Laser-based glass micromachining


By Philippe Bado, Translume

The vast majority of microdevices are built out of silicon using photolithographic processes. This activity is supported by a huge industry, which provides the necessary materials and tools. Yet, while the return from investment in silicon technology has been enormous, it is clear that for some applications silicon is not an appropriate base material - and that for a significant subset glass is better.

Many designers have contemplated glass, but fabricating microdevices out of it is not easy. Glass producers and machine tool manufacturers have not developed glass micromachining capabilities. With few exceptions (optical fibers, for example) the glass-based devices people use today are produced with fifty-year-old technology, using tools designed to fabricate much larger components. Consequently, even when glass is the most desirable material for a given application, the microdevice designer is generally forced to settle for another material.

This situation is changing. For the last decade, numerous research groups, including our group at Translume, have been working to develop processes to micromachine fused silica (a high-end, ultra pure glass). Translume has developed a manufacturing platform that relies on femtosecond lasers to micromachine fused silica substrates. This workstation has a capability to machine complex geometric contours and shapes in three dimensions.

This computer-controlled tool is able to ablate glass, to induce local permanent index of refraction changes in glass, and to locally change the chemical reactivity of glass. We typically use our index change ability to create waveguides in the glass substrate, and we rely on the other processes to shape the boundaries of the glass substrate. In addition, often we create microfluidic features using the same tool.

Initially this novel manufacturing capability was to be used to manufacture glass-based components for the telecommunication market. As this market imploded and our early customers disappeared we were forced to re-evaluate our business model. We adapted our offering. Today, Translume is commercializing its micromachining capability in the form of a foundry or contract machining service.

The mobility of a fused silica linear translation stage is provided by integrated fused silica flexures, shown here. Photo courtesy of Translume
Click here to enlarge image

In parallel, we have moved to more receptive markets - our fabrication capability is used to manufacture small instruments and sensors for customers in the biomedical, aerospace, and defense industries. Yet before our manufacturing process finds wide acceptance, we still have to overcome two psychological barriers: We work with an uncommon material and our process is a linear process.

Unlike any other foundry, we work exclusively with fused silica. This is a well-known material, which has been used for decades. Ironically, it is not considered a high-tech material, except when deposited on top of a silicon wafer. Yet, fused silica offers a set of characteristics that compare favorably with silicon. It is transparent from the deep UV to the mid-infrared. It is compatible with all industrial and biological fluids (except hydrofluoric acid). And fused silica offers excellent thermal stability (its expansion coefficient is similar to that of Invar). These characteristics are well known.

Less known, or even counter-intuitive, is the fact that fused silica is a good material for manufacturing mechanical pieces that must flex or move (what we call MEMS in the silicon world). Glass is the epitome of a breakable material, or what material scientists call a brittle material; and thus one naturally assumes it can’t be used to manufacture MEMS. At the macroscopic level this is indeed true, but at the microscopic level fused silica is rather elastic and can be used successfully to manufacture glass MEMS (GMEMS).

The other mental barrier we face is related to the linear nature of our manufacturing process. Many MEMS industry experts have predicted that our process would find no commercial acceptance, as it is a serial process (opposed to the parallel process of photolithography). In response, we point out the overwhelming commercial success of CNC machining of metal elements, including advanced small components such as cardiac stents.

The Translume process starts with a very short laser pulse, on the order of 100 femtoseconds, that is focused to a point inside a glass substrate. At the focal point, the light intensity is so great that the glass is turned instantly to plasma through nonlinear absorption. Yet, since the pulse is extremely short, the glass almost immediately re-solidifies. With the proper laser parameters, one can control the local nanostructure of the re-solidified material and modify its physical properties.

We can locally change the index of refraction of the glass, or we can also locally change the chemical reactivity of the glass. The former process is used to manufacture optical microdevices while the latter process is used to create microfluidic and micromechanical devices. By combining these two processes, one can create microdevices with a unique and highly desirable set of optical, mechanical and fluidic properties.

The Translume process uses a direct-write workstation that provides an extremely short laser pulse, on the order of 100-femtoseconds, that is focused to a point inside a glass substrate. At the focal point, the glass is turned instantly to plasma but since the pulse is so short the glass almost immediately re-solidifies. With proper laser parameters, one can control the local nanostructure of the re-solidified material and modify its physical properties. Photo courtesy of Translume
Click here to enlarge image

We see an ability to fabricate devices using a mask-less approach as a key economic advantage for some high value markets such as the aerospace industry and the military. The lack of hard tooling for a given device has important cost-savings implications for both prototype development and long-term manufacturing.

The cost of prototype development in traditional lithographic processes is high since each design iteration requires a new mask set, and there is development time lost in waiting for hard tooling to be made. In addition, some long-term costs for direct write processes are lower due to the cost of storing and maintaining masks and other hard tools. The ability to use a single manufacturing step to define both optical and mechanical features dramatically simplifies device fabrication and eliminates alignment issues associated with sequential fabrication processes.

In order to demonstrate this point we recently manufactured a fused silica mesoscale linear translation stage with characteristics similar or superior to that found in similar devices made with traditional techniques. This device provides a mesoscale displacement capability (1 mm range) combined with an integrated sub-100-nm optical sensing accuracy capability. The mobility is provided by integrated fused silica flexures acting as elastic mode compliant elements. The position is read through an array of embedded optical waveguides. The device is fully monolithic, which eliminates most assembly costs.

Activities in glass micromachining have historically been extremely limited. However the recent development of manufacturing processes based on femtosecond lasers is creating a commercial opportunity. We believe that in the next few years glass-based micro-instruments and glass-based sensors will play a much more important role than is generally envisioned today and that engineers and instrument designers may gain a competitive advantage using glass as an alternative to metal and silicon.

Philippe Bado is president and chief technical officer of Translume Inc. ( in Ann Arbor, Mich. He can be reached at