1. Introduction

Laser radiation can be used to exert significant forces on microscopic particles resulting from the transfer of photon momentum to matter. Through various applications of optical forces, one can trap particles in a liquid, manipulate them, and sort them.[1–4] Optical tweezers employ one or more laser beams that are highly focused into a solution containing microscopic particles. The momentum transfer resulting from the highly convergent rays (using a microscope objective) generates a net restoring force which retains particles at the focal point of the laser. Through translation of the surrounding medium or movement of the laser beam, particle translation and manipulation can be achieved. This method has been used to separate particles based on their appearance including size, shape, or using fluorescent labels. Recently more sophisticated and automated optical techniques have been developed to separate microscopic objects.[5, 6] In this work, arrays of optical traps in a fluid flow are used to preferentially transport microscopic objects which experience a greater optical force away from those which experience a lesser force. Other techniques involving novel beams have been developed.[7, 8]

Optical chromatography (OC) [9–11] relies on a mildly focused laser beam to propel particles along its axis of propagation. The beam is aimed directly against a fluid flow; the balance of fluid drag force and optical pressure results in the stable trapping of particles within the beam. Larger size or greater refractive index particles experience greater optical pressure and are thus propelled further than smaller or lower refractive index particles. This results in unique positions (separation) along the beam axis for particles of different size and composition (refractive index). Using this technique it has been shown that particles and microorganisms can be separated by size[9, 12], refractive index[13], shape and morphology[14], and fluid drag characteristics[14].

In this paper we demonstrate a new fractionation method based on optical chromatography for the complete separation of injected groups of particles. In traditional optical chromatography, the majority of injected particles flow past the laser beam due to the large size of the channels used and the small focal point (200–1000 µm diameter channel, and a 5–20 µm laser beam waist)[9, 13]. As such, only ≈1 % of the injected particles are optically captured by the loosely focused beam and become stably retained at their characteristic retention distance (distance from the focal point). With this new fractionation method, the channel has been decreased, and the beam waist diameter increased such that the beam completely fills the separation channel (50 µm channel). Separation efficiencies are typically greater than 95%, frequently 99–100% for 2 µm diameter polystyrene beads.

The goal with this new implementation of optical chromatography is to perform complete, high efficiency sample separations. With sufficient numbers of particles being processed, the technique can be useful for existing analytical techniques. The initial demonstration, described below, was the external collection of optically separated fluorescent and non-fluorescent particles with observation/confirmation under a microscope. The next experimental verification was the use of the OC laser filter system to purify and externally collect Bacillus anthracis (B.a.) spores for detection using a commercially available system based upon DNA amplification via real time polymerase chain reaction (RT-PCR) with fluorescence detection. RT-PCR detection is largely considered the gold standard for automated bacterial detection. However, RT-PCR can fail producing a false negative result when testing samples containing certain small molecule, protein, or microbiological interferences. This weakness was used to test the ability of the preparative optical chromatographic system to purify samples that were otherwise too contaminated for the commercially available system to detect successfully.

2. Experimental

The OC laser filter system has been adapted from previous work and is described in detail elsewhere. Briefly, the system consisted of a CW laser focused by a 1 inch diameter plano-convex 100 mm focal length lens into a microfluidic system. The laser used in this work was a 1064nm ytterbium fiber laser (IPG Photonics, Oxford, MA) and was aligned with the microfluidic flowcell using a custom-made adaptor fitting a 1 inch diameter lens tube system (Thorlabs, Inc., Newton, NJ) attached to a x-y-z positioning stage. The microfluidic network was mounted on a 5 axis positioner (New Focus, San Jose, CA) and the entire aligned optic and fluidic system was placed underneath a microscope (Lumam, LOMO America, Prospect Heights, IL), mounted on a custom x-y-z translation platform. This platform allowed for movement of the microscope used for observation and image data collection independently of the optic and fluidic components.

The microfluidic component was constructed entirely of crown glass plates, with wet-etched channels (50 µm depth and 100 µm width) to deliver the fluid and sample to the separation channel and subsequently carry it out of the microfluidic device to waste. The glass plates were drilled for fluid input and output of the etched channels, bonded together and the holes were fitted with fluidic connectors (Nanoport, Upchurch Scientific, Inc., Oak Harbor, WA) to couple both fluid and sample introduction tubing. The separation capillary channel was 50 µm in width and 500 µm in length and the beam was focused such that it completely filled the channel. This opto-fluidic approach results in a high photon density within the separation region where particles must experience the laser radiation force. The design involves two etched plates bonded on either side of a plate containing the separation channel, shown in Fig. 1(a). The clean fluid flow and the sample injection stream are located on the front side, Fig. 1(a). After the fluid flow and sample passes through the separation region in the center plate,it travels through the etched channels in the rearmost plate and subsequently into 100 µm i.d. Teflon tubing (Upchurch, Inc., Oak Harbor, WA) for collection. A top down perspective photograph of the etched fluid conduits and the separation channel is given in Fig. 1(b). Particles are retained in the channel and also against the wall in the bottom microfluidic entrance to the separation channel shown in Fig. 1(b). Some particles enter the channel and are optically accelerated against the flow toward the wall, while others enter the regions before the channel and are forced directly against the wall. Fluids were controlled with two syringe pumps (NE-1000, New Era Pump Systems, Inc, Farmingdale, NY): one for fluid flow and the other for sample injection. The flow rate used in the experiment was 3 µL/hr and the injection size was 30 nL–50 nL. For the fraction collection experiments, two 30 min fractions were collected (1.5 µL each at the flow rate used) through an open tube attached to the Nanoport on the exit channel.

 

Fig. 1. Optical chromatography glass flowcell. a) Exploded diagram of the microfluidic device showing the pathway for fluid into and out of the separation channel, and the laser beam focused through the channel in the center glass plate (rectangle region of interest). b) Microscope image of the center channel (from rectangle in A) viewed from the top down (capillary inner diameter is 50 µm).

Download Full Size | PDF

B. anthracis, Sterne strain spores (Colorado Serum Company, Denver, CO) in the spore purification experiments were detected by real time polymerase chain reaction (RT-PCR) using the commercial Ruggedized Advanced Pathogen Identification Device (R.A.P.I.D.) (Idaho Technologies, Salt Lake City, UT). This commercial RT-PCR detection system was used to confirm and test the ability of the preparative optical chromatographic system to purify spore samples. Amplifications were performed using B. anthracis – Target 1 Detection Kit (Part No. 3828, Idaho Technologies) using the fluorescence resonance energy transfer (FRET) technique that is part of the commercial R.A.P.I.D. PCR detection system used. Equal volumes of 2× concentrated reagent mixtures and samples containing B. anthracis spores were mixed. The total reaction volume was 5 µl. Samples containing B. anthracis DNA supplied with the kit as well as samples containing known spore concentrations were used as positive controls. Temperature cycling and fluorescence data collection was carried out according to the manufacturers recommendations. Humic acid was obtained from Sigma- Aldrich Co. (St. Louis, MO) and used without further purification.

Data collection and analysis were performed using ImagePro Plus version 6.0 (Media Cybernetics, Inc., Silver Spring, MD). Particles were counted using automated contrast thresholding algorithms available in the software package. This provided a reproducible method for ascertaining particle counts in injected and trapped bands. Particle imaging was best achieved by spotting 1 µL fractions onto clear poly(dimethylsiloxane) (PDMS) sheets. This served to significantly reduce the spot size (due to the hydrophobicity of the PDMS) and allowed all particles to be counted using a single image at 200× magnification (compared with 5+ images using a glass microscope slide).

3. Results

To demonstrate the potential for sample separation and preparation for other techniques, a mixture of 2 µm fluorescent polystyrene (PS) and 1 µm silica (Si) particles was injected into the system. The laser was turned on for 30 min, enough time for all the particles to have traveled through the separation channel and out of the microfluidic device into a collection vial. The laser was then turned off, and a new collection vial was placed at the outlet for an additional 30 min to collect those particles retained by the laser in the separation channel of the microfluidic device. The particles in the collected fractions (each consisted of 1.5 µL of fluid) were counted externally using brightfield and epi-fluorescence microscopy. The results of this experiment with and without the laser operating are shown in Fig. 2. With no laser operating (control experiment), the majority of PS (1068) and Si (5041) particles eluted in the first fraction, Fig. 2(a), and only a small number of PS (19) and Si (193) were counted in the second fraction, Fig. 2(b), due to variations in particle velocity in the laminar flow profile. Using 2 W of laser power resulted in the complete retention of PS particles (1246); their absence from the first fraction is shown in Fig. 2(c), where only silica particles are visible. When the laser was turned off, the polystyrene particles were released and subsequently collected in the second fraction, shown in Fig. 2(d). This fraction was contaminated by a small quantity of Si (247) particles due to the laminar flow profile and some degree of optical retention with the polystyrene particles. Effectively, the polystyrene particles were moved from fraction 1 into fraction 2 using light.

 

Fig. 2. Microscope images, collected in mixed brightfield and epi-fluorescence mode, of fluorescent polystyrene and silica particles injected into the system and collected in 30 min fractions (1.5 µL) at the outlet of the microfluidic device for the control experiment (no laser used): a) fraction 1, b) fraction 2, and the experiment with the laser (laser on during the first fraction and then turned off to release the optically retained particles into the second fraction): c) 2.0 W laser power, fraction 1, d) laser turned off, fraction 2.

Download Full Size | PDF

The ability to perform a useful purification of particles was tested using B.a. spores, the biowarfare agent that causes anthrax disease, coupled with RT-PCR DNA detection. For these experiments, the object was not to separate two particle types (such as polystyrene and silica in Fig. 2), but rather to separate a particle from a chemical interferent. The spores were injected into the system with a known PCR inhibitor (Ca²⁺) and then optically retained while the calcium ions eluted from the system leaving pure spores. When B. anthracis samples containing 1.2×10⁵ spores/µL plus 500mM CaCl2 were injected (50 nL) into the system, without laser purification, RT-PCR DNA detection was not possible. The laser was operated at 5W for 30 min (first fraction) to collect and purify the spores and then switched off to release them for collection in the second 30 min fraction (for subsequent PCR detection). Figure 3 shows the results for complete inhibition of the PCR without the laser, Fig. 3(a), and a positive PCR detection of B.a. when using the laser, Fig. 3(b). Control samples (containing B. anthracis DNA and whole, pure spores) were run for each example (with and without the laser) as given in Fig. 3. The RT-PCR results can in general only be regarded as semi-quantitative; our replicate runs were typically within 10% of one another. The data demonstrate 1) the capability to optically retain enough spores for successful PCR, and 2) the ability to purify the spores in order to obtain successful PCR amplification given a sample containing an inhibitor.

 

Fig. 3. PCR DNA analysis results a) without the laser, and b) with the laser. The positive controls used were standard B.a. DNA (positive control), whole B.a. spores (spore control) at 200 spores/µL and 400 spores/µL, and the negative control was water. The curves were offset for presentation clarity.

Download Full Size | PDF

A further challenge of the technique was performed using B.a. spores suspended in a 477 mg/L humic acid solution. Humic acid is an environmental chemical interferent derived from soil that inhibits amplification of RT-PCR. Fig. 4 shows the RT-PCR DNA analysis of a sample containing humic acid; an image of the actual sample injected into the system is shown in Fig. 4(a), inset A. The injected spore sample produced a false negative when injected without the laser operating, Fig. 4(C), and produced a positive result for injected B.a. spores when the laser was operating, Fig. 4(b). The positive result in Fig. 4(b) was possible as the spores were held against the wall by the laser and the humic acid rinsed out before the laser was blocked and the spores eluted from the system for subsequent testing via RT-PCR.

 

Fig. 4. PCR DNA analysis of a sample processed with the optical chromatographic separation system containing B.a. spores spiked with 477 mg/L humic acid (Image of the injected sample, inset a). PCR results given b) with the laser operating and c) without the laser operating. The positive control shown for standard B.a. DNA (positive control), and the negative control was water.

Download Full Size | PDF

The ultimate bulk separation capacity of this technique was investigated by making concentrated mixtures of polystyrene particles for injection to determine if optical retention was possible under such conditions. Figure 5 (Media 1) shows the injection sequence for a concentrated sample containing 2 µm polystyrene and 1 µm silica particles. The sample was prepared such that it contained approximately 120,000 particles (half polystyrene and half silica) per injection (50nL). The flow direction and laser propagation direction (focused into the channel) are shown in Fig. 5(a), before sample injection. Upon injection, polystyrene particles are forced against the wall by the laser pressure and become retained, Fig. 5(b). The large quantity of polystyrene particles can be seen curving toward the wall in Fig. 5(b) under the influence of the laser, where they begin to accumulate. Silica particles at this flow rate (50nL/min) are not retained. Midway through the experiment, the cavity outside the separation channel is half full, Fig. 5(c), and near the end the channel is almost completely full, Fig. 5(c). The injected band required approximately 2 minutes to completely pass through the system (silica particles elute and polystyrene particles are retained with the laser operating at 5W). Samples of the polystyrene particles retained in the system were collected, diluted and counted under a microscope. Using this method the efficiency of capturing 2 µm polystyrene particles was estimated to be 97%. The total throughput of this experiment was 1000 particles (polystyrene and silica) per second.

 

Fig. 5. Laser retention of an injected band of concentrated 2 µm polystyrene particles. (Media 1) a) flowcell before sample injection with the laser operating at 5 W; b) particle stream just entering the flowcell, forming a mound on the wall; c) half way through injected band – particles fill up region below channel; d) Near the end of the injected band, the region outside the separation channel is almost completely filled with particles.

Download Full Size | PDF

4. Conclusions

We have shown that preparative sample separation and purification can be achieved for entire injected bands of particles using a modified optical chromatography system that operates as a tunable (laser or flow rate) filter. This is a significant advance over previous optical chromatography experiments which interrogate only a small fraction of the injected particles. Fractionated samples have been collected external to the system for evaluation using alternative methods. Both optical microscopy and DNA analysis using PCR have been coupled with the optical chromatographic method and apparatus. In our results, the commercially available PCR detection system produced a false negative result for samples containing detectable concentrations of B.a. spores in the presence of inhibitors (CaCl₂ and humic acid). Extrapolated to a real world scenario, such a false negative could have dire consequences for first responders and/or military personnel. The preparative optical chromatographic system was able to purify the spore samples such that the PCR detection system was able to detect the B.a. spores and yield the correct true positive result.