09 Oct 2006
An optical manipulation and analysis platform that can fit onto a fingernail could signal a significant change in the field of microfluidics. Simon Cran-McGreehin, Thomas Krauss and Kishan Dholakia from the University of St Andrews reveal what's on offer.
The 21st century could herald a revolution in the way we perform biological science. Our aim is to produce microfluidic systems that analogously provide scientists with the ability to perform large-scale and parallel automated studies in the physical sciences – notably in biology.
Light is the cornerstone for such studies. It offers a powerful, non-invasive and reconfigurable way in which to image cellular samples; induce fluorescence and scattering; and move and manipulate objects at the cellular level and below. For lab-on-a-chip and microfluidic applications, we need to explore ways to actuate, move and study small volumes of analyte in extremely small and confined volumes.
The motivation behind our approach is simple. To date, physicists have relied on external macroscopic laser systems and discrete optical components to couple light into microfluidic platforms to manipulate and sense micro- and nanoparticles. This introduces problems of alignment and coupling losses, as well as placing a lower limit on the size of the apparatus – typically a bench top is needed for a conventional system.
However, recent work between the Optical Trapping and Photonic Crystal groups at the University of St Andrews, UK, bypasses these issues by combining microfluidics and integrated optics. In our monolithic optical micromanipulation and particle sensors, the lasers are integrated right next to the microfluidic flows in channels that are less than the width of a strand of human hair. Chip layout
Each laser is defined, lithographically, in a single piece of GaAs-based laser material giving perfect intrinsic alignment. Microfluidic channels are then fabricated directly on top of the laser material, allowing the light to couple directly into the sample without the need for additional optics. This drastically reduces the size of the system, making it portable and simple to incorporate into existing microscope systems.
Both the concept and device design are fairly simple. We believe that any research group with access to semiconductor processing facilities could replicate the device using materials costing just a few pounds. Indeed, we hope that this technology can be spread rapidly among the biophotonics community, giving access to the advantages of optical methods without the need for specific optics knowledge.
The aim is to make the device as user-friendly and robust as possible. Care has also been taken to avoid any problems arising from sending lasers in and out of the ports of a microscope system. To this end, the device is mounted onto a circuit board and the lasers are wire-bonded to copper tracks, which in turn are connected to a power supply.
The optical power of each laser is controlled by varying the applied voltage (and hence current), up to a maximum of around 20 mW at about 3 V (approximately 200 mA). Such electrical powers can be provided by a computer interface board, opening up the way for automated operation.
Fabricating the device
The starting point is a GaAs chip, typically 6 × 6 mm. The actual device is dominated by 2 mm-long lasers that determine the surface area of the chip.
At the heart of each laser is an AlGaAs/ GaAs singlemode heterostructure, centred 1 µm beneath the chip's surface. Grown epitaxially by German firm Nanosemiconductor of Dortmund, the structure provides vertical waveguiding and contains InAs quantum dots that emit at 1290 nm.
Horizontal waveguiding is provided by etching away the GaAs to leave a ridge 3 µm wide and 750 nm deep that supports only a single vertical transverse mode. An electrical current is injected into the ridges via gold contact pads and SU8-2000 polymer insulation on the etched GaAs confines the current to the ridges where a useful optical mode is generated.
The length of the laser cavity is defined by facets that are etched to a depth of at least 2 µm. The tight vertical confinement of the heterostructure leads to large vertical divergence of the output beam, up to as much as 40°, so the optical power density falls off quickly with distance from the facet.
Of greatest interest is the interface between the lasers and the fluids. The microfluidic channel is etched into the GaAs at right angles to the lasers and passes between pairs of facing lasers. This allows the laser beams to enter from both sides of the channel to give a dual-beam trap configuration.
The electrical activity of the lasers is insulated from the fluid by lining the microfluidic channels with SU8-2000 polymer. A thin layer covers the base and a thicker layer lines the walls, essentially determining the depth of the channel. A glass lid is then adhered to the top of the chip and sealed with adhesive to create a watertight microfluidic channel that can be fabricated in any desired configuration.
The size and position of the features can be tailored to the application. The facet spacing determines the optical power at the centre of the microfluidic channel where facing beams overlap. Closer facet spacings give higher power densities resulting in stronger trapping and larger detection signals.
Meaningful optical forces are exerted up to around 200 µm from the facets, placing an upper limit on their spacing. The lower limit on the facet spacing is determined by the resolution of the current photolithographic techniques that define the SU8-2000 lining on the walls. This places a lower limit of about 20 µm on the facet spacing.
Typical microfluidic channels measuring 30 µm high and 40 µm wide, with a facet spacing of 75 µm, allow the passage of biological cells. The channels can be arranged in various configurations, ranging from straight channels to junctions for sorting and chambers for mixing. Using water or biological buffer solutions in the microfluidic channels provides a medium in which particles can flow into the paths of the laser beams.
Using a single laser, we have guided particles over ranges of around 200 µm. Essentially, the gradient force draws objects onto the optical axis and the radiation pressure pushes them away from the facet.
Pairs of facing lasers have been used to create dual-beam traps, in which the objects are held at the equilibrium point between the two facets. Once trapped, particles can be interrogated. We have demonstrated fluorescence spectroscopy in this configuration and Raman spectroscopy is an obvious candidate for future studies.
Useful operations are possible even when the optical powers are insufficient to trap particles. This has led to the development of two detection methods, both of which make use of the fact that facing lasers feed light into one another.
In the first method, the output power of one laser is found to decrease when a particle passes between the facets. In the second, one laser is reverse-biased to create a photodetector whose photocurrent decreases when a particle passes by.
These methods use the intrinsic properties of the semiconductor lasers and require no external light source – the second method requires no external optics whatsoever. We now plan to combine these operations into more complex arrangements in which the particles are moved around a device, either by optical forces or by an externally generated fluid flow, to different interrogation sites.
A look to the future
As a first step, this new technology makes the power and versatility of optical manipulation more widely available by simplifying the operation and reducing the cost. We believe that this technology lends itself to the world of lab-on-a-chip, in which miniaturized test and measurement systems are integrated into microfluidic circuits.
By configuring the lasers and channels in order to usefully combine the functions of guiding, trapping and detection, a wide variety of multiplexed processes could be conducted, in parallel, on a single chip. This range could be extended further by using external analysis techniques such as Raman spectroscopy.
A more compact approach would modify the laser material to allow the direct, on-chip excitation and detection of fluorescence in particles, such as violet-emitting GaN-based material in conjunction with green fluorescent protein. The technology could also have a use in the realm of atom optics, creating dipole traps that could hold Bose–Einstein condensates.
• This article originally appeared in the October 2006 issue of Optics & Laser Europe magazine.