Robust PECVD SiC membrane made for stencil lithography

Stencil lithography (SL) is a shadow mask technique which allows parallel, resistless, micro- and nano- patterning of material through apertures created in a membrane (stencil) onto a substrate. The stencils are usually made of LPCVD low-stress SiN due to its outstanding physical and chemical stability. How- ever, limitations are found for some speciﬁc design cases, where membranes can be distorted because of the deformations induced by stress from the deposited materials. Here we present a recently devel- oped PECVD SiC shadow mask for applications in SL. The SiC has higher Young’s modulus than SiN, which transfers to a better performance for SiC stencil than the shadow mask made of SiN in terms of robustness to the stress-induced deformation. We show direct local etching of SiO2 through SiC stencil, otherwise impossible with SiN masks. SiC stencils with both compressive and tensile stresses were fabricated and compared to the SiN stencils. The higher robustness to deformation and better resistance to etching of the SiC membranes allow for a wider choice of the deposited materials, leading themselves to a more precise pattern duplication and less complicated resistless dry etching applications.

Jürgen Brugger, Andreea Veronica Savu, Katrin Sidler Arnet, Wei Tang, Oscar Vazquez Mena, Shenqi Xie

Stencil lithography (SL) is a shadow mask based technique which allows parallel, resistless, micro- and nano-patterning of material through apertures in a membrane (stencil) onto a substrate [1]. The stencils are usually made of LPCVD low-stress SiN due to its outstanding physical and chemical stability. However, limitations are found in some special designs, where membranes could be distorted because of the deformations induced by stress from the deposited materials. To solve this problem, the fabrication process has to become more complex by introducing corrugations on the membranes for reinforcement [2]. Here we report on a recently developed PECVD SiC based shadow mask for applications in SL. The SiC stencil demonstrated a better performance than the shadow mask made of SiN in terms of robustness to deformation and resistance to etching. The SiC was grown by PECVD at temperatures below 400 oC. The intrinsic stress of SiC can be tuned from compressive to tensile by varying the annealing time to fulfill the low stress requirement for the stencil membrane. During this process, the amorphous SiC is starting to crystallize and thus enhance its robustness, which translates into a higher Young’s modulus [3]. The fabrication of the SiC stencil (Fig 1a-d) started from a Si wafer with 540 nm thick PECVD SiC on the front side and 200 nm thick PECVD SiN on the backside. SiC was patterned by dry etching in AMS 200 DSE, following by a DRIE process on the backside to open the membrane windows. The residual 50 m thick Si left after the DRIE process was etched away by KOH. Fig 1e shows the SEM image of a released 1 mm x 1 mm area SiC membrane. The SiC stencil was then placed on a Si substrate for metal deposition. Fig 2 shows the SiC stencil and the accurately duplicated corresponding patterns on the substrate after 5 nm Ti and 50 nm Au deposition in an ebeam evaporator. The results are comparable to those from SiN stencil deposition, where the main challenge is the enlargement of the pattern, known as blurring [4]. The robustness was compared between the low-stress (110 MPa) SiC and the low-stress (200 MPa) SiN by depositing Cr film, which has a high tensile stress. Fig 3 shows the different bending behaviors of cantilever structures in the membranes when coated with 25 nm Cr. The SiC cantilevers bend 75% less than the SiN ones by measuring the maximum deformation, providing a wider possibility for the choice of the deposited material. The higher robustness to deformation of the SiC membrane indicates the possibility of having in general a more precise pattern duplication from the stencil to the substrate. Another advantage of SiC is its high chemical inertness, enabling it to be a good mask for both dry and wet etching. The SiC stencil was used as a mask to etch 500 nm SiO2 by RIE without any metal protection layer on the membrane due to its high etching selectivity (Fig 4a). By comparison, a metal protection layer is essential for SiN stencil under similar conditions for dry etching [5]. The transferred patterns (Fig 4b) show a sub 100 nm loss in diameter for a 3 m diameter aperture, as shown in Fig 4c and d. There is no evident damage to the SiC stencil after the dry etching.

Microelectronic Engineering2011

Characterisation and applications of stencil lithography

Jürgen Brugger, Nao Takano, Marc Antonius Friedrich van den Boogaart, Oscar Vazquez Mena

Increased flexibility of patterning methods is important for the engineering of multimaterial and multifunctional nano/micro-electro-mechanical systems (NEMS/MEMS), such as polymer-based electronic and sensor devices, 3D microfluidic systems, and bio-analytical systems. Stencil lithography is a good candidate as it is a resistless, single step patterning method based on direct, local deposition of material on an arbitrary surface through a solid-state membrane, e.g. a 200-nm thick silicon nitride (SiN) membrane. We present two new stencil membrane stabilisation geometries and associated MEMS processes that allow for advancing further towards a well-controlled full-wafer (100 mm) nanostencil process for reproducible, high-throughput, large-area nanopatterning of mesoscopic structures (10-9 - 10-3 m). In fact, the major challenges for well-controlled and reproducible nanostencil lithography are to control stress-induced membrane deformation, clogging of membrane apertures and the gap between stencil and substrate. The stabilized membranes show improved membrane stability (i.e. reduced out-of-plane deformation) up to 94%, resulting in an improved pattern definition of stencil-deposited structures. We have also systematically characterized the clogging of membrane apertures and the influence of the gap between stencil and substrate on the deposited structures as a function of aperture size and deposited material. A method is presented to recover nominal dimensions of stencil-deposited structures over a gap. To integrate nanomechanical structures with CMOS circuits, stencil lithography was used as a clean, CMOS-compatible method to define the mechanical structures on a CMOS circuit, using the stencildeposited metal patterns as a mask for silicon reactive ion etching (RIE). When the stencil and the CMOS wafer are put in contact, a gap exists between the layer to be structured and the stencil. We have been able to correct the blurring effect by developing a controlled dry Al etch process. The results obtained provide a clear remedy to overcome one of the major challenges in nanostencil-based surface patterning. Stencil lithography is also used in the development of novel nanotemplates of anodic porous alumina (APA). To obtain APA structures with high regularity, a method to pre-pattern the surface of the aluminum substrate before conventional anodization is a prerequisite. A pre-patterned arrangement of dots or holes on Al plates is obtained by evaporation, sputtering, electrodeposition, or chemical etching through stencils. It should result in a high regularity mono-domain APA with pore spacing comparable with the visible wavelength for large area photonic crystals for optical applications and luminescence devices.

2005

A granular model of solidification as applied to hot tearing

Stéphane Vernède

Hot cracking is a spontaneous failure of an alloy during solidification. It is a severe problem for casting industry as it reduces the productivity of cast houses and limits the range of alloys compositions that can be industrially produced. Hot tears occurs at the end of solidification, when solid grains are separated by thin liquid films. At this stage of solidification, fluid flow between the grains is difficult and the solid network is not continuous enough to transmit stresses. The material is extremely brittle. Moreover, the deformation induced by the thermal shrinkage of the material tends to localize in hot spots, i.e., in zones which solidify at last. These zones become in tension and the concomitant effect of the tensile stress with the intrinsic brittleness of the material result in a hot crack. Even though the general mechanisms of hot tearing are understood, the importance of each physical parameter remains ill defined. The present work is mainly focused on the derivation of a granular model of mushy zone (the zone where solid and liquid coexist) for aluminium alloys, i.e., a model which explicitly considers the behaviour of each grain while being sufficiently simple to allow the computation of large mushy zones. First, a solidification model based on the Voronoi diagram of a random set of nuclei is derived. This model computes the solidification in each polyhedron considering back-diffusion and coalescence. Second, a pressure drop calculation is performed in the network of connected liquid channels assuming the centre of solid grains to be fixed. This model considers a Poiseuille flow in each channel, Kirchhoff's conservation of flow at nodal points and flow Losses compensating solidification (KPL model). Finally, the displacement of the solid grain centres is considered and the mechanical behaviour of the mushy zone is computed assuming perfectly rigid solid grains. This model shows the progressive formation of grain clusters and the localization of fluid flow at high solid fraction. Therefore, it allows to define transitions in the behaviour of the mushy zone. These transitions are essential for hot cracking models, but should be introduced as parameters in continuum approaches. Finally, the mechanical model shows a strong localization of deformation in liquid channels normal to the stress direction. Such channels create ideal sites for the nucleation and the propagation of hot cracks.

EPFL2007

Robust PECVD SiC membrane made for stencil lithography

Jürgen Brugger, Andreea Veronica Savu, Katrin Sidler Arnet, Wei Tang, Oscar Vazquez Mena, Shenqi Xie

Microelectronic Engineering2011

Characterisation and applications of stencil lithography

Jürgen Brugger, Nao Takano, Marc Antonius Friedrich van den Boogaart, Oscar Vazquez Mena

2005

A granular model of solidification as applied to hot tearing

Stéphane Vernède

EPFL2007