Microelectromechanical systems (MEMS) are pervading more and more areas of our lives, from initially mainly as vehicular sensors to on-body physiological monitoring in recent years [1
]. Yet, standard photolithographic processes remain limited in their material compatibility, creating the need for cheap and scalable fabrication on unconventional substrates, such as soft and flexible materials. With respect to additive processes to add, e.g., sensing functionalities to such materials, this has led to the implementation and adaptation of various printing processes, from inkjet printing to screen printing [2
]. Although highly flexible, these processes are limited by their reliance on liquid ink and thus material quality, compared to the physical vapor deposition (PVD) of traditional MEMS.
PVD on unconventional substrates can be facilitated with the use of shadow mask lithography (SML), similar to the stencils used in screen printing [3
]. In SML, desired geometries are initially produced in a negative shadow mask—analogous to a photomask, but with physical instead of merely optical occlusions. During the PVD process, this mask is placed in close contact with the substrate so that atoms or ions can pass through the mask apertures to reproduce the intended features. The most significant advantage of SML is that it does not require coating and patterning photoresist on target substrates, which consequently eliminates substrate exposure to potentially harmful solvents, heat, or UV radiation. This makes SML an ideal procedure to pattern on not only biological samples and materials like thin plastic films that are sensitive to solvents but also water-soluble sheets. Additionally, the simpler process flow reduces both cost and time required for fabrication.
Rigid shadow masks, generally made of Si/SiN-based membranes or metal films, are the primary type in use. The former are typically fabricated using traditional bulk and surface micromachining, and allow for mask features in line with the lithography method used (i.e., down to nanometer-scale with electron beam or deep UV exposure) [3
]. Metal-based masks arise out of surface-mount technology (SMT), where these types of stencils are used to apply solder paste to printed circuit boards. They are typically fabricated using similar photolithography approaches to Si/SiN masks or by laser cutting [5
]. SML has enabled micro- and even nanoscale PVD on fragile materials, from gold metamaterial antennas on silk and paper [6
] to indium–tin-oxide (ITO)-based organic light-emitting diodes (OLEDs) on plastic films [8
]. However, the fabrication of rigid shadow masks along with their mounting in a suitable frame (mainly needed for metal films) is exceedingly expensive and time consuming, requiring a wide range of process equipment as well as relevant training. Commercial costs for a 4” wafer-suitable mask can easily exceed €1000, even at relatively low resolution (5–10 µm). This presents a significant limitation, particularly in an academic setting where rapid design iteration is often needed.
Although high mask resolution is naturally critical to nanofabrication [4
] and can also be beneficial for microfabrication [9
], we note that many applications do not actually require the high resolution afforded by traditional shadow masks. Zhao et al. recently published an innovative wearable multianalyte sensor system, enabled by fabrication on top of an adhesive conductive film [10
]. The SML PVD gold underlying all active sensor elements has a minimum feature size of ~0.5 mm. Traditional three-electrode electrochemical cells with >100 µm features are indeed a common example of SML, e.g., on permeable cell culture supports [11
] or on paper [12
]. Ishikawa et al. prepared contact pads (>100 µm) by SML on top of carbon nanotubes for algal detection [13
]. Resistive sensors on collagen films (>1 mm) [14
] and capacitive sensors in polydimethylsiloxane (PDMS; >150 µm) [15
] were similarly realized by SML.
It is in light of these limitations and needs that we consider alternative materials and fabrication methods for shadow masks (Table 1
). Photolithographic shadow mask fabrication, independent of the materials used, is not well-suited for low-cost academic prototyping. Direct photomask writers and robotic lithography systems could facilitate high turnaround, but remain resource and training intensive and uncommon in academic cleanrooms. Laser machining, on the other hand, deserves further consideration, since it is a single-step process (excepting the possible need for mounting in a frame, required to keep thin membranes under tension). Direct photoablation of a wide range of materials is possible with excimer or ultrafast lasers. This “cold” processing yields high-quality cuts (only limited by laser spot size) but equipment costs remain prohibitive [16
]. Processing with other laser systems is largely thermal (i.e., at least some of the material will undergo a solid-to-liquid transition), including the neodymium-doped yttrium aluminum garnet (Nd:YAG) lasers often used for commercial steel or nickel stencil fabrication [17
]. Notably, this also includes CO2
lasers, by far the most affordable type of laser processing available (used CO2
laser cutters can be found from ~€100; other laser types from ~€10,000).
In terms of material alternatives, nontraditional shadow masks have been introduced in recent years, e.g., from polydimethylsiloxane (PDMS) [18
] or polyimide [19
] (both examples using photolithography for mask fabrication). The driving motivation was often to obtain compliant (zero-gap) masks to alleviate the blurring effect arising from the small (10–100 µm) gap that exists between a typical shadow mask and the substrate [3
]. Furthermore, flexible materials are less prone to fracture or damage during fabrication compared to rigid masks. However, lack of stiffness also implies difficulty in maintaining in-plane dimensional stability, unless a suitable (again, laborious) frame is attached. With materials as soft as PDMS (Young’s modulus ~MPa [20
]; or too-thin plastic films), certain shadow mask geometries such as long suspended beam-like or cantilever-like features will additionally have insufficient out-of-plane stability. Some soft plastics like PDMS are moreover likely to contaminate the substrate, e.g., with low-molecular-weight silicone. Shadow masks from stiffer or thicker plastic films have also been used, including in combination with CO2
laser machining [14
]. As seen in those examples, however, the thermal cutting process causes (1) material melting at the cut edges, forming edge beads/burr, as well as potentially inducing new residual stress in the material and (2) ejection of liquified material that can redeposit as debris on the surface—both processes ultimately degrading mask quality.
Herein, we now aim to characterize laser-cut paper as a versatile and advantageous alternative to existing shadow masks (Figure 1
). Specifically, we propose that quantitative filter paper in combination with a CO2
laser cutter—an established approach to paper MEMS and paper microfluidics [22
]—offers excellent synergy for SML, overcoming many of the limitations of other material/equipment combinations considered above. First and foremost, both the necessary equipment (see above) and materials (<€0.5 per 4” disc) are highly affordable and widely available. Second, quantitative filter paper, by design and definition, burns with minimal residues; we hypothesize that this makes it an ideal material for use with thermal CO2
laser cutting (circumventing the aforementioned limitations with plastic liquefication). Third, paper has favorable mechanical properties that make it user-friendly and resilient in handling. It is not brittle like thin crystalline membranes, eliminating the risk of shattering. Its stiffness (~GPa) is in line with (or even exceeding) that of plastic or metal foils [20
], ensuring dimensional stability. Its thermal budget (~200 °C in vacuum) is higher than for many plastics [20
]. Last but not least, the overall fabrication process is truly single-step (no tension frame mounting needed) and does not require specialized training due to most CO2
laser’s plug-and-play interfaces. Masks can be turned from drawings into reality in a matter of minutes. This contrasts with extensive training, experience, and processing time needed for photolithographic approaches. More advanced laser systems (e.g., “cold-cutting” ones) can have long warm-up times and tight focusing requirements, making the process more time consuming, not to mention the often-poor software associated with typical custom-built set-ups for such lasers in academic laboratories.
We note that the use of laser-cut paper masks is not inherently novel. One very recent publication using PVD to create an electrochemical oxygen sensor (>500 µm minimum feature size) references SML with laser-cut cleanroom paper in their Methods section [27
]. Laser-cut paper stencils have also been used for screen printing applications (where material requirements differ somewhat from PVD) at least since 2013, showing down to 250 µm line width [28
]. Only one of the three examples, however, fully specifies the type of paper used (revealing it as a paper/plastic hybrid, likely suffering from some of the limitations of plastics mentioned above). Critically, none of them expand on the rationale for the material choice or provide characterization results of process quality—the paper mask as an incidental process note rather than the central object of study.
Here, our focus is the thorough characterization of paper shadow mask fabrication for exemplary interdigitated electrodes (IDE) and their use in PVD processes. Ultimately, we showcase this as an ideal approach for the rapid fabrication of shadow masks towards the wide range of applications where 100 µm resolution is sufficient.