1. Introduction

Optical tweezers provide a powerful, non-contact method of manipulating microscopic particles. An optical tweezer is a tightly focused laser beam that is used to capture and manoeuvre particles [1]. Momentum exchange occurs on refraction of the light through the particle. Hence an intensity gradient near the beam focus gives rise to a gradient force. This force opposes the radiation pressure of the laser beam and causes the particle to move into the focus. If the gradient force is strong enough, it will overwhelm any scattering or gravitational forces present and the particle will be confined in three dimensions.

Optical tweezing is now firmly established as an important and key interdisciplinary tool [2,3]. Typically the technology can readily move objects in the 1–10 micron region though trapping can occur down to 25 nm. However, as we are constrained to a diffraction limited spot size localising the particle in a small region proves difficult. The use of shorter wavelengths of light is an exciting potential route to allow much smaller trap volumes to be realized.

The development of InGaN semiconductor materials has led to the commercial availability of continuous-wave violet laser-diodes with practical lifetimes. These can be placed into extended cavities in order to force them to operate on a single longitudinal mode with a narrow linewidth [4]. Recently, frequency doubled vertical external cavity surface emitting lasers (VECSELs) operating at 488 nm have appeared on the market [5] giving a compact blue laser with high beam quality. In May 2003 violet laser diodes that had been microlensed to achieve a circular output beam became commercially available [6]. These lasers provide excellent beam quality in a compact and convenient packet. The first two classes of device are finding use in a range of studies including spectroscopy [7], flow cytometry studies [8], and confocal imaging [9], however their use is in its infancy. No reports have been made of their use in optical micromanipulation.

Optical techniques provide a powerful means to probe a wide range of important biological phenomena. New technology has facilitated research into high resolution studies allowing observation of genes, enzymes and proteins using both non-specific and specific biological tags. Many biologically important stains absorb in the blue-violet spectral region. This has led to the use of violet diode lasers as an illumination source in flow cytometry [8]. DAPI (4′-6-Diamidino-2-phenylindole) is one such stain, which non-specifically binds to all chromosomes, intercalating with their DNA. Other biological tags, such as green fluorescent protein (GFP) [10] and its derivatives such as cyan fluorescent protein (CFP), allow protein specific identification with enormous flexibility [11,12]. In general these fluorescent tags can be engineered, at the gene level, so that they are fused to a particular protein of interest. The resulting DNA construct can be transfected into living cells, where the gene of the protein of interest expresses as a GFP/CFP tagged protein. This gives a non-invasive marker, which can be exploited by new laser technology to observe and elucidate areas of cell and molecular biology; observing protein-protein interactions, cell lineage tracing and gene expression.

In this paper we demonstrate the use of compact, semiconductor laser systems operating at 410 nm, 413 nm and 488 nm in optical tweezing and for fluorescence excitation in optical micro-manipulation systems. Our work shows new studies for tweezing at these short wavelengths and demonstrates the potential for a multi-functional capability which provides the flexibility to combine the observation of biological samples with particle manipulation in an integrated optical system.

2. Optical tweezers

Optical tweezing uses a tightly focused laser beam to capture and manipulate particles in three dimensions. The particle is drawn into the region of highest local intensity producing an intensity-gradient or dipole trap near the beam focus. A truly three-dimensional trap (optical tweezers) [1] results when the gradient force exceeds the drag, scattering and gravitational forces acting on the particle as well as its thermal (Brownian) motion. Since a diffraction-limited focal spot has its minimum size constrained to λ/2, the shorter wavelength of violet laser sources permit a smaller focused spot size. The wavelength of light used for optical trapping is crucial. In the region where the particle is much smaller (factor of 20) than the wavelength, the particle is in the Rayleigh regime where it is considered as a small induced dipole that aligns itself with the light field. If the particle size is larger than the wavelength of the light trapping it, it is said to be in the Mie regime where a geometrical (ray) optics picture is appropriate. Within this regime, Mie scattering predominates over Rayleigh scattering and the dipole of the trapped particle can be neglected. However, if the particle is smaller than the focused spot size, then Rayleigh scattering predominates, and forces due to scattering have a λ⁴ dependency. Moreover, in this limit the depth of the trapping potential from the gradient force shrinks rapidly (with the volume of the particle), and tweezing is correspondingly weak. The short wavelength of violet lasers moves the trapping of particles as small as 200 nm into the Mie regime. A significantly smaller trap volume is also a potential advantage that results from the tighter focus that is possible with the violet light. The theoretical trap volume (based on [13]) for a 410 nm laser, assuming a diffraction limit spot, λ/2, is 1.1×10^-19 m³, whereas an IR laser at 1064 nm gives a trapping volume of 1.9×10^-18 m³ (over 17 times larger). The smaller volume of the violet laser trap enables a higher degree of localization of the trapped particle. Importantly, this gives the possibility of singly populating the trap with far smaller particles than would be possible within a larger trap volume. Sub-micron particles and nanoparticles have instigated new found interest due to potential applications in biophysics and imaging yet relatively little attention has been focused on them as yet from the optical trapping community where the majority of studies focus on micron sized (or greater) particles. In a violet laser trap, sub-micron sized particle can therefore be individually confined, constrained within a space comparable with its own size and manipulated in a violet laser diode trap which may not be possible with a red or near infrared source.

In order to explore this, we firstly quantify optical trapping using our short wavelength sources. The efficiency of an optical trap is usually quantified by a dimensionless parameter Q, which corresponds to its lateral trapping efficiency [14,15]. When considering lateral trapping, scattering and gravitational forces are negligible compared to the Stokes drag. Hence, Q is defined where the lateral trapping force, shown below in Eq. (1), is equal to the Stokes drag, shown below in Eq. (2).

where _nm is the refractive index of the medium surrounding the trapped particle and η the viscosity of the suspending medium, c is the in vacuo speed of light, P is the incident tweezing power at the sample, d is the particle diameter and ν the maximum lateral velocity at which the particle remains trapped. It is therefore possible to calculate Q from measuring this lateral velocity. Since Q is a measure of efficiency, a Q-value of one corresponds to a complete transfer of momentum from the photons within the light beam to the particle. We stress Q values are a complex entitity in some respects in optical tweezers as although dimensionless they do depend upon degree of light focusing and aberrations (e.g. spherical) present in the optical trapping system.

The most commonly used sources in optical tweezing are red and near infra red (IR) lasers due to their excellent beam quality and technological maturity. These lasers are ideally suited to biological applications as a consequence of their low absorption in water and biological tissue. This highly beneficial property allows biological material to be manipulated whilst minimizing the potential for damage. Laser sources at 532nm and at argon ion wavelengths have also been used effectively for optical tweezing, combining negligible water absorption with a small diffraction limited spot [e.g. 14]. Unfortunately, at shorter wavelengths, in the violet and blue spectral region, an increase in tissue absorption occurs, which can lead to tissue damage. It is, therefore, necessary to weigh the disadvantages of the increased risk of tissue damage against the advantages of a reduced trapping volume in appropriate situations.

3. Trapping particles with optical tweezers

We assess three commercially available laser systems in the blue-violet spectral region in comparison to a frequency doubled Nd:YVO₄ 532 nm solid state laser (SSL) [14] and a Nd:YVO₄ 1064 nm SSL [16]. The first system is an external cavity laser diode (ECDL) from Toptica GmbH (DL100) running in the violet providing 14 mW at 410 nm. The second was a free running violet laser diode from Blue Sky Research microlensed to provide a circular output beam of 25 mW at 413 nm. The third source was a frequency doubled vertical extended cavity surface emitting laser (VECSEL) from Novalux Inc. that provided a maximum output power of 5 mW at 488 nm

Both the VECSEL and the microlensed laser diode had circular beam profiles with good beam quality and thus required no beam-shaping. These laser sources provide M² values of less than 1.2 in both planes for the VECSEL and ${{\text{M}}_{\text{x}}}^2$ of 1.5 and ${{\text{M}}_{\text{y}}}^2$ of 1.3, for the microlensed laser diode. The diode laser used for our ECDL system exhibited poorer beam quality characteristics with measured M² values of 2.3 and 1.7 in the two planes. The elliptical output was beam shaped through the use of two cylindrical lenses (f_x=10 mm and f_x=40 mm). Each of these laser sources were used independently of the others as an optical tweezing laser in the basic design shown in Fig. 1 and the efficiency of trapping determined.

 

Fig. 1. Experimental violet ECDL tweezers set up.

Download Full Size | PDF

The beam was expanded and collimated using the first telescope. A lens relay imaged the beam onto the back of the microscope objective (Newport, x100, NA 1.25) creating a conjugate plane for beam steering. The sample was illuminated from below using a white light source and scattering from the sample allowed observation via the CCD camera. The sample cell was created from a microscope slide and cover-slip separated by an 80 µm thick spacer. Colloidal particles (0.2–10 microns in diameter, refractive index of the silica spheres=1.37 (Bangs Laboratories), refractive index of the polymer spheres=1.59 (Duke Scientific Corporation)) were deposited within the resulting 20 µl well, which led to a controlled sample volume.

The microscope objective was designed for longer wavelength visible light, and therefore, exhibited a low transmission of violet and blue wavelengths. The ECDL beam provided an available trapping power of 1.3 mW with an approximate radius beam waist of 250 nm. The VECSEL provided 0.6 mW of power with an approximate radius beam waist of 600 nm, due to it having a smaller beam size on the back surface of the objective. The microlensed laser diode provided a power of 3.5 mW which was focused to an approximate radius of 200 nm. All powers are quoted at the sample plane.

Optical tweezing with silica spheres ranging from 0.4 µm to 10 µm in diameter was achieved. Particles of 0.4 µm in size are shown being manipulated using the 410 nm ECDL in Fig. 2. Tweezing forces for this sphere size were reduced due to the strong Brownian motion of the polymer spheres and poor beam quality of the ECDL. The low available tweezing power of the 488 nm VECSEL combined with its larger beam waist, led to its inability to tweeze spheres of this size. By contrast the violet microlensed laser diode allowed tweezing of 400 nm silica particles and 300 nm green fluorescent polymer particles. Fluorescent polymer particles of 200 nm in size were also visibly affected by the optical tweezer. However, the strong thermal Brownian motion made it arduous to move these particles further than approximately 5 µm.

 

Fig. 2. CCD camera images showing particle manipulation of a 0.4 µm sphere in our optical tweezers using the ECDL

Download Full Size | PDF

The lateral trapping efficiencies of the 410 nm ECDL, 488 nm VECSEL and 413 nm microlensed laser diode were measured. Table 1 shows the results for a range of different silica sphere sizes. All Q values were obtained with the tweezed particle raised above the sample plane of diffusing particles, z-trapping close to the top of the sample surface. For 0.4 micron spheres this z-trapping was weak. These trapping efficiencies are compared to similar results obtained by O’Neill and Padgett using a 50 mW 532 nm frequency doubled Nd:YVO₄ SSL with a 100x objective [14] and Felgner et al using a 95 mW 1064 nm Nd:YVO₄ SSL [16] with a 1.3 NA 100x objective.

Table 1. Q values of the tweezer set up operated with different lasers

View Table

This data shows optical tweezing with the shortest ever reported laser wavelengths to the best of our knowledge and lateral Q values that are very comparable with other laser sources. Trapping, using the microlensed laser diode source, provided far more efficient and stronger trapping than the other two systems. The Q values show extremely strong trapping for this laser whilst trapping 1–3 micron spheres much higher than reported for tweezing of such spheres in the literature. This is in contrast with the performance of this laser with larger spheres. Tweezing of 10 µm spheres was not achieved. Whereas the weakly tweezed 5 µm spheres, were 25 times larger than the focal spot and hence reaching the size limit of particles which the system is able to manipulate. This laser had a circular focal spot as opposed to the elliptical one exhibited by the ECDL and M2 values much closer to one. Hence it is likely that beam quality issues are causing the reduced trapping efficiency of the ECDL system. The VECSEL is much closer to the microlensed laser diode in terms of beam quality and it is also shown to provide a greater trapping efficiency than the ECDL. However, it is still lower in efficiency than the microlensed diode system, due to the far lower tweezing power leading to trapping forces comparable to gravitational forces. All of the systems compared well to the frequency doubled SSL system however, only the microlensed diode tweezer system compared favorably with the fundamental SSL.

In measuring the lateral trapping efficiency it is reasonable to assume that the gravitational forces are negligible. Since gravity is working perpendicular to the trapping only the Stokes drag needs to be considered. However, in order to achieve strong axial, or z-trapping, it is necessary for the trapping force to overcome the gravitational force. The low tweezing power available at the focal region, particularly for the VECSEL and ECDL sources, meant that this was not possible for several of the sphere sizes and thus the results must be weighted appropriately. For all the lasers the gravitational force on 1 micron spheres was lower than the trapping force. At 2 microns the ECDL trapping force was comparable to the gravitational force on the particle, which with larger particles exceeded the trapping force. The three micron spheres had a gravitational force exerted on them which was an order of magnitude larger than the trapping force of the VECSEL, again at larger particle sizes the gravitational force increased while the trapping force decreased. The microlensed laser exerted a trapping force which exceeded the gravitational force up to 3 micron diameter spheres, but not for larger particles. This was observed in the tweezing of the lasers, since only the microlensed laser system readily exhibited z-trapping, the others exhibiting only weak z-trapping. It is likely that more stable z-trapping could be obtained from a tweezer set up with an inverted trapping beam.

4. Manipulation and fluorescence of biological structures

Since DAPI has a wide absorption in the violet, and GFP [10] has a wide absorption in the blue, it is possible to use the lasers discussed in the previous section to excite fluorescence in these biological markers. The ECDL was employed to excite fluorescence in both GFP, which was fused with amyloid β-peptide binding alcohol dehydrogenase (ABAD), and DAPI stained Chinese hamster ovarian (CHO) chromosomes. This laser was also utilized to induce fluorescence in, and simultaneously tweeze, blue fluorescent polymer spheres

In order to stain and prepare the chromosomes for tweezing CHO cells were treated with a chemical which prevented the cell from dividing, so that the chromosomes could be harvested. The cells were burst open by passing them through a needle and the debris was removed by washing. This left the chromosomes behind so that they could be stored in a fixative solution containing one part acetic acid and one part methanol. For tweezing, the chromosomes were resuspended in phosphate buffered saline (PBS) and then DAPI was added, one part DAPI for twenty parts PBS. Chromosomes are shown being illuminated by both violet and white light sources in Fig. 3.

 

Fig. 3. Comparison between two chromosomes under violet laser illumination, first picture and under white light illumination, last picture. The central picture shows the first picture overlaid by a red version of the last picture indicating the fluorescence is coming in part from one chromosome and in part from the other.

Download Full Size | PDF

The set up was modified from Fig. 1 to include an inverted tweezer from beneath the sample cell using a 50 mW 1064 nm SSL laser. Incident illumination from the violet ECDL from above, previously used as a tweezer, excited fluorescence in the DAPI stained chromosomes. Thus, the violet ECDL could be used for illumination of a particle optically trapped by the infrared SSL. In this geometry we retain the attributes of tight focusing of the violet light that allow a selective excitation within the biological specimen of interest. A guided chromosome is shown in Fig. 4. Since the violet laser was being using for fluorescence and not for tweezing, very low powers could be used reducing the risk of damage to the tissue.

 

Fig. 4. Guiding of one chromosome compared to a group of chromosomes.

Download Full Size | PDF

A gene of interest can be linked to a reporter gene, such as GFP, providing important information about the location of protein expression within the cell. ABAD is a protein involved in Alzheimer’s disease. Using DNA manipulation, ABAD was ligated with GFP to produce an ABAD-GFP DNA construct. This DNA construct was then used to transfect human neuroblastoma cells which were subsequently illuminated and fluoresced using the violet ECDL. The ABAD-GFP fusion protein was expressed in the cell’s mitochondria. With conventional fluorescence microscopy, fluorescence occurs wherever the ABAD-GFP is expressed, as shown in Fig. 5. The use of the violet laser, as opposed to excitation with an incoherent light source, allows a very specific area within the cell to be examined without risk of bleaching the surrounding areas. This is shown in Fig. 6. The violet laser wavelength at 410 nm is significantly far from the peak of the absorption of GFP at 488 nm. The VECSEL laser is a more suitable alternative in this instance. Blue diode lasers may be used with either cyan fluorescent protein or blue fluorescent protein, with absorption peaks at 440 and 405 nm respectively. This would potentially enable individual proteins to be identified and possibly manipulated within the cell itself using the increased functionality of our set-up.

 

Fig. 5. A human neuron (neuroblastoma) cell containing, viewed under a conventional fluorescence microscope-exhibiting fluorescence across the entire cell.

Download Full Size | PDF

 

Fig. 6. A section of a human neuron (neuroblastoma) cell containing GFP viewed with a violet diode laser fluorescence illuminator and a x100 microscope objective – exhibiting fluorescence of only a section of the cell.

Download Full Size | PDF

Finally we demonstrated that the system could be simplified by using the violet ECDL to simultaneously tweeze and fluoresce particles. In this experiment, blue fluorescent, one micron diameter polymer spheres were manipulated by the violet ECDL tweezer system, as shown below in Fig. 7. Although this experiment worked well with fluorescent spheres, it is possible that the increase in power necessary to manipulate rather than merely excite fluorescence within the trapped particle will lead to this technique being unsuitable for biological samples due to the increased risk of tissue damage.

 

Fig. 7. (1.95 Mb) Movie of the simultaneous fluorescence and manipulation of blue fluorescent 1 µm polymer spheres

Download Full Size | PDF

5. Conclusion

We have demonstrated the use of new violet laser technology in optical tweezer systems. We have shown optical trapping with the shortest wavelengths yet used for this technique. Very strong lateral trapping and high Q values were obtained for a microlensed violet diode laser for particles in the 1-3 micron range. Particles down to 300nm were readily manipulated using our system. The shorter wavelength may reduce the optical trap volume localizing the particle to a considerably smaller region of space and this aspect is the subject of further study.

We have further demonstrated a multi-functional system to manipulate and image particles and biological matter in a single set-up. The set-up combines a violet ECDL with an IR SSL optical tweezer. Within this set-up, laser-induced fluorescence and guiding of DAPI stained chromosomes has been observed, demonstrating a compact cytometry system. The fluorescence of GFP bound proteins within a human neuroblastoma has also been demonstrated using a violet ECDL source. This opens up the possibility of observing protein and gene motion through living cells in a compact diode laser based apparatus. The development of multifunctional systems, such as the system demonstrated here, has major biological implications: progressing the development of cytometry systems along with our fundamental understanding of cell and molecular biology.