Feature Article

DESIGN

Man and machine find success

The term cyborg was first used by scientists Manfred Clynes and Nathan Kline in 1960 to describe a synthesis of man and machine into a single whole functioning unit. This has largely been the realm of science fiction, but recent developments in the United States have made man–machine systems a real possibility. The goal of being able to couple patient-controlled electromechanical devices such as prosthetic limbs to injury and trauma patients is a major driver for research and development in this area. The Integrate Neural Interface Programme, led and researched by a group at the Department of Electrical and Computer Engineering, University of Utah, (Salt Lake City, Utah, USA) has achieved significant success. It has interfaced the human brain with a brain implant which, connected through an electronic control system, is being used to manipulate computers and prosthetic appendages.

A critical component of the brain– machine interface is the sensor electrode array, which actually sits as an implant in the surface of the brain (Figure 1). This consists of an array of micro needles, each of which can be as much as several millimetres long, which penetrate into the surface brain tissue or a nerve to such a depth that they pick up neuronal or nerve electrical activity. Electrical signals are detected and amplified by the control electronics, transmitted by radio frequency signals to external receivers remote from the head, and these are then interpreted by the control system to produce manipulations of prosthetic limbs or computers. Brain impulses are generated by the thought processes of the patient, but at first the patient is unable to control anything. The patient has to learn how to generate the correct mental activity to make the interfaced electromechanical system respond at all and then further learning is required to respond in a predictable and controllable way. Early results show great promise and additional research is exploring new ways of improving the implantable brain sensor array.

Figure 2: Actual implantable micro needle sensor array. (click image to enlarge)

Early needle array brain implants (Figure 2) were made using silicone dicing and etching techniques and these gave reasonable electrode length to diameter ratios (known as aspect ratio: feature depth to width ratio). There is, however, a limitation in achieving ultra high aspect ratios using the dicing technique, not least because of the strength of the needle, which becomes too delicate to use when the aspect ratio becomes extremely high. Aspect ratios of much greater than 10 present real machining and in-use difficulties for many materials when the components are micro in size. The University of Utah’s group looked at alternative techniques of making the arrays and decided that different materials may give stronger micro needles. They decided to investigate tungsten carbide, a material that is electrically conducting. This material is hard and chemically stable in the environments in which it is to be used in this application (Figure 3a). The main issue with tungsten carbide is that it is exceptionally hard to machine using conventional machining methods. As a result the group turned to microelectrodischarge machining (micro EDM), which is a noncontact thermal ablation machining process. Using microwire EDM, it is possible to manufacture the high aspect ratio needle structures that are required in tungsten carbide (Figure 3b). The square topped needles were produced by running an extremely thin EDM wire up and down across the surface of the tungsten carbide work piece so that a series of thin blades were produced, each of which were 120 microns in thickness. The work piece was then rotated through 90 degrees and the EDM wire was run in the same way over the work piece again, thus cutting the blades into square columns. The group then processed the needle array to produce fine sharp points by chemical etching and cleaning techniques. Further processing included a clever way of electrically isolating each needle with a polymer so that each needle would detect signals with a reasonable degree of vertical special resolution within the brain. The needle shafts were coated with an electrical insulation leaving a small length of needle at the tip uninsulated so that signals can be detected at accurate depth points within the brain.

The EDM process

Figure 3a: Schematic of tungsten carbide needle sensor array.

EDM has been employed in industry for many decades and is used routinely for cutting profiles and shapes in metals for items such as injection mould tools. It is understood from an application point of view and although it is well characterised for larger components, it is not so well characterised for micro work and each process requires a significant amount of process optimi-sation. When working to one or two micron tolerances and submicron surface roughnesses, micro EDM work demands much greater control and setting of the process parameters.

All EDM techniques use the principle of creating electrical sparks between an electrode and the work piece while being submerged in a dielectric fluid. A voltage difference is applied between the work piece, which must be electrically conducting or semiconducting, and the electrode. When the voltage is high enough, an electrical discharge takes place, which generates heat in the work piece or the electrode and usually both. The dielectric fluid has several functions: being of high dielectric strength it reduces the spark gap; it helps cool the work piece; and it acts as a debris flushing mechanism. As the spark discharges the electrical charge between the electrode and the work piece, intense heat is generated in a small localised area of the work piece and because each spark is of short duration a small ablation spot is created on the work piece. In micro EDM machining the pulse duration is controllable, but typically in the order of a few tens of nanoseconds; and the ablated spot size varies depending on the pulse energy used and the type of material and dielectric, but can be as small as a few hundred nanometres. The material from the impact of the spark dissipates as a plasma, which cools to a debris cloud and this is carried away by the dielectric flushing liquid. This creates a small crater or pit in the work piece, and by repeated pitting the shape of the electrode is eroded into the work piece. The two types of EDM most frequently used are:

Figure 3b: SEM image of actual tungsten carbide needle sensor array prior to point sharpening. (click image to enlarge)

Volume EDM. This uses a solid shape electrode, which is the counterpart of the material to be removed in the part. Volume EDM is useful for drilling deep holes and slots and complex shaped cavities with high aspect ratio of typically above five.Micro EDM processes can be used for milling by moving the part in the x, y or z direction.

Figure 4: Microwire EDM cutting a profile in titanium alloy.

Wire EDM. This uses a continuously running thin wire as the electrode, which is useful for cutting profiles in plates (Figure 4). It was this method of profile cutting that was used to make the needle electrode arrays. A variation of this process has been developed in which the electrode wire is used as a cutting tool for EDM turning of round components. Figure 5 shows a micro touch probe that has been turned using micro EDM turning. The spark finish is clearly visible in the magnified image.

EDM uses electrodes made from many different materials such as copper, beryllium copper, graphite, brass, tungsten and tungsten carbide. Wire EDM typically uses brass coated steel wire, tungsten wire and other materials of good conductivity, high tensile strength and high melt point. Inprocess variations are controlled effectively by ensuring that the electrode material is matched to the work piece material. It has been found that different material combinations give different spark characteristics for a given set of process parameters and for best results trials are necessary to establish the optimal combination. The spark gap can be large, typically more than 30 microns or as small as 3 microns depending on parameter settings and dielectric contamination. Achieving optimised and well controlled process conditions is essential for achieving micron tolerances.

Figure 5: Tungsten carbide electrode made by micro wire EDM turning. Ball diameter is 70 microns and shaft diameter is 45 microns. (click image to enlarge)

Micro wire EDM is particularly suited to making small parts with tight tolerances and with good quality surfaces finishes. Dual micro wire EDM machines use two wires of different diameters and the finest wire that can practically be used is 20 microns diameter, but 50 micron wire is more frequently used because it is a little easier to handle and see.

Benefits of micro EDM

Figure 6: Micro fluidics mixer. Minimum channel width is 135 microns. (click image to enlarge)

The versatility of the process and the shapes that can be produced are the main advantages of micro EDM. The process is also useful for machining and producing extremely small and delicate parts because the process is effectively noncontact. The wire does not touch the surface, therefore there should be no lateral forces on the part. Because it is a thermal process, micro EDM is excellent for machining hard substances such as hardened carbon steels or tungsten carbide or difficult to machine metals such as titanium.

EDM is used for many applications and is used extensively for making one-off or small batches of extremely small parts in specialist steels, which makes it ideal for medical applications. It is also useful for manufacturing micro injection mould tools, especially for high production volume polymer parts. One medical application is shown in Figure 6. This is a micro fluidics mixer used for mixing medical reagents in an analysis instrument. Turbulence and hence stirring of the chemicals is created because the micro channels have a high surface to volume ratio. Many fluidics devices can be generated using micro EDM techniques.

Extremely small cutters for physical removal of material are used in conventional micro machining and milling techniques. These are good for certain micro applications where the aspect ratio is low, typically in the order of 1 or 2, and the minimum feature thickness required is not less than 30 or 40 microns. The EDM process, especially wire EDM, is particularly useful for making high aspect ratio features. Holes of 200 microns diameter and 6 mm deep are possible and electrodes of high aspect ratio and 20 micron diameter and 2 mm long are possible. Laser micromachining is advancing rapidly and small features are possible. However, these also are characterised by having a limited aspect ratio especially where extremely small holes or slots are concerned. Table I shows a comparison of the micromachining techniques.

Extending possibilities

Table I: Comparison of techniques for micro machining. (click image to enlarge)

The concept of neural interfacing sounds simple, but in practice it presents many challenges that are stretching the limits of technology. One major challenge is that the brain has more than 100 billion neurons of which only a small proportion are used for motor functions, and finding and detecting impulses from these small areas of the brain is challenging. The current technology uses only a relatively small number of electrode sensors: approximately 100 needles. Each detects signals from several hundred neurons, but the drive now is to increase the number of electrodes and hence more neuron clusters from which brain activity information can be extracted. One of the essential aspects for the success of a project like this will be extending the limits of micro engineering.

Micro EDM is a useful tool for machining micro features as small as a few microns in size with high aspect ratio in many different conductive and semiconductive materials. This capability is being harnessed for the benefit of implantable neural interface device research and development.

Acknowledgements

This article was prepared with the kind permission of Prashant Tathireddy, PhD, Research Assistant Professor, and Florian Solzbacher, PhD, Associate Professor and Director of Integrated Neural Interface Programme, Department of Electrical and Computer Engineering, University of Utah, Salt Lake City, Utah, USA, in collaboration with Fraunhofer IZM and Fraunhofer IBMT. Part of this work was funded through ProfX2 Fellowship programme by Fraunhofer Gesellschaft.

MicroBridge Services Ltd (MSL), a company wholly owned by Cardiff University, has been set up to exploit the commercial potential of the micro and nano engineering capability developed at the Manufacturing Engineering Centre at Cardiff University, UK. The micro EDM machining was undertaken by one of the Centre’s researchers and MSL’s production manager, Andrew Rees.

Dr Robert Hoyle is Business Development Manager at MicroBridge Services Ltd, Manufacturing Engineering Centre, Cardiff University, The Parade, Newport Road, Cardiff CF24 3AA, UK, tel. +44 29 2087 0018, e-mail: hoylert@cardiff.ac.uk www.microbridge.cf.ac.uk