Single-Particle Scattering Spectroscopy

Nanophotonics 2021; 10(6): 1621–1655

Review

Alexander Al-Zubeidi, Lauren A. McCarthy, Ali Rafiei-Miandashti, Thomas S. Heiderscheit and Stephan Link* Single-particle scattering spectroscopy: fundamentals and applications https://doi.org/10.1515/nanoph-2020-0639 imaging techniques used to extract spectral information. Received December 4, 2020; accepted February 16, 2021; Finally, we provide a brief overview of specialized setups published online March 8, 2021 for in situ measurements of nanoparticles in liquid systems and setups coupled to scanning tip microscopes. Abstract: Metallic nanoparticles supporting a localized surface plasmon resonance have emerged as promising Keywords: dark-field; hyperspectral; nanoparticle; platforms for nanoscopic labels, sensors, and (photo-) plasmon; polarization. catalysts. To use nanoparticles in these capacities, and to gain mechanistic insight into the reactivity of inherently heterogeneous nanoparticles, single-particle character- 1 Introduction ization approaches are needed. Single-particle scattering spectroscopy has become an important, highly sensitive The localized surface plasmon resonance (LSPR) is the tool for localizing single plasmonic nanoparticles and collective oscillation of free electrons excited by incident studying their optical properties, local environment, and light (see Figure 1A) [1, 2]. The LSPR depends on the shape reactivity. In this review, we discuss approaches taken for [3–6], size [7–10], composition [11–13], and local envi- collecting the scattered light from single particles, their ronmentofthenanoparticle[14,15].Asaresult,theLSPR advantages and disadvantages, and present some recent is a sensitive probe of chemical reactions on the nano- applications. We introduce techniques for the excitation particle’ssurface[16–19], changes in the surrounding and detection of single-particle scattering such as high- refractive index [13, 20, 21], and charge [22] and energy angle dark-field excitation, total internal reflection dark- transfer [23] to nearby species. Further, the LSPR func- field excitation, scanning near-field microscopy, and tions as a nanoantenna [24, 25], focusing the energy of the interferometric scattering. We also describe methods to incident light to a nanoscopic volume, enhancing the achieve polarization-resolved excitation and detection. We catalytic activity [26–28], local heating capacity [29], then discuss different approaches for scanning, ratio- Raman scattering of molecules attached to the surface metric, snapshot, and interferometric hyperspectral [30–32], and fluorescence of nearby fluorophores [20, 33, 34]. Finally, as a result of their large scattering cross- sections, metal nanoparticles can be readily characterized Alexander Al-Zubeidi and Lauren A. McCarthy contributed equally to in a standard dark-fieldmicroscopeandlocalizedwith this work. nanometer precision [35, 36], allowing them to serve as *Corresponding author: Stephan Link, Department of Chemistry, nonbleaching labels for biological samples [37]. The Department of Electrical and Computer Engineering, Rice University, combination of the ability of the LSPR to efficiently drive 6100 Main Street, MS 60, Houston, TX 77005, USA, reactions, spectrally monitor these reactions, and localize E-mail: [email protected]. https://orcid.org/0000-0002-4781-930X them, has resulted in 30+ years of intensive research into Alexander Al-Zubeidi, Lauren A. McCarthy, Ali Rafiei-Miandashti and Thomas S. Heiderscheit, Department of Chemistry, the optical properties of metal nanoparticles, especially Rice University, 6100 Main Street, MS 60, Houston, TX 77005, those made of gold [38, 39]. USA, E-mail: [email protected] (A. Al-Zubeidi), [email protected] Tracking the LSPR on the single-particle level has (L.A.McCarthy),[email protected](A.Rafiei-Miandashti), emerged as the standard for characterizing the structure- [email protected] (T. S. Heiderscheit). https://orcid.org/0000-0001- function relationships of plasmonic nanomaterials 8124-4661 (A. Al-Zubeidi). https://orcid.org/0000-0002-9646- [40–42]. Both lithographically fabricated and especially 5333 (L. A. McCarthy). https://orcid.org/0000-0001-9838-8442 (A. Rafiei-Miandashti). https://orcid.org/0000-0003-4851-5065 chemically synthesized nanoparticles have heteroge- (T.S.Heiderscheit) neous size and shape distributions [43, 65]. Such

Open Access. © 2021 Alexander Al-Zubeidi et al., published by De Gruyter. This work is licensed under the Creative Commons Attribution 4.0 International License. 1622 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

easily averaged out and hence missed. While trends can be established, single-particle spectroscopy, often correlated with electron microscopy, is required to gain a complete understanding of nanoparticle properties [44]. To optically monitor single particles, several imaging and spectroscopic techniques have been developed including dark-field scattering [45, 46], extinction [47], photothermal imaging [48–50], scanning photoionization microscopy [51, 52], photoluminescence [53, 54], and transient absorption spectroscopy [55, 56]. Of the single- particle techniques, dark-fieldscatteringspectroscopyis arguably the most ubiquitous [40] due to its ease of implementation, high-throughput capabilities, and high spectroscopic resolution. Dark-field microscopy has become a popular research tool in the nanoscience community [40, 45, 46, 57]. Dark-field microscopy is based on the principle that the incident light is reflected, refracted, physically blocked, or confined in the near-field in a way that the only light collected by the detection optics is that scattered by the nanomaterial [40, 46, 58]. Dark-field excitation can be achieved via commercial dark-field condensers, evanescent wave excitation, subwavelength apertures, or normal incidence light with spatial filtering. Single- particle scattering spectroscopy can be implemented in adark-field microscope by attaching a spectrograph, filters, or an interferometer to characterize the intensity of the scattered light as a function of wavelength. Dark-field microscopy is nearly background-free and the large scattering cross-sections of plasmonic nanoparticles allow facile single-particle imaging, localization, and spectral acquisition. The unique capabilities of single- particle scattering spectroscopy have enabled recent advancements such as monitoring catalysis in real-time [59], measuring biomolecule adsorption [60], and quan- Figure 1: Principles of dark-field scattering. tifying single DNA modifications within cell nuclei [61]. (A) Schematic illustration of the LSPR. (B) Far-field dark-field Furthermore, in the absence of heterogeneity, dark-field scattering geometries: condenser-based (left) and objective-based scattering spectroscopy makes it possible to gain key in- HADF excitation (middle), and normal incidence dark-field excitation using a reflecting objective. Yellow represents the excitation light, sights into nanoparticle-driven reaction mechanisms [62] while orange indicates the scattering, collected either in the forward and polarized-light matter interactions [63, 64]. direction for condenser-based HADF excitation and normal This review focuses on several recently developed incidence dark-field excitation or the backward direction for single-particle scattering spectroscopy methods and is θ θ objective-based HADF excitation. The NAs and angles ( i and e) organized as follows: Section 2 covers the physical princi- listed are for excitation and collection, respectively. Angles are ples of light scattering by plasmonic nanoparticles. Section measured relative to the normal of the coverslip (white rectangles). fi (C) TIR-based near-field excitation geometries. (D) Schematic 3 describes several dark- eld excitation and collection illustration of aperture-based SNOM. geometries, with both near-field and far-field light excitation, scanning near-field optical microscopy (SNOM), nanoparticle heterogeneity makes quantitative moni- interferometric scattering techniques, and polarization- toring with ensemble spectroscopy difficult or impossible, resolved analysis. Section 4 describes spectroscopic and since contributions from a nanoparticle subpopulation imaging methods for single-particle scattering, including having potentially the most interesting properties can be those based on scanning, ratiometric, snapshot, and A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1623

interferometric detection. Section 5 covers specialized modes, the free electrons maintain collective oscillations, techniques for acquiring spectra in situ in electrochemical but in multiple antinodes around the surface of the par- cells or biological media, and dark-ﬁeld spectroscopy ticle. The determination of lmax is dependent on the size of coupled with tip-based microscopy techniques. the sphere and the cross-sections are given by Equations (1–3) [1, 13]:

2π ∞ 2 Physical principles of light σ = ( ) ∑( + ) [ + ] ext 2 2l 1 Re al bl (1) k l=1 scattering by plasmonic 2π ∞ σ = ()∑( + )|()|2 + | |2 nanoparticles scat 2 2 l 1 al bl (2) k l=1

When light is incident on a metallic nanoparticle, the free and electrons of the metal are driven to oscillate 180° out of σabs = σext − σscat (3) phase with respect to the light’s instantaneous electric field (Figure 1A) [1]. Meanwhile, the nuclei of the particle lattice where k is the wavenumber of the incident light (k = 2π/λ). attract the electrons to return to their equilibrium posi- λ is the wavelength of light in the medium and σext, σscat, tions. Just as in a harmonic oscillator, the electrons over- and σabs are the extinction, scattering, and absorption shoot equilibrium, however, and continue to oscillate back cross-sections. al and bl are expressed in terms of the ψ η and forth. When the frequency of the incoming light spherical Ricatti-Bessel functions l and l: matches the frequency of this oscillation, a resonance ⋅ ψ ()⋅ ψ'()− ψ'()⋅ ψ () m l mx l x l mx l x condition is met and the particle efficiently absorbs and al = (4) m ⋅ ψ ()mx ⋅ η'()x − ψ'()mx ⋅ η ()x scatters light [1]. The LSPR allows a particle to absorb and l l l l ′ ′ scatter light from a much larger cross-section than the ψ (mx) ⋅ ψ (x) − mψ (mx) ⋅ ψ (x) = l l l l physical size of the nanoparticle, and gives rise to a strong bl ′ (5) ψ (mx) ⋅ η′(x) − mψ (mx) ⋅ η (x) field enhancement around the particle’s surface [24]. The l l l l light that is absorbed by the particle can excite electronic where m is a refractive index parameter (m = np/nm). np is transitions leading to hot carrier generation [65–67], the refractive index of the sphere of radius, R, and nm is the followed by photoluminescence [68] and charge transfer refractive index of the medium surrounding the sphere. x is [22, 69] to the surroundings [22, 69–71]. Most of this a size parameter, defined as x = kR. absorbed energy decays nonradiatively through coupling In order to gain mechanistic insight on the influence of to nanoparticle phonons and vibrational modes of the the particle’s surroundings and volume, it is useful to surroundings, generating heat. Radiative decay of the consider the Mie theory solution for just the dipolar mode plasmon occurs through Rayleigh scattering at the incident (l = 1). While this assumption is only strictly correct for frequency [38]. These processes ultimately dampen the particles smaller than ∼10 nm in diameter [1], the general plasmon oscillation, leading to line broadening, and result conclusions are valid for a variety of particle sizes [77]. The in a spectrum of wavelengths over which the particle extinction cross-section for the dipolar plasmon mode is strongly interacts with light. given by [1, 13]: In 1908, Gustav Mie provided an analytical solution to 3/2 ’ 18πϵ V ϵ Maxwell s equations for the extinction, scattering, and σ = m 2 (6) ext λ 2 2 light absorption cross-sections (σ)ofsmall,spherical (ϵ1 + 2ϵm) + (ϵ2) particles [72]. Although simulation methods that dis- 2 where ϵm is the permittivity of the medium (ϵm = n ), cretize the particle’s shape and therefore can be applied to m ϵ1 and ϵ2 are the real and imaginary components of any nanoparticle have emerged since [73–75], Mie theory the dielectric function of the nanoparticle defined as still remains in use today for modeling and illustrates best 2 ϵ(λ)=n = ϵ1(λ)+iϵ2(λ),andV istheparticlevolume. the underlying physics [76]. The cross-sections are p Similar in form, the scattering cross-section can be expressed as a series expansion of a class of spherical approximated by: harmonics [1, 13]. The expansion is over the plasmon modes from l = 1, the dipolar plasmon mode, to truncation 24π3ϵ2 V2 (ϵ − ϵ )2 + (ϵ )2 σ = m 1 m 2 sca 4 2 2 (7) at a maximum lmax value (multipolar modes). In the λ (ϵ1 + 2ϵm) + (ϵ2) dipolar mode, the free electrons are all polarized from one sideoftheparticletotheother,whileinmultipolar and the absorption cross-section by: 1624 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

1/2 6πϵ V ϵ − ϵ NA = n sin θ (9) σ = m Im[]m (8) abs λ ϵ + 2ϵ m Condenser-based HADF excitation places a central According to Equations (7) and (8), the scattering beam stop in the excitation path, producing a hollow cone ° cross-section increases as a function of V2, whereas the at an oblique angle, such as 43 (NA =0.9)intheexample absorption cross-section increases linearly with particle given here. The forward-scattered light is collected at an ° volume. Therefore, the extinction spectra are dominated by angle of 27 (NA = 0.7), limiting the collection of the high- absorption for small particles and by scattering for large angled excitation light transmitted through the sample. particles. Moreover, for small imaginary components of the Objective-based HADF excitation uses the same dielectric function of the metal, the extinction cross-section objectivetoexcitethesampleandcollecttheback- scattered light. Excitation light is first shaped into an is maximized when ϵ1 =−2ϵm, which reflects the dependence of the LSPR peak on the dielectric function of the annulus using a central beam block and then directed medium. Finally, the Lorentzian forms of the cross-sections through an annular dark-field mirror. The light reflected by the mirror is sent into a dark-field objective designed reveal that the peak locations are primarily governed by ϵ1, with a hollow collar around the lenses. A curved mirror the real part of the dielectric function of the metal, while ϵ2 governs the line-width. Dark-field spectroscopy charac- within the collar illuminates the samples with a hollow ° terizes the light scattered by the nanoparticle, which, cone at a 75 angle (NA = 0.97). Scattered light is collected ° according to Equation (7) contains information easily inthelensoftheobjectiveat53 (NA = 0.8), while avoiding fl related to the refractive index of the surroundings, the the collection of re ected light from the surface. material, and the size of the nanoparticle. Normal incidence dark-field excitation (Figure 1B, right) illuminates nanoparticles at normal incidence and blocks the transmitted beam while collecting light scattered at higher angles. One way to block the transmitted light is to 3 Single-particle scattering use a reflecting objective, which has a beam block right excitation and collection before the primary mirror of the objective. When placed correctly, the normal incidence excitation light is blocked geometries while a large collection angle of 41° (NA = 0.65) collects the forward-scattered light from the nanoparticle. Another Due to the much smaller signal from individual nano- approach to normal incidence dark-field excitation is to use particles compared to an ensemble, a vanishing back- a lens-based objective with a central beam block located ground is necessary to measure the scattering spectra of after the back aperture of the objective lens [79]. This single nanoparticles. The principle of dark-field micro- approach is limited to well-collimated and tightly-focused scopy relies on blocking the collection of the incident excitation, typically from a laser, so that the spot size does light, while collecting the scattered light from individual notexceedthatoftheblock. nanoparticles. Numerous instrument geometries have Near-field excitation techniques limit the amount of been developed to excitenanoparticles in the far-field and unwanted excitation light detected by exciting particles near-field while following this principle [78–80]. Far-field with near-field confined light. Figure 1C shows geometries and near-field excitation techniques are introduced here, that generate an evanescent wave by total internal reflec- and are thoroughly discussed in Sections 3.1 and 3.2, tion (TIR). TIR occurs when light passes from an optically respectively. dense medium with refractive index n1 into a less dense Far-field excitation geometries avoid collection of the medium with refractive index n2 above the critical angle θc, excitation light by using high incidence angles that miss the for that interface, given by Equation (10) [81]: collecting objective or by using direct beam blocks. n2 Figure 1B shows three different examples of far-field exci- θc = arcsin ( ) (10) n1 tation based dark-field geometries. Condenser and objective-based high-angle dark-field (HADF) excitation For the geometries shown in Figure 1B, the critical angle is geometries work by using a larger excitation cone, or nu- 41° for the air-glass interface. An evanescent wave is merical aperture (NA), than the acceptance cone of the generated in the less dense medium, exciting the nano- collecting objective lens (Figure 1B, left and center). The NA particles. Due to its nonpropagating nature into the far- can be calculated from the refractive index n, of the medium field and short penetration depth, the evanescent wave and the half-angle of the incident light cone, θ [81]: does not enter the objective. A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1625

The penetration depth d0, of the evanescent field is excitation techniques employing a shallow angle can highest when the incident angle, θi is close to the critical therefore be an attractive option. In this section, we will angle and can be calculated using Equation (11) [81]: discuss specific examples and provide more details on fi λ far- eld excitation techniques enabling collection of both √̅̅̅̅̅̅̅̅̅̅̅ d0 = (11) backscattered and forward-scattered light. π 2 2θ − 2 4 n1sin i n2 3.1.1 Far-field techniques for collecting backscattered For example, 650 nm light at an air/glass interface has a light penetration depth of 71 nm when exciting at 55°. The evanescent field intensity I(z) away from the interface can Collecting the light backscattered from particles can be a be calculated using Equation (12) [81]: convenient approach for reducing background noise from d0 other components in the sample and isolating the scat- I (z) = I0 exp ( ) (12) z tering from nanoparticles. For example, spherical nanoparticles that are 100 nm in size forward- and backscatter Where I is the incident intensity of the beam and z is the 0 with approximately equal intensities, but larger media, distance from the interface. such as cells, primarily forward-scatter [98]. Therefore, Dark-field scattering based on prism-coupled TIR backscattering approaches can limit noise from extra- excitation (Figure 1C, left) places a prism on the back of a neous media in biological samples. Two key methods for coverslip, leading to TIR at the interface where the parti- collecting the light backscattered from nanoparticles are cles are deposited. Forward-scattered light from the objective-based HADF excitation, in which the collection nanoparticles is collected by the objective. Waveguide objective also gives an annulus of illumination, or HADF dark-field excitation (Figure 1C, center) couples light into excitation from a single direction using separate illumi- acoverslip[80,82]oropticalfiber [83–91] above the nation and collection optics. critical angle to confine light to the substrate. When the beam meets the interface, the beam cannot escape and 3.1.1.1 Objective-based HADF excitation instead totally internally reflects, creating an evanescent As described in Section 3 above, objective-based HADF wave. Waveguide dark-field excitation benefits from the excitation uses the objective as both the excitation and flexibility in sample location above or below the wave- collection lens. Objective-based HADF excitation can be guide without the need for a prism or condenser on the conveniently implemented, as only a single optical side opposing the objective. element needs to be aligned to the sample, increasing Near-field excitation can also be achieved using measurement stability [99]. While objective-based HADF SNOM, demonstrated in Figure 1D. SNOM generates an excitation has been successfully implemented in a number evanescent wave using an aperture that is smaller than the of studies [100–102], this technique has drawbacks limiting wavelength of light, or by scattering light off a tip [92]. The the signal-to-noise ratio. The excitation angle and direction evanescent wave excites the plasmon of nanoparticles near are not adjustable and much of the incident light is the tip and the forward-scattered light is collected by spatially filtered from reaching the sample. Further, the NA conventional far-field optics. of the collection lens is limited to lower values. The Baumberg group created an annular HADF technique for monitoring nanoparticle growth in solution using 3.1 Far-field excitation geometries a white-light laser and a high-NA objective [103]. The authors used a supercontinuum laser, expanded the beam Far-field excitation geometries are particularly convenient size, and blocked the center of the beam, creating a ring of for imaging through media such as aqueous samples in a light, which emerged from the objective as an annulus. The spectroscopic cell, as the incident light can penetrate backscattered light was collected and analyzed by a spec- through the whole cell, as opposed to near-field light, trograph. The use of a supercontinuum laser enabled short which cannot [16, 61]. Further, the polarization properties exposure times and a high signal-to-noise ratio. of the exciting light change as a function of incident angle [93, 94], potentially resulting in different scattering spectra 3.1.1.2 Unidirectional HADF excitation for steeper excitation angles or near-field excitation Unidirectional HADF excitation overcomes some of [95–97]. When preserved polarization properties are the limitations facing objective-based HADF excitation. required, normal incidence dark-field excitation or HADF Namely, there is greater control and flexibility in the 1626 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

incident angle and direction, and the light can be easily polarized, allowing optical characterization of anisotropic particles [100, 104, 105]. While objective-based HADF excitation gives an annular-symmetric cone of incident light, exciting the sample evenly from all in- plane directions, unidirectional HADF excitation has a single propagation direction. Unidirectional HADF excitation can be implemented through two methods: reflection of the incident light off the same side of the coverslip that has the nanoparticles, with the objective directly above the particles (Figure 2A) [78], or refraction at a high angle through the opposite side of the coverslip [96]. The reflection method can be achieved by simply directing light from a fiber-coupled source or laser onto the sample at an angle [106]. Alternatively, Tcherniak et al. used a dark-field objective described above and passed laser light through the hollow collar around the lenses from a singledirection[77].Inbothcases,thelightreflected at high angles is then missed by the acceptance angle of the objective, while light backscattered from the sample is collected. The reflected light approach is convenient for characterizing the scattering of nanostructures on opaque substrates[78]. Intherefractionmethod,the incidentlight is directed to the back of the coverslip at an angle below the critical angle. The light then refracts at a high angle as it crosses the glass-air or glass-water interface, excites the sample, misses the objective lens, and forward-scattered light is collected. A few studies have utilized this method Figure 2: Far-field excitation geometries. [96, 107, 108], but it is a less common approach than (A) Unidirectional HADF excitation. The incident light is reflected at reflected-light unidirectional HADF excitation because the interface at a large angle. (B) Small incident angle objective- the incident angle of the light is limited by the refractive based excitation. The same objective lens is used to focus the excitation on the sample and collect the scattering. The collection is index of the substrate and the incident polarization can be spatially filtered, removing the excitation light. (C) Single-particle distorted by the refraction. scattering spectra of a plasmonic heptamer with two different The Capasso, Halas, and Nordlander groups have excitation angles. (D) Diagram showing normal incidence dark-field utilized the reflected light unidirectional HADF excitation excitation, with incident light focused onto a beam block, while light approach to explore the effect of the incident angle on the forward-scattered to non-normal angles is collected. CM: curved single-particle scattering spectra of nanoparticle oligo- mirror, L1-L3: achromatic lenses, FS prism: fused silica prism. (E) Dark-field scattering spectrum of a single 10 nm gold particle and mers that support Fano resonances [78]. Fano resonances near-zero background spectrum. (F) Condenser-based HADF occur when bright scattering modes and dark absorbing excitation diagram and a dark-field image of a single cell nucleus plasmon modes spectrally overlap and interfere, resulting labeled with gold and silver nanoparticles functionalized with in a sharp dip in the scattering spectra [109]. They fi antibodies to detect two different cytosine modi cations on DNA. compared the scattering spectra of single heptamers with Scale bar is 10 μm. (G) Experimental and (H) simulated single- ° particle scattering spectra of three types of particles inside the cell. HADF excitation with a 70 incident angle and an alter- The Ag–Au particle is a dimer of silver and gold particles and in native approach of exciting through the objective lens at a simulations, the distance between the silver and gold particles is smaller 20° angle (Figure 2A and B) [78]. In the former, the 1 nm. (I) Images produced from the three wavelength windows based on the spectra in H for identification of gold and silver single particles and silver-gold dimers in the cell, indicating the quantity (D, E) Adapted with permission from Kukura and coworkers, and distribution of two different types of modified DNA. (J) Overlay of Copyright the American Chemical Society, 2014 [79]. (F–J) Adapted the three images in I. (A–C) Adapted with permission from Capasso with permission from Irudayaraj and coworkers, Copyright the and coworkers, Copyright the American Chemical Society, 2010 [78]. American Chemical Society, 2015 [61]. A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1627

unidirectional excitation reflected off the sample interface normal incidence light extinguished the excitation by a and only the light backscattered by the particle was factor of >107 [79]. The authors successfully employed this collected by the objective. In the latter, the excitation light technique to collect the single-particle scattering spectrum was spatially filtered and aligned to the left edge of of a 10 nm gold particle (Figure 2E), while typical dark-field thebackapertureoftheobjectivelens.Theleftsideofthe setups can only collect single-particle spectra down to objectivethenfocusedthelightontothesamplegivingthe 20 nm gold particles [40]. In addition, filter-free single- excitation a single propagation direction. The light molecule fluorescence signals can be acquired by this backscattered by the oligomer was collected with the technique. A similar approach was taken by Kirchner and same objective and the excitation light also collected was Smith et al. who focused a white-light laser through the then spatially filtered. The authors found that with the sample and onto the beam stop of a reflecting objective. traditional 70° incident angle, there were increased The primary mirror only collected light scattered at high retardation effects, caused by the phase change of light as angles from the nanoparticles, while the back of the sec- it propagated through the sample laterally [110]. These ondary mirror blocked the incident light [111]. retardation effects caused excitation of additional dark Normal incidence excitation can also be achieved by modes in the heptamer, which ultimately broadened the illuminating light through the objective using a dot mirror scattering spectra and blue-shifted the Fano resonance [112], which is essentially the opposite of an annular dark- (Figure 2C). In contrast, with a 20° incident angle, retar- field mirror (Figure 2B center). A dot mirror sits in the center dation effects were minimal as the light mostly propa- of a filter cube to reflect light in the center of a collimated gated through the shallow depth of the oligomer and beam and to allow light transmission at the sides. Light is the spectra featured sharper peaks. A simple adjustment reflected by the dot mirror to emerge at a normal incidence of incident angle through unidirectional excitation through the sample. Light backscattered at high angles by geometries can significantly alter the scattering spectra of the particle is collected by the objective, while reflected large nanoparticles and assemblies that support dark incident light from the substrate is blocked by the dot modes [78]. mirror. The Berry group has achieved a similar setup using a small rod mirror [98]. 3.1.2 Far-field techniques for collecting forward- scattered light 3.1.2.2 Condenser-based HADF excitation Condenser-based high-angle excitation is perhaps the For transparent samples containing nanoparticles free most convenient dark-field excitation geometry, as com- from strongly scattering media, far-field excitation geom- mercial dark-field condensers are readily available and etries collecting forward-scattered light can yield high typically do not require customization [5, 42]. Moreover, signal-to-noise ratios. Two commonly employed excitation many commercial dark-field condensers have annular geometries for the collection of forward-scattered light are symmetric excitation resulting in better signal-to-noise normal incidence excitation and condenser-based HADF ratios than those found with typical unidirectional excitation. Both provide symmetric excitation either excitation geometries discussed above. The Irudayaraj lab normal to the sample, or in an annulus, and require less has implemented condenser-based HADF excitation spatial filtering of the excitation light compared to the (Figure 2F) for identifying two different cytosine modifi- objective-based HADF excitation described above. cations to DNA in cell nuclei [61]. In this work, silver nanoparticles labeled 5-carboxyl-cytosine (5caC) modifi- 3.1.2.1 Normal incidence excitation cations and gold nanoparticles labeled 5-methyl-cytosine Normal incidence far-field excitation does not require (5mC) DNA modifications. By spectrally monitoring single spatial filtering of the incident light. Instead, the excitation particles in the cell nuclei and simulating spectra, the must be spatially filtered from the collected scattered light authors identified wavelength ranges corresponding to after the sample. The Kukura group implemented normal scattering from single silver and gold particles, and incidence excitation and placed a circular beam block af- dimers of gold and silver particles (Figure 2G and H) [61]. ter the back-aperture of the objective lens to block the The three spectral windows were used to form three excitation light [79]. The gold nanoparticles were excited images corresponding to the distributions of 5caC, 5mC, with a white-light laser, and the forward-scattered light and co-localized 5caC–5mC modifications within the nu- was collected with a high-NA objective (Figure 2D). While cleus (Figure 2I). The overlaid image (Figure 2J) displays many dark-field excitation geometries still have some stray the relative amounts of all cytosine modifications in the light from the excitation, this approach of directly blocking cell nucleus [61]. 1628 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

3.2 Near-field techniques and polarizers in the excitation path [117, 118]. By modi- fying the incident angle, the penetration depth of the field Near-field excitation geometries rely on using confined can be modulated [96, 116, 119, 123]. Typical incidence light that does not propagate in free space, and therefore angles range from just above the critical angle (42° for an does not reach the objective. The interaction of nano- air-glass interface) all the way to 90° (parallel to the particles with near-field light leads to far-field scattered coverslip) [123]. Prism-coupled TIR excitation can also be light, which is collected by an objective. TIR excitation achieved at glass-water interfaces for measuring nano- techniques typically offer higher signal-to-noise ratios particles in spectroscopic cells, as long as the incident compared to far-field excitation techniques [76], and allow angle is adjusted to ensure TIR at the glass-water interface new polarization geometries [96]. In this section we also (see Section 5.1) [111, 117, 120, 124]. discuss SNOM, the only technique discussed in this review Control of the incident direction and polarization that can spatially super-resolve plasmon spectra [113]. allows visualization and distinction of bonding and antibonding plasmon modes [118]. When plasmonic nanoparticles are close to each other with an edge-to-edge 3.2.1 TIR excitation separation smaller than ∼2.5 times their diameter [125], their individual plasmons can hybridize to form lower en- In a typical TIR excitation microscopy setup, nanoparticles ergy bonding modes and higher energy antibonding are immobilized on a coverslip and light is coupled into the modes, analogous to molecular orbital theory [126]. The back of the coverslip above the critical angle, leading to TIR Gwo group excited gold octahedra at different interparticle at the interface where the nanoparticles sit. Furthermore, separations with polarized light (Figure 3A and B) [118]. the NA of the objective is not limited by the excitation angle When the light was polarized along the longitudinal axis of like in HADF illumination and high-NA objectives can the dimer, the authors observed a bonding mode, which therefore be used. However, the refractive index re- red-shifted as the distance between the particles was quirements for TIR can limit its application. For example, decreased. When the authors polarized light in the trans- some studies chose to cover nanoparticles in immersion oil verse direction, they observed only the antibonding mode to create a uniform refractive index [114], which is not (Figure 3C). As the interparticle distance decreased, the possible with TIR microscopy since TIR requires an inter- spectrum blue-shifted, confirming stronger antibonding face with a changing refractive index. character. The Garini group used the exponentially decaying 3.2.1.1 Prism-coupled TIR excitation nature of the evanescent field in prism-coupled TIR exci- Prism-coupled TIR excitation is one of the most common tation to determine the conformation of DNA by tethering ways to create an evanescent field for dark-field micro- gold nanospheres to a coverslip using DNA in a liquid cell scopy and allows flexibility in choosing illumination [116]. Movements of the DNA in the x- and y-directions source, direction, polarization, wavelength, and incident could be monitored directly by a charge-coupled device angle, which changes the penetration depth, but not the (CCD) camera while movements in the z-direction were direction of the wave vector [115–121]. While it is possible to calculated from the scattering intensity, which scales with achieve TIR by shining light at a coverslip directly [122], the excitation field intensity and hence the distance from prisms are often used to control the incident angle and the surface. The probability of finding the sphere close to avoid refractions [123]. This illumination technique gives the surface of the coverslip was disproportionally low, unidirectional excitation similar to unidirectional HADF which was explained by the entropic penalty resulting excitation outlined above. In fact, prism-coupled TIR from the reduced possible conformational states as the excitation may be thought of as a special case of unidi- DNA moved closer to the substrate. rectional HADF excitation where the illumination angle is Using a parabolic prism instead of a triangular prism high enough that it leads to TIR, instead of refraction. For allows decoupling the incident angle from the spot posi- white-light illumination, halogen lamps are often used due tion [119]. One of the disadvantages of prism-based TIR to their relatively flat spectra [111]. When high power den- excitation is that a change in incident angle also leads to a sities are required, a white-light laser can be used with change in the position of the beam spot. Diez and power densities on the order of tens of kW per cm−2, coworkers eliminated this limitation by using a prism with creating high enough scattering signals to capture spectra parabolic sides [119]. The parabolic shape always focused with 1 ms exposure [111, 117]. Wavelength-dependent and the light to the same spot in the center of the prism. By polarized excitation are possible by placing spectral filters walking the beam along the flat side of the prism, the A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1629

Figure 3: TIR excitation-based near-field geometries. (A) SEM images of gold octahedra dimers with tip-to-tip separation varying from 25–125 nm. (B) Prism-based TIR excitation geometry with the incident wave vector perpendicular to the dimer axis. (C) Single dimer scattering spectra revealing antibonding mode excitation with the resonance position varying as a function of gap size. (D) Objective-based TIR excitation setup. (E, F) Waterfall plot of spectra and extracted peak resonance over time as a gold nanosphere is trapped. (G, H) Waterfall plot of spectra and extracted peak resonance over time as a gold nanosphere leaves the gap. (I) Schematic of substrate-waveguide scattering microscopy (S-WSM). (J, K) Color images collected by objective- based HADF excitation and S-WSM. (L, M) Spectra of gold nanorods and gold nanospheres taken through HADF excitation and S-WSM. (A–C) Adapted with permission from Gwo and coworkers, Copyright The American Chemical Society, 2010 [118]. (D–H) Adapted with permission from Martin and coworkers, Copyright the American Chemical Society, 2010 [127]. (I–M) Adapted with permission from Cahoon and coworkers, Copyright the American Chemical Society, 2014 [80]. authors controlled the incident angle of the beam and fluorescence microscopy where the excitation beam can be penetration depth of the evanescent field. The beam spot spectrally filtered from the Stokes shifted emission [131], did not move when the angle of incidence was adjusted objective-based TIR microscopy and spectroscopy of [119]. Similar parabolic mirrors have also been used for TIR nanoparticles have been reported by multiple groups fluorescence microscopy [123]. [87, 112, 127, 132–137]. A particular advantage is that excitation and collection are carried out on the same side of 3.2.1.2 Condenser-based TIR excitation the sample, allowing free access to the nanoparticles by Condenser-based TIR excitation enables symmetric other tools, such as an atomic force microscopy (AFM) tip excitation, similar to condenser-based HADF excitation (see Section 5.2) [138, 139]. (see Section 3.1.2.2), but with near-field confined light. To Excitation and signal collection are performed with the achieve TIR, a central beam block is placed in the same high-NA objective. In a typical setup, a beam splitter condenser, and an annulus of light is focused on the back or dark-field mirror is placed in the filter wheel of the of the coverslip above the critical angle. This annular microscope to bring light through the objective to the configuration allows excitation from all directions, giving a sample, similar to objective-based HADF excitation high signal-to-noise ratio method to study the properties of illustrated in Figure 1B. Light is focused on the side of nanoparticles [76, 128, 129]. By placing a polarizer in the the back focal plane of a high-NA oil-immersion objective. excitation path, excitation with polarized light can be The light emerges on the side of the objective through the achieved, which unlike prism-coupled TIR excitation, immersion oil at a maximum angle, θm [123]: comes from all directions [96, 130]. An oil-immersion NA condenser is required to achieve TIR at the sample. θm = arcsin ( ) (13) n1

3.2.1.3 Objective-based TIR excitation When θm > θcrit, the light undergoes TIR at the interface In objective-based TIR excitation, light is focused through between the coverslip and the surrounding medium. The the objective at an angle sufficient to cause TIR. Although reﬂected beam is recollected by the same objective, with this excitation technique is more commonly used in the scattered light collected by the center of the objective 1630 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

lens. The reflected beam is then spatially filtered from the the sample, similar to objective-based TIR excitation but scattered light. When a dark-field mirror is used, the re- without the high-NA requirements or the need to spatially flected beam is reflected off the mirror [112]. When a beam filter the reflected excitation beam from the scattering splitter is used, a beam block or iris in the detection path signal of the nanoparticles. Waveguides are especially may be used to remove the reflected beam [132]. Comparing useful in sample configurations where direct access from Equation (13) to Equation (10), it becomes clear that TIR can both sides is blocked by other instrumentation such as only be achieved when the NA of the objective exceeds the an AFM tip over the sample and an objective below the refractive index of the less dense medium [123, 140]. sample [141]. The Martin group used objective-based TIR excitation As illustrated in Figure 1C, a coverslip can be used as a to monitor optical trapping of gold nanoparticles in the waveguide for dark-field imaging when a beam of light is gaps of dimers [127]. The authors lithographically depos- coupled into the side of the coverslip with a prism above ited gold nanorods with a 25 nm gap on a coverslip and the critical angle. Light is confined to the coverslip and mounted the sample on an inverted microscope, with water continues to travel through the coverslip by successive TIR covering the nanoparticles (Figure 3D). A 1.45 NA oil- events until it can escape through a second prism, creating immersion objective was used to achieve TIR excitation evanescent fields at the coverslip interface. The spacing with white light from a halogen lamp. Trapping was ach- between the TIR sites depends on the incident angle and ieved with a laser operating at 800 nm. The reflected light the coverslip thickness: a shallow incident angle or a thick was blocked spatially using a dark-field mirror. Addition- coverslip will lead to a larger distance traveled between the ally, scattered laser light was eliminated with a notch filter. top and the bottom of the coverslip, resulting in a larger The light scattered from the nanoparticles was directed to a spacing between the TIR sites. Calculations for coverslip spectrograph and spectra were acquired. When a colloidal thickness and incident angle can be found in an article by solution of 20 nm gold spheres was added, the plasmon Burghardt [142]. To improve light confinement, materials of red-shifted from 690 to 740 nm as nanoparticles were high refractive indices such as tantalum pentoxide [143] trapped by the 800 nm laser (Figure 3E and F, indicated by and silicon nitride [144], both with a refractive index of times T1 and T2). When the laser was turned off, the authors approximately 2.2, have been used. Instead of using a observed a gradual shift back to 690 nm as the 20 nm prism, Matsuda used the high viscosity and high refractive particles were no longer trapped and diffused away index of glycerin to couple white light into a coverslip from (Figure 3G and H, indicated by time T3). an optical fiber [145]. Objective-based TIR excitation has been reported to Waveguide illumination can also be achieved by show a higher signal-to-noise ratio and a lower back- coupling light from a focused halogen lamp into the side at ground than through-the-objective normal incidence dark- an angle that leads to total internal reflection (Figure 3I) field excitation. Noji and coworkers built a hybrid setup [80, 82]. A comparison between objective-based annular that can achieve objective-based TIR excitation and reflected HADF excitation and waveguide TIR excitation of through-the-objective normal incidence dark-field excita- gold nanorods is shown in Figure 3J and K. The Cahoon tion with minimal modifications [112]. Objective-based TIR group clearly demonstrated an improved signal-to-noise excitation was achieved with a 1.45 NA objective and a ratio for TIR excitation compared to HADF excitation in the 532 nm laser using a dark-field mirror to direct light images (Figure 3J and K) and the spectra (Figure 3L and M), through the side of the objective and to spatially filter the consistent with other TIR illumination techniques. The scattered light from the reflected beam. Normal incidence authors also confirmed that polarized excitation is possible excitation was achieved by replacing the dark-field mirror using their setup. with a dot mirror and steering the beam to the center of the The non-collimated nature of LED light can be used to dot mirror, which allowed the beam to emerge normal achieve very simple, yet effective waveguide illumination through the objective, while only collecting backscattered [146–150]. When the LED is placed next to a coverslip light at high angles. The authors note a higher signal-to- without any focusing lenses, some of the light from the LED noise ratio and a lower, more uniform background for naturally meets the coverslip at angles between 90° and the objective-based TIR excitation compared to normal inci- critical angle, resulting in some light being coupled into dence dark-field excitation. the coverslip. Cladding, such as cardboard or rubber, between the LED and the microscope objective may be used 3.2.1.4 Waveguide TIR excitation to block stray light that is not coupled into the coverslip When the substrate is used as a waveguide, it can be illu- from reaching the objective. While this method only uses minated from the side, allowing access to the other side of part of the LED light, the high contrast of TIR excitation A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1631

techniques still enables a good signal-to-noise ratio. [165, 167]. Aperture-SNOM setups have been employed to Furthermore, this method is easy to incorporate into image nanoparticles [169, 170] and spectroscopically existing microscopy setups. monitor reshaping of silver nanoparticles [171]. Fritzsche Optical fibers have also been used as waveguides for and coworkers used an aperture with a 80 nm diameter in dark-field illumination [83–91]. Optical fibers are essen- transmission mode of a microscope to detect DNA binding tially cylindrical waveguides: light is coupled into the fiber to individual nanoparticles [172]. from the end and prevented from escaping by TIR. The Homogeneous line-widths of single and coupled outside of the fiber is surrounded by an evanescent field, nanoparticles were determined using an aperture-SNOM exciting nanoparticles deposited on the optical fiber. The setup by Feldmann and coworkers [156]. Individual 40 nm fiber is mounted to the sample holder of the microscope to nanospheres embedded in a titania film were imaged by allow the objective to focus on the fiber [83–91]. A slight scanning a wavelength tunable laser coupled to an optical disadvantage is the cylindrical shape of the fiber, as the fiber with an opening of 80 nm (Figure 4A). The transmitted curvature can lead to loss of focus for particles along the light was directed to a photodiode and single-particle radial direction, making the effective measurable field of spectra were collected by recording transmitted light in- view long but narrow [83]. tensities at different excitation wavelengths (Figure 4B). The resulting spectra had homogenous line-widths and were visibly narrower than the broadened ensemble 3.2.2 SNOM-based excitation and detection spectrum of the film composite. The relative transmission was larger than unity, indicating a transmission enhance- SNOM breaks the diffraction limit of light by scanning ment, as expected from plasmonic nanoparticles with large across a sample with near-field confined light using a sub- scattering cross-sections. The collected spectrum of parti- – wavelength sized aperture or scattering tip [151 156]. The cle one also compared well to Mie theory of a 40 nm gold diffraction limit of light is a fundamental limitation in nanosphere (Figure 4B). However, as evident in the spectra how well an imaging system can distinguish two nearby of particles two and three, heterogeneity in line-width and fi λ λ emitters and is de ned as 2⋅NA, where is the excitation spectral shape were observed (Figure 4C). The authors wavelength. In 1928, E. H. Synge proposed that the attributed the differences in line-width to variations in diffraction limit could be overcome by illuminating the local refractive index due to imperfections in the titania sample through a subwavelength hole [157]. In 1984, D.W. and to variations in particle size. The spectrum shown in Pohl realized this technique by puncturing a hole in an Figure 4D was determined to be that of a coupled dimer and AFM tip and collecting the near-field signal scattered from agreed well with the simulated spectrum of two 20 nm gold a transparent sample [158, 159]. This work inspired future spheres with a 6 nm inter-particle separation. studies featuring a variety of excitation and collection In apertureless SNOM, also called scattering-type geometries, and SNOM is now a powerful technique for SNOM (s-SNOM), scattering is collected in the near-field near-field imaging [160–162]. at the surface of a specimen that is illuminated by far-field Near-field instruments are usually separated into two optics [173]. The sample is illuminated with a laser categories depending on the type of tip used: with or and subwavelength resolution is obtained by scanning a without aperture [163, 164]. Aperture-SNOM encompasses metallic tip while recording the field scattered by the tip configurations where a nanoaperture is used either as a [174]. The desired signal originating from the local inter- nanocollector [165] or as a local source of illumination action between the tip and the object of interest is filtered [152]. The resolution of these configurations is defined by from the background through vertical modulation of the tip the size of the aperture and the tip-sample separation [164]. position in tapping mode and demodulation of the optical When the aperture is used as an illumination source, light signal at the tip oscillation frequency Ω, or one of its from a laser is guided through an optical fiber and emerges higher harmonics [174, 175]. s-SNOM has been used from the subwavelength aperture at the fiber tip to locally more frequently for imaging [176–179] and spectroscopy illuminate the sample, such as organic molecules [151, 166] [155, 180] than aperture-SNOM. Hillenbrand, Aizpurua, and nanostructures [156, 167]. Light scattered, reflected, or and coworkers, used the fact that the substrate-tip distance transmitted by the sample is detected in the far-field with can be controlled to study weak and strong coupling be- an objective [156] or guided back through the aperture tween gold nanodisks and scattering tips [181]. [168]. In other geometries incident light is illuminated from The Suchowski group employed s-SNOM coupled with the side or at normal incidence through an objective and interferometric detection to collect super-resolved spectra the near-field signal is collected via the SNOM aperture from gold nanorods [113]. Half of the light from a 1632 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

supercontinuum laser was focused onto an AFM tip, while the other half was diverted to a Michelson interferometer (Figure 4E). The laser pulse at the AFM tip excited the nanoantennas, and the tip then interacted with the sample’s electric near-field. Spectral information was obtained from the interference between the extracted near-field signal and a reflected reference beam (Figure 4F and G). The spectral information was collected at each location as the tip was scanned across the sample, forming a spectral image (Figure 4F). The setup allowed a spatial resolution of 30 nm and spectral resolution of 50 cm−1 [113]. Recent reviews covering the application of SNOM for sensing and spectroscopy of nanoparticles have been published by Zhang and coworkers [182], and Fanchini and coworkers [183].

3.3 Interferometric scattering techniques

The detection of very small particles can be achieved by interferometric scattering (iSCAT) techniques [184, 185]. Interferometric scattering signals are obtained from the interference between scattered light from nanoparticles and light reflected from a glass coverslip, as reported by Sandoghdar and coworkers [184]. A beam splitter was used to excite nanoparticles deposited on a glass coverslip at normal incidence through an air-spaced objective (Figure 5A). Approximately 5% of the light was reflected by the air-glass interface [186] while the rest of the light excited the particle. The scattered light from the particle and the reflected light from the coverslip were collected by the same objective and directed towards a spectrograph. The reflected field, Er, was used as the reference and can be described by:

−iπ Er = r Ei e 2 (14)

where r is a measure of the reﬂectivity of the beam at

the air-coverslip interface and Ei is the electromagnetic ﬁeld of the incident light. The scattered ﬁeld Es, is then described by:

film. (C) Single-particle spectra of two particles and corresponding fits. (D) Spectrum of a coupled nanostructure and theoretical spectrum of two 20 nm spheres with a 6 nm separation. (E) s-SNOM Instrument diagram based on an AFM coupled with an interferometer for spectral detection. (F) Scattering intensity across a gold nanorod. Every pixel contains a spectrum. The arrow indicates the polarization of the excitation pulse. (G) Scattering spectra at Figure 4: Sub-diffraction limited scattering spectroscopy using SNOM. points a and b, and the far-field transmitted light. (A–D) Adapted (A) Schematic representation of aperture-SNOM. (B) Transmission with permission from Feldmann and coworkers, Copyright APS spectrum of a single 40 nm particle together with a Mie theory Publishing, 1998 [156]. (E–G) Adapted with permission from spectrum and the far-field transmission spectrum of the composite Suchowski and coworkers, Copyright OSA Publishing, 2019 [113]. A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1633

ϕ i s Es = Ei|s|e (15) Equation (16). For larger nanoparticles (diameter > 30 nm), the scattering signal dominates over the interference term. where ϕ is the phase of the scattered beam. s is a measure s However, as the particle size decreases, the interference of the polarizability of the particle and is proportional to term only shrinks linearly, rather than with the square of the volume of the particle. The signal measured at the the volume, allowing characterization of particles that are detector, I , is then described by: m otherwise below the detection limit for noninterferometric 2 2 2 2 Im = |Er + Es| = |Ei| (r +|s| − 2r|s|sinϕ) (16) scattering techniques. In 2004, the Sandoghdar group acquired spectra of The detected signal is thus the sum of the reflected beam 5 nm plasmonic nanoparticles using the interference (first term), the scattered light (second term), and the mechanism described above with a supercontinuum laser interference between the scattered and reflected electric beam in confocal scanning mode (Figure 5A–C) [184]. For fields (third term). While the pure scattering signal scales 60 nm particles, the signal was dominated by scattering with the volume squared, interference can now be seen to and the authors observed a bright signal on a dark back- scale linearly with volume, shown by the third term of ground, and spectra with a positive spectral intensity after background subtraction (Figure 5C). For 20, 10, and 5 nm particles, however, the signal was dominated by the interference signal and the particles appeared as dark spots with negative spectra (Figure 5B and C). A 31 nm particle exhibited intermediate behavior. Related interferometric techniques have been developed where the reference beam is created by different mechanisms to allow independent attenuation and improvements in signal-to-noise ratio. From Equation (16), the interference signal can be increased by increasing the intensity of the incident ﬁeld. However, the background (e.g., the reﬂected light) will then also be increased. Therefore, the signal-to-noise ratio is increased when the incident light intensity is increased while attenuating the reference beam [187, 188]. Particles as small as 2 nm [187, 188] have been detected and interferometric contrast techniques have found application for scattering labels in biological imaging [185, 189– 191]. The technique has been used to determine the size [192], dielectric function [193], and with polarized excitation, the orientation [194] of nanoparticles. Extensive reviews were published by Sandoghdar [195] and Kukura [196].

3.4 Polarized excitation dark-field microscopy

Polarized dark-field scattering spectroscopy is a powerful tool to optically characterize the orientation and geometric properties of anisotropic and chiral nanostructures. The Figure 5: Interferometric scattering techniques. scattering of anisotropic plasmonic nanostructures can (A) Schematic of the iSCAT setup. In this setup, a photonic crystal be modulated by adjusting the incident light’s linear fiber (PCF) was used as the light source and a spectrometer with a polarization, revealing single-particle orientation. Chiral photomultiplier tube (PMT) as the detector. (B) Confocal scan of a nanoparticles are those that are not superimposable with sample containing 5 nm gold nanospheres, with two particles highlighted (i and ii). (C) Normalized spectra of differently sized gold their mirror images. The handedness of the chirality can nanospheres. (A–C) Adapted with permission from Sandoghdar and be assigned using circular differential scattering (CDS), coworkers, Copyright the American Physical Society, 2004 [184]. which measures the differential intensity of light scattered 1634 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

under left- and right-handed circularly polarized light (LCP and RCP, respectively) [197–199]. The polarization state of the incident light can be easily tuned by placing a series of polarizers and waveplates in the excitation path. Linearly polarized light is achieved by adding a linear polarizer and circularly polarized light is generated by placing a linear polarizer and a quarter-wave plate in the excitation path.

3.4.1 Linearly polarized excitation

TheWilletsgroupobservedanunexpectedanisotropyin nanoprisms under a polarized excitation field generated by TIR [132]. The excitation light, with wavelengths selected by a liquid-crystal tunable ﬁlter (LCTF), was sent through a linear polarizer before reaching the sample. Figure 6A shows the representation of the coordinate system for the polarization-resolved setup with objective-based TIR excitation. The blue arrows indicate two incident polarization components: p and s. p-polarized excitation has an electric ﬁeld vector parallel to the plane of incidence, while that of s-polarized light is perpendicular to the plane of incidence. The single- particle scattering spectrum of a nanoprism acquired by scanning the incident wavelengths in 1 nm steps and recording the scattered intensity at each wavelength is showninFigure6B.Tocharacterizetheanisotropyofthe nanoprisms, Willets and coworkers adjusted the polarization of the incident light in 10° steps from 0 to 360° while holding the excitation wavelength constant. The recorded scattering intensities were analyzed in a polar plot and overlaid on scanning electron microscopy (SEM) images

scattering spectrum of a silver nanoprism. (C, D) Associated wavelength-dependent polarization anisotropy plots of the nanoprism acquired at 660 and 650 nm, respectively, overlaid on the correlated SEM image. All scale bars are 50 nm. (E) Schematic diagram of a CDS setup to characterize gold nanorod-BSA aggregates under circularly polarized light. (F) Distribution of subpopulations of gold nanorod-BSA aggregates sorted by type, with corresponding percentages of CDS-active and CDS-inactive nanostructures. Experimental single-particle CDS spectrum of a (G) chiral dimer and (H) achiral dimer with CDS originating from electromagnetic interactions with the chiral BSA. The pink envelope represents the experimental error in these CDS measurements. (I) Schematic of chemically synthesized chiral gold nanostructures. The panel on the right shows the averaged g-factor of 26 L-handed Figure 6: Polarized excitation dark-field scattering spectroscopy. (red), 26 achiral (black), and 26 D-handed nanoparticles (blue). TIR microscopy of plasmonic nanoprisms using evanescent wave (A–D) adapted with permission from Willets and coworkers, excitation. Copyright the American Chemical Society, 2012 [132]. (E–H) adapted (A) Schematic of the coordinate system for polarization-resolved with permission from Link and coworkers, Copyright Science, 2019 objective-based TIR excitation. The blue arrows indicate the p- and [63]. (I) Adapted with permission from Hentschel and coworkers, s-polarization components of the incident light. (B) Single-particle Copyright the American Chemical Society, 2019 [198]. A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1635

of the nanoprisms (Figure 6C and D). The nanoprisms suggesting the possibility of single-molecule chirality showed an unexpected wavelength-dependent anisot- sensing reported through the CDS of plasmonic particles. ropy. For example, the plasmon modes of the nanoprism Karst et al. used CDS to compare the optical activity at 660 and 650 nm were orthogonally polarized (Figure 6C of chiral helicoids and found that minute morphological and D). However, under condenser-based HADF excita- differences can manifest themselves in pronounced tion, the authors observed that the scattering from the chiroptical differences [198]. The left panel in Figure 6I nanoprisms was isotropic. Therefore, the origin of the shows the schematic of chemically synthesized chiral gold anisotropy was assigned to the unique polarization nanostructures. The diagram includes left-handed (L), properties of the evanescent wave. Due to the exponential achiral, and right-handed (D) gold nanostructures. The decay of the evanescent field, light that was initially authors measured the scattering, S, of the helicoids under p-polarized becomes elliptically polarized in the (x, z) LCP and RCP excitation, which they used to calculate the plane (Figure 6A) [93]. Meanwhile s-polarized light asymmetry, or g-factor: maintained its polarization with only a y-component. SRCP − SLCP Depending on the nanoprism’s orientation, the polariza- g = 2 SRCP + SLCP tion properties of the evanescent wave lifted the de- generacy of the orthogonal modes, clearly resolving them The right panel in Figure 6I shows the averaged g-factor of in the wavelength-dependent polar plots. 26 L-handed (red), 26 achiral (black), and 26 D-handed nanoparticles. The averaged single nanoparticle g-factor 3.4.2 Circularly polarized excitation agreed with that of the ensemble measurement. However, due to the heterogeneity of the nanoparticles prepared by The Landes and Link groups utilized CDS measurements of colloidal synthesis, a large variance in g-factors was observed single particles to uncover the origin of chirality from in the single-particle spectra. One of the advantages of single- bovine serum albumin (BSA)-aggregated gold nanorods particle CDS is the ability to distinguish minute differences in [63]. Chiral assemblies were prepared by aggregating gold nanoparticle morphologies that are otherwise inaccessible nanorods with low concentrations of BSA and spin-coating using other characterization methods [198]. the complexes onto transparent conductive substrates. In the CDS measurements, unpolarized white light from a tungsten-halogen lamp was sent through a linear polarizer 3.5 Polarization resolved detection of and a quarter-wave plate before entering a dark-field single-particle scattering condenser (Figure 6E). Then, scattering spectra from single aggregates were acquired under LCP and RCP excitation. While polarized excitation can significantly modulate the Calculating the differences between the two spectra, with scattering of single particles, analysis of the polarization appropriate corrections for linear artifacts [199], gave CDS state of light scattered by the nanoparticle can similarly spectra (Figure 6G and H). By repeating CDS measurements provide valuable information about the orientation and of ∼100 single aggregates, the authors obtained distribu- geometric properties of anisotropic particles [130, 199, tions of chiral and achiral subpopulations of gold nanorod- 201–203]. For example, by placing a polarizing beam BSA complexes (Figure 6F). The results revealed that single splitter in the detection path, the orientations of nanorods gold nanorods coated with BSA do not have measurable in biological media have been tracked [204]. The scat- CDS, while aggregates do, suggesting that aggregated tering intensity of a single dipole scatterer like a gold BSA-gold nanorods were the origin of the chirality in an nanorod varies with a cos2Ψ dependence where Ψ is the ensemble sample. Single-particle CDS measurements were angle between the long axis of the nanorod and the angle then correlated with transmission electron microscopy to of the linear polarization [204]. In a typical experiment, understand the origin of the chirality from each individual nanoparticles are immobilized on a substrate and a linear aggregate (Figure 6G and H). By combining CDS mea- polarizer is rotated in the detection path. A single-particle surements with correlated tomographic reconstruction and scattering spectrumorimageiscollected at eachangle.By electromagnetic simulations, structural chirality was fitting the resulting intensity of the LSPR to a cos2Ψ determined to predominately give rise to CDS in aggregates dependence, it is possible to determine the nanorod’s (Figure 6G). However, CDS was also observed from aggre- orientation using only optical techniques. Polarization- gates found to be structurally achiral (Figure 6H). The resolved detection of single-particle scattering can often origin of this signal was attributed to chiral dipole-dipole be easier to implement without the polarization distor- coupling [200] between the gold nanorods and the BSA, tions induced by the high-NA optics used in dark-field 1636 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

excitation [199, 205]. However, the polarization information captured in the image plane is a two-dimensional projection of that contained in the scattered light, limiting the ability to track out-of-plane polarization phenomena, such as tilt angle [206].

3.5.1 Linear polarization detection

Käll and coworkers have implemented an interesting variant of orientation sensing by placing a linear polarizer with a fixed angle in the detection path but utilizing a circularly polarized optical trap to rotate a single nanorod (Figure 7A and B) [207]. The transfer of angular momentum from the circularly polarized light caused the nanorod to rotate. By characterizing the intensity of the scattered light that passed through the fixed linear polarizer with an avalanche photodiode (APD) and autocorrelator, the authors could characterize incredibly fast rotation rates of the nanorod of up to 42 kHz (∼2.5 million revolutions per minute) depending on the laser power of the trapping beam (Figure 7C). A significant portion of the optical torque on the nanorod was attributed to resonant scattering of the circularly polarized light from the optical trap (Figure 7D).

3.5.2 Stokes polarimetry

The Stokes parameters form a vector, S, that contains the complete polarization state of light:

S0 ⎜⎛ ⎟⎞ ⎜ S1 ⎟ S = ⎝⎜ ⎠⎟ S2 S3

S0 describes the full intensity of the polarized light, S1 describes if the linear polarization components are more

aligned at 0 or 90°, S2 is similar to S1, but for 45 and 135°, Figure 7: Polarization characterization of light scattered by and S3 describes any right- or left-handed circular polari- nanoantennas. zation [208]. Stokes polarimetry can easily be implemented (A) Experimental setup to achieve and measure the spinning of gold nanorods in water. A gold nanorod is trapped by 830 nm circularly polarized laser excitation, causing it to rotate. The gold nanorod is also illuminated with white light for linear polarization other near the transverse and longitudinal plasmon modes. characterization, detected by either a CMOS camera or an APD and (E) Schematic of dark-field Mueller matrix spectroscopy. PSG: autocorrelator for rotation rate assessment. LCTF-based spectral polarization state generator, PSA: polarization state analyzer, characterization can be added to the white light source. (B) P1,2: linear polarizers 1 and 2, QWP1,2: quarter-wave plates 1 and Experimental cartoon of the rotating sample. (C) Dependence of 2, s: spectrometer. (F) Single-particle spectrum of a nanorod, NR, gold nanorod rotation frequency on laser power for varying with unpolarized detection and analysis. (G) The 4 × 4 nanorod lengths with nearly constant diameters. The lengths and spectroscopic Mueller matrices of a single nanorod are acquired diameters for samples (1–5) are as follows: (1) 134 ± 10, 64 ± 5nm through sixteen calibration-optimized PSG and PSA settings. (A–D) (2) 147 ± 10, 65 ± 5 nm (3) 157 ± 10, 65 ± 5 nm (4) 169 ± 10, 66 ± 5nm Adapted with permission from Käll and coworkers, Copyright the (5) 173 ± 8, 65 ± 5 nm. (D) Calculated ratios of absorption and American Chemical Society, 2015 [207]. (E–G) Adapted with scattering cross-sections (σabs and σsca) and absorption and permission from Ghosh and coworkers, Copyright Nature scattering torques (Mabs and Msca).Thetworatiosareclosetoeach Publishing Group, 2016 [205]. A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1637 with common microscopy techniques with the addition of a 4 Spectral characterization of quarter-wave plate and linear polarizer in the detection path. Stokes polarimetry is often combined with Fourier- single particles plane imaging, also called k-space imaging, which captures the back-focal plane of the objective lens. Fourier- In Section 3, we discussed different approaches of plane imaging characterizes the momentum-space of the detecting scattering from single nanoparticles. While emission or scattering of an analyte [209]. Commonly many applications simply seek to localize nanoparticles implemented in photoluminescence studies, k-space as labels, more information can be gained from the polarimetry simultaneously captures the polarization state plasmon spectrum. LSPR position for example can give and direction of the emission of dyes [210–212]. The valuable information on the shape, size, and local envi- Koenderink group has utilized k-space polarimetry to ronment of the nanoparticle, while the homogeneous line- characterize the polarization state of light scattered by a width of a single nanoparticle can be used to directly lithographically-patterned plasmonic antenna shaped like study the dephasing of the plasmon and thus different a bullseye as a function of the incident light polarization damping mechanisms including charge and energy [201]. Using this technique, the researchers observed that transfer [215]. In this section, we will describe several plasmonic bullseyes can convert linearly polarized light to techniques that have been developed to spectrally resolve circularly polarized light, and convert from right-to left- thescatteringofsingleparticles,includingpush-broom handed polarizations [201]. scanning hyperspectral imaging, wavelength scanning spectral imaging, multiple-channel spectral imaging, snapshot hyperspectral imaging, and interferometric 3.5.3 Mueller matrix spectroscopy detection. Broadly, single-particle scattering spectra can be acquired by dispersing the wavelengths of scattered TheGhoshgrouphasimplementedanadvancedpolari- light with a diffraction grating, or by filtering the incident zation analysis approach by performing Mueller matrix light, and recording the intensity of the scattered light as a spectroscopy in a dark-field microscope (Figure 7E) [205]. function of the wavelength. Table 1 provides an overview The Mueller matrix is the 4 × 4 matrix that transforms the of the key techniques for spectral characterization along initial light polarization state (starting Stokes parameters) with their typical time and spectral resolutions. to the final polarization state [213]. Mueller matrix spectroscopy of a single nanoparticle then quantitatively describes how a particle interacts with polarized light and alters its polarization state. The authors carefully cali- 4.1 Scanning single-particle spectral brated the polarization state generation and analysis imaging methods values required to correct for the high-NA imaging system using polarizers as standards. The resulting Mueller A dark-field microscope combined with a spectrometer or a matrix captured only the polarization information of wavelength-tunable excitation source is a powerful tool for the scatterer. With this technique, the authors recorded characterizing the scattering spectra of single nano- the Mueller matrix of a single gold nanorod with signifi- particles. Particle-by-particle acquisition approaches have cant intensity in the transverse mode (Figure 7F). The been successfully implemented [224]. In these techniques, Mueller matrix demonstrated linear diattenuation an intensity integrated image is typically first acquired to (Figure 7E – blue elements) resulting from the differential locate individual particles, followed by centering a single excitation of both the transverse and longitudinal modes. particle in the microscope field of view. A slit or pinhole is Linear retardance (Figure 7E – red elements) described the placed in the detection path to spatially filter the scattering phase retardation between the orthogonal plasmon of the particle of interest from that of neighboring particles. modes. Finally, the black elements of the Mueller matrix The scattered light is then directed to a spectrograph, were evidence of depolarization due to excitation of the which disperses the collected light with spectral resolution nanorod with a focused beam, averaging over the planes defined by the chosen grating. The dispersed light is then of azimuthal scattering as well as over the collection angle imaged onto a CCD camera or other array detector to of the objective lens [205]. The Ghosh group has since measure the intensity of the scattering as a function of applied this successful proof-of-principle demonstration wavelength. Once the spectrum of one particle is collected, to oligomers with Fano resonances [214]. the next particle is centered in the field of view and its 1638 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

Table : Summary of key single-particle spectral characterization techniques with typical time and spectral resolution.

Technique Method of wavelength Widefield Typical time resolution Typical spectral separation acquisition for complete image resolution possible? acquisition

Push-broom scanning Diffraction grating in detection No ∼ min [] ≤ nm [] hyperspectral path Wavelength scanning Tunable ﬁlter in excitation path Yes < min [, ] ≤ nm [, ] Multiple channel single-particle Fixed ﬁlters in detection or Yes . []–. s[] Select wavelengths spectral imaging excitation path only Snapshot hyperspectral imaging Diffraction grating in detection Yes  ms [] ≤ nm [] path Interferometric detection Interferometer in detection path Yes  s[]  nm [, ]

Note that the time resolution for push-broom hyperspectral imaging is partially limited by the size of the scanning area, as an image must be acquired at each step of the scan. The time resolution given here is for the acquisition of a  ×  μm image with × magnification []. For widefield acquisition techniques, the entire field of view is captured, allowing larger image sizes such as  ×  μm with × magnification for interferometric detection []. spectrum is acquired in the same fashion. This process is of interest. The Landes and Link groups achieved hyper- repeated until the spectra of all particles of interest are spectral detection by scanning a spectrograph and CCD sequentially acquired. Particle-by-particle accumulation camera across the field of view to acquire 3D data cubes techniques provide high spectral resolution but, unless (Figure 8B) [76, 197, 199, 217, 229]. This technique has the fully automated [225], can be prohibitively time consuming benefit of being less expensive to implement than stage- when acquiring the spectra of a high number of particles. scanning, as the scanning step can be larger by the amount As a result, several spectral acquisition techniques based of objective magnification, and thus can be accomplished on automated position or wavelength scanning have by a simple linear actuator. Alternatively, scanning the been implemented to efficiently achieve high spectral stage often requires the finer step sizes of a piezoelectric resolution of many individual particles in parallel instead actuator. While a piezoelectric stage has reduced hystere- of consecutively. sis relative to a linear actuator, the precision of a linear actuator is often more than sufficient to perform hyper- 4.1.1 Push-broom scanning hyperspectral imaging spectral imaging. Byers et al. were able to utilize this hyperspectral technique to quantify changes in LSPR One popular implementation of push-broom single- maxima across tens of particles as a potential was applied particle spectral acquisition is hyperspectral imaging, in an electrochemical cell (Figure 8C) [217]. which collects a three-dimensional (3D) data cube con- A push-broom method implemented by Ringe and taining the x- and y-positions of the nanoparticle, as well as coworkers scanned a piezoelectric stage in 0.8 µm steps to the scattering spectra at each (x, y) location [61, 226–228]. In acquire a new 2D matrix with each step (Figure 8D–F) scanning hyperspectral dark-field imaging, either the forming the 3D data cube (Figure 8G) [216]. The hyper- sample stage [216] or spectrometer [217] is scanned across spectral image size was determined by the magnification, the field of view in the y-direction. The scattered light from theheightoftheslit,andthescanningrangeofthestage. the nanoparticles is spatially filtered by a slit aperture The diffraction-limited spatial resolution was determined placed before the spectrograph, and the dispersed light by the number of pixels in the CCD camera and the from the slit is imaged onto a two-dimensional (2D) CCD scanning step size of the stage, while the spectral reso- camera (Figure 8A). The images acquired at each step are lution was determined by the grooves/mm of the chosen 2D matrices containing the x-location of the particle along diffraction grating. The acquired data cube was integrated the slit and the intensity as a function of wavelength along the wavelength axis to produce a spectral image recorded along the z-axis. Scanning the stage or spec- (Figure 8H) from which single-particle spectra were trometer over many steps in the y-direction allows for the extracted (Figure 8K). The technique by Ringe and co- collection of a three-dimensional (3D) data cube from workers also had the flexibility to characterize trans- which spectra can be extracted by summing the spectral mission and reflectance by altering the excitation data from a square region of pixels containing the particle geometry (Figure 8I and J) [216]. A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1639

4.1.2 Wavelength scanning spectral acquisition

Another scanning method for single-particle spectral imaging is scanning the wavelength of the excitation with a tunable filter, such as an acousto-optical tunable filter (AOTF) or LCTF while holding the stage and camera location constant [132]. Typically, wavelength-scanning approaches are implemented in a widefield fashion, in which a 2D array detector captures images of the field of view at each wavelength step, building up a 3D data cube similar to the push-broom hyperspectral techniques discussed above [218, 219, 230, 231]. While push-broom methods often have higher spectral resolution, wavelength scanning approaches can have faster acquisition times [218]. The Mulvaney group successfully implemented a wavelength-scanning approach to achieve a higher signal-to-noise ratio than lamp-based push-broom hyperspectral techniques [107]. The particles were excited by a spectral slice selected from a white-light laser with an AOTF. The excitation wavelength was scanned in 1 nm steps at a rate of 0.5 nm/s (Figure 8L). The scattered intensity was spatially filtered by a 50 µm pinhole and recorded by a photomultiplier tube (PMT) in single- photon counting mode with the resulting spectrum shown in Figure 8M. The combination of high-intensity laser excitation and PMT detection offered high signal-to-noise ratios. Moreover, in this approach, the particles are only briefly excited on resonance with intense illumination, which is useful for reducing particle deformation due

(E) Similar to the schematic in (A), light scattered from the nanoparticles is collected by the objective, spatially ﬁltered by a slit, then dispersed by a diffraction grating, and imaged onto a CCD camera. Moving the stage enables collection of multiple slices across the ﬁeld of view. (F) Directly acquired 2D matrices of

intensity as a function of x-position along the slit (Nx)and wavelength (Nλ). Multiple matrices are acquired by scanning the stage in the y-direction. (G) Representation of the 3D hyperspectral data cube resulting from stacking 2D matrices of spectral and x-position information. (H) Summing along the Nλ axis gives a spectrally integrated image. Extracting the spectral information from a region of interest (ROI) yields the representative single- particle dark-field scattering spectra shown in (K). (I, J) Absorbance and reflectance spectra can similarly be acquired through a change Figure 8: Scanning hyperspectral dark-field spectroscopy. in excitation geometry. (L) Single-particle spectra are acquired by (A) Schematic of an optically transparent electrochemical cell successive monochromatic illumination steps using a white-light allowing for dark-field spectroscopic monitoring of chemical laser and an AOTF with a spectrum shown in (M). Excitation is reactions and charge density tuning on single 50 nm gold spheres delivered to the sample by a multimode (MM) fiber, a mirror (M1) as a potential is applied. ITO serves as the working electrode (WE) and a dark-field condenser (DFC). (A–C) Adapted with permission while silver wires serve as the reference and counter electrodes (RE from Landes and coworkers, Copyright The American Chemical and CE). (B) Scanning the spectrograph across the field of view Society, 2014 [217]. (D–K) Adapted with permission from Ringe and allows construction of a hyperspectral data cube. (C) Single- coworkers, Copyright Springer, 2018 [216]. (L, M) Adapted with particle scattering spectra are extracted from the data cube. permission from Mulvaney and coworkers, Copyright The American (D) Condenser-based HADF excitation of gold nanoparticles. Chemical Society, 2018 [107]. 1640 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

to melting. Finally, holding the excitation wavelength 4.2.2 Multi-channel spectral imaging with digital constant allowed for direct monitoring of changes in single-lens reflex (SLR) cameras particle intensity with a 1 ms bin time [107]. Based on principles similar to ratiometric imaging, a digital SLR camera can function as a three-channel spectrometer 4.2 Multiple channel single-particle by measuring the ratio between the red (R), green (G), and spectral imaging blue (B) intensities of the image [237–240]. Odom and coworkers have implemented such a design with a digital SLR Multiple channel spectral imaging techniques enable camera in a standard dark-field microscope equipped with characterization of all particles within the field of view a spectrometer for calibration (Figure 9E) [236]. Optional simultaneously [36, 221, 232–234]. Push-broom hyper- filters and polarizers can be added for even more sensitive spectral imaging methods offer high spectral resolution spectral separation of different particles. This approach [216, 217], but spectral acquisition of ∼20 particles can often was applied to gold nanopyramids on a substrate orien- take 1–2 min. Multiple channel spectral imaging tech- tated either up or down, thereby impacting their spectral niques utilize filters or digital color analysis to acquire characteristics. Specifically, the tip-down orientation was spectral information in a single exposure without a slit or found to scatter with greater intensity than the tip-up pinhole with high time-resolution [235]. orientation in the near-infrared (NIR) wavelength range (Figure 9F). To rapidly differentiate nanoparticles in these 4.2.1 Ratiometric imaging two orientations, the authors inserted an 850 nm bandpass filter before a digital camera optimized for NIR. Then, the Ratiometric imaging records the ratio of the scattering in- ratio of R/G intensity values from the digital camera image tensity at two well-separated wavelengths for the entire was plotted (Figure 9H) demonstrating a clear threshold field of view [235]. Therefore, this technique is ideally between tip-up and tip-down orientations with R/G = 3.2. suited for quickly characterizing significant spectral shifts, Note that intensity was recorded in the G channel at NIR such as those that occur upon coupling to nearby nano- wavelengths, potentially due to NIR transmission of the particles, especially in a biological context for studying Bayer filter used in digital color cameras [241]. This dual-labeled DNA [234] and protein interactions [220]. threshold was applied to a widefield image and the Ratiometric imaging (Figure 9A) has been successfully orientations were unambiguously assigned (Figure 9G). implemented by the Reinhard group to perform plasmon Confirmation via SEM verified the assignment with 100% coupling microscopy [220]. Plasmon coupling microscopy accuracy and the authors applied the same principles to relies on the principle that the LSPR red-shifts as one or distinguish between particles with different compositions, more neighboring particles increase in proximity resulting and with the insertion of a polarizer, were able to identify in plasmon coupling (Figure 9C) [125]. In ratiometric im- the orientation of the long axis of a nanostar. Moreover, the aging, the particles are excited with a white-light source, use of a digital SLR camera as a three-channel spectrometer often from a dark-field condenser, and the scattering from decreases the cost of the detection path by more than an the particles is spectrally separated into long and short order of magnitude, increasing the ease of implementation wavelengths by a beam splitter. The resulting light is then of dark-field scattering spectral characterization. filtered and imaged onto two separate halves of a CCD camera. One imaging channel is centered around the LSPR of a single particle, and one is centered around the LSPR of 4.3 Snapshot hyperspectral imaging a closely spaced dimer. Reinhard and coworkers successfully employed this technique to track the formation of gold Snapshot hyperspectral imaging techniques combine the nanoparticle dimers in HeLa cells below the diffraction high spectral resolution of push-broom hyperspectral limit (Figure 9B) [220]. The ratio of the intensities of the two techniques with the high temporal resolution of ratiometric channels allowed for assigning the time at which two techniques by taking entire large-area spectra in just one particles coalesced, even within a diffraction-limited point snapshot [111]. The goal of snapshot hyperspectral imaging spread function (Figure 9D). For example, upon dimer is to capture entire spectra of multiple nanoparticles in one formation, the intensity of the red channel (I580) increased shot, to achieve the temporal resolution necessary to at the same time as the intensity of the green channel (I530) monitor dynamic chemical or electrochemical changes of decreased, signifying the particles were within the plas- nanoparticles [111, 242], and to quickly monitor changes in mon coupling regime. multiple nanoparticles’ surroundings for biosensing [243]. A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1641

Figure 10A shows an instrument diagram of the snapshot hyperspectral imaging approach taken by Kirchner and Smith et al. [111]. Scattered light from the tube lens of the microscope was directed through a 30:70 beam splitter. 30% of the light was refocused onto a complementary metal-oxide-semiconductor (CMOS) detector (labeled CMOS1) to find the zeroth-order scattering (i.e., the position) of the particle (Figure 10B). The remaining 70% was directed to a 300 groves/mm transmission grating. The first order diffracted light was refocused onto a second CMOS detector (labeled CMOS2) to give spectral information (Figure 10C). A slit was placed in the first image plane at the focal point of the tube lens (Figure 10A) and standard lamps were used to calibrate the distance between the position given by CMOS1 and each pixel of the first order diffraction recorded on CMOS2. The calibration was then used to convert the difference in pixel position between the zeroth order and each pixel in the first order diffraction into a wavelength, with a spectral resolution of 0.21 nm per pixel. The slit was removed for measuring nanoparticles. By using a white-light laser, enough photons were scattered off the particles to give a temporal resolution of 1 ms, limited by the frame rate of the camera. Figure 10D shows a representative spectrum of a single gold nanorod. When particles were close together, their spectra overlapped, and appeared as two merged Lorentzian curves (Figure 10F). To avoid overlapping spectra, a relatively low particle density had to be used. The major disadvantage of snapshot hyperspectral imaging compared to push-broom hyperspectral imaging, other than the lower particle density, is a

(A) Schematic of ratiometric dark-field imaging. Scattered light from gold nanoparticles is separated using a dichroic longpass filter, then passed through bandpass (BP) filters at 580 and 530 nm. (B) Images of gold particle-labeled HeLa cells captured on both channels. (C) Distance change between the particles can be monitored ratiometrically. At large separations, the particles scatter near 530 nm; at short distances, the particles scatter near 580 nm.

(D) Intensities of the two channels, I580 nm (red) and I530 nm (green), and total intensity (black) recorded during the collapse of a pair of DNA tethered particles. The arrow marks the collapse. (E) Schematic of digital camera-based three-channel spectral imaging with a spectrometer included for calibration. (F) Dark-field spectra of gold nanopyramids with tip-up (U, red squares) and tip-down (D, black triangles) orientations. (G) Image of nanopyramids acquired with a digital SLR camera with the built-in NIR/ultraviolet filter removed, which the authors referred to as an NIRcam, with an additional 850 nm bandpass (BP) filter. (H) Histogram of R/G intensity ratios of nanopyramids allows clear widefield assignment of U- and D- oriented nanopyramids in G. (A–D) Adapted with permission from Reinhard and coworkers, Copyright the American Chemical Society, 2008 [220]. (E–H) Adapted with permission from Odom and Figure 9: Multiple channel single-particle spectral imaging. coworkers, Copyright The American Chemical Society, 2011 [236]. 1642 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

higher background as the absence of a slit allows light from the entire field of view to enter the detection path. Snapshot hyperspectral imaging with a single detector can be achieved by using a diffraction grating with fewer grooves per millimeter, dispersing light less and capturing the zeroth and first orders on the same detector chip, as implemented by the Yeung and Gai groups [244, 245]. The setup reported by Kirchner and Smith et al. was also used in a one camera setup by Al-Zubeidi and Hoener et al. by replacing the 300 groves/mm grating with a 75 grooves/mm grating and capturing both the zeroth and first order diffracted light with CMOS2 [117]. This setup allowed a higher number of particles to be measured simultaneously by dispersing the light less and hence allowing a higher particle density. Furthermore, a higher signal-to-noise ratio was achieved since the lower dispersion resulted in more photons per pixel, while sacrificing some spectral resolution (1 nm per pixel). Su et al. developed a snapshot hyperspectral imaging setup by placing a diffraction grating between the sample and the objective and directing the light to a camera [243]. As the light passed through the diffraction grating, it was split into its components and the objective then captured both the zeroth and first order scattering of the particles. The captured image consisted of the particle with its spectrum next to it. Figure 10G shows the intensity distribution for a few rows of pixels. The big spike on the right corresponds to the zeroth order scattering of the particle. The wider, less intense peak on the left gives the spectrum. The authors used this setup to monitor refractive index changes surrounding 50 nm gold nanoparticles with 1 s time resolution. When the nanoparticles were functionalized with biotin, binding of streptavidin was sensed in solution by tracking LSPR shifts.

2 = focusing lens 2, and f1 and f2 are the focal lengths of the tube lens and FL 2, respectively. α and β are the angles between the transmission grating and the incident and transmitted light, respectively. (B, C) Images captured with the two detectors. (D) Spectrum of a single particle. (E) Spectrum of a particle whose spectrum overlaps with that of another particle. (F, G) Similar results were achieved by Su and coworkers by placing a diffraction grating between the sample and the objective. (H, I) The spectrum of a single nanorod and the changes in resonance energy for many nanorods tracked over time using the fast single-particle spectroscopy method developed by the Sönnichsen group. (A–E) Adapted with permission from Link and coworkers, Copyright the American Chemical Society, 2018 [111]. (F, G) Adapted with Figure 10: Snapshot hyperspectral imaging techniques. permission from Su and coworkers, Copyright the Royal Society of (A) Snapshot hyperspectral setup employing two CMOS detectors. Chemistry, 2011 [243]. (H, I) Adapted with permission from M = microscope, FL 1 = focusing lens 1 (f3 is twice the focal length Sönnichsen and coworkers, Copyright the American Chemical of FL 1), CL = collimating lens, TG = transmission grating, FL Society 2007 [242]. A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1643

A dynamic hyperspectral imaging technique that Sung and coworkers used a home-built Michelson combines the time resolved nature of snapshot techniques interferometer to measure single-particle spectra (Figure 11A) with the high signal-to-noise ratio of slit or pinhole-based [222]. Light from a dark-field microscope was sent through a techniques is fast single-particle spectroscopy, developed beam splitter. The reflected beam was sent to a stationary by the Sönnichsen group [242]. The setup is based on a mirrorM2,whilethetransmittedbeamwassenttoamirror commercial spectrograph but replaces the slit at the image M1 mounted on a piezoelectric stage. Reflected beams plane with a liquid-crystal device (LCD) shutter to allow from M2 and M1 were recombined by the beam splitter and light transmission from only a selected subset of particles, sent to a CCD detector. By carefully controlling the posi- essentially forming an array of switchable pinholes. In a tion of M1 and hence the beam path of the transmitted typical measurement, all LCD pixels were first made light, interferograms were collected, allowing the spec- transparent, the spectrograph grating was moved to the trum at any point of the field of view to be reconstructed zeroth order, and an image of all particles was taken. (Figure 11B, green). A hyperspectral data cube of a Particles were then selected and grouped into horizontally 659 × 496 μm fieldofviewwascollectedin5s,asignifi- nonoverlapping groups of up to 20 particles. All particles in cant improvement over push-broom hyperspectral imag- one group were imaged at the same time, while groups ing. The authors compared their interferometer with a were imaged consecutively. The response time of the LCD commercial spectrograph (Figure11B, red line) for 40 nm shutter was 40 ms, allowing fast scanning across the gold spheres and found a significant improvement in groups. Areas close to the particles and white-light scat- signal-to-noise ratio for their system compared to the terers (dust or large silica beads) were used for correction of commercial spectrograph. Stranik and coworkers used a the background and the incident light spectrum, respec- similar setup to measure single-particle spectra of gold tively. This setup was employed to monitor the growth of nanospheres in water and glycerol, demonstrating the use gold nanoparticles in real time. Gold nanorods were placed of interferometer-based detection for refractive index in a liquid cell containing a growth solution that consisted sensing (Figure 11C). of cetrimonium bromide, silver ions, gold ions, and ascorbic acid. Figure 10H shows the evolution of the LSPR of a gold nanorod, illustrating that growth occurred pref- 5 Specialized sample erentially on the sides, leading to a blue-shift in the LSPR peak as particles became more spherical. The peak LSPR configurations for 20 nanoparticles measured simultaneously in 30 s intervals demonstrates that nearly all particles became Much of the research and applications involving nano- more spherical as the reaction proceeded (Figure 10I). particlesfocusesontheuseofnanoparticlesassensors, scattering labels, and catalysts [120, 247–249]. These measurements typically require the particles to be sub- 4.4 Interferometric detection merged in solutions or dispersed on electrodes. Addi- tionally, combining single-particle dark-field or SNOM Another way to achieve hyperspectral imaging is by using spectroscopy with tip-based scanning techniques pro- an interferometer [222, 223, 246]. Similar to snapshot vides complementary information on the reactivity and hyperspectral imaging techniques, interferometric detec- sensing-relevant adsorption properties of nanoparticles tion aims to achieve spectral imaging of the entire field of [113,139,250,251].Tomeasureparticlesinsolutions, view at a higher temporal resolution compared to scanning liquid filled spectroscopic cells are commonly used techniques [222]. Typically, the collected light from the [111, 120], which can be used on most microscopy setups objective lens is separated by a beam splitter with portions with small modifications of the excitation angle. Recently, sent to both a permanent mirror and a movable mirror [222]. some groups have opted to use water-immersion objec- The light is then recombined at the beam splitter and tives for measuring in liquid to increase the sensitivity of collected by a detector. Since both beams originate from their setups [252–254]. Liquid cells and water-immersion the same source, they can constructively or destructively setups designed for electrochemical measurements are interfere, depending on the path difference. By changing particularly relevant for catalyst research, due to the the position of the movable mirror, an interferogram of the growing interest in using plasmonic nanoparticles as field of view is formed by measuring the interferometric antennas for photo-electrochemical reactions [111, 117, signal at each pixel. A Fourier transform generates a 252, 253, 255]. Correlated scattering spectroscopy and AFM spectrum at each pixel position [222]. or scanning electrochemical cell microscopy (SECCM) is 1644 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

possible by combining an inverted microscope with a scanning tip [113, 139, 250].

5.1 Measurements in liquid

To study reactions on plasmonic nanoparticles or use plasmonic nanoparticles as scattering labels, spectroscopic cells are often used [124, 256–260]. Spectroscopic cells sandwich the nanoparticles between two coverslips with a spacer between them and can be filled with liquids. Figure 12A shows an example of a flow cell on an upright microscope used by the Sönnichsen group [215]. Nano- particles were immobilized on a coverslip with six holes in it, three on each side forming three flow channels. Parafilm with three laser-cut flow channels was placed over the coverslip, with the flow channels matching the holes. A second coverslip was added on top and the parafilm was placed on a hot plate to melt the parafilm and create airtight channels with the coverslip. Tubing was added to the holes to flow liquid through the cell [215]. Figure 12B shows an example of a spectrum recorded using this cell. This setup has been used to investigate the mechanism of chemical interface damping in gold nanorods [215] and to monitor conformational dynamics of single proteins [261]. In the latter work, heat shock protein 90 was studied, which is known to exist in open and closed configurations. Gold nanospheres were attached to the ends of the protein. As the protein changed between the open and closed configurations, the coupling between the gold nanospheres increased and decreased, allowing the authors to monitor the conformation by tracking the scattering intensity over time. The orientation of the substrate determines which excitation and collection geometries can be used. If the substrate forms the side of the spectroscopic cell that is positioned further away from the collecting objective, as illustrated in Figure 12A, both HADF excitation and prism- coupled TIR excitation are possible. The option to use prism-coupled TIR excitation opens up the opportunity to use a white-light laser for excitation, giving high enough Figure 11: Interferometric detection. intensities to collect spectra with millisecond time resolu- (A) Instrument diagram of a hyperspectral imaging system based tion [111, 117]. The disadvantage of this configuration is that on a Michelson interferometer, which uses a beamsplitter (BS) and the scattered light needs to penetrate the entire water layer a movable (M1) and a stationary (M2) mirror. M1 was mounted to a piezoelectric translator (PZT) (B) Spectra of the same gold nanosphere acquired using a commercial spectrograph and the spheres is shown in black. (A, B) Adapted with permission from home-built interferometer compared to the spectrum simulated by Sung and coworkers, Copyright OSA Publishing, 2011 [222]. Mie theory. (C) Spectra of three gold nanospheres (red, green, (C) Adapted with permission from Stranik and coworkers, blue) in water (solid) and glycerol (dashed). The average of all Copyright Elsevier, 2016 [223]. A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1645

and the opposing coverslip, undergoing multiple refractive index changes and possible scattering events, and requiring an objective with a long working distance to image through the bottom coverslip of the spectroscopic cell. If the particles are placed on the other coverslip close to the objective (i.e., the top coverslip in the upright configuration in Figure 12A), light only needs to travel through the substrate coverslip, reducing the number of interfaces and the working distance of the objective, making it even possible to employ an oil-immersion objective. However, the excitation light now has to travel through the medium before reaching the sample, making TIR excitation impossible. Typically, HADF excitation is used in these cells [76]. Spectroscopic cells can be used to optically monitor electrochemical reactions by using conductive substrates [76, 124, 229, 258]. The nanoparticles need to be deposited on a conductive transparent substrate such as indium tin oxide (ITO), which is used as the working electrode. A platinum wire may be placed in the cell slightly away from the field of view to function as a counter electrode. Alter- natively, the other coverslip can be coated with ITO and used as the counter electrode. In both cases, a platinum pseudo-reference electrode may be added in the cell. The Landes and Link groups used a spectroelectrochemical cell to probe the dependence of electrochemiluminescense enhancement through plasmonic nanoparticles [76], to characterize electrodissolution inhibition of gold nanorods through oxoanions [229], and to track the reversible formation of charge transfer plasmons by electrochemical oxidation and reduction of silver in a chloride solution [262]. A flow cell designed for TIR excitation with a white- light laser enabled electrochemical reactions to be studied with millisecond time resolution using snapshot hyperspectral imaging (see Section 4.3) [111, 117]. To avoid the added noise and background associated with multiple refractive index changes, some groups use

normalized spectrum of the laser before (blue line) and after Figure 12: Specialized configurations for nonstatic dark-field passing through a moveable mask at two different positions (solid imaging. and dotted red lines). (F) The colors of silver-coated gold nanorods (A) Spectroscopic flow cell for plasmon sensing and (B) an example are monitored during a reaction with 100 nM sodium hydrosulfide. of a spectrum obtained with this technique. (C) Water-immersion (G) Objective-based TIR excitation for combined dark-field objective for monitoring single-particle electrochemistry and spectroscopy and AFM. (H) AFM images and spectra of two gold (D) spectra obtained with this technique for a gold nanosphere nanorods measured with the setup described in (G). (A, B) Adapted when aniline was adsorbed and after electro-polymerization of with permission from Sönnichsen and coworkers, Copyright the aniline to polyaniline (PANI). WE, CE, and RE stand for working American Chemical Society, 2017 [215]. (C, D) Adapted with electrode, counter electrode, and reference electrode, permission from Maier and coworkers, Copyright the American respectively. (E) Schematic of a tunable light-sheet illumination Chemical Society, 2019 [253]. (E, F) Adapted with permission from setup with multiple spectral bands. S: shutter, SP: short-pass He and coworkers, copyright the American Chemical Society, 2018 filter, AL and ACL achromatic and achromatic cylindrical lenses, IL: [249]. (G, H) Adapted with permission from Zijlstra and coworkers, illumination lens, OL: objective lens, RL: relay lens system. Inset: Copyright the American Chemical Society, 2018 [139]. 1646 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

water-immersion objectives and place the objective single gold nanorods. The authors observed LSPR red- directly into the analyte solution [253–255]. Figure 12C shifts upon binding of single proteins, which were also shows a spectroelectrochemical setup used by the Maier detected by AFM. The Booksh group measured plasmon and Cortés groups, with a spectrum recorded using this coupling between a nanoparticle and a nanohole array by setup given in Figure 12D [253, 255]. They used a dark-field attaching a gold nanoparticle to an AFM tip and scanning condenser to excite nanoparticles and collect light through the tip across a gold film containing nanoholes [266]. The a water-immersion objective [253]. Since the electro- authors found 10 nm shifts in the LSPR when the tip chemical cell does not need to be sealed, the authors had approached the surface close enough for the plasmon of enough space next to the objective to use a saturated the nanoparticle to couple to the plasmon modes of the calomel electrode as a reference electrode, which provides nanohole array. a more stable reference potential than a pseudo-reference Dark-field spectroscopy coupled with single-particle electrode. electrochemical sensing was reported by the Hill group, The He group employed light-sheet illumination for who combined SECCM with a dark-field microscope to dark-field imaging, to sense hydrogen sulfide at a sensi- investigate the correlation between electrochemical reactivity of 0.1 nM [249]. Light-sheet microscopy may be tivity and particle shape [250, 267]. The authors first took a viewed as a specialized case of unidirectional HADF exci- hyperspectral image of a region of interest and then used tation as described in 3.1.1.2 where a cylindrical lens in the the hyperspectral image to guide the SECCM tip to scan all excitation path is used to focus a laser beam into a sheet of particles of interest. In this way, the electrochemical light [263]. The light sheet is maneuvered to dissect the reactivity was correlated to the spectrum and hence the sample by illuminating the entire field of view in the x-y particle shape. Notably, they found no correlation between direction but only a thin limited cross-section in the axial particle size and electrochemical reactivity, further dimension. Light-sheet illumination has become a popular confirmed by correlated SEM, indicating that the often re- choice for fluorescence microscopy [264, 265]. In the work ported interparticle heterogeneity in reactivity is not purely done by He, a supercontinuum laser was sent through a due to variations in size. Hill and coworkers attribute the wavelength selector to create discrete bands of light and variations in surface reactivity to surface restructuring and then shaped into a light sheet using a cylindrical lens variations in ligand density. (Figure 12E). The inset of Figure 12E illustrates the spectrum of the supercontinuum and the wavelength filtered beam. The light scattered from silver-coated gold nanorods was 6 Conclusions and outlook collected and sent to a color camera. Spectral changes were monitored as changes in color. Upon exposure to hydrogen Overall, we provided an overview of single-particle sulfide, the silver layer oxidized to silver sulfide, leading to scattering techniques and discussed different ways to a change in refractive index and a visible color change excite, detect, and measure plasmon scattering spectra of from red to green, enabling the optical detection of 0.1 nM single nanoparticles. Far-field HADF techniques rely on sodium hydrosulfide, as shown in Figure 12F. excitation at high angles to avoid collecting the excitation beam with the objective and only collect the forward- or backscattered light. Normal incidence excitation allows 5.2 Integrated multi-modality setups excitation perpendicular to the sample by blocking the incident beam and collecting scattering at high angles. Dark-field spectroscopy can be combined with tip-based These techniques excite particles in the far-field and allow scanning probe techniques such as AFM. The Zijlstra group excitation of nanoparticles in a wide variety of angles and implemented an objective-based TIR setup coupled to an refractive indices, including in refractive index-matched AFM (Figure 12G) [139]. A white-light source was focused on immersion oil. Near-field techniques rely on creating the side of the back aperture of a high-NA oil-immersion evanescent fields through prisms, objectives, condensers, objective to achieve TIR excitation. The reflected excitation waveguides, fibers, tips, or apertures and typically show beam was spatially filtered from the scattered light, which very high signal-to-noise ratios but have some constraints was directed to an electron multiplying CCD camera. when it comes to refractive indices and excitation geom- Bandpass filters were inserted in the detection path before etries. Detection of nanoparticles down to 2 nm can be the camera to collect spectra of the nanoparticles. The top achieved by interferometric scattering techniques, which of the coverslip allowed access to an AFM tip. Figure 12H measure the interference of the scattered light with a shows AFM topography maps and spectra of the same reference beam, rather than the pure scattering signal. A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1647

Far-field and near-field techniques can be used to detect supporting A.A.-Z., the Department of Energy (Grant No. the interaction of nanoparticles with polarized light, with DE-SC0016534), supporting T.S.H., and the National some setups polarizing in the excitation path, while Science Foundation (Grant No. CHE-1903980), supporting others place polarizers the detection path, or both. A.R.-M. S.L thanks the Robert A. Welch Foundation (Grant Numerous detection methods were introduced, from No. C-1664) for support. L.A.M. acknowledges support from high-resolution scanning techniques to millisecond the National Science Foundation Graduate Research resolved snapshot and interferometric approaches. Sub- Fellowship Program (1842494) and from the Lodieska nanometer spectrally resolved and high signal-to-noise Stockbridge Vaughn Fellowship, Rice University. ratio measurements of single nanoparticles can be ach- Conflict of interest statement: The authors declare no ieved by scanning a spectrograph with a slit or pinhole conflicts of interest regarding this article. across the field of view. However, this method usually takes on the order of minutes. Ratiometric and snapshot techniques can achieve millisecond time resolution, while References sacrificing some spectral resolution. Interferometers are less commonly used but have shown great potential for [1] C. F. H. Bohren and R. Donald, Absorption and Scattering of Light acquiring entire hyperspectral data cubes in seconds. by Small Particles, New York, Wiley-Interscience, 1998. Finally, we described how the different excitation and [2] S. A. Maier, Plasmonics: Fundamentals and Applications, New detection schemes have been combined with liquid cells, York, NY, Springer, 2007. electrochemical cells, and AFM and SECCM probes to use [3] B. J. Wiley, S. H. Im, Z. Y. Li, J. McLellan, A. Siekkinen, and Y. N. Xia, “Maneuvering the surface plasmon resonance of silver plasmonic nanoparticles as sensors and to study their role nanostructures through shape-controlled synthesis,” J. Phys. as catalysts. Chem. B, vol. 110, pp. 15666–15675, 2006. We expect single-particle scattering spectroscopy to [4] K. L. Kelly, E. Coronado, L. L. Zhao, and G. C. Schatz, “The optical continue to grow as an important tool in nanoscience. properties of metal nanoparticles: the influence of size, shape, ” Single-particle spectroscopy is a relatively new field that and dielectric environment, J. Phys. Chem. B, vol. 107, pp. 668–677, 2003. has grown rapidly over the last 20 years, made possible [5] J. J. Mock, M. Barbic, D. R. Smith, D. A. Schultz, and S. Schultz, by advancements and falling costs of microscopy and “Shape effects in plasmon resonance of individual colloidal silver spectroscopy components, especially detector chips. In nanoparticles,” J. Chem. Phys.,vol.116,pp.6755–6759, 2002. particular, CMOS chips are becoming cheaper, with [6] C. J. Orendorff, T. K. Sau, and C. J. Murphy, “Shape-dependent higher frame rates and higher contrast. At the same time, plasmon-resonant gold nanoparticles,” Small, vol. 2, – interest in plasmonic nanoparticles as highly sensitive pp. 636 639, 2006. [7] C. J. Murphy, T. K. San, A. M. Gole, et al., “Anisotropic metal optical sensors and as catalysts or light-harvesting an- nanoparticles: synthesis, assembly, and optical applications,” tennas is growing. To continue pushing the detection limit J. Phys. Chem. B, vol. 109, pp. 13857–13870, 2005. for biological sensors, single-particle scattering tech- [8] E. Ringe, M. R. Langille, K. Sohn, et al., “Plasmon length: A niques will continue to be an important tool. The strong universal parameter to describe size effects in gold ” – push towards renewable fuels has put plasmonic nano- nanoparticles, J. Phys. Chem. Lett., vol. 3, pp. 1479 1483, 2012. [9] S. E. Lohse and C. J. Murphy, “The quest for shape control: A particles into the spotlight due to their high photon history of gold nanorod synthesis,” Chem. Mater., vol. 25, absorption cross-section and hence applications as pp. 1250–1261, 2013. catalytic substrates driving reactions through both hot [10] C. Xue and C. A. Mirkin, “pH-switchable silver nanoprism growth carriers and heat. While ensemble studies have revealed pathways,” Angew. Chem. Int. Ed., vol. 46, pp. 2036–2038, great potential, researchers are increasingly making 2007. use of single-particle studies to gain a mechanistic [11] M. W. Knight, N. S. King, L. F. Liu, H. O. Everitt, P. Nordlander, and N. J. Halas, “Aluminum for plasmonics,” ACS Nano, vol. 8, understanding of these heterogeneous processes. The pp. 834–840, 2014. examples given here clearly demonstrate the need for [12] S. Link, Z. L. Wang, and M. A. El-Sayed, “Alloy formation of gold- single-particle approaches to avoid ensemble averaging silver nanoparticles and the dependence of the plasmon and thereby missing important information. absorption on their composition,” J. Phys. Chem. B, vol. 103, pp. 3529–3533, 1999. [13] K. M. Mayer and J. H. Hafner, “Localized surface plasmon Author contributions: All the authors have accepted resonance sensors,” Chem. Rev.,vol.111,pp.3828–3857, responsibility for the entire content of this submitted 2011. manuscript and approved submission. [14] J. J. Mock, D. R. Smith, and S. Schultz, “Local refractive index Research funding: The authors acknowledge funding from dependence of plasmon resonance spectra from individual the Army Research Office (Grant No. W911NF1910363), nanoparticles,” Nano Lett., vol. 3, pp. 485–491, 2003. 1648 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

[15] L. J. Sherry, R. C. Jin, C. A. Mirkin, G. C. Schatz, and [33] V. Giannini, A. I. Fernandez-Dominguez, S. C. Heck, and R. P. Van Duyne, “Localized surface plasmon resonance S. A. Maier, “Plasmonic nanoantennas: fundamentals and their spectroscopy of single silver triangular nanoprisms,” Nano use in controlling the radiative properties of nanoemitters,” Lett., vol. 6, pp. 2060–2065, 2006. Chem. Rev., vol. 111, pp. 3888–3912, 2011. [16] B. S. Hoener, S. R. Kirchner, T. S. Heiderscheit, et al., “Plasmonic [34] K. A. Willets, A. J. Wilson, V. Sundaresan, and P. B. Joshi, “Super- sensing and control of single-nanoparticle electrochemistry,” resolution imaging and plasmonics,” Chem. Rev., vol. 117, Inside Chem., vol. 4, pp. 1560–1585, 2018. pp. 7538–7582, 2017. [17] C. Novo, A. M. Funston, and P. Mulvaney, “Direct observation of [35] P. Zhang, S. Lee, H. Yu, N. Fang, and S. H. Kang, “Super-resolution chemical reactions on single gold nanocrystals using surface of fluorescence-free plasmonic nanoparticles using enhanced plasmon spectroscopy,” Nat. Nanotechnol., vol. 3, dark-field illumination based on wavelength-modulation,” Sci. pp. 598–602, 2008. Rep., vol. 5, 2015, https://doi.org/10.1038/srep11447. [18] L. Shi, C. Jing, W. Ma, et al., “Plasmon resonance scattering [36] J. Ando, A. Nakamura, M. Yamamoto, C. H. Song, K. Murata, and spectroscopy at the single-nanoparticle level: real-time R. Iino, “Multicolor high-speed tracking of single biomolecules monitoring of a click reaction,” Angew. Chem. Int. Ed., vol. 52, with silver, gold, and silver-gold alloy nanoparticles,” ACS pp. 6011–6014, 2013. Photonics, vol. 6, pp. 2870–2883, 2019. [19] D. Seo, G. Park, and H. Song, “Plasmonic monitoring of catalytic [37] Y. Wu, M. R. K. Ali, K. C. Chen, N. Fang, and M. A. El-Sayed, “Gold hydrogen generation by a single nanoparticle probe,” J. Am. nanoparticles in biological optical imaging,” Nano Today, vol. Chem. Soc., vol. 134, pp. 1221–1227, 2012. 24, pp. 120–140, 2019. [20] A. B. Taylor and P. Zijlstra, “Single-molecule plasmon sensing: [38] V. Amendola, R. Pilot, M. Frasconi, O. M. Marago, and M. A. Iati, current status and future prospects,” ACS Sens., vol. 2, “Surface plasmon resonance in gold nanoparticles: A review,” pp. 1103–1122, 2017. J. Phys. Condens. Matter, vol. 29, 2017, https://doi.org/10. [21] H. J. Chen, X. S. Kou, Z. Yang, W. H. Ni, and J. F. Wang, “Shape- 1088/1361-648x/aa60f3. and size-dependent refractive index sensitivity of gold [39] E. C. Dreaden, A. M. Alkilany, X. H. Huang, C. J. Murphy, and nanoparticles,” Langmuir, vol. 24, pp. 5233–5237, 2008. M. A. El-Sayed, “The golden age: gold nanoparticles for [22] C. Goldmann, R. Lazzari, X. Paquez, et al., “Charge transfer at biomedicine,” Chem. Soc. Rev., vol. 41, pp. 2740–2779, 2012. hybrid interfaces: plasmonics of aromatic thiol-capped gold [40] J. Olson, S. Dominguez-Medina, A. Hoggard, L.-Y. Wang, nanoparticles,” ACS Nano, vol. 9, pp. 7572–7582, 2015. W.-S. Chang, and S. Link, “Optical characterization of single [23] M. El Kabbash, A. R. Rashed, K. V. Sreekanth, A. De Luca, plasmonic nanoparticles,” Chem. Soc. Rev., vol. 44, pp. 40–57, M. Infusino, and G. Strangi, “Plasmon-exciton resonant energy 2015. transfer: across scales hybrid systems,” J. Nanomater., [41] E. Ringe, J. M. McMahon, K. Sohn, et al., “Unraveling the effects vol. 2016, 2016. http://dx.doi.org/10.1155/2016/4819040. of size, composition, and substrate on the localized surface [24] C. F. Bohren, “How can a particle absorb more than the light plasmon resonance frequencies of gold and silver nanocubes: A incident on it,” Am. J. Phys., vol. 51, pp. 323–327, 1983. systematic single-particle approach,” J. Phys. Chem. C, vol. 114, [25] P. Biagioni, J. S. Huang, and B. Hecht, “Nanoantennas for visible pp. 12511–12516, 2010. and infrared radiation,” Rep. Prog. Phys., vol. 75, 2012, https:// [42] M. Hu, C. Novo, A. Funston, et al., “Dark-field microscopy doi.org/10.1088/0034-4885/75/2/024402. studies of single metal nanoparticles: understanding the [26] J. B. Sambur and P. Chen, “Approaches to single-nanoparticle factors that influence the linewidth of the localized surface catalysis,” Annu. Rev. Phys. Chem., vol. 65, no. 65, pp. 395–422, plasmon resonance,” J. Mater. Chem., vol. 18, pp. 1949–1960, 2014. 2008. [27] S. Yu and P. K. Jain, “Plasmonic photosynthesis of C1–C3 [43] S. M. E. Peters, M. A. Verheijen, M. W. J. Prins, and P. Zijlstra, hydrocarbons from carbon dioxide assisted by an ionic liquid,” “Strong reduction of spectral heterogeneity in gold bipyramids Nat. Commun., vol. 10, p. 2022, 2019. for single-particle and single-molecule plasmon sensing,” [28] J. L. Brooks, C. L. Warkentin, D. Saha, E. L. Keller, and Nanotechnology, vol. 27, 2016, https://doi.org/10.1088/0957- R. R. Frontiera, “Toward a mechanistic understanding of 4484/27/2/024001. plasmon-mediated photocatalysis,” Nanophotonics, vol. 7, [44] A. I. Henry, J. M. Bingham, E. Ringe, L. D. Marks, G. C. Schatz, and pp. 1697–1724, 2018. R. P. Van Duyne, “Correlated structure and optical property [29] Y. Liu, B. M. Crawford, and T. Vo-Dinh, “Gold nanoparticles- studies of plasmonic nanoparticles,” J. Phys. Chem. C, vol. 115, mediated photothermal therapy and immunotherapy,” pp. 9291–9305, 2011. Immunotherapy, vol. 10, pp. 1175–1188, 2018. [45] H. H. Wang, T. Zhang, and X. C. Zhou, “Dark-field spectroscopy: [30] A. Campion and P. Kambhampati, “Surface-enhanced Raman development, applications and perspectives in single scattering,” Chem. Soc. Rev., vol. 27, pp. 241–250, 1998. nanoparticle catalysis,” J. Phys. Condens. Matter, vol. 31, 2019, [31] J. Kim, S. Yoo, J.-M. Kim, et al., “Synthesis and single-particle https://doi.org/10.1088/1361-648x/ab330a. surface-enhanced Raman scattering study of plasmonic tripod [46] T. Li, X. Wu, F. Liu, and N. Li, “Analytical methods based on the nanoframes with Y-shaped hot-zone,” Nano Lett., vol. 20, light-scattering of plasmonic nanoparticles at the single particle pp. 4362–4369, 2020. level with dark-field microscopy imaging,” Analyst, vol. 142, [32] A. Bhattarai, B. T. O’Callahan, C.-F. Wang, S. Wang, and P. Z. El- pp. 248–256, 2017. Khoury, “Spatio-spectral characterization of multipolar [47] E. Vinegrad, D. Vestler, A. Ben-Moshe, A. R. Barnea, plasmonic modes of Au nanorods via tip-enhanced Raman G. Markovich, and O. Cheshnovsky, “Circular dichroism of scattering,” J. Phys. Chem. Lett., vol. 11, pp. 2870–2874, 2020. single particles,” ACS Photonics, vol. 5, pp. 2151–2159, 2018. A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1649

[48] M. A. van Dijk, A. L. Tchebotareva, M. Orrit, et al., “Absorption [65] S. Mukherjee, F. Libisch, N. Large, et al., “Hot electrons do the and scattering microscopy of single metal nanoparticles,” Phys. impossible: plasmon-induced dissociation of H-2 on Au,” Nano Chem. Chem. Phys., vol. 8, pp. 3486–3495, 2006. Lett., vol. 13, pp. 240–247, 2013. [49] M. Yorulmaz, S. Nizzero, A. Hoggard, et al., “Single-particle [66] L. V. Besteiro, X. T. Kong, Z. M. Wang, G. Hartland, and absorption spectroscopy by photothermal contrast,” Nano Lett., A. O. Govorov, “Understanding hot-electron generation and vol. 15, pp. 3041–3047, 2015. plasmon relaxation in metal nanocrystals: quantum and classical [50] D. Boyer, P. Tamarat, A. Maali, B. Lounis, and M. Orrit, mechanisms,” ACS Photonics, vol. 4, pp. 2759–2781, 2017. “Photothermal imaging of nanometer-sized metal particles [67] G. V. Hartland, L. V. Besteiro, P. Johns, and A. O. Govorov, among scatterers,” Science, vol. 297, pp. 1160–1163, 2002. “What’s so hot about electrons in metal nanoparticles?,” ACS [51] V. Schweikhard, A. Grubisic, T. A. Baker, I. Thomann, and Energy Lett., vol. 2, pp. 1641–1653, 2017. D. J. Nesbitt, “Polarization-dependent scanning photoionization [68] C. M. Andolina, S. E. Crawford, A. M. Smith, et al., “Near-infrared microscopy: ultrafast plasmon-mediated electron ejection photoluminescence from small copper, silver, and gold dynamics in single Au nanorods,” ACS Nano, vol. 5, nanoparticles,” ChemNanoMat, vol. 4, pp. 265–268, 2018. pp. 3724–3735, 2011. [69] C. Boerigter, U. Aslam, and S. Linic, “Mechanism of charge [52] V. Schweikhard, A. Grubisic, T. A. Baker, and D. J. Nesbitt, transfer from plasmonic nanostructures to chemically attached “Multiphoton scanning photoionization imaging microscopy for materials,” ACS Nano, vol. 10, pp. 6108–6115, 2016. single-particle studies of plasmonic metal nanostructures,” [70] X. C. Ma, Y. Dai, L. Yu, and B. B. Huang, “Energy transfer in J. Phys. Chem. C, vol. 115, pp. 83–91, 2011. plasmonic photocatalytic composites,” Light Sci. Appl., vol. 5, [53] Y.-Y. Cai, J. G. Liu, L. J. Tauzin, et al., “Photoluminescence of gold 2016, https://doi.org/10.1038/lsa.2016.17. nanorods: purcell effect enhanced emission from hot carriers,” [71] S. K. Cushing, J. T. Li, F. K. Meng, et al., “Photocatalytic activity ACS Nano, vol. 12, pp. 976–985, 2018. enhanced by plasmonic resonant energy transfer from metal to [54] C. M. Hill, R. Bennett, C. Zhou, S. Street, J. Zheng, and S. L. Pan, semiconductor,” J. Am. Chem. Soc., vol. 134, pp. 15033–15041, “Single Ag nanoparticle spectroelectrochemistry via dark-field 2012. scattering and fluorescence microscopies,” J. Phys. Chem. C, [72] G. Mie, “Beiträge zur Optik trüber Medien, speziell kolloidaler vol. 119, pp. 6760–6768, 2015. Metallösungen,” Ann. Phys., vol. 330, pp. 377–445, 1908. [55] M.-N. Su, P. D. Dongare, D. Chakraborty, et al., “Optomechanics [73] E. M. Purcell and C. R. Pennypacker, “Scattering and absorption of single aluminum nanodisks,” Nano Lett., vol. 17, of light by nonspherical dielectric grains,” Astrophys. J., vol. pp. 2575–2583, 2017. 186, pp. 705–714, 1973. [56] H. Staleva and G. V. Hartland, “Vibrational dynamics of silver [74] B. T. Draine and P. J. Flatau, “Discrete-dipole approximation for nanocubes and nanowires studied by single-particle transient scattering calculations,” J. Opt. Soc. Am. A, vol. 11, absorption spectroscopy,” Adv. Funct. Mater., vol. 18, pp. 1491–1499, 1994. pp. 3809–3817, 2008. [75] J. Zhao, A. O. Pinchuk, J. M. Mcmahon, et al., “Methods for [57] Y. H. Peng, B. Xiong, L. Peng, H. Li, Y. He, and E. S. Yeung, describing the electromagnetic properties of silver and gold “Recent advances in optical imaging with anisotropic plasmonic nanoparticles,” Accounts Chem. Res., vol. 41, pp. 1710–1720, nanoparticles,” Anal. Chem., vol. 87, pp. 200–215, 2015. 2008. [58] E. Ringe, B. Sharma, A. I. Henry, L. D. Marks, and [76] T. S. Heiderscheit, M. J. Gallagher, R. Baiyasi, et al., R. P. Van Duyne, “Single nanoparticle plasmonics,” Phys. Chem. “Nanoelectrode-emitter spectral overlap amplifies surface Chem. Phys., vol. 15, pp. 4110–4129, 2013. enhanced electrogenerated chemiluminescence,” J. Chem. [59] C. Jing, F. J. Rawson, H. Zhou, et al., “New insights into Phys., vol. 151, p. 144712, 2019. electrocatalysis based on plasmon resonance for the real-time [77] A. Tcherniak, J. W. Ha, S. Dominguez-Medina, L. S. Slaughter, monitoring of catalytic events on single gold nanorods,” Anal. and S. Link, “Probing a century old prediction one plasmonic Chem., vol. 86, pp. 5513–5518, 2014. particle at a time,” Nano Lett., vol. 10, pp. 1398–1404, 2010. [60] S. Song, J. U. Lee, J. Kang, K. H. Park, and S. J. Sim, “Real-time [78] J. A. Fan, K. Bao, J. B. Lassiter, et al., “Near-normal incidence monitoring of distinct binding kinetics of hot-spot mutant p53 dark-field microscopy: applications to nanoplasmonic protein in human cancer cells using an individual nanorod- spectroscopy,” Nano Lett., vol. 12, pp. 2817–2821, 2012. based plasmonic biosensor,” Sensor. Actuator. B Chem., [79] A. Weigel, A. Sebesta, and P. Kukura, “Dark field vol. 322, p. 128584, 2020. microspectroscopy with single molecule fluorescence [61] X. L. Wang, Y. Cui, and J. Irudayaraj, “Single-cell quantification sensitivity,” ACS Photonics, vol. 1, pp. 848–856, 2014. of cytosine modifications by hyperspectral dark-field imaging,” [80] D. J. Hill, C. W. Pinion, J. D. Christesen, and J. F. Cahoon, ACS Nano, vol. 9, pp. 11924–11932, 2015. “Waveguide scattering microscopy for dark-field imaging and [62] S. Alekseeva, I. I. Nedrygailov, and C. Langhammer, “Single spectroscopy of photonic nanostructures,” ACS Photonics, vol. particle plasmonics for materials science and single particle 1, pp. 725–731, 2014. catalysis,” ACS Photonics, vol. 6, pp. 1319–1330, 2019. [81] E. Hecht, Optics, 4th ed. San Francisco, CA, USA, Pearson, 2002. [63] Q. F. Zhang, T. Hernandez, K. W. Smith, et al., “Unraveling the [82] J. Olson, A. Manjavacas, T. Basu, et al., “High chromaticity origin of chirality from plasmonic nanoparticle-protein aluminum plasmonic pixels for active liquid crystal displays,” complexes,” Science, vol. 365, pp. 1475–1478, 2019. ACS Nano, vol. 10, pp. 1108–1117, 2016. [64] J. Karst, N. Strohfeldt, M. Schäferling, H. Giessen, and [83] A. Tcherniak, A. Prakash, J. T. Mayo, V. L. Colvin, and S. Link, M. Hentschel, “Single plasmonic oligomer chiral spectroscopy,” “Fluorescence correlation spectroscopy of magnetite nanocrystal Adv. Opt. Mater., vol. 6, p. 1800087, 2018. diffusion,” J. Phys. Chem. C,vol.113,pp.844–848, 2008. 1650 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

[84] H. Kuang, X. Chen, C. Hao, W. Ma, L. Xu, and C. Xu, “Immuno- explore the single nanoparticle plasmon resonances,” Opt. driven plasmonic oligomer sensor for the ultrasensitive detection Express, vol. 26, pp. 6439–6445, 2018. of antibiotics,” RSC Adv., vol. 3, pp. 17294–17299, 2013. [102] L. Y. Jiang, T. T. Yin, Z. G. Dong, et al., “Accurate modeling of [85] J. S. Crosby, D. Lucas, and C. P. Koshland, “Fiber optic based dark-field scattering spectra of plasmonic nano structures,” evanescent wave sensor for the detection of elemental mercury ACS Nano, vol. 9, pp. 10039–10046, 2015. utilizing gold nanorods,” Sensor. Actuator. B Chem., vol. 181, [103] L. O. Herrmann and J. J. Baumberg, “Watching single pp. 938–942, 2013. nanoparticles grow in real time through supercontinuum [86] J. Cao, M. H. Tu, T. Sun, and K. T. V. Grattan, “Wavelength-based spectroscopy,” Small, vol. 9, pp. 3743–3747, 2013. localized surface plasmon resonance optical fiber biosensor,” [104] X. X. Lu, J. Wu, Q. N. Zhu, et al., “Circular dichroism from single Sensor. Actuator. B Chem., vol. 181, pp. 611–619, 2013. plasmonic nanostructures with extrinsic chirality,” Nanoscale, [87] G. Maire, H. Giovannini, A. Talneau, P. C. Chaumet, K. Belkebir, vol. 6, pp. 14244–14253, 2014. and A. Sentenac, “Phase imaging and synthetic aperture super- [105] N. J. Greybush, V. Pacheco-Pena, N. Engheta, C. B. Murray, and resolution via total internal reflection microscopy,” Opt. Lett., C. R. Kagan, “Plasmonic optical and chiroptical response of vol. 43, pp. 2173–2176, 2018. self-assembled Au nanorod equilateral trimers,” ACS Nano, [88] Y. Li, L. Xu, and B. Li, “Gold nanorod-induced localized surface vol. 13, pp. 1617–1624, 2019. plasmon for microparticle aggregation,” Appl. Phys. Lett., [106]J.A.Fan,C.H.Wu,K.Bao,etal.,“Self-assembled plasmonic vol. 101, p. 053118, 2012. nanoparticle clusters,” Science, vol. 328, pp. 1135–1138, [89] J.-T. Lin, D.-C. Cheng, M. Jiang, Y.-S. Chiang, and H.-W. Liu, 2010. “Analysis of scaling law and figure of merit of fiber-based [107] S. S. E. Collins, X. Wei, T. G. McKenzie, A. M. Funston, and biosensor,” J. Nanomater., vol. 2012, p. 154736, 2012. P. Mulvaney, “Single gold nanorod charge modulation in an ion [90] Y. Li, Y. J. Hu, and Q. Wu, “Nanofiber-excited plasmonic gel device,” Nano Lett., vol. 16, pp. 6863–6869, 2016. manipulation of polystyrene nanospheres,” RSC Adv., vol. 5, [108] X. L. Nan, P. A. Sims, and X. S. Xie, “Organelle tracking in a pp. 76202–76205, 2015. living cell with microsecond time resolution and nanometer [91] W. Ni, H. Chen, X. Kou, M. H. Yeung, and J. Wang, “Optical fiber- spatial precision,” ChemPhysChem, vol. 9, pp. 707–712, 2008. excited surface plasmon resonance spectroscopy of single and [109] M. F. Limonov, M. V. Rybin, A. N. Poddubny, and Y. S. Kivshar, ensemble gold nanorods,” J. Phys. Chem. C, vol. 112, “Fano resonances in photonics,” Nat. Photonics, vol. 11, pp. 8105–8109, 2008. pp. 543–554, 2017. [92] Y. D. Wilde and P. A. Lemoine, “Review of NSOM microscopy for [110] J. P. Kottmann and O. J. F. Martin, “Retardation-induced materials,” AIP Conf. Proc., vol. 931, pp. 43–52, 2007. plasmon resonances in coupled nanoparticles,” Opt. Lett., [93] R. M. A. Azzam, “Circular and near-circular polarization states vol. 26, pp. 1096–1098, 2001. of evanescent monochromatic light fields in total internal [111] S. R. Kirchner, K. W. Smith, B. S. Hoener, et al., “Snapshot reflection,” Appl. Opt., vol. 50, p. 6272, 2011. hyperspectral imaging (SHI) for revealing irreversible and [94] K. Y. Bliokh, A. Y. Bekshaev, and F. Nori, “Extraordinary heterogeneous plasmonic processes,” J. Phys. Chem. C, momentum and spin in evanescent waves,” Nat. Commun., vol. 122, pp. 6865–6875, 2018. vol. 5, p. 3300, 2014. [112] H. Ueno, S. Nishikawa, R. Iino, et al., “Simple dark-field [95] M. W. Knight, J. Fan, F. Capasso, and N. J. Halas, “Influence of microscopy with nanometer spatial precision and microsecond excitation and collection geometry on the dark field spectra of temporal resolution,” Biophys. J., vol. 98, pp. 2014–2023, individual plasmonic nanostructures,” Opt. Express, vol. 18, 2010. pp. 2579–2587, 2010. [113] U. Arieli, M. Mrejen, and H. Suchowski, “Broadband coherent [96] K. W. Smith, L. A. McCarthy, A. Alabastri, et al., “Exploiting hyperspectral near-field imaging of plasmonic evanescent field polarization for giant chiroptical modulation nanostructures,” Opt. Express, vol. 27, pp. 9815–9820, 2019. from achiral gold half-rings,” ACS Nano, vol. 12, [114] C. Sönnichsen, T. Franzl, T. Wilk, et al., “Drastic reduction of pp. 11657–11663, 2018. plasmon damping in gold nanorods,” Phys. Rev. Lett., vol. 88, [97] N. S. King, Y. Li, C. Ayala-Orozco, T. Brannan, P. Nordlander, p. 077402, 2002. and N. J. Halas, “Angle- and spectral-dependent light [115] K. Marchuk and N. Fang, “Three-dimensional orientation scattering from plasmonic nanocups,” ACS Nano,vol.5, determination of stationary anisotropic nanoparticles with pp. 7254–7262, 2011. sub-degree precision under total internal reflection scattering [98] Y. Sowa, B. C. Steel, and R. M. Berry, “A simple backscattering microscopy,” Nano Lett., vol. 13, pp. 5414–5419, 2013. microscope for fast tracking of biological molecules,” Rev. Sci. [116] M. Lindner, G. Nir, S. Medalion, H. R. C. Dietrich, Y. Rabin, and Instrum., vol. 81, 2010, https://doi.org/10.1063/1.3495960. Y. Garini, “Force-free measurements of the conformations of [99] S. J. Kim, P. C. Blainey, C. M. Schroeder, and X. S. Xie, DNA molecules tethered to a wall,” Phys. Rev. E, vol. 83, “Multiplexed single-molecule assay for enzymatic activity on p. 011916, 2011. flow-stretched DNA,” Nat. Methods, vol. 4, pp. 397–399, 2007. [117] A. Al-Zubeidi, B. S. Hoener, S. S. E. Collins, et al., “Hot holes [100] G. C. Li, Y. L. Zhang, and D. Y. Lei, “Hybrid plasmonic gap modes assist plasmonic nanoelectrode dissolution,” Nano Lett., in metal film-coupled dimers and their physical origins vol. 19, pp. 1301–1306, 2019. revealed by polarization resolved dark field spectroscopy,” [118] S.-C. Yang, H. Kobori, C.-L. He, et al., “Plasmon hybridization Nanoscale, vol. 8, pp. 7119–7126, 2016. in individual gold nanocrystal dimers: direct observation of [101] A. Movsesyan, A. L. Baudrion, and P. M. Adam, “Extinction bright and dark modes,” Nano Lett.,vol.10,pp.632–637, measurements of metallic nanoparticles arrays as a way to 2010. A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1651

[119] R. Schneider, T. Glaser, M. Berndt, and S. Diez, “Using a quartz reflection dark-field microscopy,” PloS One, vol. 7, p. e49208, paraboloid for versatile wide-field TIR microscopy with sub- 2012. nanometer localization accuracy,” Opt. Express, vol. 21, [136] P. von Olshausen and A. Rohrbach, “Coherent total internal pp. 3523–3539, 2013. reflection dark-field microscopy: label-free imaging beyond the [120] I. Ament, J. Prasad, A. Henkel, S. Schmachtel, and diffraction limit,” Opt. Lett., vol. 38, pp. 4066–4069, 2013. C. Sönnichsen, “Single unlabeled protein detection on [137] M. Liu, J. Chao, S. Deng, K. Wang, K. Li, and C. Fan, “Dark-field individual plasmonic nanoparticles,” Nano Lett., vol. 12, microscopy in imaging of plasmon resonant nanoparticles,” pp. 1092–1095, 2012. Colloids Surf, vol. 124, pp. 111–117, 2014. [121] C. Sönnichsen, S. Geier, N. E. Hecker, et al., “Spectroscopy of [138] D. Eschimèse, P. Hsia, F. Vaurette, et al., “Comparative single metallic nanoparticles using total internal reflection investigation of plasmonic properties between tunable microscopy,” Appl. Phys. Lett., vol. 77, pp. 2949–2951, 2000. nanoobjects and metallized nanoprobes for optical [122] C. Haiden, T. Wopelka, M. Jech, F. Keplinger, and spectroscopy,” J. Phys. Chem. C, vol. 123, pp. 28392–28400, M. J. Vellekoop, “Sizing of metallic nanoparticles confined to a 2019. microfluidic film applying dark-field particle tracking,” [139] M. A. Beuwer, B. van Hoof, and P. Zijlstra, “Spatially resolved Langmuir, vol. 30, pp. 9607–9615, 2014. sensitivity of single-particle plasmon sensors,” J. Phys. Chem. [123] D. Axelrod, “Total internal reflection fluorescence microscopy C, vol. 122, pp. 4615–4621, 2018. in cell biology,” Traffic, vol. 2, pp. 764–774, 2001. [140] K. N. Fish, “Total internal reflection fluorescence (TIRF) [124] K. Namink, X. Meng, M. T. M. Koper, P. Kukura, and S. Faez, microscopy,” Curr. Protoc. Cytom., vol. 50, p. 12.18.1–12.18.13, “Electric-double-layer-modulation microscopy,” Phys. Rev. 2009. Appl., vol. 13, p. 044065, 2020. [141] S. Ramachandran, F. T. Arce, N. R. Patel, A. P. Quist, [125] P. K. Jain, W. Huang, and M. A. El-Sayed, “On the universal D. A. Cohen, and R. Lal, “Structure and permeability of ion- scaling behavior of the distance decay of plasmon coupling in channels by integrated AFM and waveguide TIRF microscopy,” metal nanoparticle pairs: A plasmon ruler equation,” Nano Sci. Rep., vol. 4, p. 4424, 2014. Lett., vol. 7, pp. 2080–2088, 2007. [142] T. P. Burghardt, A. D. Hipp, and K. Ajtai, “Around-the-objective [126] E. Prodan, C. Radloff, N. J. Halas, and P. Nordlander, “A total internal reflection fluorescence microscopy,” Appl. Opt., hybridization model for the plasmon response of complex vol. 48, pp. 6120–6131, 2009. nanostructures,” Science, vol. 302, p. 419, 2003. [143] H. M. Grandin, B. Städler, M. Textor, and J. Vörös, “Waveguide [127] W. Zhang, L. Huang, C. Santschi, and O. J. F. Martin, “Trapping excitation fluorescence microscopy: A new tool for sensing and and sensing 10 nm metal nanoparticles using plasmonic dipole imaging the biointerface,” Biosens. Bioelectron., vol. 21, antennas,” Nano Lett., vol. 10, pp. 1006–1011, 2010. pp. 1476–1482, 2006. [128] B. Ostovar, M.-N. Su, D. Renard, et al., “Acoustic vibrations of Al [144] R. Diekmann, Ø. I. Helle, C. I. Øie, et al., “Chip-based wide nanocrystals: size, shape, and crystallinity revealed by single- field-of-view nanoscopy,” Nat. Photonics,vol.11, particle transient extinction spectroscopy,” J. Phys. Chem., vol. pp. 322–328, 2017. 124, pp. 3924–3934, 2020. [145] Z.-M. Qi, N. Matsuda, T. Yoshida, H. Asano, A. Takatsu, [129] L. J. Tauzin, Y.-Y. Cai, K. W. Smith, et al., “Exploring the and K. Kato, “Optical waveguide spectrometer based on relationship between plasmon damping and luminescence in thin-film glass plates,” Opt. Lett.,vol.27,pp.2001–2003, lithographically prepared gold nanorods,” ACS Photonics, vol. 2002. 5, pp. 3541–3549, 2018. [146] M. Qi, C. Darviot, S. Patskovsky, and M. Meunier, “Cost- [130] K.W.Smith,J.Yang,T.Hernandez,etal.,“Environmental effective side-illumination darkfield nanoplasmonic marker symmetry breaking promotes plasmon mode splitting in gold microscopy,” Analyst, vol. 144, pp. 1303–1308, 2019. nanotriangles,” J. Phys. Chem. C,vol.122,pp.13259–13266, 2018. [147] S. Ramachandran, D. A. Cohen, A. P. Quist, and R. Lal, “High [131] L. A. Warning, Q. Zhang, R. Baiyasi, C. F. Landes, and S. Link, performance, LED powered, waveguide based total internal “Nanoscale surface-induced unfolding of single fibronectin is reflection microscopy,” Sci. Rep., vol. 3, p. 2133, 2013. restricted by serum albumin crowding,” J. Phys. Chem. Lett., [148] Y. Kawano, C. Higgins, Y. Yamamoto, et al., “Darkfield adapter vol. 11, pp. 1170–1177, 2020. for whole slide imaging: adapting a darkfield internal reflection [132] K. A. Koen, M. L. Weber, K. M. Mayer, E. Fernandez, and illumination system to extend WSI applications,” PloS One, vol. K. A. Willets, “Spectrally-resolved polarization anisotropy of 8, p. e58344, 2013. single plasmonic nanoparticles excited by total internal [149] Y. Kawano, C. Otsuka, J. Sanzo, et al., “Expanding imaging reflection,” J. Phys. Chem. C, vol. 116, pp. 16198–16206, 2012. capabilities for microfluidics: applicability of darkfield internal [133] D. Talaga, M. Comesaña-Hermo, S. Ravaine, R. A. L. Vallée, and reflection illumination (DIRI) to observations in microfluidics,” S. Bonhommeau, “Colocalized dark-field scattering, atomic PloS One, vol. 10, p. e0116925, 2015. force and surface-enhanced Raman scattering microscopic [150] L. Wang, C. Darviot, J. Zapata-Farfan, S. Patskovsky, D. Trudel, imaging of single gold nanoparticles,” J. Opt., vol. 17, and M. Meunier, “Designable nanoplasmonic biomarkers for p. 114006, 2015. direct microscopy cytopathology diagnostics,” J. Biophotonics, [134]Y.Zhu,J.He,C.Shang,etal.,“Chiral gold nanowires with vol. 12, p. e201900166, 2019. boerdijk–coxeter–bernal structure,” J. Am. Chem. Soc., vol. [151] J. A. Veerman, M. F. Garcia-Parajo, L. Kuipers, and 136, pp. 12746–12752, 2014. N. F. Van Hulst, “Single molecule mapping of the optical field [135] S. Enoki, R. Iino, N. Morone, et al., “Label-free single-particle distribution of probes for near-field microscopy,” J. Microsc., imaging of the influenza virus by objective-type total internal vol. 194, pp. 477–482, 1999. 1652 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

[152] M. F. Garcia-Parajo, J.-A. Veerman, S. J. T. van Noort, noble metal Au nanoparticles in the sub-diffraction limit,” B. G. de Grooth, J. Greve, and N. F. van Hulst, “Near-field optical Phys. Chem. Chem. Phys., vol. 18, pp. 23680–23685, 2016. microscopy for DNA studies at the single molecular level,” [170] J. Abed, F. Alexander, I. Taha, N. Rajput, C. Aubry, and Bioimaging, vol. 6, pp. 43–53, 1998. M. Jouiad, “Investigation of broadband surface plasmon [153] H. G. Frey, F. Keilmann, A. Kriele, and R. Guckenberger, resonance of dewetted Au structures on TiO2 by aperture-probe “Enhancing the resolution of scanning near-field optical SNOM and FDTD simulations,” Plasmonics, vol. 14, microscopy by a metal tip grown on an aperture probe,” Appl. pp. 205–218, 2019. Phys. Lett., vol. 81, pp. 5030–5032, 2002. [171] M. Beleites, C. Matyssek, H.-H. Blaschek, and G. Seifert, “Near- [154] F. Keilmann, “Scattering-type near-field optical microscopy,” field optical microscopy of femtosecond-laser-reshaped silver Microscopy, vol. 53, pp. 187–192, 2004. nanoparticles in dielectric matrix,” Nanoscale Res. Lett., vol. 7, [155] A. Mikhailovsky, M. Petruska, K. Li, M. Stockman, and p. 315, 2012. V. I. Klimov, “Phase-sensitive spectroscopy of surface [172] N. Jahr, M. Anwar, O. Stranik, et al., “Spectroscopy on single plasmons in individual metal nanostructures,” Phys. Rev. B, metallic nanoparticles using subwavelength Apertures,” vol. 69, p. 085401, 2004. J. Phys. Chem. C, vol. 117, pp. 7751–7756, 2013. [156] T. Klar, M. Perner, S. Grosse, G. Von Plessen, W. Spirkl, and [173] R. J. Hermann and M. J. Gordon, “Nanoscale optical microscopy J. Feldmann, “Surface-plasmon resonances in single metallic and spectroscopy using near-field probes,” Annu. Rev. Chem. nanoparticles,” Phys. Rev. Lett., vol. 80, p. 4249, 1998. Biomol. Eng., vol. 9, pp. 365–387, 2018. [157] E. H. Synge XXXVIII, “A suggested method for extending [174] R. Hillenbrand and F. Keilmann, “Complex optical constants on microscopic resolution into the ultra-microscopic region,” a subwavelength scale,” Phys. Rev. Lett., vol. 85, p. 3029, Philos. Mag. A, vol. 6, pp. 356–362, 1928. 2000. [158] U. Durig, D. W. Pohl, and F. Rohner, “Near-field optical- [175] N. Zhou, Y. Li, and X. Xu, “Resolving near-field from high order scanning microscopy,” J. Appl. Phys., vol. 59, pp. 3318–3327, signals of scattering near-field scanning optical microscopy,” 1986. Opt. Express, vol. 22, pp. 18715–18723, 2014. [159] D. W. Pohl, W. Denk, and M. Lanz, “Optical stethoscopy: image [176] D.-S. Kim and Z. H. Kim, “Role of in-plane polarizability of the recording with resolution λ/20,” Appl. Phys. Lett., vol. 44, tip in scattering near-field microscopy of a plasmonic pp. 651–653, 1984. nanoparticle,” Opt. Express, vol. 20, pp. 8689–8699, 2012. [160] D. Courjon, J.-M. Vigoureux, M. Spajer, K. Sarayeddine, and [177] A. Cricenti, M. Luce, D. Moroni, O. Salvetti, and M. D’Acunto, S. Leblanc, “External and internal reflection near field “Ultra small clusters of gold nanoshells detected by SNOM,” microscopy: experiments and results,” Appl. Opt., vol. 29, Opto-Electron. Rev., vol. 23, pp. 39–45, 2015. pp. 3734–3740, 1990. [178] P. Vaccarello, L. Tran, J. Meinen, C. Kwon, Y. Abate, and [161] H. F. Hamann, A. Gallagher, and D. J. Nesbitt, “Enhanced Y.-S. Shon, “Characterization of localized surface plasmon sensitivity near-field scanning optical microscopy at high resonance transducers produced from Au25 nanoparticle spatial resolution,” Appl. Phys. Lett., vol. 73, pp. 1469–1471, multilayers,” Colloids Surf, A, vol. 402, pp. 146–151, 2012. 1998. [179] G. A. Wurtz, N. M. Dimitrijevic, and G. P. Wiederrecht, “The [162] C. Girard and M. Spajer, “Model for reflection near field optical spatial extension of the field scattered by silver nanoparticles microscopy,” Appl. Opt., vol. 29, pp. 3726–3733, 1990. excited near resonance as observed by apertureless near-field [163] B. Hecht, B. Sick, U. P. Wild, et al., “Scanning near-field optical optical microscopy,” Jpn. J. Appl. Phys., vol. 41, pp. L351–L354, microscopy with aperture probes: fundamentals and 2002. applications,” J. Chem. Phys., vol. 112, pp. 7761–7774, 2000. [180] S. Benrezzak, P. M. Adam, J. L. Bijeon, and P. Royer, [164] A. Lereu, A. Passian, and P. Dumas, “Near field optical “Observation of nanometric metallic particles with an microscopy: A brief review,” Int. J. Nanotechnol., vol. 9, p. 488, apertureless scanning near-field optical microscope,” Surf. 2012. Sci., vol. 491, pp. 195–207, 2001. [165] J. Ingham, M. Pilling, T. Craig, et al., “An evaluation of the [181] A. García-Etxarri, I. Romero, F. J. G. de Abajo, R. Hillenbrand, application of the aperture infrared SNOM technique to and J. Aizpurua, “Influence of the tip in near-field imaging of biomedical imaging,” Biomed. Phys. Eng. Express, vol. 4, nanoparticle plasmonic modes: weak and strong coupling p. 025011, 2018. regimes,” Phys. Rev. B, vol. 79, p. 125439, 2009. [166] E. Betzig and R. J. Chichester, “Single molecules observed by [182] K. Minn, B. Birmingham, and Z. Zhang, “New development of near-field scanning optical microscopy,” Science, vol. 262, nanoscale spectroscopy using scanning probe microscope,” pp. 1422–1425, 1993. J. Vac. Sci. Technol., vol. 38, p. 030801, 2020. [167] S. Baral, A. Rafiei Miandashti, and H. H. Richardson, “Near-field [183] P. Bazylewski, S. Ezugwu, and G. Fanchini, “A review of three- thermal imaging of optically excited gold nanostructures: dimensional scanning near-field optical microscopy scaling principles for collective heating with heat dissipation (3D-SNOM) and its applications in nanoscale light into the surrounding medium,” Nanoscale, vol. 10, management,” Appl. Sci., vol. 7, p. 973, 2017. pp. 941–948, 2018. [184] K. Lindfors, T. Kalkbrenner, P. Stoller, and V. Sandoghdar, [168] F. Intonti, V. Emiliani, C. Lienau, et al., “Quantum mechanical “Detection and spectroscopy of gold nanoparticles using repulsion of exciton levels in a disordered quantum well,” supercontinuum white light confocal microscopy,” Phys. Rev. Phys. Rev. Lett., vol. 87, p. 076801, 2001. Lett., vol. 93, p. 037401, 2004. [169] A. K. Sivadasan, K. K. Madapu, and S. Dhara, “The light–matter [185] K. Holanová, M. Vala, and M. Piliarik, “[INVITED] Optical interaction of a single semiconducting AlGaN nanowire and imaging and localization of prospective scattering labels A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1653

smaller than a single protein,” Opt. Laser. Technol., vol. 109, [202] S. Biswas, D. Nepal, K. Park, and R. A. Vaia, “Orientation pp. 323–327, 2019. sensing with color using plasmonic gold nanorods and [186] J. B. Heaney, K. P. Stewart, and G. Hass, “Transmittance and assemblies,” J. Phys. Chem. Lett., vol. 3, pp. 2568–2574, 2012. reflectance of crystalline quartz and high- and low-water [203] J. B. Lassiter, H. Sobhani, J. A. Fan, et al., “Fano resonances in content fused silica from 2 μm to 1 mm,” Appl. Opt., vol. 22, plasmonic nanoclusters: geometrical and chemical tunability,” pp. 4069–4072, 1983. Nano Lett., vol. 10, pp. 3184–3189, 2010. [187] F. V. Ignatovich and L. Novotny, “Real-time and background- [204] C. Sönnichsen and A. P. Alivisatos, “Gold nanorods as novel free detection of nanoscale particles,” Phys. Rev. Lett., vol. 96, nonbleaching plasmon-based orientation sensors for no. 1, p. 013901, 2006. polarized single-particle microscopy,” Nano Lett., vol. 5, [188] C.-Y. Cheng, Y.-H. Liao, and C.-L. Hsieh, “High-speed imaging pp. 301–304, 2005. and tracking of very small single nanoparticles by contrast [205] S. Chandel, J. Soni, S. K. Ray, et al., “Complete polarization enhanced microscopy,” Nanoscale, vol. 11, pp. 568–577, 2019. characterization of single plasmonic nanoparticle enabled by a [189] K. J. Mickolajczyk, N. C. Deffenbaugh, J. Ortega Arroyo, novel Dark-field Mueller matrix spectroscopy system,” Sci. J. Andrecka, P. Kukura, and W. O. Hancock, “Kinetics of Rep., vol. 6, 2016, https://doi.org/10.1038/srep26466. nucleotide-dependent structural transitions in the kinesin-1 [206] K. Marchuk, J. W. Ha, and N. Fang, “Three-dimensional high- hydrolysis cycle,” Proc. Natl. Acad. Sci. U.S.A., vol. 112, resolution rotational tracking with superlocalization reveals p. E7186, 2015. conformations of surface-bound anisotropic nanoparticles,” [190] Y.-H. Lin, W.-L. Chang, and C.-L. Hsieh, “Shot-noise limited Nano Lett., vol. 13, pp. 1245–1250, 2013. localization of single 20 nm gold particles with nanometer [207] L. Shao, Z. J. Yang, D. Andren, P. Johansson, and M. Kall, “Gold spatial precision within microseconds,” Opt. Express, vol. 22, nanorod rotary motors driven by resonant light scattering,” pp. 9159–9170, 2014. ACS Nano, vol. 9, pp. 12542–12551, 2015. [191] M. De Brabander, G. Geuens, R. Nuydens, M. Moeremans, and [208] E. Collett, “Description of polarization in classical physics,” J. De Mey, “Microtubule-dependent intracellular motility Am. J. Phys., vol. 36, pp. 713–725, 1968. investigated with nanometer particle video ultramicroscopy [209] I. Sersic, C. Tuambilangana, and A. F. Koenderink, “Fourier (nanovid ultramicroscopy),” Ann. N. Y. Acad. Sci., vol. 466, microscopy of single plasmonic scatterers,” New J. Phys., vol. pp. 666–668, 1986. 13, 2011, https://doi.org/10.1088/1367-2630/13/8/083019. [192] L. Zhang, S. M. Christensen, P. M. Bendix, V. K. Bhatia, S. Loft, [210] M. Cotrufo, C. I. Osorio, and A. F. Koenderink, “Spin-dependent and D. Stamou, “Interferometric detection of single gold emission from arrays of planar chiral nanoantennas due to nanoparticles calibrated against TEM size distributions,” lattice and localized plasmon resonances,” ACS Nano, vol. 10, Small, vol. 11, pp. 3550–3555, 2015. pp. 3389–3397, 2016. [193] P. Stoller, V. Jacobsen, and V. Sandoghdar, “Measurement of [211] A. Mohtashami, C. I. Osorio, and A. F. Koenderink, “Angle- the complex dielectric constant of a single gold nanoparticle,” resolved polarimetry of antenna-mediated fluorescence,” Opt. Lett., vol. 31, pp. 2474–2476, 2006. Phys. Rev. Appl., vol. 4, 2015, https://doi.org/10.1103/ [194] I.-B. Lee, H.-M. Moon, J.-H. Joo, K.-H. Kim, S.-C. Hong, and physrevapplied.4.054014. M. Cho, “Interferometric scattering microscopy with [212] P. Pachidis, B. M. Cote, and V. E. Ferry, “Tuning the polarization polarization-selective dual detection scheme: capturing the and directionality of photoluminescence of achiral quantum orientational information of anisotropic nanometric objects,” dot films with chiral nanorod dimer arrays: implications for ACS Photonics, vol. 5, pp. 797–804, 2018. luminescent applications,” ACS Appl. Nano Mater., vol. 2, [195] R. W. Taylor and V. Sandoghdar, “Interferometric scattering pp. 5681–5687, 2019. microscopy: seeing single nanoparticles and molecules via [213] E. Collett, “Mueller-Stokes matrix formulation of fresnels Rayleigh scattering,” Nano Lett., vol. 19, pp. 4827–4835, 2019. equations,” Am. J. Phys., vol. 39, pp. 517–528, 1971. [196] G. Young and P. Kukura, “Interferometric scattering [214] S. Chandel, A. K. Singh, A. Agrawal, et al., “Mueller matrix microscopy,” Annu. Rev. Phys. Chem., vol. 70, pp. 301–322, spectroscopy of fano resonance in plasmonic oligomers,” Opt. 2019. Commun., vol. 432, pp. 84–90, 2019. [197] K. W. Smith, H. Q. Zhao, H. Zhang, et al., “Chiral and achiral [215] B. Foerster, A. Joplin, K. Kaefer, S. Celiksoy, S. Link, and nanodumbbell dimers: the effect of geometry on plasmonic C. Sönnichsen, “Chemical interface damping depends on properties,” ACS Nano, vol. 10, pp. 6180–6188, 2016. electrons reaching the surface,” ACS Nano, vol. 11, [198] J. Karst, N. H. Cho, H. Kim, et al., “Chiral scatterometry on pp. 2886–2893, 2017. chemically synthesized single plasmonic nanoparticles,” ACS [216] A. Kumar, E. Villarreal, X. Zhang, and E. Ringe, “Micro- Nano, vol. 13, pp. 8659–8668, 2019. Extinction Spectroscopy (MExS): A versatile optical [199] L.-Y. Wang, K. W. Smith, S. Dominguez-Medina, et al., “Circular characterization technique,” Adv. Struct. Chem. Imag., vol. 4, differential scattering of single chiral self-assembled gold 2018, https://doi.org/10.1186/s40679-018-0057-6. nanorod dimers,” ACS Photonics, vol. 2, pp. 1602–1610, 2015. [217] C. P. Byers, B. S. Hoener, W.-S. Chang, M. Yorulmaz, S. Link, and [200] J. M. Slocik, A. O. Govorov, and R. R. Naik, “Plasmonic circular C. F. Landes, “Single-particle spectroscopy reveals dichroism of peptide-functionalized gold nanoparticles,” Nano heterogeneity in electrochemical tuning of the localized surface Lett., vol. 11, pp. 701–705, 2011. plasmon,” J. Phys. Chem. B,vol.118,pp.14047–14055, 2014. [201] C. I. Osorio, A. Mohtashami, and A. F. Koenderink, “K-space [218] M. A. Beuwer, M. W. J. Prins, and P. Zijlstra, “Stochastic protein polarimetry of bullseye plasmon antennas,” Sci. Rep., vol. 5, interactions monitored by hundreds of single-molecule 2015, https://doi.org/10.1038/srep09966. plasmonic biosensors,” Nano Lett., vol. 15, pp. 3507–3511, 2015. 1654 A. Al-Zubeidi et al.: Single-particle scattering spectroscopy

[219] W. Ye, S. Celiksoy, A. Jakab, et al., “Plasmonic nanosensors [236] C. M. Sweeney, C. L. Nehl, W. Hasan, et al., “Three-channel reveal a height dependence of MinDE protein oscillations on spectrometer for wide-field imaging of anisotropic plasmonic membrane features,” J. Am. Chem. Soc., vol. 140, nanoparticles,” J. Phys. Chem. C,vol.115,pp.15933–15937, 2011. pp. 17901–17906, 2018. [237] M. Ouyang and D. Di Carlo, “Nanoplasmonic swarm biosensing [220] G. X. Rong, H. Y. Wang, L. R. Skewis, and B. M. Reinhard, using single nanoparticle colorimetry,” Biosens. Bioelectron., “Resolving sub-diffraction limit encounters in nanoparticle vol. 132, pp. 162–170, 2019. tracking using live cell plasmon coupling microscopy,” Nano [238] J. Zhou, T. Yang, W. He, Z. Y. Pan, and C. Z. Huang, “A galvanic Lett., vol. 8, pp. 3386–3393, 2008. exchange process visualized on single silver nanoparticles via [221] L. Lermusiaux, V. Maillard, and S. Bidault, “Widefield spectral dark-field microscopy imaging,” Nanoscale, vol. 10, monitoring of nanometer distance changes in DNA-templated pp. 12805–12812, 2018. plasmon rulers,” ACS Nano, vol. 9, pp. 978–990, 2015. [239] J. Zhou, G. Lei, L. L. Zheng, P. F. Gao, and C. Z. Huang, “HSI [222] T.-Y. Tseng, P.-J. Lai, and K.-B. Sung, “High-throughput colour-coded analysis of scattered light of single plasmonic detection of immobilized plasmonic nanoparticles by a nanoparticles,” Nanoscale, vol. 8, pp. 11467–11471, 2016. hyperspectral imaging system based on Fourier transform [240] Y. Liu, J. Ling, and C. Z. Huang, “Individually color-coded spectrometry,” Opt. Express, vol. 19, pp. 1291–1300, 2011. plasmonic nanoparticles for RGB analysis,” Chem. Commun., [223] D. Zopf, J. Jatschka, A. Dathe, N. Jahr, W. Fritzsche, and vol. 47, pp. 8121–8123, 2011. O. Stranik, “Hyperspectral imaging of plasmon resonances in [241]. J. Wang, T. Xue, J. T. Barron, and J. Chen, “Stereoscopic dark metallic nanoparticles,” Biosens. Bioelectron., vol. 81, flash for low-light photography,” in 2019 IEEE International pp. 287–293, 2016. Conference on Computational Photography (ICCP), Tokyo, [224] W.-S. Chang, L. S. Slaughter, B. P. Khanal, P. Manna, Japan, 2019, https://doi.org/10.1109/iccphot.2019.8747337. E. R. Zubarev, and S. Link, “One-dimensional coupling of gold [242] J. Becker, O. Schubert, and C. Sönnichsen, “Gold nanoparticle nanoparticle plasmons in self-assembled ring growth monitored in situ using a novel fast optical single- superstructures,” Nano Lett., vol. 9, pp. 1152–1157, 2009. particle spectroscopy method,” Nano Lett., vol. 7, [225] C. Rosman, J. Prasad, A. Neiser, A. Henkel, J. Edgar, and pp. 1664–1669, 2007. C. Sönnichsen, “Multiplexed plasmon sensor for rapid label-free [243] K. H. Chen, J. Hobley, Y. L. Foo, and X. Su, “Wide-field single analyte detection,” Nano Lett., vol. 13, pp. 3243–3247, 2013. metal nanoparticle spectroscopy for high throughput localized [226] S. A. Lee and J. S. Biteen, “Spectral reshaping of single dye surface plasmon resonance sensing,” Lab Chip, vol. 11, molecules coupled to single plasmonic nanoparticles,” J. Phys. pp. 1895–1901, 2011. Chem. Lett., vol. 10, pp. 5764–5769, 2019. [244] J. Cheng, Y. Liu, X. Cheng, Y. He, and E. S. Yeung, “Real time [227] I. V. Novikov, C. R. Smallwood, Y. Gong, et al., “Multimodal observation of chemical reactions of individual metal hyperspectral optical microscopy,” Chem. Phys., vol. 498, nanoparticles with high-throughput single molecule spectral pp. 25–32, 2017. microscopy,” Anal. Chem., vol. 82, pp. 8744–8749, 2010. [228] P. Z. El-Khoury, A. G. Joly, and W. P. Hess, “Hyperspectral dark [245] X. Liu, Q. Zhang, Y. Tu, W. Zhao, and H. Gai, “Single gold field optical microscopy of single silver nanospheres,” J. Phys. nanoparticle localized surface plasmon resonance spectral Chem. C, vol. 120, pp. 7295–7298, 2016. imaging for quantifying binding constant of carbohydrate– [229] C. Flatebo, S. S. E. Collins, B. S. Hoener, Y.-Y. Cai, S. Link, and protein interaction,” Anal. Chem., vol. 85, pp. 11851–11857, C. F. Landes, “Electrodissolution inhibition of gold nanorods with 2013. oxoanions,” J. Phys. Chem. C, vol. 123, pp. 13983–13992, 2019. [246] W.-S Tsai and Y.-C. Lin, “High-throughput optical biosensing [230] F. Yesilkoy, E. R. Arvelo, Y. Jahani, et al., “Ultrasensitive arrays detection using white light fourier transform method,” in hyperspectral imaging and biodetection enabled by dielectric 2014 Conference on Lasers and Electro-Optics (CLEO) – Laser metasurfaces,” Nat. Photonics, vol. 13, pp. 390–396, 2019. Science to Photonic Applications, San Jose, California, United [231] J. Juan-Colas, I. S. Hitchcock, M. Coles, S. Johnson, and States, 2014, https://doi.org/10.1364/CLEO_AT.2014.JTh2A.93. T. F. Krauss, “Quantifying single-cell secretion in real time [247] T. Tatsuma and H. Nishi, “Plasmonic hole ejection involved in using resonant hyperspectral imaging,” P. Nat.l Acad. Sci. U. S. plasmon-induced charge separation,” Nanoscale Horiz., vol. 5, A., vol. 115, pp. 13204–13209, 2018. pp. 597–606, 2020. [232] H. Y. Wang, G. X. Rong, B. Yan, L. L. Yang, and B. M. Reinhard, [248] C. Zhan, Z.-Y. Wang, X.-G. Zhang, et al., “Interfacial “Optical sizing of immunolabel clusters through multispectral construction of plasmonic nanostructures for the utilization of plasmon coupling microscopy,” Nano Lett., vol. 11, the plasmon-excited electrons and holes,” J. Am. Chem. Soc., pp. 498–504, 2011. vol. 141, pp. 8053–8057, 2019. [233] S. Kim, J. E. Park, W. Hwang, et al., “Optokinetically encoded [249] X. Cao, G. Lei, J. J. Feng, Q. Pan, X. D. Wen, and Y. He, “A novel nanoprobe-based multiplexing strategy for MicroRNA color modulation analysis strategy through tunable multiband profiling,” J. Am. Chem. Soc., vol. 139, pp. 3558–3566, 2017. laser for nanoparticle identification and evaluation,” Anal. [234] G. X. Rong, H. Y. Wang, and B. M. Reinhard, “Insights from a Chem., vol. 90, pp. 2501–2507, 2018. nanoparticle minuet: two-dimensional membrane profiling [250] P. Saha, J. W. Hill, J. D. Walmsley, and C. M. Hill, “Probing through silver plasmon ruler tracking,” Nano Lett., vol. 10, electrocatalysis at individual Au nanorods via correlated pp. 230–238, 2010. optical and electrochemical measurements,” Anal. Chem., [235] L. X. Wu and B. M. Reinhard, “Probing subdiffraction limit vol. 90, pp. 12832–12839, 2018. separations with plasmon coupling microscopy: concepts and [251] B. T. O’Callahan, K.-D. Park, I. V. Novikova, et al., “In liquid applications,” Chem. Soc. Rev., vol. 43, pp. 3884–3897, 2014. infrared scattering scanning near-field optical microscopy for A. Al-Zubeidi et al.: Single-particle scattering spectroscopy 1655

chemical and biological nanoimaging,” Nano Lett., vol. 20, [260] J. Zhou, S. R. Panikkanvalappil, S. Kang, et al., “Enhanced pp. 4497–4504, 2020. electrochemical dark-field scattering modulation on a single [252] S. Hu, J. Yi, Y.-J. Zhang, et al., “Observing atomic layer hybrid core–shell nanostructure,” J. Phys. Chem. C, vol. 123, electrodeposition on single nanocrystals surface by dark field pp. 28343–28352, 2019. spectroscopy,” Nat. Commun., vol. 11, p. 2518, 2020. [261] W. Ye, M. Götz, S. Celiksoy, et al., “Conformational dynamics of [253] E. Pensa, J. Gargiulo, A. Lauri, S. Schlücker, E. Cortés, and a single protein monitored for 24 h at video rate,” Nano Lett., S. A. Maier, “Spectral screening of the energy of hot holes over vol. 18, pp. 6633–6637, 2018. a particle plasmon resonance,” Nano Lett., vol. 19, [262] C. P. Byers, H. Zhang, D. F. Swearer, et al., “From tunable core- pp. 1867–1874, 2019. shell nanoparticles to plasmonic drawbridges: active control of [254] S.-C. Huang, J.-Z. Ye, X.-R. Shen, et al., “Electrochemical tip- nanoparticle optical properties,” Sci. Adv., vol. 1, p. e1500988, enhanced Raman spectroscopy with improved sensitivity 2015. enabled by a water immersion objective,” Anal. Chem., vol. 91, [263] P. A. Santi, “Light sheet fluorescence microscopy: A review,” pp. 11092–11097, 2019. J. Histochem. Cytochem., vol. 59, pp. 129–138, 2011. [255] S. Simoncelli, E. L. Pensa, T. Brick, et al., “Monitoring [264] A.-K. Gustavsson, P. N. Petrov, and W. E. Moerner, “Light sheet plasmonic hot-carrier chemical reactions at the single particle approaches for improved precision in 3D localization-based level,” Faraday Discuss., vol. 214, pp. 73–87, 2019. super-resolution imaging in mammalian cells [Invited],” Opt. [256] A. Carattino, S. Khatua, and M. Orrit, “In situ tuning of gold Express,vol.26,pp.13122–13147, 2018. nanorod plasmon through oxidative cyanide etching,” Phys. [265] A. K. Gustavsson, P. N. Petrov, M. Y. Lee, Y. Shechtman, Chem. Chem. Phys., vol. 18, pp. 15619–15624, 2016. andW.E.Moerner,“3D single-molecule super-resolution [257] V. Thambi, A. Kar, P. Ghosh, and S. Khatua, “Light-controlled microscopy with a tilted light sheet,” Nat. Commun., in situ bidirectional tuning and monitoring of gold nanorod vol. 9, 2018, https://doi.org/10.1038/s41467-017- plasmon via oxidative etching with FeCl3,” J. Phys. Chem. C, 02563-4. vol. 122, pp. 24885–24890, 2018. [266] L. L. Kegel, S.-S. Kim, B. Mizaikoff, C. Kranz, and K. S. Booksh, [258] M. Chirea, S. S. E. Collins, X. Wei, and P. Mulvaney, “Position dependent plasmonic interaction between a single “Spectroelectrochemistry of silver deposition on single gold nanoparticle and a nanohole array,” Plasmonics, vol. 9, nanocrystals,” J. Phys. Chem. Lett., vol. 5, pp. 4331–4335, 2014. pp. 1229–1237, 2014. [259] T. Miyazaki, R. Hasegawa, H. Yamaguchi, et al., “Electrical [267] J. D. Walmsley, J. W. Hill, P. Saha, and C. M. Hill, “Probing

control of plasmon resonance of gold nanoparticles using electrocatalytic CO2 reduction at individual Cu nanostructures electrochemical oxidation,” J. Phys. Chem. C, vol. 113, via optically targeted electrochemical cell microscopy,” J. Anal. pp. 8484–8490, 2009. Test., vol. 3, pp. 140–149, 2019.