Optical trapping of optical nanoparticles: Fundamentals and applications

Plasmonic nanoparticles

The optical trapping forces experienced by plasmonic nanoparticles (PNPs) are usually larger than for dielectric nanoparticles of the same size due to their larger polarizability (7.33 × 10⁻¹⁷ cm³ for 36 nm diameter gold nanosphere)⁴³. Such superior polarizability obeys to the coupling of the electric field of light and the surface plasmons, i.e. the collective resonant oscillations of the electrons of the conduction band of the metal⁴⁴. Indeed, when the wavelength of incident light matches the plasmon resonance of the PNP, it may lead to enhanced light absorption, Rayleigh light scattering, and intense local field enhancement near the surface⁴⁴. PNPs’ large polarizability is not only wavelength-dependent, but it also varies with the particle shape, volume, frequency-dependent dielectric response of the metal, and refractive index of the environment. The optical gradient force results large and attractive when the metallic particle is trapped by light with a lower frequency than the plasmon resonance and repulsive if it has a larger frequency. Therefore, the selection of the optical trapping laser wavelength is critical for this type of nanomaterials. Another crucial point for this type of ONP is their inherent high light-to-heat conversion efficiency. When the collective motion of electrons is excited, the subsequent electron relaxation via Joule effect leads to heat generation. This may lead to deleterious Brownian thermal motion and reduce the stability in the optical trap. An increase in the local temperature may reduce the viscosity of the medium surrounding the PNP. This in turn reduces the hydrodynamic resistance acting on the PNP, allowing it to escape from the OT. Thus, local heating may interfere with the ongoing research or application^{44, 45}.

Despite the difficulties listed above, OT of PNPs has been achieved by different groups in a great variety of configurations. Typical applications of optically trapped metallic nanoparticles are based on: spinners⁴⁶, single-molecule surface-enhanced Raman spectroscopy (SERS)⁴⁷, particle-particle interaction (dimers)⁴⁸, and temperature sensing⁴⁹. In this section we limit the scope to overview the main results obtained in the last two of those applications, as they are the only ones based on the optical properties of the trapped PNPs. For that purpose, usually darkfield spectroscopy or imaging devices are coupled to the optical tweezer setup. That can be easily achieved by adding illumination with a cone of light and then spectrally analyzing the light scattered by the PNP or by performing hyperspectral imaging of the optical trap, respectively^{50, 51}. In that way, single PNP scattering spectra can be recorded^{52, 53}.

Temperature sensing with optically trapped plasmonic nanoparticles

Some nanoparticles, including plasmonic nanoparticles, behave as optical nanothermometers. An optical nanothermometer is a nanoparticle whose optical properties are strongly temperature dependent so the analysis of the radiation absorbed, scattered, or emitted by the nanoparticle can be used to obtain a remote thermal readout. As we will describe in Section Thermal sensing with an optically trapped lanthanide-doped nanoparticle and Section Fluorescent nanodiamonds, lanthanide-doped nanoparticles and nanodiamonds can also behave as thermal sensors when being optically trapped. The comparison of the different nanothermometers can be found in previous reviews^54-56.

As it was mentioned previously, the plasmon resonance wavelength of a PNP depends on the refractive index of surrounding medium, which in turn depends on temperature. Thus, measuring the variation of the plasmon resonance wavelength of a single PNP provides information about the medium temperature so that the PNP becomes a remote thermal sensor. Usually, an increment in the medium temperature is detected by a blueshift of the plasmon resonance. The easiest way to determine the plasmon resonance wavelength of an individual PNP is to immobilize it on a substrate. However, the temperature of a PNP deposited over a substrate may be different depending on the material of the substrate, which may act as a heat dissipator and have a variation of refractive index with temperature different from that of the medium⁵⁷. For that reason, if a PNP is to be used as a thermal sensor, it should be kept away from the substrate, i.e. the PNP has to be isolated and analyzed while being suspended in the medium by an OT. Optical manipulation of individual PNPs opens the possibility of an alternative way for remote thermal sensing through the scattering spectra of the trapped PNP, which can be monitored by darkfield spectroscopy. However, in these cases there is a risk that their large light-to-heat conversion efficiency may alter the temperature reading by artifactually increasing the temperature of the sensor itself⁵⁰.

Alternatively, temperature can be estimated by the rotation of the PNP within the optical trap, as the spinning velocity is determined by the temperature-dependent medium viscosity. Spinning originates because light can exert torque on an object through absorption and scattering. In the case of the spinning of PNPs, the angular momentum is in general maximized for trapping wavelengths close to the plasmon resonance⁴⁶. However, the self-heating of PNPs due to their absorption of the trapping laser may alter the viscosity of the fluid around it, and therefore its rotation speeds may experience artefacts in temperature estimation^{49, 58}.

The most representative example of the potential of optically trapped PNP for thermal sensing is the work published by D. Andrén et al.⁴⁹ They employed gold nanorods as optically trapped nanothermometers. They monitored both the longitudinal plasmon wavelength by darkfield spectroscopy and the spinning velocity to determine the temperature of an individual gold nanorod within an optical trap. Both methods rendered similar results for the estimated temperature (Fig. 1(a, b)). Through this process they observed a slight deviation in the tendency of the plasmon wavelength with temperature. They observed that the gold nanorod was producing so much heating over certain threshold of trapping laser power, that it experienced a progressive reshaping, leading to rods of a lower aspect ratio.

Figure 1.  (a) Time evolutions of the longitudinal surface plasmon resonance (LSPR) peak position (blue points) and the rotation frequency (red) as the laser power (green) is swept in sawtooth ramps with linearly increasing maxima. The dashed blue line is fit to the low laser power spectral peak positions, illustrating the continuous reshaping behavior. The grey area indicates a change in the slope of the linear relation between the laser power and λ_LSPR, where nucleation of vapor is suggested to occur. (b) Temperature determination based on LSPR spectroscopy (red data points, based on water’s refractive index (RI) temperature variation and subsequent λ_LSPR variation) and rotational Brownian dynamics (blue data points, based on water’s viscosity temperature variation and subsequent rotation frequency change) obtained from the data in (a). Figure reproduced with permission from ref.⁴⁹, Copyright 2017 American Chemical Society.

DownLoad: Full-Size Img PowerPoint

Nanoparticle-nanoparticle interaction

When two PNPs approach each other forming a dimer, their coupling leads to a shift of the resulting surface plasmon resonance towards lower energy (Fig. 2(a–f)) and to a significant increment of the intensity of the scattered light. This phenomenon also creates a hot spot, i.e., a region of intense electric field in between the interacting PNPs. Such a hot spot can be used to enhance the Raman signal or the fluorescence of a fluorophore, making plasmonic dimers useful for applications in single-molecule spectroscopy and sensing^{47, 59}. Next we highlight the most relevant works dealing with the use of optical traps for the study of the formation, dynamics, and applications of plasmonic dimers.

Figure 2.  Plasmonic nanoparticles for investigation of nanoparticle-nanoparticle interaction. (a) Time evolution of the darkfield spectrum of optically trapped silver nanoparticles. Observing intensity steps at the silver plasmon resonance (~450 nm) allowed to count the number of nanoparticles entering inside the trap. (b) Experimental darkfield spectra of one (black), two noncoupled (blue), and three (two of them coupled) (red) trapped silver nanoparticles. (c) Theoretical plane wave scattering cross section spectra of one (black), two noncoupled (blue), and two coupled (distance = 2 nm) plus one noncoupled (red) silver nanoparticles, assuming randomly changing orientation of nanoparticles in space. Evolution of scattering color at the optical trap (true color images) with time when two silver nanoparticles are simultaneously trapped: (d) initially, (e) 10 s after (d), (f) 60 ms after (e). No other particles were observed entering the trap during the color evolution process. (g) Darkfield images at different times (left and center) and time-lapse image (right) illustrating the optical transport of several 60 nm diameter silver nanospheres around a 3 μm radius ring-shaped optical trap. Darker blue for monomers, lighter blue for close-standing nanospheres, light green for dimers. (h) Illustration of the thermo-responsive behavior of the core (purple)-satellite (orange) plasmonic nano-assemblies incorporating thermo-responsive polymers. Figure reproduced with permission from: (a–f) ref.⁴⁸, Copyright 2011 American Chemical Society; (g) ref.⁶², Copyright 2021, Chinese Laser Press.

DownLoad: Full-Size Img PowerPoint

Optical traps are an ideal tool to study the formation of plasmonic dimers, as they allow the controlled approaching of the interacting PNPs and their simultaneous characterization by darkfield spectroscopy. A. Ohlinger et al.⁴⁸ observed the sequential incorporation of silver nanospheres to an optical trap (Fig. 2(a)) and recorded a spectral shift of over 100 nm (Fig. 2(b)), supported by the simulated spectra in (Fig. 2(c)), due to the formation of a dimer of PNPs within the optical trap. Such a large variation of the plasmon resonance wavelength can be detected even by simply observing the color of the light scattered by the PNPs in the optical trap by means of a CCD camera coupled to the set-up (Fig. 2(d–f)). However, the formation of dimers also implies an enhancement of the heating produced by the trapped particles, which destabilizes them in the optical trap. Later, H. Kermani et al.⁶⁰ demonstrated that a dimer of silver nanospheres orients along the electric field of a linearly polarized trapping laser. Therefore, it can stably spin in the plane perpendicular to the trapping laser propagation direction if the polarization direction rotates. They developed a method to obtain the interparticle distance and the azimuthal angle of the dimer with an accuracy of a few degrees independent of the wavelength or the distance between the interacting nanoparticles. In this way, they could explore the effect of spinning speed on the interparticle distance of the PNPs dimer.

Dimers can be also exploited as sensors of the dynamics and interactions of PNPs in an optical trap⁶¹. J. A. Rodrigo et al.⁶² achieved the controlled transport of multiple silver PNPs within a freestyle optical trap (structured single-beam trap shaped as a diffraction-limited arbitrary curve). They could detect regions of slower motion of the PNPs along the freestyle optical trap path due to the formation of dimers when PNPs circulating through the trap approached each other as they experienced less propulsion in those regions. They directly observed the formation of dimers in real time by means of a CCD camera (Fig. 2(g)). They observed that PNPs dimers move on average 2.8 times faster than single PNPs. This is due to the large extinction cross-section of the dimers (2.6 times larger than for the monomer), which therefore should experience an optical propulsion force about 2.6 times stronger than a single PNP.

Furthermore, as the plasmon resonance wavelength of a dimer is highly sensitive to the separation among the interacting PNPs, dimers are therefore also called plasmonic rulers^{63, 64}. F. Han et al.⁶⁴ optically trapped a core-satellite plasmonic system joined by a thermo-responsive polymer (Fig. 2(h)). They modified the heating achieved by the trapped plasmonic system by varying the applied trapping laser power. Then, when the collapse temperature (T_c) of the thermo-responsive polymer was reached, the system underwent the irreversible approach of the core and the satellites PNPs, which coupled experiencing a shift of their plasmon resonance wavelength. However, some core-satellite plasmonic systems tended to disassociate into the constituent gold PNPs when irradiated at large laser powers. The authors attributed that to the instability of the thiol bond used to join the thermo-responsive polymer to the PNPs.

Lanthanide-doped nanoparticles

The specific electronic configuration of lanthanide-doped nanoparticles leads to narrow emission bands and long lifetimes, typically between micro- and milliseconds. Fluoride-based crystalline matrices are being more frequently used as hosts because fluorides are characterized by low-energy phonons. Consequently, the number of phonons required to depopulate the excited energy levels of Ln³⁺ ions is larger than in other hosts and so the probability of non-radiative decays is minimized. Among Ln³⁺ doped fluoride nanocrystals, Er³⁺-Yb³⁺ co-doped ONPs (so-called upconverting nanoparticles, UCNPs) have attracted special attention due to their ability to convert infrared radiation into visible and ultraviolet (UV) emission. This allows to excite UCNPs in the infrared biological transparency windows for deep tissue imaging⁶⁵. Moreover, they can be excited using low power and inexpensive near-infrared (NIR) lasers and their luminescence can be detected with cost-effective silicon detectors⁶⁶. Optical trapping of UCNPs was firstly demonstrated by P. Haro et al., where dielectric NaYF₄:Er³⁺,Yb³⁺ nanoparticles were trapped and excited using a 980 nm single-laser beam⁶⁷. As shown in Eq. (1), the optical force experienced by ONP is determined by their electronic polarizability. For dielectric nanoparticles, as lanthanide-doped nanoparticles, the polarizability is defined as³⁴:

where $r$ and ${n}_{{\rm{NP}}}$ are the radius and the refractive index of the trapped nanoparticle, respectively, and ${n}_{{\rm{m}}}$ is the refractive index of the surrounding medium. As polarizability is proportional to ${r}^{3}$, typical UCNPs undergo trapping forces of femtonewtons. Although these forces can be used to trap particles, they may be too weak for applications in viscous media such as intracellular controlled trapping. Thus, an enhancement of optical forces is needed. As an increase in trapping beam intensity can damage the cells by heating and an increase in particle size can reduce the ability of UCNPs to be internalized in living cells, innovative strategies have been developed. First, polarizability can be optimized by increasing $\alpha ´$ through UCNPs surface modification. The effect of a coating covering a nanoparticle on the electronic polarizability is given by³⁴:

where ${\epsilon }_{1}$ and ${\epsilon }_{2}$ are the permittivity of the nanoparticle and the coating, respectively. A suitable coating leads to an increment of $\alpha ´$ and, consequently, an enhancement of the optical force. This strategy has been used by several authors to easily manipulate UCNPs^{33, 34}. An enhancement of optical forces can also be achieved by increasing the doping of lanthanide ions. X. Shan et al.⁶⁸ found that highly doped UCNPs can generate a strong oscillation resonance effect in the electromagnetic field, resulting in a significant enhancement of the optical trapping force. Both strategies are essential for 3D controlled optical trapping of UCNPs, allowing simultaneously imaging and sensing.

Thermal sensing with an optically trapped lanthanide-doped nanoparticle

In addition to the above-described properties, UCNPs are also sensitive to temperature. Such ability is based on the specific energy level diagram of UCNPs (Fig. 3(a))⁶⁹. After excitation of Yb³⁺ with infrared (980 nm) radiation, energy is transferred to Er³⁺. Subsequent absorption of 980 nm photons from excited states of Er³⁺ ions populate their high energy ⁴F_7/2 level from which both ⁴S_3/2 and ²H_11/2 states are populated. Radiative de-excitation from these states to the ground level are responsible for the green emission bands at 520 and 540 nm (²H_11/2→⁴I_15/2 and ⁴S_3/2→⁴I_15/2, respectively). These two states are so close in energy that they are thermally coupled, i.e., their relative populations follow the Boltzmann statistics. Consequently, the temperature dependence of relative intensities of the emissions generated at 520 and 540 nm makes determining temperature possible from a ratiometric analysis of the emission generated by Er³⁺-Yb³⁺ co-doped UCNPs^{66, 70-75}. Indeed, NaYF₄:Er³⁺,Yb³⁺ UCNPs are the most commonly used lanthanide-doped nanothermometers because of their superior luminescence brightness, good chemical and physical stability, and low toxicity^{66, 76-79}.

Figure 3.  Lanthanide-doped nanoparticles for thermal sensing. (a) Energy scheme of Er³⁺-Yb³⁺ co-doped UCNP. Black continuous arrows indicate ⁴F_7/2 level population by an energy transfer process. Grey discontinuous arrows indicate energy transfer processes between Er³⁺ and Yb³⁺ ions. Dark green (²H_11/2→⁴I_15/2), light green (⁴S_3/2→⁴I_15/2), and red (⁴F_9/2→⁴I_15/2) continuous arrows indicate radiative emissions centered at 520 nm, 540 nm, and 660 nm, respectively. (b) Schematic representation of the experimental setup used for thermal scanning in the surroundings of a HeLa cell subjected to a plasmonic photothermal treatment. The arrow indicates the scanning direction. (c) On the top, the temperature decay measured from cell surface for three different distances from the substrate. The symbols are the experimental data, and the lines are a guide for the eyes. On the bottom, the control experiment in absence of the 800 nm heating laser. (d) Functional scheme of the active dual-wavelength optical multitrap setup. The 980 nm beams (orange cones) for trapping of the cell and trapping/reading of the optical microthermometers, while the 1064 nm optical trap (green cone) for optical trapping/heating. (e) Microscope image of the cell, silica microparticles and optical microthermometers being trapped, activated, and read. Red rectangles represent 980 nm trapping laser, while green rectangles represent 1064 nm trapping laser. (f) Time lapse images of the heated cells at 10, 30, and 65 s after switching on the heating. On the bottom-right, graphs of relative Trypan Blue (TB) accumulation in the dying cell and the corresponding temperature (right axis). Figure reproduced with permission from: (b, c) ref.¹⁴, Copyright 2016 John Wiley and Sons; (d–f) ref.⁸⁰, Copyright 2017 American Chemical Society.

DownLoad: Full-Size Img PowerPoint

Thermal sensing from a single UCNP requires simultaneous trapping and emission acquisition. This becomes feasible since one of the most common fiber-coupled lasers used for trapping works at 980 nm, so it can be used for simultaneous excitation of UCNPs. As it is mentioned in Section Lanthanide-doped nanoparticles, the main challenge is the low optical forces typically acting on UCNPs. Indeed, long term trapping of a single UCNP is not easy. Therefore, it is not surprising that the first demonstration of simultaneous trapping and thermal sensing was demonstrated in Er³⁺-Yb³⁺ co-doped upconverting microparticles (UCΜPs). P. Rodríguez-Sevilla et al.¹⁴ have used NaYF₄:Er³⁺,Yb³⁺ UCΜPs to determine the thermal gradient induced in the surrounding of a live cancer cell being subjected to a photothermal therapy. In their experiments, authors favoured cell heating by incorporating gold nanorods and exciting them with an 800 nm laser focused within the treated cell. As it is mentioned in Section Plasmonic nanoparticles, since gold nanorods are PNPs, they have a high light-to-heat conversion efficiency. The 800 nm laser was used as a light source to induce heat generation process. The heat generated will gradually diffuse into the cancer cell. For the estimation of the thermal gradient in the surroundings of a cancer cell, NaYF₄:Er³⁺, Yb³⁺ UCΜPs were added to the cell culture medium. A second 980 nm laser focused within the culture medium was used to trap and excite a single NaYF₄:Er³⁺,Yb³⁺ UCΜP (Fig. 3(b)). Translation of the 980 nm laser beam allowed to scan the upconverting particle away from the cell while performing continuous acquisition of emission spectra, i.e., simultaneous thermal sensing. Figure 3(c) shows the temperature gradient at different horizontal distances from the cells obtained at three different heights. In this figure, the temperature profile obtained in a control experiment (800 nm laser beam was switched off) is also included revealing a complete absence of heating. From experimental data, author concluded that during photothermal treatment of individual cells the thermal gradient created in the surroundings extends over 10 µm, affecting surrounding cells and hindering single-cell treatments.

In 2017, S. Drobczynśki et al.⁸⁰ also scanned the thermal gradient in the vicinity of the diffuse large B-cell lymphoma (DLBCL) cells due to locally induced hyperthermia. In this case authors used optical microthermometers (OμTs) fabricated by encapsulating NaYF₄:Er³⁺, Yb³⁺ UCNPs with amorphous silica shells of about 6 µm in diameter. For heating, mesoporous silicon microparticles (SiμPs) were used. DLBCL cells, microthermometers, and SiμPs particles could be simultaneously trapped by an active dual-wavelength optical multitrap (Fig. 3(d)), where two different 980 nm laser beams (orange cones) were used to trap the DLBCL cells and OμTs and a 1064 nm laser beam (green cone) is used to trap the SiμPs particles, simultaneously converting light into heat. To quantify the damage to the cells due to temperature increase, the accumulation of Trypan Blue (TB) inside the cells was simultaneously measured. Optical microscope imaging was used to locate the OμTs close to the cells for thermal reading (Fig. 3(e)). As it is shown in Fig. 3(f), the temperature in the cells environment increases up to almost 100 °C, thermally injuring them. That is also manifested by a TB accumulation in the damaged cells 15 seconds after the temperature rise. It took another 15 seconds to reach TB saturation in the damaged cells, so the damaged cells respond attempting to alleviate the damage provoked by heating.

Intracellular viscosity sensing with an optically trapped lanthanide-doped nanoparticle

Intracellular viscosity is an essential parameter that plays a crucial role in the diffusion in biosystems⁸¹. It is strongly related with the mechanical deformability of the cytoplasm⁸². In turn, this affects many cellular and subcellular processes such as the translocation of organelles, partially controlled by their frictional drag and consequently by the viscoelastic properties of the cytoplasm^83-86. At cellular level, viscosity is also critical for protein–protein interactions and transportation of small solutes, macromolecules, and other cellular organelles in living cells⁸⁷. Thus, changes in intracellular viscosity are associated to different diseases, as diabetes⁸⁸, atherosclerosis⁸⁹, and Alzheimer’s disease⁹⁰.

Besides the superior luminescent properties of NaYF₄:Er³⁺,Yb³⁺, the non-centrosymmetric crystalline phase causes the luminescence to be polarized^91-94. The spectral shape of both green and red bands depends on the propagating direction and polarization of detected emission. The emission spectra can be explained in terms of three main polarization states: σ (light propagating along the optical axis (z-axis) with a polarization perpendicular to it (parallel to x- or y-axes)), α, and π (light propagates perpendicularly to the optical axis of the UCNP with a polarization perpendicular (α) or parallel (π) to it) (Fig. 4(a)). Figure 4(b) shows, as an example, the red emission generated by a NaYF₄:Er³⁺,Yb³⁺ UCNPs as obtained in each of these states¹⁵. Once the emission spectra corresponding to each of these configurations is known, it is possible to determine the orientation of an UCNP by just analysing their spectral shape^{92, 95}. Not only that, but a real time analysis of emission spectra also generated by UCNPs can be used to determine the rotation velocity of the particle while rotating from its initial position (optical axis parallel to the propagation direction of laser beam) to its steady state position (optical axis perpendicular to the propagation direction of laser beam). For a fixed torque (fixed laser trapping power) the rotation velocity depends on the medium viscosity, so that the analysis of the emission spectra can be used to determine the medium viscosity. This possibility was explored by P. Rodriguez et al. using submicrometric NaYF₄:Er³⁺,Yb³⁺ particles within HeLa cells (Fig. 4(c))¹⁵. An analysis of the polarization state of the red emission of Er³⁺ ions allowed the authors to access to the orientation of the particle within the cell as well as its rotation velocity (${w}_{l}$, what authors called active determination of intracellular viscosity.) The intracellular viscosity was obtained from the angular velocity of the particle by just applying:

Figure 4.  Lanthanide-doped particles for intracellular viscosity sensing. (a) Diagram of the three excitation configurations. σ-polarization state, defined by propagating light along the optical axis (z-axis) with a polarization perpendicular to it (parallel to x- or y-axes) and α- and π-polarization states, where light propagates perpendicularly to the optical axis of the UCNP with a polarization parallel (π) or perpendicular (α) to it. (b) Emission spectra along the three polarization states. (c) Schematic representation of the rotation of a UCNP inside a HeLa cell. (d) Measured viscosity as a function of the rotation frequency. Inset is the log−log representation. Blue data correspond to the values measured by the active method while red data is the averaged value obtained from passive method experiments. The rotation frequency for passive method is estimated by dividing the accumulated rotated angle by the elapsed drag. Black dashed lines correspond to the best fitting. Figure reproduced with permission from ref.¹⁵, Copyright 2016 American Chemical Society.

DownLoad: Full-Size Img PowerPoint

where ${I}_{{\rm{L}}}$ is the laser intensity. Additionally, intracellular viscosity was also determined from the oscillations of the trapped particle around its stable position. This is what authors named passive method, where the mean square angular displacement of the particle in the trap (MSAD) is analysed as a function of the lag time ($\tau$). The intracellular viscosity is then given by:

where $f/f_0$ is the Perrin friction factor for the trapped particle defined by the friction factor of the particle $(f$) and the friction factor of a spherical body ${(f}_{0})$, $R$ is the ideal gas constant, $T$ represents the temperature of the medium, ${V}_{{\rm{p}}}$ stands for the volume of the particle, and $\alpha$ determines if the movement of the particle is restricted ($\alpha$ < 1) or if it experiences superdiffussion ($\alpha$ > 1) (in this case, they set $\alpha$ = 1). As it can be observed in Fig. 4(d), a good agreement between both methods was obtained. Furthermore, authors were able to measure both the static and dynamic intracellular viscosity. They could also determine how the intracellular viscosity varies with the rotation frequency, as it can be expected from Eq. (4). A mean static value of 2.51 Pa·s has been obtained for the static viscosity in the cytoplasm, which is very close to the value obtained (2.3 Pa·s) by S. Chakraborty et al.⁹⁶, who used fluorescence correlation spectroscopy measurements of gold nanocluster, to determine its diffusion time in HeLa cells. Both experimental results are similar to the one obtained by analysing the movement of fluorescence magnetic nanoparticles inside the cells under a local field (1.45 Pa·s), method developed by I. Ramazanova et al.⁹⁷

Chemical sensing with an optically trapped lanthanide-doped nanoparticle

Adequate surface functionalization of upconverting particles can convert them into chemical sensors, in particular of pH. First UCNPs pH-sensors were obtained by coating NaYF₄:Er³⁺,Yb³⁺ with polyglutamic porphyrin-dendrimers¹⁶. The energy transfer from the UCNP to the porphyrin chromophore exhibits a pH-dependent absorption at 660 nm, leading to a decrease of the red luminescence emission of the UCNP with decreasing pH and so, to a pH-dependent red-to-green emission ratio⁹⁸. First demonstration of pH sensing based on a single UCMP was published by K. Suresh et al.¹⁷ In their pioneering work authors demonstrated how an optically trapped polyethyleneimine capped UCMP can act as a broad range pH probe. With this specific coating, the ratio between red and green emissions of Er³⁺ ions becomes strongly pH-dependent. The ratio between red and green emissions of a trapped UCMP shows a linear dependence on pH values (Fig. 5(a)). This behaviour was confirmed using different buffer solutions (Fig. 5(b)). Nevertheless, it must be considered that the emission ratio is different for each buffer solution, thus this pH probe is not independent on the composition of the surrounding medium. In conclusion, the authors have developed a robust reproducible pH sensor, based on red to green emission ratio analysis of a single UCMP trapped in a microchannel, even though it must be calibrated for each medium to be measured.

Figure 5.  Lanthanide-doped particles for chemical sensing. (a) Red to green emission ratio depending on pH value for a single UCMP in a buffer solution. Blue dots are experimental data. Grey line is the linear fitting. (b) Red to green emission ratio of the optically trapped UCMP vs. pH in various buffer solutions. Figure reproduced with permission from ref.¹⁷, under a Creative Commons Attribution 3.0 International License.

DownLoad: Full-Size Img PowerPoint

Fluorescent polymeric nanoparticles

Dielectric particles like polymeric particles are widely used in OTs experiments since they are commercially available. An advantage of polymeric particles is their higher refractive index at NIR wavelengths, which enhances optical forces and therefore facilitates more stable optical manipulation in water. However, pure polymeric nanoparticles are non-luminescent, so visualizing them within the trap is challenging. Incorporation of dyes into the polymeric nanoparticles is an easy way to make them luminescent and easy to track within the optical trap. As it is explained in next sections, the tracking of the polymeric nanoparticles makes it possible to study single particle dynamics and develop biological applications.

Investigation of single particle dynamics by optically trapped polymeric nanoparticles

There have been some reports on the dynamics of microparticles in optical traps^{99, 100}. For ONPs application in aqueous environment, it is also essential to understand the dynamic process of a nanoparticle stepping into an optical trap. The visualization of ONPs facilitates the study of dynamics at the nanoscale.

The effect of their own luminescence on the optical trapping forces is of high interest. R. Bresolí-Obach et al.¹⁰¹ studied the optical resonance effect (ORE) between a single dye-labeled polystyrene nanoparticle and the trapping laser. A 1064 nm laser was used for trapping, and a 488 nm laser was used for dye excitation. They found a 4-fold trapping stiffness enhancement due to resonant excitation under 488 nm irradiation. As shown in Fig. 6(a), upon 488 nm excitation, S₁ was populated, and the 1064 nm laser was cyclically resonant with the S₁−S₂ excited state electronic transition. The 1064 nm photons absorption in the S₁−S₂−S₁ cycle enhanced the polarizability of the system, leading to an increase in the induced radiation force that was responsible for the observed ORE. Theoretical calculations in Fig. 6(b) indicate the increasing S₁(v₀) and S₂(v₀) state population for resonance. When the 488 nm excitation was off (Fig. 6(c)) or the particle was labeled with fully photobleached dye (Fig. 6(d)), the trapping stiffness enhancement was no longer observed.

Figure 6.  (a) Jablonsky diagram of the electronic states involved in the optical resonance effect. (b) Theoretical calculations of the population of each state vs the 488 nm excitation power. (c) Changes in the trapping stiffness vs the excitation wavelength, there is no fluorescence or resonance effect without 488 nm excitation. (d) Changes in the trapping stiffness vs the 488 nm laser power. Figure reproduced with permission from ref.¹⁰¹. Copyright 2021 American Chemical Society.

DownLoad: Full-Size Img PowerPoint

In order to study how a particle incorporates to the circularly polarized laser trap, B. Louis et al.¹⁰² directly mapped the 3D incoming trajectory of NPs toward a stable optical trapping spot using a 200 nm fluorescent polystyrene particle. As shown in Fig. 7(a), they used a widefield multiplane microscope to achieve multiplane fluorescence imaging. The particles followed a Brownian motion outside the irradiated area and then exhibited a helicoidal motion once they entered the “trapping region” (Fig. 7(b)). The shallow cone shape incorporation of most of the ONPs in OT was observed (blue lines). However, not all the ONPs followed the same shallow cone trajectory. There were some ONPs that happened to be trapped for a short time in a metastable position around 2 μm below micrometers below the optical trapping focus before they were pushed to the focal spot (red lines). This metastable position indicates that there is a secondary focal point around 2 μm below the main focal point. It is induced by the interference of the focusing beam. The author confirmed the existence of secondary focal point by the calculated electric field intensity distribution. S. Ito et al.¹⁰³ demonstrated the photoswitching of radiation forces acting on a particle by using photochromic reactions of fluorescent diarylethene derivative (DE1). DE1 was put into 300 nm PMMA (poly(methylmethacrylate)) particles by swelling method. Photochromic reaction was controlled by UV irradiation, as Fig. 7(c) illustrates. The vertical Z position of a trapped particle was dependent on the scattering, absorption, and gradient forces. The scattering and absorption forces were proven to be affected by dramatical photochromic reaction, thus leading to the Z displacement of the particle (Fig. 7(d, e)).

Figure 7.  Fluorescent polymeric particles for investigation of single particle dynamics. (a) Left: Schematic of multiplane microscope with optical tweezer. Right: Time frame of fluorescent images from different planes. (b) 3D incorporation trajectories of 200 nm fluorescent polystyrene particles in an optical trap. Blue and red lines represent the incorporation trajectories going through the inner and outer cones, respectively. (c) Schematic of mesoscopic mechanical motions induced by the photochromic reaction of a single PMMA particle. (d) Time course of the positional change along the Z axis (red line) and the fluorescence intensity (blue line) of a trapped particle with the continuous wave 532 nm laser. (e) Z displacement of a fluorescent particle under trapping. (f) Profile of radial flow component u_r for various particle numbers (N) and diameters (d_p). (g–h) Snapshots of the fluorescence images of different numbers of 1 μm polystyrene particles around the optical trap. Figure reproduced with permission from: (a, b) ref.¹⁰², under a Creative Commons Attribution 4.0 International License; (c–e) ref.¹⁰³, Copyright 2023 American Chemical Society; (g–h) ref.¹⁰⁴, Copyright 2020 American Chemical Society.

DownLoad: Full-Size Img PowerPoint

OTs are usually employed in flow conditions. C. Hosokawa et al.¹⁰⁴ revealed the fluid flow in an OT by particle-image-velocimetry of fluorescent polymeric particles. They proposed a new mechanism based on the ability of scattering force to alter the fluid dynamics through the interaction with the laser-induced particle motion. Authors were even capable of distinguishing the relevance of particle number and size in the process. To demonstrate it, they recorded the fluorescence images of 500 nm and 1 μm fluorescent polystyrene particles around optical traps every few seconds. The radial component of flow vector ${u}_{{\rm{r}}}$ was analyzed (Fig. 7(f)). The case of 500 nm nanoparticles exhibited significant smaller flow speed compared with the case of 1 µm ones. Moreover, for the 1 µm nanoparticles, the increasing number of particles increased the flow speed (Fig. 7(g, h)). It denied the conventional thermal convection, which is independent of sizes and number of particles. Instead, they found that the fluid dynamics result from the direct impact of the optical forces acting on the suspended particles. Authors also found that the optical force and the resulting particle motion can be used to induce the large-scale fluid convection to accumulate the particles to the laser focal spot.

These works show us the dynamic motion of single or multiple ONPs in an optical trap in a visual way. They facilitate a more thorough comprehension of optical and mechanical interaction between trapping laser and optical particles. The measurement errors that may arise from such dynamics need to be considered during the practical application of optically trapped ONPs.

Optical trapping of polymeric nanoparticles for bioapplications

There are some attempts of using optically trapped polymeric nanoparticles for the study and characterization of biological samples. It was proposed in 1996 by E. L. Florin et al.¹⁰⁵ Authors demonstrated the potential application of fluorescent latex beads for topological imaging of biological systems. In 1997, they used an optically trapped 200 nm fluorescent latex bead as a sensing unit for scanning probe microscopy¹⁰⁶. When the trapping laser was fixed vertically and scanned horizontally, the intensity of the fluorescent signal varied because the particles were displaced away from the trapping laser focus when scanning the surface of the bio-object (Fig. 8(a)). Therefore, the intensity of the fluorescence emission from latex bead depended on its displacement. This feature was utilized to reconstruct a map of the height of the bio-object. In particular, the authors demonstrated how a two-dimensional image can be formed by laterally scanning the trapped latex bead across biological samples while recording the two-photon-induced fluorescence intensity. They successfully obtained a scanning image of the outer surface of a small neurite from a cultured rat hippocampal neuron and converted it to a height profile.

Figure 8.  Fluorescent polymeric particles for biological applications. (a) Scheme of the scanning procedure: a latex bead is trapped by the focused laser beam (red arrows). The fluorophore in the latex beads is excited by laser light via a two-photon absorption process. When the trapped bead is moved across an obstacle such as a cell (grey arrow), it is pushed out of the laser focus and the fluorescence intensity decreases. (b) A 500 nm polystyrene bead is tethered to cover glass surface by a 1010 base-pair DNA molecule, consisting of a long strand (black) joined to a shorter 15- base-pair duplex region (red). (c) Simultaneous records of force (red trace) and fluorescence (blue trace). Rupture occurred at 2 s at an unzipping force of 9 pN. The dye unquenched at the point of rupture, and later bleached at 9 s. Figure reproduced with permission from: (b, c) ref.¹⁰⁷, under a Creative Commons Attribution 4.0 International License.

DownLoad: Full-Size Img PowerPoint

Later, M. J. Lang et al.¹⁰⁷ developed an instrument combining OTs and single-molecule fluorescence. They used a 500 nm optically trapped dye-labeled polystyrene bead as the mechanical load to separate the strands of individual DNA molecules labeled with fluorescent dyes. A DNA duplex with an overhanging segment was attached at one end to a polystyrene bead. Its distal end was annealed at to a 15 base-pair oligonucleotide, anchored directly to the glass surface. Tetramethylrhodamine dyes conjugated to nucleotides were placed on complementary bases, one on the 3-end of the shorter 15-mer and the other on the 5-end of the long strand (Fig. 8(b)). Quenching of fluorescence happened due to the proximity of dye. The DNA-bead complex was trapped, and the microscope stage was moved at constant velocity to pull the bead from the center of the trap, which increased the load until rupture. Both rupture-force and fluorescence were monitored simultaneously. The rupture-force was converted from bead displacement from the trap center. The rupture of DNA molecules was monitored thorough the intensity of dye fluorescence. Before rupture the dye was quenched. However, upon rupture the fluorescence level increased abruptly as the two fluorophores were separated. This dye photobleached in several seconds and the light signal returned to background level (Fig. 8(c)). These results demonstrated the instrument capable of studying single molecule structure and mechanics. They successfully applied this method to measure the force needed to unzip or shear duplex DNA molecules¹⁰⁸.

In summary, as polymeric nanoparticles have been widely used for bioimaging and theragnostic, the combination of optical trapping of polymeric nanoparticles with fluorescence microscopy or scanning probe microscopy are promising strategies to broaden their application.

Semiconductor nanoparticles

Semiconductor quantum dots (QDs), nanorods, and nanowires have recently gained great attention thanks to their special photoluminescence properties such as tunable emission, lower susceptibility to photobleaching, high quantum yields, and chemical stability. Semiconductor nanoparticles are widely used in biomedicine as molecular sensors^109-111 and photostable substitutes for fluorophores in bioimaging¹¹². The potential application of semiconductor nanoparticles can be even expanded if they can be isolated and manipulated individually by optical traps, as later described in this section. Optical trapping of individual semiconductor nanocrystals was demonstrated for the first time in 2007 by L. Y. Pan et al.¹¹³, who gave some hints to make trapping easier such as to work on D₂O medium to prevent the thermal increase of Brownian motion and therefore increase the trapping stability. This pioneering work motivated other researchers to work on the optical trapping of QDs for different purposes and in different experimental conditionsl conditions^114-118. As it has been described for birefringent upconverting particles, simultaneous trapping and rotation of no-spherical semiconductor nanocrystals was also possible as demonstrated by C. R. Head et al.¹¹⁹ Optical trapping of semiconductor nanocrystals has been also used for the fabrication of complex structures¹²⁰. Among the experimental demonstrations of optical trapping of semiconductor nanocrystals in this review, we summarize and highlight those works in which the optical response of the trapped semiconductor nanoparticle is of relevance, such as those studying single particle spectroscopy or achieving cell imaging.

Optical trapping of semiconductor nanocrystals for single particle spectroscopy

Before applying the semiconductor nanocrystals to in vitro applications, it is essential to understand and improve the luminescent properties of single particles or small clusters. P. Pinapati et al.¹²¹ proposed the trapping of 35 nm silica-shelled quantum dots on graphene oxide (GO) to study the quenching behavior of QDs. They achieved the first optical trapping of QDs at low intensities, on the order of few μW/μm², in the presence of GO. The emission quenching of the QD@SiO₂ trapped on GO was studied (Fig. 9(a)). As shown in Fig. 9(b), the emission of the trapped QD@SiO₂ remained unquenched for 120 s in the absence of GO. This strong quenching behavior has been attributed to a nonradiative energy transfer mechanism from the QD to the GO.

Figure 9.  Fluorescent nano-semiconductors for single particle spectroscopy. (a) Time series fluorescence snapshots of trapping and releasing events of 35 nm QDs on graphene oxide. (b) Plot of luminescence intensity of trapped QD with time. The decrease in the emission of the QD on graphene oxide with time can be clearly observed. Evolution of emission peak intensity for trapped (c) QD@SiO₂ with only SiO₂ shell and (d) pQD@SiO₂ with sulfur and SiO₂ shells. Insets: diagram of trapped (c) QD@SiO₂ and (d) pQD@SiO₂. Figure reproduced with permission from: (a, b) ref.¹²¹, Copyright 2020 American Chemical Society; (c, d) ref.¹²², Copyright 2017 American Chemical Society.

DownLoad: Full-Size Img PowerPoint

Photobleaching may hinder the use of semiconductor nanoparticles in OTs. To tackle that problem, H. Rodríguez-Rodríguez et al.¹²² demonstrated the photobrightening and bleaching dynamics of optically trapped silica encapsulated QDs clusters. They studied two types of silica-encapsulated quantum dots excited upon two-photon absorption in an optical trap. The first type consists of alloyed CdSeZnS quantum dots were covered with a silica shell (QD@SiO₂). The second type had an extra intermediate sulfur passivating shell (pQD@SiO₂). Both tolerated irradiance doses 3 orders of magnitude higher than those reported for bare core/shell architectures. Meanwhile, the extra sulfur layer allowed a more gradual intensity decrease under continuous irradiation. They concluded that the sulfur layer can provide better passivation, suppress bleaching, and enable higher luminescence stability (Fig. 9(c, d)).

Optical trapping of semiconductor nanocrystals for cell imaging

Optically trapped semiconductor nanocrystals can act as energy converters for local in vitro fluorescence activation. P. J. Pauzauskie et al.¹²³ demonstrated the possibility to scan a cylindrical GaN nanowire across the cell membrane into living Hela cell by optical trapping. The position of GaN nanowire could be manipulated and maintained for arbitrary time. They chose a 1064 nm trapping wavelength because the cells can tolerate limited infrared laser. GaN emitted ultraviolet light under the two-photonexcitation by the trapping laser. In this way, GaN nanowire could act as a local subwavelength ultraviolet nanolaser to excite local fluorophores for precisely positioned cellular imaging. The feasibility of nanowires as local point-like illumination source was further demonstrated by Y. Nakayama et al.¹²⁴. The emission from an optically trapped KNbO₃ nanowire was used to excite the two-photon absorption of a fluorescent dyed polystyrene bead in near field. H. Rodríguez-Rodríguez et al.¹²⁵ also used optically trapped QDs as localized emitters to induce photoluminescence of dye-labeled cells. As shown in Fig. 10(a), the silica-coated colloidal QDs were simultaneously excited and manipulated to vicinity of labeled Jurkat cell by the trapping laser. The radiative energy transfer happened between QDs (donors) and fluorescent dyed cells (acceptors). When the QD was trapped far (~10 μm) from the Jurkat cell, the emission line shape (black) equaled that of QD (red) in the spectrofluorometer (Fig. 10(b)). The same line shape was collected even when the QD540@SiO₂ approached the Jurkat cell up to a few microns (Fig. 10(c)). If the QD was trapped close to the cell within 1 μm, Alexa Fluor 546 emission from the cell dominated the recorded signal due to radiative energy transfer (Fig. 10(d, e)). Despite the low efficiency of this method, it is effective to collect fluorescence signals in far field at relatively low dye concentration. The low energy transfer efficiency also prevents the biological damage and dye photobleaching effects.

Figure 10.  Fluorescent nano-semiconductors for cell imaging. (a) Sketch of the indirect excitation of a dye labeled cell. An optically trapped QD₅₄₀@SiO₂ cluster is positioned in the vicinity of a labeled Jurkat cell with Alexa Fluor 546 to induce its photoluminescence. Bright field microscopy images and collected spectra (black curves) from a QD₅₄₀@SiO₂ cluster trapped (b) ~10 μm away from the Jurkat cell, (c) close (a few microns) to the labeled Jurkat cell membrane, (d) in the vicinity (within 1 μm) of the Jurkat cell membrane, and (e) above the whole Jurkat cell structure. The optical trap position roughly matches the reticle center. Red curves represent in all cases the emission of the original QD₅₄₀@SiO₂. Figure reproduced with permission from ref.¹²⁵, Copyright 2019 American Chemical Society.

DownLoad: Full-Size Img PowerPoint

Fluorescent nanodiamonds

The fluorescence of nanodiamonds (NDs) is caused by point-defects in the diamond structure, known as color centers. The nitrogen-vacancy (NV⁻), the most studied defect in NDs, consists of a nitrogen atom replacing a carbon atom, and a vacancy on a neighboring lattice site. Spin-preserving optical transitions between the ground and the excited states in the NV⁻ center have different relaxation pathways depending on the initial spin state, allowing them to be distinguished through optical spectroscopy. The NV⁻ of a ND can be excited by a light source with a specific wavelength and emit stable fluorescence. In addition, as the existence of extra magnetic field can cause NV⁻ ground energy level splitting, the optical properties can couple with external magnetic field. The spin-dependent fluorescent readout of ground-state electron-spin resonance (ESR) transitions in NV⁻ has been used for magnetic^{126, 127}, electric¹²⁸, and thermal¹²⁹ metrology at the nanoscale. The photostability and biocompatibility of NDs^{130, 131} have also permitted to image and characterize NV⁻ centers within living cells¹³², pointing to potential applications in sensing, tracking, and tagging in submicron biophysical systems.

Combination of OTs and the luminescence generated by ND has allowed quantum sensing at the single particle level. The first demonstration of optical trapping of NDs ensembles by optical trapping in liquid was published by V. R. Horowitz et al.¹³³. The commercial 100-nm NDs were irradiated and annealed to contain ~500 NV⁻ centers in one nanocrystal. The optical trapping was performed with a 1064 nm continuous wave laser while a separate 532 nm continuous wave laser was used for optical excitation of the NV⁻ centers. They were able to measure spectral changes in the presence of an external magnetic field, thereby demonstrating the first magnetometry-based sensing on an ensemble of NDs, obtaining a direct current (DC) magnetic field sensitivity of $50\;{\rm{\mu T}}/\sqrt{{\rm{Hz}}}$ at a magnetic field of 5G. R. Quidant research group applied optical trapping to deterministically trap and to manipulate in three dimensions an individual ND hosting a single NV⁻ center¹³⁴. The optical trapping of a single ND was proven by using a high sensitivity fluorescence camera that allows real time monitoring of the diffusion of the ND. Quidant and co-workers were able to perform vectorial magnetometry on a single electron spin, by varying the three-dimensional magnetic field orientation, used to determine the NV⁻ axis, while the NV⁻ axis was also rotated using the electric field polarization of the trapping beam. The achieved DC magnetic field sensitivity in this configuration was improved to $40\;{\rm{\mu T}}/\sqrt{{\rm{Hz}}}$.

For temperature sensing application, B. Li group followed an interesting approach based on the NDs aggregation into a microsphere¹³⁵. The authors demonstrated that the NDs can be aggregated using optical forces in liquid medium to form a microsphere, whose size, rates of formation and stabilization are related to the NDs concentration, trapping laser power, and duration of irradiation. Then, the NDs microaggregates remained stable after the laser was turned off. To perform intracellular thermometry, the NDs were incorporated into the cell by endocytosis process and then, the NDs microaggregates were formed intracellularly by using optical forces. The main advantage of the NDs aggregation and electromagnetic resonance between individual nanodiamonds is that they lead to a sevenfold intensity enhancement of 546-nm laser excitation. OT allows precise patterning and positioning the NDs microaggregates within the cell. Here, the optical trapping was performed with a 1064 nm laser and an additional 546 nm beam was used for optical excitation of the NV⁻ centers. Comparing to liquid solution, to form a stable NDs microaggregates in the complex intracellular environment, it required a higher trapping laser power (150 mW) and longer duration of irradiation (180 s). Under this condition, the NDs aggregates also remained stable after the laser was turned off. Then, intracellular thermal probing was demonstrated by measuring the fluorescence signal from aggregated NDs microspheres at different locations within a human brain microvascular endothelial cell (HMEC-1). First, after 24 h (Fig. 11(a)), the endothelial cells were marked to visualize the mitochondria (Fig. 11(b)). Then, the NDs were intracellularly aggregated at different locations after 180 s of irradiation (Fig. 11(c)). Temperature calibration inside cell was performed by measuring the luminescence of several NDs microspheres, between 20 °C and 60 °C. The zero-phonon line of the NDs aggregated provided a thermal shift of 25 °C nm⁻¹, with a minimum temperature difference $({\rm{\Delta}} {T}_{\rm{min}})$ of 0.5 °C at the desired locations. In addition, temperature reading was performed at different positions inside the cell: close to the nuclear membrane and the outer membrane, indicated by I and II in Fig. 11(c), respectively. This experiment revealed that the temperature at the nucleus was 2.5 °C higher than that at the membrane, what the authors explained because most of the energy-generating mitochondria stained were gathered around the nucleus¹³⁶.

Figure 11.  Fluorescent nanodiamonds for cell imaging. (a) Confocal microscope image of a human brain microvascular endothelial cell (HMEC-1). (b) Fluorescence image of the cell with mitochondria stained by Mito-Tracker Green FM (excitation: 490 nm, emission: 516 nm). (c) Fluorescence image of the nanodiamonds microaggregates at different locations in the HMEC-1. Blue and pink arrows pointed to the microspheres near the nuclear membrane and cell membrane, respectively. Figure reproduced from ref.¹³⁵, under a Creative Commons Attribution License.

DownLoad: Full-Size Img PowerPoint