This PDF file contains the editorial “Next-Generation Lithography” for JM3 Vol. 6 Issue 04

Inverse lithography mask design and double-exposure lithography are two technologies that have gained a lot of attention in the recent past. Inverse lithography consists of synthesizing the input mask that leads to the desired output wafer pattern by inverting the mathematical forward model from mask to wafer. Double-exposure lithography uses two pairs of mask and exposure to print a single (desired) wafer pattern. It usually involves splitting the latter into two parts. In this work, we present some preliminary results in our unique effort to combine the previous two powerful techniques. The goal is to use the inverse imaging approach to automatically synthesize the masks required to print the desired wafer pattern employing double-exposure lithography. We employ the pixel-based mask representation, analytically calculate the gradient, and use a cyclic coordinate descent optimization algorithm to synthesize the two masks. We present results for chromeless phase-shift masks for an idealized case of a coherent imaging system (σ=0) using the Kirchhoff approximation. The results indicate that our algorithm automatically splits the target pattern into two (overlapping) parts, which are used separately during the individual exposures. Furthermore, the proposed approach is also capable of resolving the phase conflicts. The comparison with a single-exposure case indicates a superior contrast and no hot-spots.

Two designs of grazing incidence collectors for extreme ultraviolet (EUV) lithography are described as alternative solutions to the type-I Wolter configuration. The main purposes of the designs are the improvement of collection efficiency and the increase in flexibility with which the design can be adapted and adjusted to the boundary specifications set by the source and the illuminator. With reference to a specific scenario, examples of these designs and their performances are presented, discussed, and compared to what can be achieved with a Wolter collector. In this scenario, one of the designs offers the possibility to achieve large collection efficiency with a limited number of mirrors, as opposed to the Wolter case where high values of the collection efficiency are possible, provided the number of mirrors is increased.

Physical models are developed to investigate the following conditions relevant to discharge-produced plasma (DPP) devices under development for extreme ultraviolet (EUV) lithography: gaseous jet propagation in the chamber, removal of neutral particles with a gaseous jet, and deviation of charged particles with a magnetic field. Several geometries of the mitigation systems are considered for removing debris during the EUV lithographic process. The design of a mitigation system is proposed and simulated with the computer models. The behavior of Xe, Li, and Sn debris in Ar and He jets is simulated by using the high energy interaction with general heterogeneous target systems (HEIGHTS) integrated package. Final energy and local distributions are calculated using experimental debris data from current EUV facilities.

Of great importance in post-optical lithographies, such as electron beam (EB) and extreme ultraviolet, is the improvement of line edge roughness or line width roughness of patterned resists. We provide an exposure dose dependence on LER of a latent image in chemically amplified EB resist from 1 to 50 μC/cm². By using a Monte Carlo simulation and empirical equations, the effects of exposure dose and amine concentration on LER are investigated in terms of shot noise and image contrast. We make clear the correlation between LER and the fluctuation of the initial number of acid molecules generated in resists.

The impact of various parameters such as photoacid generator (PAG) concentration, acid diffusion length, and polymer size on the finally obtained line edge roughness (LER) in chemically amplified photoresists are investigated with a stochastic simulator. A new aspect of the simulations is to start with a polymer matrix modeled by molecular dynamics simulation and subsequently simplify the description of the resist composition for mesoscopically simulating the post-exposure bake (PEB) and development steps. The results show that decreasing the molecular weight (MW) of chain-like polymers does not necessarily lead to lower roughness values. Acid-breakable polymers are simulated as well showing that they can lead to improved LER characteristics.

Isolated dots and lines with 6 nm width are written in 20-nm-thick hydrogen silsesquioxane (HSQ) layers on silicon substrates, using 100-keV electron beam lithography. The main factors that might limit the resolution, i.e., beam size, writing strategy, resist material, electron dose, and development process, are discussed. We demonstrate that, by adjusting the development process, a very high resolution can be obtained. We report the achievement of 7 nm lines at a 20-nm pitch written in a 10-nm-thick HSQ layer, using a potassium-hydroxide (KOH)-based developer instead of a classical tetra-methyl-ammonium hydroxide (TMAH) developer. This is the smallest pitch achieved to date using HSQ resist. We think that the resolution can be improved further, and is presently limited by either the beam diameter (which was not measured separately) or by the not-fully-optimized development process.

Polymerase chain reaction (PCR) is a molecular biological method for in vitro amplification of nucleic acids. Our objective was to design a micro-PCR system that included a Rayleigh-Bénard convection PCR chip, measurement circuits, and circuits to control the temperature. A Rayleigh-Bénard convection PCR chip was easily fabricated by using microelectromechanical system technology, and the sample solution could be put in to finish the completed PCR cycling within several minutes. The flow stream, velocity, and temperature profile in a micro-PCR system are important to achieve the successful PCR, but they are not easily observed with an experimental method. Thus, a CFDRC simulation of Rayleigh-Bénard convection PCR was undertaken to determine the above important parameters. The duration of one cycle, the extension time, and total duration of 25 cycles can be calculated to achieve the optimal design. Finally, using agarose-gel electrophoresis, we verified the practicability of this system. The comparison study of PCR experiments performed with a commercial PCR machine and our chips showed that our chips can greatly decrease the duration of the reaction. By comparing the simulation and PCR experiment results with varied designed sizes, a user can set the parameters and computational fluid dynamics results for optimal designs and decrease the total duration of future reactions.

Dynamic analysis of sandwiched driving diaphragms, made of ionic conductive polymer (ICP), and an innovative fabrication technique to micromachine them are reported. The dynamics of ICP film (ICPF) is theoretically studied and described by governing equations via chemical and mathematical analysis. The purpose of deriving the governing equations is to explore the explicit relation of the induced lateral deflection of ICPF with respect to an applied electrical field. This is the first major issue of this report. The other major issue is ICPF fabrication technology. The device structure of the micropump unit, made of polydimethylsiloxane (PDMS), and the sandwiched ICPF are successfully integrated as a monolithic driving module by the proposed innovative fabrication technique. From the viewpoint of mass production, there are three challenging targets to be achieved. First, a novel method to fabricate ICPF actuators is developed to overcome the conventional defect: insufficient adhesion of coated electrodes upon ICPF. Second, the ICPF has to be restrained considerably to keep it straight and flat as it is fabricated, even though its thickness is far less than its length. Last, the ICPF has to operate in an aqueous environment. The proposed fabrication technique fully takes this into consideration.

A monolithic inkjet print head, fabricated with silicon micromachining technology and capable of generating microscale liquid droplets, is developed and shown to function successfully. The print head uses a dense array of thermal bubble inkjet devices, made on a single silicon wafer. Each device is made of a Pt heater stack, a small, shallow fluid chamber, and a refilling channel formed by a Ge-sacrificial etching process, a deep-etched through-wafer feeding hole, and a micron-scale nozzle opened in a thin nitride membrane by plasma etching. Experimental results with a high resolution video imaging system show that this print head is capable of generating water droplets as small as 3 µm in diameter (0.014 pL), about 1/70th the volume of the droplets produced by existing inkjet systems. The printing process is also found to be stable, uniform in droplet size and velocity, and free of satellite droplets at optimum operation conditions. At small distances between the print head and substrate, droplet spreading is also small. This print head is then a capable tool for ultra-high-resolution inkjet printing and can be used in research areas where delivery of micron-scale fluid droplets is desired.

In this work, a newly developed optical lens cementing technology is reported. A fluoride material is used as an optical cement that can reduce damage from deep-ultraviolet (DUV) radiation. The degradation of transmittance and the surface quality of the cemented optical elements, including adhesive used for cementing, are evaluated after prolonged DUV irradiation. It is shown that with a 248-nm wavelength, this cement works quite well, up to 1000 h of operation, and the change in transmittance is negligible where average irradiation power is within 27 to 37 mW/cm². Hence for all practical purposes, the use of this cement in microscope objectives is quite acceptable for 248-nm applications, thus confirming that this cementing technology is satisfactory and meets the performance requirement of DUV inspection systems.

Finite element (FE) modeling of electrostatically actuated torsional and flexural-torsional microelectromechanical systems (MEMS) micromirrors is investigated by means of three different tools available in ANSYS FE modeling software, and numerical results are compared with static analytical solutions and existing experimental data. These methods include (a) a sequential coupled electrostatic and structural field tool, (b) a directly coupled electrostatic and structural field tool employing one-dimensional transducer element, and (c) a directly coupled electrostatic and structural field tool utilizing a two- or three-dimensional reduced order model. The torsional micromirror is 1000×250 μm², and the flexural-torsional micromirror is 100×100 μm². We present the advantages and disadvantages of these tools for MEMS micromirror modeling. These comparisons allow a selection to be made of the most suitable tool for a given modeling task and assess the accuracy of analytical solutions.

The availability of recently developed microelectromechanical system (MEMS) micro-mirror technology provides an opportunity to replace macroscale actuators for free-space laser beam steering in light detection and ranging and communication systems. Precision modeling of mirror pointing and its dynamics are critical to the design and control of MEMS beam steerers. Beginning with Hornbeck's torque approach, we present a first-principles, analytically closed-form torque model for an electrostatically actuated two-axis (tip-tilt) MEMS mirror structure. The torque expression is a function of the mirror's physical parameters, such as angles, voltages, and size. An Euler dynamic equation formulation describes the gimballed motion as a pair of damped harmonic oscillators with a coupled torsion function. Static physical parameters such as MEMS mirror dimensions and voltages are inputs to the model as well as dynamic harmonic oscillator parameters, such as damping and restoring constants, which are calculated or fitted to measurements. A Taylor series expansion of the torque function provides insights into MEMS behavior, including operational sensitivities near "pull-in." MATLAB and SIMULINK simulations illustrate performance sensitivities, controllability, physical limitations, and other important considerations in the design of precise pointing systems. Commercial-off-the-shelf micromirror measurements confirm the model's validity in steady state and dynamic scanning operations.

In surface micromachined structures, many parameters like geometry and Young's modulus depend on the process steps and need to be measured for accurate prediction of their functionality. This work discusses simple electrical measurement techniques on surface micromachined cantilever beams to determine Young's modulus, the gap between the beam and the substrate, and the thickness of a deposited aluminum layer on the beam. Cantilevers are ubiquitous in most microelectromechanical system (MEMS) sensors and actuators, and hence are ideal test structures. Pull-in, and a novel resonance frequency measurement based on the pull-in technique, are done on oxide anchored doped polysilicon beams at the wafer level, and some of the device and material properties are extracted from these measurements. The extracted values are compared with those determined from established methods like vibrometry and surface profiler measurements, and show good agreement. Since the measurements are all electrical, they can be part of standardized testing and are also suitable for packaged devices.

In order to successfully integrate bottom-up fabricated nanostructures such as carbon nanotubes or silicon, germanium, or III-V nanowires into microelectromechanical systems on a wafer scale, reliable ways of integrating catalyst dots are needed. Here, four methods for integrating sub-100-nm diameter nickel catalyst dots on a wafer scale are presented and compared. Three of the methods are based on a p-Si layer utilized as an in situ mask, an encapsulating layer, and a sacrificial window mask, respectively. All methods enable precise positioning of nickel catalyst dots at the end of a microcantilever, while avoiding contamination of the used cleanroom process equipment. The methods are suitable for fabrication of scanning probe tips and nanoelectrodes for advanced characterization probes.

We demonstrate the direct photolithographic patterning of a grossly nonplanar substrate by creating 62-μm helical tracks on a 22-mm-high cone. The projection of focused light onto the 3-D surface is achieved using a computer-generated hologram (CGH) suitably illuminated so as to create the required pattern on the photoresist-coated surface. The approach adopted forms the basis of a novel method for patterning nonplanar structures. We address the key challenges encountered for the implementation of holographic photolithography in three dimensions, including mask design and manufacture, exposure compensation, mask alignment, and chemical processing. Control of linewidth and resolution over the nonplanar surface is critical. We describe the methods adopted and critically assess the structures created by this process. The bihelical cone is representative of a broadband, high-frequency coil-like structure, known in wireless communications as a log-periodic antenna.

The ablation of a polymethyl methacrylate (PMMA) substrate with a 248-nm long-pulsed KrF excimer laser is studied. PMMA is ablated at a laser pulse repetition rate (PRR) of 2, 5, and 10 Hz, and fluence varying from 0.2 to 1 J/cm². The coupling effects of multiple shots, PRR, and fluence on the etching depth and topography of PMMA are discussed. It is observed that the etching rate and the ablation depth increase with the increase in fluence. The increase in the ablation depth with fluence is more significant at a higher number of shots. It is also observed that the etching rate decreases with the number of shots for a given fluence. An optimal combination is selected for the fabrication of a 100-μm-depth microchannel using styrene mask. The current microfabrication technique also demonstrates the use of low-cost masks.