A university project is underway, aiming to develop a machine tool capable of attaining 50nm accuracy with 5nm repeatability. In addition, it aims to make that machine perhaps one-tenth the size of anything before and at one-tenth of the price, as Andy Joslin of Delta Tau explains.
A four-year university project - which involved 36 partnering organisations from 13 countries with eight of the participants drawn from the UK - has been focusing on developing multi-axis machine tools capable of producing complex 3D geometries to nanoscale tolerances.
Two types of machines are being developed, a five-axis machine that is essentially like a conventional machining centre; and a three-axis test bench that, in principle, is a turning centre. Development is driven by the increasing miniaturisation of systems and also by the new features, functions and benefits that micro-surfaces can deliver. These include anything from sophisticated lubrication designs in which oil can be retained and directed to the surfaces needing it using highly accurate microgrooves, to the manufacture of miniature filters for portable blood testing apparatus in which the micropores are laser machined to incredible accuracies.
Recently, new demands in the fabrication of miniature/micro-products have appeared, such as the manufacture of microstructures and components with complex 3D shapes or freeform surfaces. These microstructures possess some special properties, including light guiding, anti-reflecting and self-cleaning, among others, and are useful in microlenses, micromoulds, laser targets and so forth.
Take, for example, the matrix of tiny prismatic lenses that reflect light within a mobile telephone's screen; while this functionality exists before finishing the part at a nanoscale, the efficiency is increased by nanomachining, which, in turn, reduces the number of LEDs required to light the screen and so lowers power consumption accordingly.
Micromachining for multiple functions
The fabrication of real 3D miniaturised structures and freeform surfaces is also driven by the integration of multiple functions in one product; for example, adding light guides to existing mechanical components or aiding lubrication methods by using micromachined surface geometries. As one example, the smoothness of lenses in a camera or telescope increases the accuracy, effectiveness and quality of the resultant image. A tedious manual process of machining (lapping), grinding and polishing traditionally produces glass lenses but, in ultra-precision machining, using a single-point diamond cutting tool, a surface roughness of less than 10nm can be achieved. This is especially important today in producing optical microstructures such as DVD, camera and photonics lens, and optical fibre for telecommunications. Many of these lenses can now be moulded from plastic materials as a result of the ultrasmooth finish that can be achieved within the mould tools.
Currently, design advances in micro electro-mechanical systems (MEMS) devices, together with their increasing acceptance by industry, is one of the major driving forces for making microcomponents. Silicon is the traditional material for making MEMS or microsystems, but many other materials, including a range of polymers and even some metals, have emerged for the increasing number of applications that are becoming relevant for micro-products. While early MEMS relied primarily on laser micromachining, the silicon surfaces, which is largely a 2.5D application with interpolated geometry in the horizontal plane but with a fixed depth of cut, more recently complex 3D geometries have been deployed in devices. For example, life sciences are an emerging application area for MEMS requiring the manufacture from glass, ceramics, metal and plastics - rather than just silicon - of microcomponents for disposable blood testing cartridges.
But herein lies a small problem - at the nanoscale the current high-accuracy machines cannot produce 3D components with sufficiently reliable repetition, thus leading to high scrap levels of expensive parts. The issues to be tackled include both the behaviour of materials under machining and the structure and motion control system of a machine tool.
Like machining jelly
On the material side, the University observed hard ceramic materials that exhibit ductility at the machining face when removing material at nanoscale, which is therefore likened to machining jelly. Further work was required to model material behaviour at the nano-level.
On the machine side, no single element can be treated in isolation: feedback impinges on gain, gain on stability, inertia on dynamics, friction on stiffness, and so on. The machine currently most advanced in its development is a preliminary three-axis turning machine. The current construction is mounted on a high-density granite platform and uses a direct-drive rotary servo for the c-axis, a linear motor for the cross slide (y-axis) and a piezo-electric actuator to give fine motion in the x-axis.
A Delta Tau UMAC controller provides motion control and also factors in environmental conditions (temperature, humidity and dust), and then communicates according to MASMICRO framework protocols. Tool handling and inspection, component handling, inspection, placement and lubrication must all be integrated into this holistic platform.
A new motor was designed for the spindle, and an air bearing and a linear motor have been adopted to obtain the level of stiffness required for nanoscale machining. In parallel, the MASMICRO team at the University provided extensive FE (finite element) analysis of the machine design and critical parts in order to come up with a final design.
Sub-micron tool tip radii
Cutting tool production makes use of existing methods of laser machining diamond tool tips to sub-micron accuracy. Tools can now be made with tip radii measured in nanometres, but a reliable method of measurement is still to be developed.
If the MASMICRO project achieves its objectives, then Europe will be at a distinct advantage in the nanotechnology field, and the UK potentially a major player.
For nano-level machining, mechanical stiffness is of overriding importance, so not only must a control system react to NC commands, but also respond to external influences - and both in a very short time. To deliver a stiff (responsive) NC axis, the speed at which current can be injected into a motor in order to provide torque or force is critical. Torque figures in excess of 30N/micron are not untypical for directly driven systems using linear motors. For an NC system there are three feedback control loops that must be closed before action can be taken: position (axis), velocity (motor speed) and current (motor torque). An error in position calls for more or less speed, and an error in speed calls for more or less torque (current).
Delta Tau has for a long time maintained that the best control philosophy is to incorporate as many of the control loops in one processor as is practically possible. This approach has seen the company migrate the velocity loop into the position control with the current loop similarly integrated.
Positional resolution measured in picometres
The net effect of this development, together with the introduction of some very-high-resolution laser interferometers and interpolation, is to produce positioning control systems that are potentially capable of resolving to just 10-12pm (10-12picometres). To achieve any level of positional accuracy, it is necessary to resolve at least an order of magnitude lower. The speed at which information about any loop is transmitted is clearly relevant too, and Delta Tau has eliminated one of the biggest sources of delay in processing feedback data, digital-to-analogue (DAC) and analogue-to-digital (ADC) conversions at the amplifier/position control interface. This is achieved by employing direct pulse-width-modulation (PWM) in the servo amplifier. By doing this, and deploying high-accuracy, stable-current sensing devices to control the current loop, Delta Tau has attained exceptionally high loop gains that increase stiffness or the speed when the system reacts to feedback signal errors. Another important factor in positional error control is how often the position loop is closed and errors recalculated. This is achieved by loop closure frequencies measured in just a few microseconds.
The UMAC system provides motion control with interface for six axes of direct PWM. There is also a six-axis interpolator card that offers 4096 times multiplication or interpolation of the incoming feedback signal to deliver high accuracy at high speed. By using the internal interpolation calculations, as opposed to an interpolator built into the feedback device, the system avoids errors caused by corruption of the high-frequency positional data feedback from the encoder (1nm resolution easily gives thousands of megahertz frequency for position feedback).
Nanoscale in context
To get a sense of proportion of the accuracies required, it must be realised that a full stop on a printed page could cover an area of 600 microns, while a typical human hair is about 200 microns in diameter. A red blood cell is about 8 microns. A nanometre is nine orders of magnitude smaller than a metre (10-9 m), which approaches atomic measurements; a single atom of gold, for instance, is 0.1441nm across. The naked eye cannot detect particles smaller than 10-5m, so the human eye loses its efficacy four orders of magnitude above the units proposed as manufacturing tolerances. Taking this into account just demonstrates the amazing results and achievements of this multi-agency and company collaboration.
Delta Tau products are distributed in the UK by Micromech Ltd.