Optical trapping and manipulation for single-particle spectroscopy and microscopy

In the last two decades, developments in wet chemistry have dramatically increased the availability of colloidal nanoparticles with well-controlled sizes, shapes, and compositions.^1–3 Rational design and controlled synthesis of these nanoparticles have led to novel physical and chemical properties and applications.⁴ Fundamental investigations of nanoscale properties by optical spectroscopy and microscopy usually require exquisite control of the positions and orientations of single nanoparticles.^5,6 However, such investigation often relies on drop-casting and drying a colloidal nanoparticle solution on a substrate, then looking for suitable single nanoparticles or their assemblies formed by chance.^7–13 This procedure obviously lacks control and reproducibility. In addition, to best understand the biomedical, biophysical, and optomechanical properties of nanoparticles, it is often necessary to study the dynamics and properties of single nanoparticles in bulk liquid environments.^10,14,15 However, colloidal nanoparticles are normally associated with strong Brownian motion in solution and, thus, are difficult to control with precision. Optical tweezers offer a unique tool to address this problem.

The momentum of light can exert force on an object even though the force is seldom detectable. Classical optical tweezers, invented by Ashkin in 1986,¹⁶ manifest this effect by using a focused laser beam to trap microscopic objects ranging from micrometers to angstroms in size.¹⁷ This technique has led to remarkable breakthroughs in cooling atoms^18–20 and studying biological systems,^21–23 e.g., by trapping microparticles that can serve as force transducers and/or reporters to investigate biomolecules. Computer-generated holography further increased the capacity of optical tweezers by shaping a laser beam into structured optical fields,²⁴ which provide more freedom for optical trapping and manipulation. Classical and holographic optical tweezers have been applied to study a variety of particles, including atoms, molecules, colloidal particles, and living cells.^17,21,25 By combining optical forces with other types of forces, such as the drag force in a microfluidic flow,²⁶ various optical techniques for trapping, assembling, sorting, and transporting microscopic objects have been developed.^27,28

Although Ashkin’s seminal paper reported optical trapping of silica particles with nominal diameters of ∼25 nm,¹⁶ optical tweezers have been predominantly used to trap microparticles.²³ Optical trapping and manipulation of metal and semiconductor nanoparticles started to attract more attention in the 2000s, largely due to the booming studies on plasmonics and nanophotonics.^25,29 These studies have provided a better understanding of the roles of optical forces in controlling the position and orientation of nanostructures, which can facilitate the investigation of their properties in different environments by optical microscopy and spectroscopy. In this perspective, we will introduce the mechanism of optical trapping and manipulation of single particles and discuss how that can be combined with different spectroscopy and microscopy techniques to enable single-particle studies and applications.

In the cross section of the beam, the intensity gradient force is the main optical force in classical optical tweezers, i.e., the horizontal component of F_grad in Fig. 1. The force comes from the intrinsic intensity profile of the beam, which is normally a two-dimensional (2D) Gaussian profile, so F_grad exists along the radial direction in the cylindrical coordinate system (ρ, φ, z) and will bring the particle to the cylindrical axis. The polarization of light can also create optical forces in the cross section of the beam. Early study has shown that a circularly polarized laser beam can induce the spin of a microparticle due to the transfer of spin angular momentum.³² More recently, optical spinning of nanoparticles has triggered strong interest as the rotation speed can be record-high, e.g., on the order of kHz in liquid³³ and GHz in vacuum.^34,35 For aspherical particles, linear polarization can also generate optical force to rotate the particles in certain directions,³² which provides a unique route to control the orientation of single nanostructures.^36,37 Generally, the polarization force only applies to the azimuthal direction of the polar coordinates. Recent studies further revealed a new type of optical force arising from the phase gradient of light.^38–40 When the laser beam has an extended profile in its cross section, the phase gradient in this plane can redirect the axial radiation pressure of a laser beam to its transverse plane.³⁸ A similar transverse scattering force can also arise in a circularly polarized optical field with nonuniform helicity,³¹ which has been described by the third term of Eq. (1). Therefore, the optical force in the beam cross section can be generated by intensity, phase, and polarization of light, where F_ρ is contributed by both intensity and phase gradients, and F_φ is mainly contributed by polarization. By tuning the laser wavelength, intensity, phase, and polarization, one may generate suitable optical forces to control the position and orientation of a microscopic particle.

Table I summarizes the optical trapping of representative metal and dielectric nanoparticles, where the trap stiffness is determined by the optical forces. The trap stiffness in the table mainly represents that in the cross section of the beam, while the axial trap stiffness is normally smaller due to the axial F_scat.⁴¹ The trap stiffness is proportional to the volume of the particle when the F_scat is small, e.g., for Au nanoparticles with diameters less than 100 nm,⁴² and polystyrene beads with diameters less than 500 nm.⁴¹ The trapping efficiency of a metal nanoparticle is generally higher than a dielectric nanoparticle of the same size due to the larger polarizability of metals, e.g., the trap stiffness for a 100 nm Au nanoparticle is six-fold of that for a 110 nm polystyrene nanoparticle.⁴³ However, metal particles will become more difficult to trap in the submicrometer size range, which will be discussed in Sec. II B.

Traditionally, optical trapping means stable positional control of a particle in a bulk liquid environment,¹⁶ i.e., 3D optical trapping. However, optical trapping of particles near a substrate surface is very common in recent literature,^25,29,56 which is 2D optical trapping but was often not clearly stated. It is thus worth clarifying the differences between the two cases.

In 3D optical trapping, F_z = 0 and F_ρ = 0, and the trapping potential well is deep enough to restrict the escape of the particle under thermal fluctuations. The trapping potential normally needs to be at least 10k_BT, where k_BT is the Boltzmann constant and T is the absolute temperature of the particle.¹⁶ However, sometimes it is challenging to satisfy these criteria. First, from Eqs. (2) to (4), we can see that the magnitude of gradient force scales down with the particle volume. This becomes an issue for small nanoparticles (e.g., when r < 10 nm), especially in dielectric materials, where the trapping potential well is usually too shallow to resist thermal fluctuations (note that potential energy is the integral of force over distance). The trapping potential can be enhanced by increasing the laser intensity, but at the risk of damaging the particles due to laser heating. The heating can also increase the thermal fluctuations and cause other issues, e.g., the formation of bubbles. So far, optical trapping of small nanoparticles mostly relies on plasmon nano-optical tweezers⁵⁷ and their derivatives, such as the opto-thermoelectric nanotweezers⁵⁸ and opto-thermo-electrohydrodynamic tweezers,⁵⁹ but the manipulation is largely confined to the surface of plasmonic substrates. Second, for large particles, the scattering force increases quickly since C_sca scales with r⁶. In particular, for plasmonic metal nanoparticles, the scattering cross sections often increase dramatically at larger sizes (e.g., when r > 100 nm) or with aspherical shapes (e.g., nanocubes and nanowires) due to the arising of multipole plasmon resonances. As a result, the F_scat becomes too strong and can hardly be compensated by the F_grad, making 3D optical trapping of certain nanostructures (e.g., silver nanowires⁶⁰) impossible by classical optical tweezers. Third, when holographic optical tweezers generate structured light (e.g., vortex beam⁶¹ and Bessel beam³⁷) for optical manipulation, the beam is no longer tightly focused along the beam propagation direction, and the light intensity of an extended optical pattern is much weaker than that of a focused Gaussian beam. The two factors often make the axial F_grad too small to trap nanoparticles.

For these reasons, 2D optical trapping is very common in literature. In this case, the gradient force is not strong enough in the beam propagation direction, so a particle is simply pressed against a substrate surface by the scattering force.^{33,36,37,39,62–64} This can cause permanent adhesion of the particle to the substrate due to van der Waals interactions, which is favorable for optical printing of a nanoparticle on the substrate,^65,66 but unfavorable for optical manipulation since the optical force can no longer move the nanoparticle. A simple solution to this problem is to induce surface charges on the substrate (e.g., treating a cover slip with UV/ozone or plasma) and coat the nanoparticles with a surfactant of the same charge. The strong scattering force will then be balanced by the electrostatic repulsive force between the charged surfaces of nanoparticle and substrate,⁶⁷ leading to a small separation to prevent surface adhesion.

Raman spectroscopy can provide rich structural information of materials based on their characteristic vibrational modes. However, the Raman scattering associated with those vibrational modes is normally very weak. The localized surface plasmon resonances (SPRs) of metal nanostructures can significantly enhance the electromagnetic fields near the surfaces of the nanostructures, enabling surface-enhanced Raman spectroscopy (SERS) for detecting various molecules. These metal nanostructures, often referred to as SERS substrates, are not limited to the widely used rough metal surfaces and aggregated nanoparticle layers, but also include colloidal nanoparticles in solutions.⁶⁸ The latter is particularly important for studying biomolecules and living systems in their original environments, especially for detecting nanoparticle–biomolecular (such as protein and polycyclic aromatic hydrocarbons) complexes.^14,69 However, the solution-based SERS can hardly obtain consistent and reproducible spectra because of the strong Brownian motion of nanoparticles. The diffusion time of a single nanoparticle in the detection region of the conventional confocal Raman spectrometer is normally on the order of microseconds (∼μs). The short diffusion time does not permit the capturing of strong SERS signals from the single nanoparticle in a relatively long signal collection time (>1 s).⁷⁰ 

Optical trapping provides a promising method to overcome this challenge since a nanoparticle can be stably trapped in solution for seconds or longer.⁵⁶ This will enable unique applications, such as using trapped nanoparticles as intracellular probes to detect biomolecules and sequencing single DNAs in cells.^71,72 SERS of single molecules on single nanoparticles was achieved in 1997,⁷³ where individual silver colloidal nanoparticles dried on a cover slip allowed the detection of single rhodamine 6G molecules. The ultrahigh sensitivity indicated the possibility of SERS with single nanoparticles in an optical trap. This possibility was examined theoretically by Calander and Willander in 2002,⁷¹ who predicted that the SPRs at laser-illuminated metal tips or nanoparticles can trap molecules and allow single-molecule Raman spectroscopy. 2D optical trapping and SERS of silver nanoparticles was studied in 2006, but SERS signals were only detected from nanoparticle pairs and not from single nanoparticles.⁶² 

Optical tweezers have been more successful in making aggregated colloidal nanoparticles for SERS, where plasmonic hotspots can arise in the gaps of nanoparticles that provide very strong electromagnetic field enhancement. A SERS substrate can be fabricated in situ either on a cover slip or in solution by accumulating multiple nanoparticles from the solution into an optical trap, providing opportunities for solution-phase SERS sensing. This method has been applied to detect biomolecules. For example, Tanaka et al. amplified the SERS signals of pseudoisocyanine molecules by absorbing the molecules on Ag nanoparticles and aggregated the nanoparticles in a 3D optical trap.⁷⁴ Bernatová et al. used a focus laser beam to aggregate gold nanorod-bovine serum albumin complexes on a cover slip surface inside a microfluidic channel, and achieved ultrasensitive detection of the proteins down to pM concentration (see Fig. 2).⁷⁵ Similarly, Yin et al. used optical tweezers to trap Ag nanoaggregates and rapidly detected 4-aminothiophenol by SERS.⁷⁶ Ag nanoparticles have also been trapped with bacillus subtilis bacteria together for enhanced Raman spectroscopy.⁷⁷ Besides biomolecules, SERS with optically trapped nanoaggregates is useful in the detection of toxicants and pollutants, such as dyes, polycyclic aromatic hydrocarbons, and chemical toxins, for environmental monitoring and food quality control. For example, Huai et al. combined optical tweezers with SERS spectroscopy to monitor marine biotoxin down to 16 nM concentration, where silver nanoparticles were dissolved in a saxitoxin solution and SERS signals of saxitoxin adsorbed on optically trapped nanoparticles could be collected in just 2 s.⁷⁸ 

While it is still challenging to realize SERS with single nanoparticles in 3D optical trapping, it is feasible to perform single-particle SERS on the micrometer level by 3D optical trapping of single microparticles with nanostructured surfaces. Such microparticles can be aggregated by plasmonic nanoparticles around spherical templates, e.g., attaching Ag or Au nanoparticles to the surface of a silane-monolayer-coated silica bead, which can be placed and scanned near living cells [see Fig. 3(a)].⁷⁹ Stetciura et al. also designed composite SERS-active microparticles that can be optically trapped as probes for detecting cellular composition.⁸⁰ The composite microparticles were formed by coating silica microparticles with layers of silver or gold nanoparticles and astralen (a Raman active marker). The researchers demonstrated that a composite microparticle can be optically trapped and placed onto the surface of an L929 mouse fibroblast cell, and the many microspheres engulfed by a cell can be used to map the intracellular composition based on the spatial distribution of SERS signals. However, no optical trapping and manipulation was performed on the microspheres in the cells in this study. It is very likely that the microspheres with large sizes cannot be freely manipulated inside a cell due to steric hindrance by the cytoskeleton meshwork. Later, Spadaro et al. synthesized hollow microparticles with Au nanoparticles embedded in the inner wall of mesoporous silica shells (i.e., plasmonic mesocapsules), and demonstrated that a single mesocapsule in an optical trap can serve as a local SERS probe to detect methylene blue molecules at 10⁻⁵M concentration in a liquid environment.⁸¹ Very recently, Dai et al. used optical tweezers to tune the separation of two Ag nanoparticle-coated silica microbeads and, thus, control the hotspots between the two beads, leading to tunable SERS enhancements with single-molecule level sensitivity [Fig. 3(b)].⁸² 

The combination of optical tweezers and single-molecule fluorescence has been of great interest to the biophysics community.^23,83,84 The enhanced photobleaching of dye molecules in the intense optical trap was a problem for the combination of two techniques, but the problem was largely solved by optimizing the experimental systems and conditions in the early 2000s.^85–88 The optical traps can generate piconewton forces to induce biomechanical transitions of macromolecular complexes (e.g., DNA), and their conformation and structure changes can be revealed by single-molecule fluorescence of fluorescent probes attached to molecular subdomains. The time trace of the fluorescence intensity can be monitored by a photodetector or a camera. The intensity change has been used to study the photobleaching of fluorophores coated on a silica bead trapped by optical tweezers,⁸⁶ and the optical-force-induced strand separation of a dye-labeled double-stranded DNA.⁸⁷ However, most of the studies in the field applied optical tweezers to assist single-molecule microscopy, which we will discuss in Sec. IV.

The luminescence of some nanomaterials is sensitive to temperature, making it possible to perform nanothermometry with single nanoparticles. Nanothermometers are valuable in cell biology, especially preclinical diagnostics and therapeutics.⁸⁹ For example, nanothermometers can be used to measure intracellular temperature in the early detection of tumors, and they can also benefit photothermal therapy for precise treatment of cancer and minimization of the damage to normal cells.^90–92 Traditional nanothermometers are based on fluorescent molecules, whose performance may be influenced by the overlapping of their emission with the tissue autofluorescence. Lanthanide-doped upconversion nanoparticles are new types of luminescent materials for nanothermometry.^89,93 Upconversion nanoparticles, such as NaYF₄ doped with Yb³⁺ and Er³⁺ ions, have large anti-Stokes shifts to allow emission in the visible spectrum under near-infrared (NIR) excitation. These nanoparticles can serve as nanothermometers because their emission peaks are affected by temperature in several ways, including changes in emission intensity, peak width, and luminescence lifetime. The intensity change is the easiest to measure, but that can also be affected by other factors, especially the fluctuation of excitation power. This limitation can be overcome by calculating the intensity ratio of two adjacent emission peaks, where the thermal dependence of the intensity ratio is very stable.⁹⁴ This technique was applied to measure the intracellular temperature associated with thermally induced death of HeLa cervical cancer cells, where the nanoparticles entered the cells by incubating the cells in the nanoparticle solution.⁹⁵ 

However, the incubation and uptake method lacks spatial control of the nanoparticles and, thus, can only passively measure the temperature at unspecified locations. Rodríguez‐Sevilla et al. demonstrated thermal scanning at the cellular level by an optically trapped upconverting fluorescent microparticle.⁹⁶ As shown in Fig. 4, they used optical tweezers to perform 3D manipulation and scanning of a single Yb/Er doped NaYF₄ microparticle in the surrounding medium of a HeLa cell and measured the temperature distribution using the ratio of two Er³⁺ emission bands centered at 525 and 550 nm. Upconversion microparticles were used in this study to generate enough optical forces but that limited the resolution. Upconversion nanoparticles will provide better resolution for thermometry, but nanothermometry with the use of optically trapped upconversion nanoparticles has not been reported.

Single-molecule fluorescence is naturally suitable for high-resolution imaging, so the combination of optical tweezers and single-molecule fluorescence can provide opportunities for both single-particle spectroscopy and microscopy. In the early studies, optical tweezers were combined with total internal reflection fluorescence microscopy and epifluorescence microscopy to study biomolecules, such as the interactions of RNA polymerase and DNA single molecules⁹⁷ and DNA strand separation.⁸⁷ A protocol to construct an optical trap on a commercial fluorescence optical microscope was also provided.⁹⁸ However, traditional fluorescence microscopy is limited by the diffraction of light, so there is a strong need to improve the resolution of optical microscopy beyond the diffraction limit. Recently, confocal and stimulated emission depletion (STED) fluorescence microscopy has been combined with optical tweezers, allowing super-resolution imaging of proteins at high coverage density on DNA [Fig. 5(a)].⁹⁹ In this study, a double-stranded DNA linked between two traditional microspheres was controlled by two optical traps to allow super-resolution imaging of proteins on the DNA.⁹⁹ A similar strategy has been used to perform nanoscopy of non-adherent bacterial cells, where the position and orientation of the cells can be directly controlled by multiple optical traps to extend acquisition time and enable super-resolution far-field fluorescence microscopy.¹⁰⁰ These studies share the same feature that the optical traps are used to control the object of study (either directly or via force transducers) rather than the fluorescent probes.

Currently, most super-resolution microscopy uses fluorescent proteins and organic dyes as probes, which are hard to be trapped directly by optical tweezers. In addition, nearly all previous studies used submicrometer- or micrometer-sized silica or latex particles as force transducers. The big particles can ensure large trap stiffness at relatively low laser power, which can minimize the photodamage of biological samples but at the expense of spatial and temporal resolutions. The latter is affected by the large hydrodynamic drag of a big particle, which is not suitable for revealing the fast or small details in the mechanochemistry of cellular processes. Very recently, Sudhakar et al. addressed this issue by trapping Ge nanospheres as force transducers.¹⁰¹ Ge has a very high refractive index that leads to large polarizability, so even at small diameters of ∼70 nm, the nanoparticles can still allow piconewton force measurements. A classic example of such measurements is to detect the mechanical work against piconewton loads associated with the movement of kinesin motors along microtubules.¹⁰² The movement takes 8 nm steps coupled with each adenosine 5′-triphosphate (ATP) hydrolysis, which was first revealed by optical trapping experiments using 0.6 μm silica beads as force transducers in 1993.¹⁰² Re-examination of this process using the Ge nanoparticles revealed that each hydrolysis cycle is actually broken up into two 4-nm center-of-mass substeps,¹⁰¹ which opened a new temporal window to uncover hidden dynamics in molecular machines.

New types of inorganic probes such as semiconductor quantum dots and rare-earth-doped lanthanide nanoparticles have been developed for super-resolution microscopy.¹⁰³ They are typically brighter and more stable than organic probes for prolonged subcellular imaging, and they are better candidates for optical trapping. Liu et al. reported a super-resolution STED nanoscopy strategy using amplified stimulated emission in Yb/Tm co-doped NaYF₄ upconversion nanoparticles [Fig. 5(b)].¹⁰⁴ The nanoscopy relies on the metastable energy levels of doped Tm³⁺ ions, which can be excited by a 980 nm focused Gaussian beam to establish a population inversion on the metastable ³H₄ level. The 808 nm doughnut depletion laser then discharges the ³H₄ level in the tails of the Gaussian profile, allowing blue luminescence only at the center of the Gaussian spot. Using this strategy, a lateral resolution of 28 nm was achieved.¹⁰⁴ More types of super-resolution microscopies with upconversion nanoparticles have been explored in recent years, e.g., by using structured illumination microscopy and multiphoton (nonlinear) fluorescence microscopy.^105–108 Other types of lanthanide nanoparticles have also been developed for super-resolution microscopy. For example, Liang et al. synthesized Nd-doped NaYF₄ nanoparticles with downshifting luminescence at 850–900 nm, which showed zero photobleaching when illuminated with an 808 nm laser.¹⁰⁹ The nanoparticles thus allowed hours-long, autofluorescence-free STED super-resolution imaging in all-NIR windows with a lateral resolution of below 20 nm. So far, lanthanide nanoparticles have only been used as fixed luminescent labels for super-resolution microscopy. However, since these nanoparticles can be trapped and manipulated using optical tweezers,^54,55 it should be feasible to use them as movable probes and perform super-resolution microscopy with a single nanoparticle.

Scanning probe microscopy, such as the atomic force microscopy (AFM) and scanning electrochemical microscopy,^110,111 can also generate high-resolution images by scanning a probe over the sample surface. In fact, scanning probe microscopy often provides better resolution than super-resolution fluorescence microscopies. However, it is challenging to scan soft surfaces such as biological membranes without deforming the surfaces because they cannot apply enough force to displace the probe before deformation occurs. Photonic force microscopy may address this issue. Photonic force microscopy uses a freestanding particle as the probe and remotely controls it in solution by optical tweezers, allowing the probe to be displaced with minimal force. The probe is moved to scan the sample surface by translating the position of the laser trap or the sample holder in three dimensions. During the scanning, the displacements of the probe are recorded and used to reconstruct the surface profile of the sample. The probe position can be detected by the scattered or emitted light from the probe, and more advanced particle tracking techniques, such as the combination of a time-shared twin-optical trap and 3D interferometric particle tracking,¹¹² can further improve the scanning resolution. Compared to the atomic force microscopy, the photonic scanning probe is reconfigurable, and multiple probes may even work simultaneously by using an array of optical traps.

The concept of photonic scanning probe microscopy was first introduced in the early 1990s by Malmqvist and Hertz, who optically trapped single SiO₂ or KNbO₃ nanoparticles as probes to scan the surface profiles of solid samples.^113,114 Resolution of about 500 nm was demonstrated with the KNbO₃ nanoparticles, which emitted visible light by the second harmonic generation of the NIR trapping laser.¹¹⁴ Ghislain and Webb also developed the idea of photonic scanning-force microscope nearly at the same time.¹¹⁵ They used optical tweezers to trap and detect the small fluctuating forces on a nonspherical probe with a sharp tip scanning over a polymer surface. In 2007, a single KNbO₃ nanowire was used as the photonic scanning probe to scan subwavelength surface features of <100 nm.¹¹⁶ The resolution of scanning microscopy is limited by the probe size. Although nanowires naturally provide shape tips for photonic scanning probe microscopy, the relatively low stability of a nanowire in an optical trap limits the scanning resolution. Since the trapping force scales with the trapped volume, the force on a nanowire is normally weak due to the small portion of a nanowire in the optical trap. Strong fluctuations are thus caused by Brownian motion and other effects, such as the rotational motion arising from the transverse components of the radiation pressure on an optically trapped ultrathin Si nanowire.¹¹⁷ 

The trapping stability of photonic scanning probe microscopy can be enhanced by increasing the trapping volume of the probe while maintaining a sharp tip. For example, complex scanning probes with sharp tips have been fabricated by two-photon polymerization, which led to a lateral resolution of 200 nm and a depth resolution of ∼10 nm.¹¹⁸ More recently, Desgarceaux et al. developed a nanofabrication protocol to fabricate optically trappable quartz microparticles as scanning probes.¹¹⁹ As shown in Fig. 6, the probe is designed as a micrometer-sized truncated cone with a sharp tip, whose radius of curvature is down to 35 nm. This type of probe ensures stable optical trapping and, thus, high resolution. Images of hard surfaces with resolution below 80 nm were obtained, which were comparable to those scanned by AFM. More importantly, the new probe allows scanning and capturing submicrometer features on the soft membrane of living cells, e.g., knobs of ∼100 nm size on malaria-infected red blood cells, while other scanning techniques can hardly do without killing the cells.

Optical trapping and manipulation can control the position and orientation of a freestanding micro-/nanostructure, making it possible to perform single-particle spectroscopy and microscopy without fixing the particle to a substrate. This can facilitate the characterization of single particle properties.^52,120 More importantly, an optically trapped particle can serve as a probe to detect and scan the physical and chemical properties of its surrounding environment. Such nanoprobes will be particularly valuable for studying cellular biology, e.g., to accurately map the spatial distribution of intracellular temperature; measure the nonuniform temperature change induced by biochemical reactions in cytoplasm based on single-nanoparticle thermometry; and reveal nanoparticle interactions with cell membrane, organelle, and proteins during cellular uptake of nanoparticles based on single-nanoparticle SERS. However, challenges still exist to combine optical trapping and manipulation of nanoparticles with spectroscopy and microscopy, and new opportunities will emerge once these challenges are addressed.

Stable control of nanoparticles is the foundation for single-nanoparticle spectroscopy in solution, but 3D optical trapping of nanoparticles is challenging due to either small gradient force or large scattering force. The small gradient force issue may be tackled by computer-generated holography. Stronger intensity gradient can be created by shaping the Gaussian beam into a structured optical field, thus increasing the gradient force. The scattering force issue can be solved by balancing the radiation pressure in the beam propagation direction using two counter-propagating laser beams.¹²¹ In this geometry, the gradient force component dominates the optical trap, making it possible to trap nanoparticles at very low powers.

For dielectric nanoparticles, the weak gradient force is normally the limiting factor for stable optical trapping. In addition to making advanced optical traps, one can also tune the nanoparticle structure to enhance the gradient force, e.g., in one of our early studies, we coated thick silica shells on CdSe/ZnS quantum dots to increase their polarizability, so it was easier to trap these quantum dots.⁶⁴ Moreover, for nanoparticles with resonance absorption, the trapping stability can be further enhanced by tuning the laser wavelength close to the resonance wavelength. It is worth noting that optical forces, especially the scattering force, can be enhanced by tuning the laser wavelength to specific resonances of atoms and molecules as pointed out by Ashkin’s seminal paper on laser radiation pressure¹²¹ and several theoretical papers in the 1970s.^125–127 Successful experimental demonstrations^128,129 of this mechanism has led to wide studies of laser cooling and trapping of neutral atoms.¹⁹ Nearly two decades ago, two theoretical studies applied this concept to optical trapping of nanoparticles near resonance absorption, and predicted dramatic enhancement of optical trapping forces.^130,131 The resonance optical trapping has been demonstrated in fluorophore-labeled antibodies under resonant excitation of the fluorescent dye.¹³² The phenomenon has been also been observed in dye-doped polystyrene microparticles, either by directly tuning the wavelength of the trapping laser^133,134 or using a second excitation laser¹³⁵ close to the absorption band of dyes. More recently, this mechanism was used to enhance the optical trapping of upconversion nanoparticles by tuning the laser frequency to the dipole oscillation frequency of doped lanthanide ions.⁵⁵ The resonance optical trapping may provide additional opportunities to enhance the optical trapping of other nanoparticles (e.g., the Ge nanoparticles¹⁰¹) by coating them with resonant dyes or upconversion nanoparticles, meanwhile making it easy to visualize the nanoparticles. Fluorescent molecules may also aggregate into nanoparticles and be trapped in resonant excitation. For example, Fang et al. synthesized near-infrared fluorescence nanoparticles based on the hemicyanine structure HD-Br and used the nanoparticles for super-resolution imaging of mitochondria–lysosome interactions.¹³⁶ Since these organic nanoprobes have excellent biocompatibility and photostability, optical trapping of such nanoprobes will be useful for super-resolution microscopy of intracellular processes.

For noble metal nanoparticles, the gradient force can also be enhanced by tuning the laser wavelength slightly above the SPRs of the nanoparticles. This phenomenon was first observed by Pelton et al. in optical trapping of nanorods, where the nanorods could be stably trapped for several minutes when the laser was detuned to the long-wavelength side of the resonance.⁴⁶ It is worth noting that the laser wavelength for trapping plasmonic nanoparticles should be selected carefully. If the wavelength is below the resonance, the gradient force will be repulsive, and no trapping can happen.⁴⁶ In addition, for relatively larger plasmonic nanoparticles, the scattering force and thermal effects will be very strong if the laser wavelength is close to resonance, making the stable optical trapping impossible. In that case, the laser wavelength needs to be tuned away from the resonance.

So far, a clear demonstration of SERS based on 3D optical trapping of a single nanoparticle is still missing, but if that can be achieved, it would be interesting to use the nanoparticle as a probe to trap and detect single molecules in solution and to study cellular biology. Performing SERS with an optically trapped nanoparticle in bulk solution is associated with several challenges. First, strong SERS signals rely on local field enhancement, which normally requires the excitation wavelength close to SPR of the nanoparticle. However, resonant excitation of a plasmonic nanoparticle will increase its scattering and absorption cross sections and the corresponding forces, making the optical trapping unstable. Second, the increased absorption may also induce significant heating of the nanoparticle. The optothermal effects on a stationary nanoparticle may be useful. For example, a heated Au nanoparticle deposited on a substrate can serve as a thermoplasmonic trap to attract other plasmonic nanoparticles from the solution, forming an aggregate to enable single molecule SERS.¹³⁷ However, a heated nanoparticle in an optical trap will have stronger thermal fluctuation and be more likely to escape.¹³⁸ In addition, the heating may induce the photothermal release of molecules from the nanoparticle surface,¹³⁹ making it hard to record a stable Raman spectrum (the heating can be useful for controlled drug release though). Improving the trapping stability at a low trapping power using the aforementioned methods will alleviate these problems. Third, Ag and Au nanospheres have been most widely studied for optical trapping, but the Raman signal enhancement from a nanosphere is moderate. Plasmonic nanoparticles with complex shapes and structures, such as Au nanostars and nanocages,^140,141 are better candidates for single-particle SERS.⁶⁸ However, optical trapping behaviors of aspherical plasmonic nanostructures are more complex than nanospheres,^37,48,142 and more studies are needed in order to use them as optically trapped nanoprobes. In addition, tunable and optically trapped plasmonic nano-aggregates have also been fabricated by linking a number of 40 nm Au nanoparticles to single 80 nm Au nanoparticles with thermo-responsive polymers, forming core–satellite assemblies whose configuration can be adjusted by plasmonic heating of the nanoparticles.¹⁴³ These nano-aggregates could be candidates for tunable SERS-active nanoprobes, but the polymers inside the gaps between the nanoparticles may impact the measurements.

The combination of optical tweezers and nanoparticle luminescence will provide a new approach to perform nanothermometry using single-particle probes, especially single upconversion nanoparticles whose structure and composition can be tuned to achieve nanoscale ultrasensitive temperature sensing, e.g., by inducing lattice self-adaptation of core/shell upconversion nanoparticles.¹⁴⁴ However, it is generally difficult to trap upconversion nanoparticles unless using high laser powers, which are unfavorable for cellular studies. The recent discovery of enhanced optical forces by resonant optical trapping of upconversion nanoparticles solved this problem,⁵⁵ and we shall see the demonstration of 3D single-particle nanothermometry with optically trapped upconversion nanoparticles in the near future. In addition, other types of luminescent nanoparticles and thermal sensing techniques are worth studying for nanothermometry in optical traps, e.g., by measuring the fluorescence lifetime of semiconductor or halide perovskite nanoparticles,^145–148 or using the anti-Stokes emission from gold nanorods irradiated by a laser in resonance.¹⁴⁹ 

While optical traps can control the positions of individual particles, the continuous-wave lasers normally used for optical trapping cannot provide temporal information for spectroscopy. Coupling ultrafast lasers into the optical traps will provide new opportunities for time-resolved spectroscopy. It is technically feasible to integrate ultrafast lasers with optical tweezers, which was done a decade ago for studying the damping of acoustic vibrations of single gold nanoparticles in water.¹²⁰ Femtosecond lasers can be used to stimulate Raman scattering and enable time-resolved Raman spectroscopy with the pump–probe technique^150,151 and that can be further combined with SERS over plasmonic nanoparticles to perform surface-enhanced femtosecond stimulated Raman spectroscopy.¹⁵² Moreover, femtosecond lasers can be used to perform ultrafast Raman thermometry by comparing relative intensities of Stokes and anti-Stokes scattering of specific modes, and this technique has been used to study the role of plasmonic heating in plasmon-driven photocatalysis over Au nanoparticles on the picosecond time scale.¹⁵³ Ultrafast lasers can further enable other types of Raman spectroscopy, such as the coherent anti-Stokes Raman scattering (CARS) and stimulated Raman scattering (SRS),¹⁵⁴ which can be enhanced by plasmonic nanostructures. In a recent study, a single gold dumbbell was used as a SERS substrate for visualizing a single molecule vibration through time-resolved CARS.¹⁵⁵ However, all these studies were performed with stationary nanoparticles placed on 2D substrate surfaces.^152–154 Integration of these techniques with optical tweezers will provide more functionalities to optically trapped probes, where the probes can be moved and positioned on demand in a 3D space. Femtosecond lasers can also benefit microscopy, e.g., second harmonic generation imaging, CARS/SRS microscopy, and multiphoton excitation microscopy. Therefore, we envision that the integration of ultrafast lasers, optical tweezers, and state-of-the-art spectroscopy and microscopy techniques will provide enormous opportunities for single-particle studies.

This material is based upon work supported by the National Science Foundation under Grant No. 2131079. An acknowledgment is made to the Donors of the American Chemical Society Petroleum Research Fund for partial support of this research.

The authors have no conflicts to disclose.

Data sharing is not applicable to this article as no new data were created or analyzed in this study.