How Lasers Are Boosting Glass Microfabrication

They're enabling further advances in MEMS and microfluidic diagnostic applications.

Glenn Ogura, Resonetics

Borosilicate or its tempered trademarked version "PYREX" is widely used in MEMs and microfluidic diagnostic applications. (Although fused silica, which is formed by pure silica, without the addition of metal oxides, has superior optical transmission, it is much more expensive than borosilicate glass owing to more complex processing requirements associated with its very high melting point.)

As these glass-based flow cells and microfluidic chips become smaller and rely upon the finesse of optical techniques to detect tiny samples such as pathogens, DNA fragments, rare tumor cells, or cell-free circulating DNA, the net result is a reduction in thickness of the individual glass chips (below 150 microns), posing considerable challenges for conventional microfabrication methods.

Conventional Glass Microfabrication Methods

The microfabrication of glass-based microfluidic chips requires drilling, cutting and etching to create features such as entry via holes, micro-channels or flow cells.

Aside from mechanical CNC drilling and milling of large features (typically less than 500 microns), abrasive machining is also a common technique to drill holes, employing a contact or conformal mask (i.e.; metal or polymer resist) to define the hole pattern, followed by a high pressure particle jet directed at the surface. However this technique cannot drill hole diameters below 500 microns, has limited aspect ratio capability (e.g. 2.5:1), an associated taper angle of 22 degrees (or more), and most important, this technique is challenged by the handling of thin glass slides due to the high pressure flow of the  particles and carrier fluid.  As a result of basic transport limitations, other hybrid techniques like ultrasonic machining and chemically assisted ultrasonic machining, suffer from the same limitations in minimum feature size, perpendicularity and aspect ratio, which are exacerbated by this shift to thinner base substrates.

Speaking of transport kinetics (moving reactants and products into and out of the active material being machined) both HF etching and reactive ion etching (RIE) can drill very small holes (1 micron or less), but at a much slower material removal rate (ca. <10 um/min). All etch processes require a conformal mask to define the hole pattern, creating additional process steps. HF etching is an isotropic process resulting in the holes or channels being etched near the tops of the features nearly twice as wide as deep.  All etch processes suffer from imperfect selectivity between target substrate and mask material in addition to variability in isotropy, limiting the practical aspect ratio to 1:1 with HF or 5:1 with RIE processes, even when special chemistry is employed to combat the otherwise U-shaped or V-shaped structures created that restrict designers from placing holes or channels close together. Deep reactive ion etching (DRIE) has a much higher aspect ratio (50:1) but is constrained by a slow etch rate of ~250-600 nm per minute that translates into higher costs due to lower throughput and more expensive vacuum and gas handling systems. Ultrasonic drilling benefits from being a direct process (no conformal mask) but cannot reach feature sizes below 75 microns and the vibratory nature of the process poses difficulties for drilling and handling of thin glass slides.  In all cases, the roughness of the final parts is often too high for medical and diagnostic applications because of restricted flow dynamics and surface friction.  In those cases where it is acceptable for the application, the final performance of the device is limited due to the roughness and potential reactivity of the resulting surfaces.

Laser Microfabrication of Glass

Over the years, CO₂ lasers have been used for drilling and dicing of thick glass sheets and slides, limited by the hole diameter (recently as small as 50 -70 microns), but more importantly by the thermal damage and debris generation resulting from a variety of industrial and laboratory processes. The introduction of ultrafast lasers, whose pulse duration (the amount of time the laser flash exists in time, typically measured in picoseconds (10^-12) or femtoseconds (10^-15)) is six to nine orders of magnitude shorter than typical CO₂ laser pulses has opened up exciting opportunities for the fabrication of thin glass microfluidic chips. 

Unlike abrasive machining, HF etching and RIE/DRIE etching, ultrafast laser machining is a single step, direct process, not requiring a conformal mask. Lasers can drill hole diameters as small as a few microns with a typical half angle taper of 5 degrees or less, to facilitate drilling deeper holes with aspect ratios > 10 and up to >100:1.  Ultrafast lasers can also cut thin glass (thickness of less than100 microns) without the inherent process drawback of particle blasting or ultrasonic drilling.  Ultrafast laser pulses exist for such brief periods of time that each successive pulse is able to excite the target material before the previous pulse has fully dissipated and long before the heat of the pulse radiates out into the bulk. Particularly interesting are so-called non-linear processes where the short pulses are employed to form very limited channels of ionized gas within the target material. These channels are rarely larger than 5 um in diameter and thus produce features that can be used to singulate glass sheets or to machine curvilinear features within the sheets.  

The advanced development of faster, cleaner glass cutting processes, especially with smaller and thinner glass microfluidic chips and MEMs devices, opens up opportunities for laser microfabrication to be an attractive, cost-effective high volume manufacturing process.

Glenn Ogura is the SVP Market Development at Resonetics (Nashua, NH).

[Image courtesy of Resonetics]