A message from the Editor-in-Chief.

The semiconductor and hard disk drive industries are investigating nanoimprint for future high volume manufacturing of memory devices and patterned media. Nanoimprint, a form of 1× contact lithography, is one of the few technologies capable of meeting the resolution requirements for next generation electronic and storage devices. Its ability to produce small features with low line width roughness and critical dimension uniformity has been demonstrated by multiple sources. Significant improvements in defectivity have been shown; overlay has improved to within a factor of 2 of that required by the International Technology Roadmap for Semiconductors for 22 nm node flash memory devices; and next generation tools, templates, and processes are being commercialized and tested at end-user sites. Defectivity, throughput, and infrastructure remain as critical challenges, but each has experienced marked improvements in the past year. This technology review and assessment covers critical aspects of nanoimprint for both semiconductor and patterned media manufacturing. It focuses on jet and flash imprint lithography, the type of nanoimprint most often considered for these two applications. The requirements and current status of nanoimprint with respect to high volume manufacturing are presented, and critical aspects are discussed.

In the last years, interest in reducing linewidth roughness (LWR) in EUV lithography through a dedicated process step has significantly increased. Various post-litho processing techniques to improve LWR without compromising resolution or sensitivity have been proposed. While these techniques are giving smoothing levels up to 30% before etch, the important question is, of course, how efficient they are in the full patterning process. To evaluate the effectiveness of the smoothing techniques on the EUV resist process and the post-etch pattern, a few of the most promising techniques have been selected for evaluation. Post-develop rinse smoothing and solvent vapor smoothing can reduce the LWR by 10% to 15%. Ion-beam smoothing is giving higher smoothing values, but suffers some important limitations for its application. Two case studies of post-litho smoothing followed by a standard etch process reveal that a large portion of the LWR smoothing can remain after etch, but the LWR gain may also be completely lost. Finally, a plasma smoothing process combined with a layer deposition is proposed to optimize the etch process itself. Analysis of LWR in the spatial frequency domain at the different stages of the patterning process gives a better insight into the impact of the different steps.

Laser produced plasmas (LPPs) are currently a promising source of an efficient extreme ultraviolet (EUV) photon source production for advanced lithography. Optimum laser pulse parameters with adjusted wavelength, energy, and duration for a simple planar or spherical tin target can provide 2% to 3% conversion efficiency (CE) in laboratory experiments. Additional effects such as targets with complex geometry and tin-doped targets using prepulsing of laser beams can significantly increase CE. Recent studies showed that such improvements in an LPP system are due to reduction in laser energy losses by decreasing photons transmission (higher harmonic of Nd:yttrium-aluminum-garnet laser) or photons reflection (for CO₂ laser). Optimization of target heating using prepulses or ablating low-density and nanoporous tin oxide can further improve LLP sources by creating more efficient plasma plumes and, as a result, increase CE, the most important parameter for EUV sources. We investigated the combined effects of prepulsing with various parameters and different target geometries on EUV conversion efficiency and on energetic ions production. The much higher reflectivity of CO₂ laser from a tin target leads to two possible ways for system improvement using prepulses with shorter laser wavelengths or using more complex targets geometries with special grooves as developed previously by the authors.

This work examines aspects of source mask polarization optimization (SMPO). Cases are described where unusual polarization conditions, beyond the familiar and expected XY or TE, are beneficial. These results are best understood by examining the diffraction patterns and noticing that image formation is influenced by interference between higher diffraction orders when the zeroth order is relatively weak. Examples include hole patterns with attenuated phase shift masks and also dark field illumination. Further analysis shows, however, that for some problems, perhaps most, aggressive mask optimization can transform the imaging problem in such a way that zeroth order light again dominates. Under these conditions familiar two beam imaging can be obtained with off-axis illumination, and then TE polarization will probably provide the best solution. From a computational point-of-view SMPO procedures are hampered by the fact that polarization affects the required optical proximity correction (OPC), the OPC affects the diffraction pattern, and the diffraction pattern and polarized source light interact to determine lithographic performance. This makes co-optimization of source, mask, and polarization conditions difficult, and a useful intermediate approach was found to be source mask optimization with a scalar imaging calculation.

In this paper the Arrhenius behavior of blur upon extreme ultraviolet (EUV) exposure is investigated through variation of the post-exposure bake (PEB) temperature. In this way, thermally activated parameters that contribute to blur (such as acid/base diffusion) can be separated from nonthermally activated parameters (such as secondary electron blur). The experimental results are analyzed in detail using multiwavelength resist modeling based on the continuum approach and through fitting of the EUV data using stochastic resist models. The extracted blur kinetics display perfectly linear Arrhenius behavior, indicating that there is no sign for secondary electron blur at 22-nm half pitch. At the lowest PEB setting the total blur length is ∼4 nm, indicating that secondary electron blur should be well below that. The stochastic resist model gives a best fit to the current data set with parameters that result in a maximum probability of acid generation at 2.4 nm from the photon absorption site. Extrapolation of the model predicts that towards the 16-nm half pitch the impact on sizing dose is minimal and an acceptable exposure latitude is achievable. In order to limit the impact on linewidth roughness at these dimensions it will be required to control acid diffusion to ∼5 nm.

In this study, an influence of the structural parameters of a three-dimensional (3D)-Fresnel-based lens which was designed as a primary optic on the narrowing of light-emitting diode (LED) light beams has been investigated and proposed to gather the emitted light into a narrow-angle light beam. Most conventional LEDs are not able to create a narrow-angle light beam without using a reflector cup or secondary optics, because the LED lighting angle is usually wide (±30 deg from the centerline of the light source). Nevertheless, the energy of the LED light within the range ±16 deg from the centerline of the light source can be greatly enhanced by the focusing effect of a 3D-Fresnel-based lens design as shown in this study. The illumination and the luminous flux of the LED light beam in comparison with the conventional LED have been respectively enhanced up to 61% and 20% respectively. The main contribution of this study is to adjust the height/width ratio of the ladder-shaped structure of a 3D-Fresnel-based lens in order to focus the LED light into a stronger beam, and to meet some specific working circumstances.

We have developed compact devices to control electrostatically driven resonant micromirrors with one and two axes. For stable oscillation with large amplitude, operation close to resonance must be ensured under varying environmental conditions. Our devices feature optical position sensing and driver electronics with closed loop control. In this contribution, we present in much detail the novel two-dimensional device and highlight specific aspects of this system.

We build up an active triangulative metrology system with a high dynamic range for measuring the three-dimensional shape of microstructures and for reverse engineering. The triangulation is done using a stereomicroscope to project dot shaped measurement labels through one port. The second port is used to locate their positions. In a calibrated system, three-dimensional coordinates are allocated to the measured positions. Using a digital mirror device (DMD) the generation of user definable patterns is very quick and flexible. Furthermore, the DMD features high brightness and contrasts values. This is important to recognize and distinguish projected measurement labels. The illumination is done by a white high power light-emitting diode, which is collimated and coupled to the DMD with a total internal reflection prism. The acquisition of measurement data from bright and dark surface areas results in a high change in brightness of measurement labels. To measure bright and dark surfaces in one measurement, a high dynamic range imaging technique is required. To identify measurement labels a temporal coding is used. Identification enables the adaption of the brightness of each projected measurement label. This results in a homogeneous brightness distribution of measurement labels, which enables their measurement in a second step. As a result, the dynamic of the measurement system is expanded by the dynamic of the projection device.

We formerly developed the "assembly method" for analyzing the line-edge and line-width roughness (LER/LWR) that has a long-range correlation beyond the conventional analysis limit, as reported in a previous work. In that method, we repeatedly assembled virtual long lines by gathering line segments, which were arbitrarily disposed on actual long lines and by randomly changing their combination and order, permitting the assembled lines to share the same line segments. Then, we obtained the power spectral density (PSD) of the LER/LWR of the assembled lines considering the lines as seamless. We also derived an analytic formula of the assembled-line PSDs for use in the PSD fitting method. This formula agreed very well with experimental PSDs. In this report, we propose guidelines for suppressing the statistical-noise effect on the assembly method for the purpose of accurately analyzing the long-range-correlated LER/LWR. The guidelines will greatly help shed light on the long-range correlation, which causes the variability even in large devices but has long been veiled due to the lack of metrology.

We discuss the future of resist materials for sub-20-nm lithography and believe that polymer-bound PAG-based resists will be used to 16-nm node. There has been enough progress in resolution and sensitivity to justify the use of these materials. Polymer-bound PAG resists have shown that the principal demerit of acid diffusion can be overcome through attachment of the PAG anion to the lithographic polymer. Since the introduction of this chemically amplified resist approach, we have seen steady improvement in resolution, sensitivity, and LWR. We have also seen improvement in OOB response, outgassing, and pattern collapse. There is no doubt that continuous improvement is still required for these resist systems. We believe that increasing the overall resist quantum yield for acid generation substantially improves the shot-noise problem thereby leading to faster high-resolution resist materials. Using a 0.30-NA extreme ultraviolet tool with dipole, we can achieve 22-nm hp resolution, with a 12-mJ dose and a 4.2-nm LWR.

As integrated circuit design dimensions continue to shrink, previously ignored three-dimensional (3-D) mask effects have become significant for the accurate prediction and correction of proximity effects. Optical proximity correction (OPC) process models must consequently take into account 3-D mask effects. The state-of-the-art model-based OPC methodology, which is called delta-chrome OPC (DCOPC), needs the repeated computation of the mask perturbation to facilitate its convergence. The increasing complexity of OPC process models challenges this DCOPC methodology because each computation of the mask perturbation becomes prohibitively expensive. In this study, a new model-based OPC methodology, which is called non-delta-chrome OPC (non-DCOPC), is proposed without introducing any mask perturbations. It only requires image intensity information to achieve convergence using classical control techniques, and its effectiveness is demonstrated, showing that the run time with and without considering 3-D mask effects can be significantly improved. In addition, the correction accuracy of the DCOPC and non-DCOPC methodologies without considering 3-D mask effects is quite comparable. However, the correction accuracy considering 3-D mask effects can be slightly improved by the non-DCOPC methodology.

Extreme ultraviolet lithography (EUVL)-embedded phase-shift mask (EPSM) for 16-nm half-pitch node technology and beyond generations will provide improvements in process window and low shadowing effect as compared to those of the conventional EUVL binary mask. In a previous study, we experimentally demonstrated the advantages of the EUVL EPSM for the dense lines and contacts as compared to those of the standard EUVL mask. In this study, we focused more on systematic analysis and comparison of the process windows between EUVL EPSM and conventional EUVL binary mask for various feature types. In addition, we investigated the EUVL EPSM performances as a function of phase errors in order to further define the EUVL EPSM absorber thickness uniformity requirements. Both the EUVL EPSM and binary mask were exposed at the Alpha Demo Tool at IMEC, and the wafer level data show features in the EPSM have a larger process window, lower dose-to-target value, and reduced shadowing effect as compared to those of the conventional EUVL binary mask.

Evanescent-wave imaging is demonstrated using solid-immersion Lloyd's mirror interference lithography at λ = 325 nm to produce 44-nm half-pitch structures (numerical aperture, NA = 1.85). At such an ultrahigh NA the image depth is severely compromised due to the evanescent nature of the exposure, and the use of reflections from plasmonic underlayers is discussed as a possible solution. Simulations and modeling show that image depths in excess of 100 nm should be possible with such a system, using silver as the plasmonic material. The concept is scalable to 193 nm illumination using aluminium as the plasmonic reflector, and simulation results are shown for 26-nm half-pitch imaging into a 37-nm thick resist layer using this scheme.

A micropump of biomedical-microelectric mechanical system (Bio-MEMS) for drug delivery systems and health monitoring systems using a new biocompatible piezoelectric thin film was developed. We first assessed the performance of our piezoelectric thin film pump of a newly designed microfluid system using the finite element method. Numerical results show a sufficient transportation ability of our micropump system for blood tests. We generated a multilayer MgSiO₃ thin film on Cu/Ti/Si (100) substrate using RF-magnetron sputtering. We measured the crystallographic orientation and piezoelectric property of the thin film and confirmed that MgSiO₃ (101) crystal grew well. The strain constant d₃₃ was calculated as 179.4 pm/V using the displacement-voltage curve. Furthermore, the deflection and eigenfrequency of the monomorph-actuator, fabricated using a micromachining process, were measured using the laser Doppler vibrometer, which revealed that the deflection linearly increased with applied voltage: it was 82.6 nm with applied voltage of 15 V. We measured the flow rate of a micropump using the luminance difference measurement method. Results showed that the maximum flow rate was 7.1 nl/s at the applied voltage of 15 V. We concluded that our newly developed MgSiO₃ monomorphic micropump can be a comprehensive Bio-MEMS device.

As miniaturization technologies have improved for the past few decades, electronic devices are becoming smaller and the technology needed to fabricate complex three-dimensional (3D) microstructures becomes important. Photolithography is a relatively simple and low-cost method. However, photolithography is, in general, a fabrication method for planar two-dimensional microstructures. In this study, a technique using photolithography is developed to fabricate three-dimensional blades for axial microfan applications. Fabrication of a five-layer SU-8 microstructure for a miniature axial fan of 5 mm diameter using ultraviolet (UV) lithography is presented. Each SU-8 layer was fabricated with a small shift relative to the previous layer. Hence, 3D SU-8 blades with oblique sections have been successfully fabricated using this relatively simple UV lithography. A multiple resist coating and exposure with a single development stage was employed. It was observed that the exposure time for the second to fifth layers was reduced by a factor of 0.7 as compared to that needed for the first layer. A relatively efficient method for the fabrication of multilayer oblique walls was developed. It is probably the first time that UV lithography is used for creating 3D sloping microstructures.

This paper describes the development of a superzone solid immersion diffraction vortex lens and sub-wavelength antireflection surfaces for analyzing dense integrated circuitry through the silicon substrate. The diffractive optic is used with the scanning laser microscope imaging system at a wavelength of 1064 nm. The silicon substrate consists of wafer grade silicon, 10¹⁵ cm⁻³P-type, with an integrated memory array at the 90 nm technology node as the target structures. The solid immersion diffractive optic is designed using computer generated holography, and the functionality is simulated using a finite difference time domain numerical method. The blazed phase profile and sub-wavelength structures of the diffractive optic are fabricated using a two step procedure of gray scale gallium implantation with the focused ion beam, followed by the reactive ion etching using CHF₃ chemistry. The developed fabrication conditions for the optical structures investigate implant doses ranging from 0 to 1000 pC μm⁻² using the 1000 pA beam current, with etch times from 0 to 25 min. Experimental results demonstrate functional optical vortex lenses that achieve selective suppression of signals from target IC structures to enhance the contrast of the surrounding circuitry. To reduce reflection losses, sub-wavelength, antireflection surfaces targeting the 1064 and 3300 nm wavelengths are investigated. Spatial periods of 140 and 500 nm are achieved, with fill factors of 0.3 to 0.4 over the quarter wavelength antireflective height. For resolution enhancement, a second order superzone solid immersion lens with a numerical aperture of 3.4 is fabricated and demonstrates a 1.6× improvement in image resolution over the first order solid immersion lens with a numerical aperture of 1.7, which in turn achieves a 3.5× improvement over no immersion lens.

This paper discusses a novel fabrication process that uses a combination of negative and positive photoresists with positive tone photomasks, resulting in masking layers suitable for bulk micromachining high-aspect ratio microelectromechanical systems (MEMS) devices. MicroChem's negative photoresist Nano™ SU-8 and Clariant's image reversal photoresist AZ 5214E are utilized, along with a barrier layer, to effectively convert a positive photomask into a negative image. This technique utilizes standard photolithography chemicals, equipment, and processes, and opens the door for creating complementary MEMS structures without added fabrication delay and cost. Furthermore, the SU-8 masking layer is robust enough to withstand aggressive etch chemistries needed for fabrication research and development, bulk micromachining high-aspect ratio MEMS structures in silicon substrates, etc. This processing technique was successfully demonstrated by translating a positive photomask to an SU-8 layer that was then utilized as an etching mask for a series of trenches that were micromachined into a silicon substrate. In addition, whereas the SU-8 mask would normally be left in place after processing, a technique utilizing Rohm and Haas Microposit™ S1818 as a release layer has been developed so that the SU-8 masking material can be removed post-etching.

Angle resolved Mueller polarimetry implemented as polarimetric imaging of a back focal plane of a high NA microscope objective has already demonstrated a good potential for CD metrology. Here we present the experimental and numerical results indicating that this technique may also be competitive for the measurements of overlay error δ. A series of samples of superimposed gratings with well controlled overlay errors have been manufactured and measured with the angle resolved Mueller polarimeter. The overlay targets were 20-μm wide. When the overlay error is δ is equal to 0, absolute values of elements of real 4×4 Mueller matrix M are invariant by matrix transposition. Otherwise this symmetry breaks down. Consequently, we define the following overlay estimator matrix as E = |M| − |M|<sup>t. The simulations show that matrix element E₁₄ is the most sensitive to the overlay error. The scalar estimator of E₁₄ was calculated by averaging the pixel values over a specifically chosen mask. This estimator is found to vary linearly with δ for overlay values up to 50 nm. Our technique allows entering small overlay marks (down to 5-μm wide). Only one target measurement is needed for each overlay direction. The actual overlay value can be determined without detailed simulation of the structure provided two calibrated overlay structures are available for each direction.

Acceptance of imprint lithography for manufacturing will require demonstration that it can attain defect levels commensurate with the defect specifications of high-end memory devices. We summarize the results of defect inspections focusing on two key defect types: random nonfill defects occurring during the resist filling process and repeater defects caused by interactions with particles on the substrate. Nonfill defectivity must always be considered within the context of process throughput. The key limiting throughput step in an imprint process is resist filling time. Repeater defects typically have two main sources: mask defects and particle-related defects. Previous studies have indicated that soft particles tend to cause nonrepeating defects. Hard particles, on the other hand, can cause either resist plugging or mask damage. We use an Imprio 500 20- wafer per hour development tool to study both defect types. By carefully controlling the volume of inkjetted resist, optimizing the drop pattern, and controlling the resist fluid front during spreading, fill times of 1.5 s are achieved with nonfill defect levels of ∼1.2/cm². Longevity runs were used to study repeater defects, and a nickel contamination was identified as the key source of particle-induced repeater defects.

The stochastic nature of extreme ultraviolet (EUV) resist exposure leads to variations in the resulting acid concentration, which leads to line-edge roughness (LER) of the resulting features. Using a stochastic resist simulator, we predicted the mean and standard deviation of the acid concentration for an open-frame exposure and fit the results to analytical expressions. The EUV resist exposure mechanism of the PROLTIH Stochastic Resist Simulator is first order, and an analytical expression for the exposure rate constant C allows prediction of the mean acid concentration of an open-frame exposure to about 1% accuracy over a wide range of parameter values. A second analytical expression for the standard deviation of the acid concentration also matched the output of the simulator to within about 1%. Given the assumptions of the PROLTIH Stochastic Resist Simulator, it is possible to use the results of this paper to predict the stochastic uncertainty in acid concentration for EUV resists, thus allowing optimization of resist processing and formulations and contributing to a comprehensive LER model.

Multispectral imaging is becoming a new powerful tool in a wide range of biological studies by adding spectral, spatial, and temporal dimensions to tissue abnormity and the underlying biological processes. Conventional spectral imaging systems are bulky, expensive, require multiple exposures, or extensive post-processing to align multiple images of pure spectral components. Recently, a break-through technology has emerged to instrument multispectral imaging technology into handheld real-time devices using miniaturized filter mosaic containing densely patterned micro-arrayed multiple channel bandpass optical filters. The filter mosaic can be directly placed near the focal plane, immediately in front of the imaging sensor of an off-shelf charged-coupled device/complementary metal-oxide-semiconductor camera, with the potential of one element of such a micro-arrayed filter to cover one pixel of the imaging sensor. This paper reveals the technical details of how such a micro-arrayed multichannel optical filter is fabricated using traditional multifilm vacuum deposition and the modern microlithography technologies. The selection of different coating materials, their structures, and effects to the adhesive forces between film and substrate, the spatial resolution, width of passing band, and the transmittance of the resulting miniaturized optical filter are discussed.

We provide new yield-aware mask strategies to mitigate emerging variability and defectivity challenges. To address variability, we analyze critical dimension variability with respect to reticle size and its impact on parametric yield. With a cost model that incorporates mask, wafer, and processing cost, considering throughput, yield, and manufacturing volume, we assess various reticle strategies (e.g., single-layer reticle, multiple-layer reticle, and small and large size) considering field-size-dependent parametric yield. To address defectivity, we compare parametric yield due to extreme ultraviolet mask blank defects for various reticle strategies in conjunction with reticle floorplan optimizations such as shifting of the mask pattern within a mask blank to avoid defects being superposed by performance-critical patterns of a design.