Fabricating tools for optically actuated microrobotics

Over the past decade, nanofabrication technologies such as direct laser writing (DLW) have made it possible to build near-arbitrarily shaped 3D microstructures.¹ If we disperse these structures in aqueous media, it is possible to pick up and dynamically actuate them in 3D using optical tweezers: focused beams of light that can trap and manipulate dielectric particles.² Such capabilities herald a variety of novel applications, including new ways to explore, image, and characterize the nanoscale world.^3–6

To increase the technology's potential, we can hand over control of these microscopic tools to a computer. By automatically monitoring their positions in real time, and guiding the locations of the optical traps using feedback, we can program the microtools to react to their environment, and can choreograph their motion with nanoscale finesse. This represents a new form of optically actuated microrobotics,^{7, 8} which relies on the union of cutting-edge fabrication, manipulation, and tracking technologies.

We build bespoke microstructures using a commercially available DLW system (Nanoscribe Photonic Professional), as shown in Figure 1. DLW relies on the local solidification of a photoresist at the focus of a laser beam using a two-photon polymerization process.¹ By sweeping the beam through the photoresist, optical-quality 3D structures are reproducibly “drawn” with feature sizes down to ∼100nm. When we have fabricated an array of structures on a surface, we immerse them in a droplet of water, gently disengage them from the substrate, and inject them into a microscope sample cell.⁹

Figure 1. Microscopic tools fabricated using direct laser writing.³(a) An optical microscope image of three different designs immersed in water in a sealed sample. Inset: A schematic of how a microtool is stably optically trapped. (b, c) Scanning electron microscope (SEM) images of two of the microtools shown in (a).

Once they are in the sample chamber, we can pick up and actuate the microstructures using a microscope equipped with holographic optical tweezers (HOT).¹⁰ The tweezers use the high-intensity gradients produced in a tightly focused laser beam to trap micron-sized dielectric particles. If the position of the laser is translated through the sample, the particle moves with it. By incorporating a spatial light modulator (a device that can shape light) into the HOT system, we can split a single laser beam into many separate optical traps in the sample, each of which we can independently control in 3D in real time.^{11, 12} Therefore, microtools are held stably using several optical traps, and we can control both their position and orientation in 3D. Using this technology, it is possible to literally ‘fly’ the tools around the 3D volume of the sample.

To close the loop, the computer controlling the optical traps requires information about the current position of the microtools. We achieve this using stereoscopic video tracking, where viewing an object from two directions simultaneously enables recovery of the object's 3D movement from its apparent 2D movement in each viewpoint using parallax. Employing this principle in a modified microscope,^{14, 15} we track the 3D position and orientation of each microtool to nanometer-level accuracy (enough to observe them undergoing Brownian motion¹⁶), in real time at a rate of up to several kilohertz. We then use a feedback loop to control the position of the optical traps manipulating the microtool, based on the tool's current position.¹⁷ This enables preprogrammed autonomous behavior of the microstructures and represents a step toward optically actuated microrobotics.

Figure 2 shows how we have used this technology to develop a new form of scanning probe microscope (SPM).^{7, 18,19} Here, we automatically raster-scanned a microtool along the side of a vertical wall in the sample, building up an image by feeling the bumps in the surface.^20–23 As shown in Figure 2(c), it is not possible to see the side of the test object directly from the microscope image, and we reconstruct the shape of the surface, shown in Figure 2(b), from the tracked motion of the probe as it slides over the undulations. Unlike traditional SPM, the probe can operate inside a sealed chamber. The sample can be accessed from any orientation (in this case, the side), allowing investigation of the topography of areas previously inaccessible to SPM, with a resolution that could approach that of scanning electron microscopy.⁷

Figure 2. Surface imaging using a microtool.^7,13 (a) SEM image of the test object to be imaged. (b) The image of the test object recorded using the microtool in aqueous conditions. (c) Optical microscope view of the microtool during the image process. (d) SEM image of the microtool. (e) Schematic showing how the microtool is optically trapped. All scale bars are 5μm in length (∼1/10 the width of a human hair).

The flexibility of DLW offers much freedom in the choice of microtool shape,^{24, 25} and we are investigating a range of applications. For example, Figure 3 shows a prototype of a microscale rotator, designed to rotate individual cells about an axis parallel to the focal plane, enabling increased imaging resolution.⁹

Figure 3. A prototype microscale rotator. It is shaped like a crankshaft, with a claw at one end, and held using three optical traps (the positions of which are labeled with colored circles). When the pink trap is rotated in and out of plane, the device spins on its axis. By holding a cell between two opposing rotators, it can be controllably rotated.

In summary, we have described how we can combine recent advances in nanofabrication, optical micromanipulation, and high-resolution 3D particle tracking technologies to create a new form of optically actuated microrobotics. We have shown how this enables a novel type of SPM that will complement existing techniques, and we described work toward a new technique to control the orientation of single cells. Looking forward, we aim to take advantage of the inherent multiple beam capabilities of HOT, and scale up to multiple devices operating in parallel. This would speed up surface imaging rates. It would also enable multiple tools to perform separate preprogrammed tasks within the same field of view, such as simultaneous mechanical stimulation and force measurement on cells at independent locations. The field of microrobotics is still in its infancy, and we believe there is a wide range of potential applications to be realized in the future.

We wish to acknowledge the Universities of Bristol and Glasgow, where this work was undertaken. We also wish to thank the UK Engineering and Physical Sciences Research Council for financial support. We wish to acknowledge Stephen Simpson, Miles Padgett, Mervyn Miles, Graham Gibson, Richard Bowman, John Rarity, and the other members of the team with whom this work was a collaboration.