily due to the compression of the PEG layer, because the Au core shape was not modified [82].

**Figure 6.** Pressure effect on the lattice constant *a* for (**a**) AuNP@PEG5k and (**b**) AuNP@PEG2k in each chloride salt solution of 2 M (blue crosses for NaCl, red squares for KCl, orange disks for RbCl, and purple triangles for CsCl). All of the figures are reprinted (adapted) with permission from [82], Copyright 2018 American Chemical Society.

In the same research group, Lehmkühler et al. studied the kinetics of the supercrystal formation induced by pressure [83]. The authors have taken the same radius of 6 nm than previously for the gold spherical nanoparticles (AuNPs) functionalized with *α*-methoxypoly(ethylene glycol)-*ω*-(11-mercaptoundecanoate) ligands (PEGMUA). The molecular weight of PEGMUA is 5000 g.mol−1. These PEGMUA-coated AuNPs were disseminated in an 2 M chloride salt solution (RbCl). Authors observed that the time of the supercrystal formation has decreased when the jump from initial pressure (below the crystallisation pressure) to final pressure (beyond the crystallisation pressure) was more important. The time scale of this supercrystal formation has varied from 25 s to 0.3 s with the increasing of the pressure jump. This effect is linked to an improvement of the crystal quality caused by a larger speed of supercrystal formation [83].

Finally, Schroer et al. (same research group) also reported on the supercrystal formation of Au nanorods (AuNRs) stimulated by high pressure. The Au nanorods were functionalized with *α*-methoxypoly(ethylene glycol)-*ω*-(11-mercaptoundecanoate) ligands (PEGMUA2k). The dimensions of AuNRs (see Figure 7a) were 75 nm for the length and 22 nm for the width, and PEGMUA2k had a molecular weight of 2000 g·mol−1. First, authors have recorded SAXS patterns for AuNR@PEGMUA2k in a RbCl solution of 2 M for two pressures: 1 bar and 4000 bar. For SAXS pattern that was obtained for the pressure of 1 bar, they observed a same behavior than in the case of Au spherical nanoparticles seen previously, i.e., the AuNR@PEGMUA2k solution was in a liquid state (see Figure 7b). For the second SAXS pattern at the pressure of 4000 bar, Debye–Scherrer rings were distinguished showing the formation of supercrystals under the form of a 2D hexagonal superlattice (see Figure 7b). They also observed that the formation was very fast (a few seconds) and also reversible [84]. The authors also calculated the effective compressibilities *κeff* from the dependence of the interparticle distance to the pressure (see Figure 7c). They found *<sup>κ</sup>eff* ,*liquid* = 10.6 × <sup>10</sup>−<sup>5</sup> bar−<sup>1</sup> and *<sup>κ</sup>eff* ,*supercrystal* = 6.8 × <sup>10</sup>−<sup>5</sup> bar<sup>−</sup>1.

**Figure 7.** (**a**) TEM picture of an AuNR@PEGMUA2k assembly. (**b**) SAXS patterns of the AuNR@PEGMUA2k in a RbCl solution of 2 M recorded for a pressure of 1 bar (at **top**), 4000 bar (at **bottom**) with the corresponding scheme of the structural assembly (liquid state and supercrystal with a two-dimensional (2D) hexagonal superlattice, respectively). (**c**) Interparticle distance for AuNR@PEGMUA2k versus pressure. The red and blue data correspond to the liquid state and supercrystal formation, respectively. The black data correspond to the switching from liquid state to supercrystals. All the figures are reprinted (adapted) with permission from [84], Copyright 2019 American Chemical Society.

To finish this section, the same authors demonstrated that, by changing the shape of gold nanoparticles, they obtained supercrystals under the form of a different superlattice as a face-centered cubic lattice with the spherical Au nanoparticles [82], and a 2D hexagonal lattice with the Au nanorods [84].

#### *2.3. Detection of Molecules and Phase Transitions in High Pressure Environment*

In this section, we present studies regarding the detection of molecules and phase transitions in high pressure environment (see Table 3).

**Table 3.** Detection of molecules and phase transitions in high pressure environment (RI = Refractive Index, AuNPs = Gold nanoparticles; MoS2 NFs = Molybdenum disulphide nanoflowers).


Martin-Sanchez et al. demonstrated the detection of the refractive index of a methanol-ethanol (4:1) mixture in high pressure environment with gold spherical nanoparticles (AuNPs) of 20-nm diameter by following their LSPR shift [85]. First, the authors observed the LSPR shift of AuNPs in the methanol-ethanol (4:1) solution as a function of pressure (see Figure 8a–c). In the hydrostatic regime (from 0 to 10 GPa), they recorded a redshift of the AuNP LSPR due to a larger compressibility of solvent when compared to this of gold. In the non-hydrostatic regime (from 10 to 60 GPa), a blueshift of the AuNP LSPR was observed caused by the plasmon compression, which is more important than this of solvent in this case.

**Figure 8.** (**a**) Extinction spectra of AuNPs in the methanol-ethanol (4:1) solution at different pressures. (**b**) LSPR wavelength versus pressure. The dashed line represents the limit between the hydrostatic and non-hydrostatic regimes. (**c**) TEM picture of Au spherical nanoparticles with a diameter of 20 nm. (**d**) Refractive index of the methanol-ethanol (4:1) solution versus pressure. All of the colored points correspond to data found in the literature. The gray line corresponds to experimental RI measurements obtained with the expression (3). All the figures are reprinted (adapted) with permission from [85], Copyright 2020 American Chemical Society.

Subsequently, they studied the variations of the refractive index (RI) of the methanol-ethanol (4:1) solution with pressure. The authors described these RI variations with pressure by using the expression of Murnaghan type:

$$n = n\_0 \left(\frac{P\alpha}{\beta} + 1\right)^{1/\alpha} \tag{3}$$

where *n*<sup>0</sup> corresponds to the RI of the methanol-ethanol (4:1) solution taken at ambient pressure (*P* = 0). *α* and *β* correspond to parameters of fit. These parameters were obtained by fitting the expression (1)

of the LSPR wavelength at the pressure *P* by employing the expressions (2) and (3) in order to depict the variations of electron density of gold and dielectric function of solvent, respectively. They obtained *α* = 19.3 and *β* = 4.3 in the whole range of pressure (0–60 GPa) by taking *K*<sup>0</sup> = 190 GPa and *K* <sup>0</sup> = 6 (see Section 2.1 and references [85,86]). Afterwards, the authors compared their RI values as function of pressure to the literature [89–93], and these latter were generally in good agreement with this literature (see Figure 8d). Furthermore, the authors of this same research group investigated the detection of the refractive index of water in its liquid, ice VI, and ice VII phases by measuring the LSPR shift of aqueous solutions of Au nanorods at different high pressures [86] (see Figure 9). The dimensions of Au nanorods were 45.7 nm for the length and 13.4 nm for the width. By using the expressions (1)–(3), the RI values for each water state were calculated from the LSPR wavelength of Au nanorods. For the liquid phase (*P* = 0–1.8 GPa), the authors have taken *n*<sup>0</sup> = 1.33, *α* = 26, and *β* = 6. For the ice VI phase (*P* = 1.5–2.2 GPa), the values of *n*0, *α*, and *β* were equal to 1.40, 34, and 14, respectively. For the ice VII phase (*P* = 2.2–9 GPa), these values of *n*0, *α*, and *β* were equal to 1.43, 13.7, and 30, respectively. For all of the water states, *α* and *β* were obtained with the same method as described previously (with *K*<sup>0</sup> = 190 GPa and *K* <sup>0</sup> = 6), and *n*<sup>0</sup> corresponds to the RI of each water state at ambient pressure. Subsequently, these RI values of each water state were compared to the literature [92,94–96], and a good agreement between them was obtained [86].

**Figure 9.** (**a**) TEM picture of Au nanorods employed for experiments (scale bar = 200 nm). (**b**) LSPR wavelength versus pressure. The red, pink and green points correspond to experimental values for the liquid state, ice VI phase, and ice VII phase, respectively. The gray line represents the values determined with the Mie–Gans model. (**c**) Refractive index of the water versus pressure. All of the colored points correspond to data referenced in the literature. The gray line corresponds to experimental RI measurements obtained with the expression (3), and the dashed gray line corresponds to the values extrapolated with this model. All of the figures are reprinted (adapted) with permission from [86], Copyright 2019 American Chemical Society.

In addition, Runowski et al. demonstrated the detection of phase transitions of different media by measuring the LSPR shift of gold nanorods with pressure [87]. The dimensions of Au nanorods were 100 nm for the length and 40 nm for the width. Authors have found the phase transition from liquid water to ice VI for a pressure of 1 GPa due to an abrupt redshift of the LSPR of both longitudinal and transversal modes of Au nanorods caused by a large jump of the refractive index of water (from liquid state to ice VI). Subsequently, they observed the transition from ice VI to ice VII at a pressure of 2.2 GPa due to a short blueshift for both plasmonic modes caused by the presence of both ice VI and ice VII. Then, authors investigated the detection phase transitions of urea [87]. A transition pressure was observed at 0.5 GPa for the phase from urea I to urea III measured by an abrupt redshift in the LSPR of Au nanorods due to the crystal lattice change from the tetragonal structure (phase I) to the orthorhombic one (phase III) corresponding to a significant deformation between these phases [97]. A second transition pressure was observed at 2.8 GPa for the transition from urea III to urea IV characterized by a smaller redshift of the Au nanorod LSPR due to the crystal lattice change from the

orthorhombic phase III to the orthorhombic phase IV corresponding to a weaker deformation between these phases [97].

In the work of Sun et al., the authors reported on the surface enhanced Raman scattering (SERS) enhancement induced by high pressure with semiconducting nanoflowers (MoS2 NFs) decorated with gold nanoparticles (AuNPs) [88]. The diameters of the MoS2 NFs and AuNPs were 700 nm and 10 nm, respectively (see Figure 10a). They used the rhodamine 6G (R6G) molecules as SERS probe in the experimental measurements. The authors recorded the SERS spectra of R6G molecules on AuNPs/Mo2 NFs at the excitation wavelength of 532 nm for pressures varying from 0 to 8.38 GPa. The highest enhancement of the SERS signal was recorded for the pressure of 2.39 GPa (see Figure 10b). At this pressure, the SERS enhancement was due to a better alignment of the energy levels between MoS2 NFs, Au, and R6G molecules. This better alignment was obtained by the reduction of the energy of the band gap of MoS2 NFs, and gap of the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) levels for R6G molecules, when the applied pressure increased. The Fermi energy level of gold was kept almost constant when the pressure increased [88]. Thus, at this pressure of 2.39 GPa, two charge transfers (CTs) have occurred for enhancing the SERS signal (see Figure 10c).

**Figure 10.** (**a**) SEM picture of AuNPs/MoS2 NFs (scale bar = 500 nm). (**b**) SERS spectra of MoS2/Au/R6G system at different pressures. (**c**) Charge transfer mechanism for the SERS enhancement induced pressure. All of the figures are reproduced from [88] with permission from the Royal Society of Chemistry.

The first CT mechanism was an electron transfer from the HOMO level of R6G to the conduction band minimum (CBM) of MoS2 NFs. Subsequently, the second one was a two-step mechanism: (i) electron transfer from the HOMO level of R6G to the Fermi energy level of Au and (ii) electron transfer from the Fermi energy level of Au to CBM of MoS2 NFs. In this study, the band gap energy of MoS2 NFs was determined by using the following expression [98]:

$$E\_{\%} = 1.68 - 0.07P + 0.00113P^2 \tag{4}$$

where *Eg* is the band gap energy of MoS2 NFs (in eV), and *P* is the pressure (in GPa). Thus, the *Eg* value for MoS2 NFs was calculated at the pressure of 2.39 GPa, and this latter was equal to 1.51 eV, which was weaker than its value at ambient pressure (1.68 eV) proposing that CBM and valence band maximum (VBM) levels of MoS2 NFs became smaller. Furthermore, the SERS mechanism at ambient pressure was a two-step mechanism: (i) electron transfer from the HOMO level of R6G to the Fermi energy level of gold enabled by the excitation laser wavelength (532 nm) and (ii) transfer of hot electrons produced by the plasmon resonance of AuNPs to the CBM of MoS2 NFs (see Figure 10c). Finally, for the pressures superior to 2.39 GPa, the SERS intensity decreased due to the fact that the HOMO level of R6G molecules has exceeded the Fermi energy level of gold [88].

#### **3. Future Directions**

The nanoplasmonics in high pressure environment has generally been studied for gold. It would be interesting to apply this type of studies to other plasmonic materials that are well-known, such as silver, copper, palladium, and aluminum [39,40,43] in order to have the influence of the nature of the plasmonic material on the LSPR shift in high pressure environment. We can extend this investigation type to other alternative plasmonic materials, such as transition-metal nitride nanoparticles [44], transparent conductive oxides [46], and iron carbide nanoparticles encapsulated by graphene [99], which are materials at lower costs having a better temperature stability. Moreover, the domain of the nanoplasmonics in high pressure environment can be applied to sensing of analytes, pollutants in high pressure media as the marine medium, for instance. Another future direction of this domain is the SERS effect induced by high pressure, which represents a novel frontier in the SERS field.

#### **4. Conclusions**

In this review, we discussed the emerging topic of the nanoplasmonics in high pressure environment. First, we spoke about the effect of high pressure on the localized surface plasmon resonance and the optical density of gold nanoparticles with different shapes and sizes. LSPR shifts were observed for gold nanoparticles and were dependent on the shape, size, pressure regime (hydrostatic and non-hydrostatic), and surrounding medium. Subsequently, we presented studies on the supercrystal formation in a high pressure environment. The supercrystal formation was dependent on the shape of gold nanoparticles influencing the form of the crystal superlattice. The phenomenon of the supercrystal formation was reversible and very fast (a few seconds). To finish, we reported on the detection of phase transitions and refractive index variations for different liquids as water and urea, by measuring the LSPR shift of gold nanoparticles in these liquids. Moreover, we also reported on the detection of chemical molecules by using the SERS effect that was induced by high pressure. In summary, the nanoplasmonics in high pressure environment can be very useful for obtaining structural information on solvents or studying optical, thermodynamic properties of several liquids (organic and inorganic) or solids.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


© 2020 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### MDPI

St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Photonics* Editorial Office E-mail: photonics@mdpi.com www.mdpi.com/journal/photonics

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com

ISBN 978-3-0365-3704-7