Search

Integrated circuits are becoming more sensitive to overlay errors between the most critical layers. This paper focuses on inter-transistor overlay variations, which are defined as the short-range variations of overlay between transistors separated by distances of 1 micrometers to 100 micrometers . Many circuits are particularly sensitive to these inter- transistor variations. However, inter-transistor variations are difficult to measure using conventional techniques of metrology. We have developed an active electrical metrology method using on-chip test circuitry to map inter-transistor overlay variations. Test chips were designed and fabricated on a commercial HP 0.35 micrometers process. An array of 127 x 64 active electrical overlay test structures was measured. The array has an area of 856.8 micrometers x 705.6 micrometers , with uniform sampling spacing of 6.8 micrometers x 11.2 micrometers . A measurement speed of 5 microsecond(s) per site was achieved with an accuracy of 6.5 nm (3-sigma). The measured overlay variations between gate poly and diffusion were found to be made up of alignment errors probably associated with the wafer stepper operation combined with short-range overlay variations probably contributed primarily by the mask. With 3-sigma values of 20-30 nm, the inter-transistor overlay variations are surprisingly large when viewed in the context of the typical overall overlay budget for a 0.35 micrometers process. Contour plots and Fourier analysis show that they have an obvious periodicity of 102.4 micrometers in y direction, which can be related to the writing stripes of the raster-scanned mask lithography system used to fabricate the masks. Intra- stripe and stripe-to-stripe overlay variations are then decomposed by spatial frequency filtering, and the intra- stripe variations are further analyzed.

A pixel array readout integrated circuit (ROIC) containing per-pixel analog-to-digital conversion (ADC) and digital-to- analog conversion (DAC) for infrared detectors is presented with design and test result details. Fabricated in a standard 0.35 micron, 3.3 volt CMOS technology. the prototype consists of a linear array of 64 pixels, containing over 100 transistors per 30 by 30 micron pixel. The 8-bit per-pixel ADC is a Nyquist-rate single-slope design consisting of a three stage comparator and an 8 bit memory. This fully pixel- parallel ADC architecture operates in full-frame 'snapshot' mode and can reach over 1,000 frames per second. Each pixel also contains cascoded current source, globally biased to subtract an identical, fixed amount of current from each pixel in order to remove a common background signal by 'charge skimming.' It operates over more than 3 decades of current cancellation (approximately 10 pA to > 10 nA). As well, each pixel contains a 4 to 6+ bit current-mode DAC, intended to trim-out pixel-to-pixel variations in background current. It consists of 16 unit-cells of switched cascoded current sources per pixel, organized as two separately biased weights and controlled by a 16-bit per-pixel memory. The DAC operates over more than 4 decades of current cancellation (< 10 pA to approximately equals 100 nA) per least significant bit (LSB).

Recently global space charge effects were found to seriously limit the performance of high throughput projection electron beam lithography systems. A fundamental analytical theory describing the global space charge lens has been developed using the variational principle. A modified paraxial ray equation and a set of aberration integrals is derived. All the space charge effects, including the defocus, the magnification variation, and the third order aberrations, are found to be proportional to the perveance, the weighted integrals of the normalized current density distribution, and a combination of the system parameters, such as: the field size and the convergence angle. A simple scaling law for the space charge aberrations, where the aberrations at the image edge scale linearly with the radial dimension, has been found for the telecentric projection configuration. Both the theoretical calculations and the Monte Carlo simulations have been done for a SCALPEL configuration with 25 (mu) A beam current and two scaled systems, whose radial dimensions are scaled from SCALPEL by twice and 3 times, and current by 4 times and 9 times respectively at 10 kV. From both theoretical and simulation results, the space charge aberrations at the edge show approximately a 63 times increase in the twice-scaled system, and about 95 times increase in the three-time-scaled system. All these aberration scaling factors are very close to those predicted from the simple scaling law. An experiment based on the simple scaling law was designed to verify the theory and evaluate the performance of the high throughput projection system. In the scaled testing column which has a 1.6 mA beam current, 5 kV beam voltage, 40 cm column length, 8 mm field diameter, and 4 mrad convergence angle at the mask, the space-charge-aberration blur to be measured will be as large as 200 micrometer. Then the final image blur of the real high throughput system can be derived by using the scaling law.

Electron beam exposure of masks and wafers results in charging of the insulting resist film. This charging results in an electric field which deflects incoming electrons and can be a serious source of pattern placement error in electron-beam lithography. In earlier work (Ingino et al. 1992) the surface potential was found to be positive or even zero under certain conditions. In this study, a model is developed to explain this effect and the surface potential is measured by an independent method, a Kelvin probe non-contacting electrostatic voltmeter. This new study confirms qualitatively the findings of the first study. An area of PBS resist measuring a few square millimeters is exposed using a Gaussian focused probe and moved under the Kelvin probe immediately after exposure to measure the surface potential. Thicker resist tended to charge more negatively. The model and experiments confirm early studies that the surface potential is a function of resist thickness, and that there may exist a resist thickness where the surface charge is essentially zero.

Resist coating on alignment marks can be asymmetrical due to the spinning motion. Alignment errors may result from this asymmetrical coating because the alignment signal can contain asymmetry when the optics detects the mark through the resist. The amount of signal asymmetry depends on the mark topography, resist thickness, spin speed, location on the wafer and the alignment technique. We have proposed a Brewster angle illumination technique (BAIT) as an effective means to greatly reduce the resist-induced alignment error by reducing alignment signal reflections from the surface of the resist. In this paper, we present the experimental results. We built an interferometric alignment system to implement BAIT. By scanning a resist-coated mark in close proximity to an exposed reference mark, drift and other system offsets are minimized, allowing an offset detectability of 6 nm. Resist-induced alignment errors with a radial dependence on mark position and magnitudes up to 25 nm have been measured when BAIT is not used. However, when BAIT is employed the resist-induced errors at all positions on the wafer are reduced to less than 10 nm and are close to the resolution limit of the alignment setup. The validity of the data was established by changing the polarization of the light from TM to TE, thereby destroying the BAIT condition while holding all other parameters constant. A study of the dependence of resist-induced alignment error on spin speed reveals that the amount of asymmetry in the resist profile increases monotonically as the spin speed increases. This effect is due to two factors: that the resist is thinner and that the spin speed is also faster. Further study using a thinner to dilute the resist isolated these factors and showed that the resist-induced alignment errors depend more sensitively on spin speed than the resist thickness. In most of the cases, the BAIT alignment system is immune to resist coating asymmetry and the error magnitude stays below 10 nm. However, the BAIT cannot work effectively when resist thickness becomes thinner than 0.6 micrometer. This limitation of BAIT technique was attributed to the ripples in the resist profile causing local impinging angles to deviate significantly from the Brewster angle.

One of the most crucial emerging challenges in Lithography is achieving rapid and accurate alignment under a wide variety of conditions brought about by different overlying films occluding the marks. The problem is exacerbated by planarizing processes such as Chemical Mechanical Polishing (CMP) and asymmetric processes such as metal deposition and photoresist coating. These processes give rise to displacement of the perceived position of the alignment mark. Thus, any effective algorithm must be based on the history of such displacements. A new approach based on subspace decomposition of the alignment signals is described. The method only applies to imaging and/or scanning based alignment signals. The main assumption is that the process-induced asymmetries are small enough such that only linear effects need to be considered. We first find the subspace of alignment signals using a set of alignment signals with pre-known positions. The positions of the new signals are measured based on the fact that, if shifted correctly, they will lie in the same subspace as the previous signals. Current alignment algorithms assume symmetric alignment signals. Since this method exploits the structure of the signals, it results in more accurate measurement of the position than the current algorithms. Simulation results show that the alignment error is about an order of magnitude smaller than that achieved with conventional Maximum Likelihood or phase-fitting approaches. The computational complexity also increases linearly with the dimension of the subspace and is linearly proportional to signal bandwidth.

Proposed high throughput maskless multi-parallel electron beam systems require a large current, which can also potentially destroy the imaging quality due to Coulomb interaction effects. By using a Monte Carlo approach, we have studied the beam current limitation set by Coulomb interaction among electrons at 0.1 micrometers resolution under a variety of column lengths and acceleration voltages for both a blanker aperture array system and a photocathode system. By taking into account the limits from the resist sensitivity and the freedom of system parameter selection, the throughput performance is evaluated in terms of an upper limit and a lower limit, and the feasible system configurations are suggested for achieving the desired throughput as well as 0.1 micrometers resolution. Some results for SCALPEL are also obtained, and the comparison of different lithography tools are discussed.

One of the most crucial emerging challenges in lithography is achieving rapid and accurate alignment under a wide variety of conditions brought about by the different overlying films occluding the marks. The problem is exacerbated by planarizing processes such as chemical mechanical polishing that reduce the topographical contrast used to view the marks and by distortion of the wafer and of the stage. Thus an effective learning process is needed to rapidly acquire the best possible positional information from an array of the marks across the wafer. In this paper, a neural network model for global alignment is described. This method has significant advantages including robustness to measurement noises and other random disturbances. Stage distortion can also be easily included in the model, resulting in more accuracy in the presence of a known repeatable distortion. The process can learn to identify and ignore any alignment marks that yield significantly erratic signals. Different approaches for training of the system will be discussed. A few common distortion functions will also be used to test the model. Preliminary simulation results show very accurate alignment in the presence of noise and distortion. In addition the simulations show that stage distortion has no effect on the results. In other words, increasing stage distortion amplitude will not change the positional accuracy in a given time.

The molecular organization in ultrathin polymer films (thicknesses less than 1000 angstroms) and thin polymer films (thicknesses between 1000 and 10,000 angstroms) may differ substantially from that of bulk polymers, which can lead to important differences in resulting thermophysical properties. Such constrained geometry films have been fabricated from amorphous poly(3-methyl-4-hydroxy styrene) (PMHS) and semicrystalline poly(di-n-hexyl silane) (PD6S) by means of spin-casting. The residual solvent content is substantially greater in ultrathin PMHS films, which suggests a higher glass transition temperature that results from a stronger hydrogen-bonded network as compared with that in thicker films. Crystallization of PD6S is substantially hindered in ultrathin films, in which a critical thickness of 150 angstroms is needed for crystalline morphology to exist and in which the rate of crystallization is initially slow but increases rapidly as the film approaches 500 angstroms in thickness.

A two-dimensional self-calibration experiment obtains Cartesian traceability for high-precision tools. The calibration procedure incorporates group theory principles to solve our industry's two-dimensional calibration problem. With group theory, a Cartesian system is obtainable through mathematics; thus, eliminating the need for any certified standards. The calibration algorithm was developed by Jun Ye at Stanford University and funded by the Semiconductor Research Corporation (SRC) with collaboration from Hewlett Packard (HP) and IBM. The data was collected from Leica's LMS2000 and LMS2020 systems.© (1996) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Retarding field immersion lenses are coming into wide-spread use for focusing low energy electron and ion beams. One disadvantage is that any sample non-flatness will distort the focusing field. A more recent variant is a non-immersion retarding field lens, in which the sample is located outside the retarding field. We describe here a simple model and analysis which predicts that the performance of such a lens is competitive with the retarding immersion lens. With this lens design, it should be possible to focus a 1 keV electron beam with an energy spread of 1 eV into a spot of 4 nm diameter using realistic values of maximum voltage and retarding field strength. The effective resolution under these conditions is estimated to be 1.4 nm.

Brightness, energy spread and emission area are key parameters of electron sources for instruments such aselectron microscopes and electron beam lithography tools. In developing transmission-mode NEAphotocathodes as sources for these applications, these characteristics have been measured in specializedsealed tubes. Average lateral energies were measured at 63 meV for a 1 .5 micron thick photocathode, and83meV for a 0.5 micron thick photocathode, which was known to be emitting "hot" electrons. A currentdensity of 841A/cm2 was obtained from a 1 .7 micron diameter emission area. This high cunent density canbe explained in terms of lateral drift and diffusion of surface trapped electrons. Combined angular and current density data indicate a brightness of —108A/cm2-sr at 3kV. Keywords: negative electron affinity, GaAs, Cs, brightness, energy spread, emission area, angulardistribution I . INTRODUCTIONThe emission of photostimulated electrons from a negative electron affinity surface was first demonstratedby Scheer and Van Laar 1969.l Since then, many applications have developed which take advantage ofunique characteristics of negative electron affinity photocathodes —

We have explored two-level, or `twin' masks as a mask-based means of increasing depth of focus. Simulations have shown that the technique offers substantial gains for a variety of pattern types. To verify this experimentally, we have fabricated a test mask containing two- level as well as conventional mask patterns. We have performed through-focus series of photoresist exposures and demonstrated the expected improvement in focus latitude.

Residual solvent present in resist films following spin casting can significantly influence dissolution behavior. We have used radiolabeling to measure solvent remaining in spin cast films after they have undergone various annealing protocols. Films of poly(3-methyl-4-hydroxystyrene) were found to contain a significant fraction of propylene glycol methyl ether acetate (as much as 50 percent by weight) depending on the annealing conditions. 1100 angstrom films contained a higher percentage of residual solvent than 1.2 micrometers films. A diffusion plus evaporation model was used to deduce the diffusion coefficients of residual solvent as a function of film thickness and annealing temperature. We found that diffusion coefficients of residual solvent in 1100 angstrom polymer films are more than two orders of magnitude less than in 1.2 micrometers thick films.

The optical performance of Markle-Dyson projection optics is now well established. Here we describe options for 1X reflective optical masks that might achieve the desired linewidth control. One option is the use of aluminum as the reflecting material. A film less than 50 nm thick has nearly twice the reflectivity of the silicon used until now, and so it should be possible to develop an etching process (for such a thin film) that is adequately precise. Moreover options exist for repairing both opaque and clear defects. An interesting alternative configuration, that eliminates the need to etch the aluminum, is to use a patterned absorber on the substrate and to deposit the aluminum over the patterned absorber.

Residual casting solvent present in resist films can have a profound effect on their dissolution behavior. We have used a radiolabeling technique to make quantitative measurements of solvent remaining in spin cast films after they have undergone different baking protocols. Films of poly (3-methyl-4-hydroxystyrene) were found to contain a significant fraction of propylene glycol methyl ether acetate (as much as 50% by weight) depending on the baking conditions. Films that were 1000 angstroms thick contained a higher percentage of residual solvent than films that were 1 micrometers . The solvent concentration in films containing a mixture of polymer and photoactive compound was a maximum for films containing 5 wt% PAC.

In a previous paper we described a miniature unity-magnification system that employed magnetic focusing solely and indicated that a minimum beam diameter of about 150 angstroms should be possible at 100 V. However, such a system has proved difficult to realize in practice because of the difficulty of providing a suitable beam extraction system for the field emission source. The scanning tunneling microscope has proven capable of high-resolution patterning at ultra-low energy (< 100 eV), but the sensitivity of electron-beam resists tends to drop at these energies. Here we describe a variety of alternative designs that appear to be more practical. One possible configuration, an acceleration/retardation system with superposed magnetic field, should allow us to obtain a minimum beam diameter of about 100 angstroms at 200 V. In another configuration which we have implemented experimentally, we employ a standard commercial SEM as our source and introduce a confined magnetic lens as the final focusing element. The magnetic field is provided by a NdFeB permanent magnet, which allows for reasonably high field strength (0.6 - 0.8 T) and compact size. This system offers an effective means of extending the range of operation of any SEM to low energies.© (1993) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Even at 0.5 micrometers design rules, the specifications on 5X reticles are set extremely tightly, on the grounds that ULSI patterns are so complex that tight specifications are essential to obtain acceptable yield. It is usually assumed that these specifications scale with the design rules, and that they should be even tighter for 1X reticles. As a consequence, it has been argued that 1X reticles for 0.25 micrometers design rules are impracticable. A statistical analysis, starting from first principles, and assuming point independent, normally distributed errors, supports the way in which mask specifications are currently set. The assumptions of spatial invariance and normal distribution are crucially important in the analysis. However, it is far from clear that they are valid. Consequently, mask specifications in general, as they are currently set, may be unnecessarily severe.

Understanding the electronic structures of silicon (Si) nanostructures is essential for exploring their potential applications in opto-electronics. Well behaved and ordered Si nanostrctures are desirable for facilitating the characterization process and the interpretation of experimental results. A combination of high resolution electron beam lithography, anisotropic reactive ion etching (RIE), and thermal oxidation was shown to yield such well behaved Si nanostructures [1][2]. Etched Si nano-wires with diameters less than 20 nm and aspect ratios greater than 20 were fabricated using a RIE process based on NF3 plasma. Thermal oxidation was demonstrated to be a feasible technique for reducing the diameter of the Si nano-wires to below 5 nm. A transmission electron microscopy (TEM) technique was developed to obtain precise structural information from these oxidized Si nanostructures and to study other interesting oxidation related phenomena. A feasible electrical contact technique was proposed and preliminary contact characteristics were obtained. All of these constitute an important first step toward understanding the intrinsic properties of Si nanostructures.

The critical operation needed to achieve monolithic three-dimensional integrated circuits is obtaining single-crystal, device-quality semiconductor material for upper layer active circuits without exceeding the thermal budget of underlying devices. Previous attempts at using pulsed laser crystallization of amorphous films for upper layer devices failed to provide large enough single crystals at a low enough temperature. Here, the authors demonstrate a more direct approach to realizing high-quality, single-crystal Si (100) and Ge (100) islands (3–3000 μm across) on amorphous SiO2 substrates. The technique is a form of hydrophilic fusion bonding featuring low temperatures (≤400 °C), chemical mechanical polishing, and chemical surface activation (using NH4OH). The bonds are strong enough (>1 J/m2) to withstand SmartCut® removal of the donor wafer. By bonding arrays of islands rather than one large contiguous layer, the authors effectively avoided the formation of thermally induced voids at the bonding interface, and so dramatically improved yield.

Dose correction is commonly used to compensate for the proximity effect in electron lithography. The computation of the required dose modulation is usually carried out using 'self-consistent' algorithms that work by solving a large number of simultaneous linear equations. However, there are two major drawbacks: the resulting correction is not exact, and the computation time is excessively long. A computational scheme, as shown in Figure 1, has been devised to eliminate this problem by the deconvolution of the point spread function in the pattern domain. The method is iterative, based on a steepest descent algorithm. The scheme has been successfully tested on a simple pattern with a minimum feature size 0.5 micrometers , exposed on a MEBES tool at 10 KeV in 0.2 micrometers of PMMA resist on a silicon substrate.

Electron beam lithography for the fabrication of 1X x-ray masks and optical lithography system such as Markle-Dyson system requires precision of edge placement of mask less than 100 nm. In this paper we report experimental results with low voltage exposures (2 keV). 100 nm features were delineated in 66 nm resist in both sparse and dense patterns on bare silicon and 300 nm Au/Si confirming the predicted dramatic reduction in proximity effects. In addition, both sparse and dense 100 nm patterns were resolved over doses from 16 to 28 (mu) C/cm2 on bare silicon and 14 to 27 (mu) C/cm2 on gold substrate illustrating the high dose tolerance enjoyed at low voltages. Electromagnetic interference and the lower electron beam brightness are not limiting factors for the low voltage exposure.

The inhibition mechanism in diazonaphthoquinone positive resists has been widely studied. The ability of the photoactive compound to retard the dissolution of the matrix resin has been attributed to both chemical interactions between resist components as well as spatial distribution of sensitizer in a resist film. Our work has focused on understanding the physical distribution of the photoactive compound on the development behavior of this class of positive resists. We have used Langmuir-Blodgett deposition to fabricate resist structures with highly stratified sensitizer distributions. Dissolution measurements using a quartz crystal microbalance have demonstrated that preferential placement of very thin layers of sensitizer both at the surface of a resist film and embedded between two layers of polymeric resin can significantly affect development.

Fabricating well controlled nanostructures and obtaining precise structural, electrical, and optical information from them are essential for understanding die intrinsic properties of silicon (Si) nanostructures, which in turn is important for exploring the potential of quantum confinement induced light emission from crystalline Si. A combination of high resolution electron beam lithography, anisotropic reactive ion etching (RIE), and thermal oxidation has been successfully applied to obtain sub-5 nm Si columnar structures [1]. A transmission electron microscopy (TEM) technique has also been used to characterize die precise structural dimensions of these columns [1]. To obtain the electrical and optical information, a process based on polyimide planarization was developed to establish electrical contacts to these nanostructures. The same process is also applicable for fabricating device structures to study electrically pumped optical response. Preliminary transport studies have confirmed current conduction through die Si nano-pillars and yielded an estimate of die conductivity.

Recent work has demonstrated the high resolution optical performance possible with simple IX mostly-reflective optics: using 248nm light from a mercury arc lamp, 0.25pm features were delineated across a 2mm radius semicircular field, and much larger fields are possible with a scaled up version.[1] The mask required for this system consists of a quartz substrate, a patterned thin film reflector and a non-reflective backing which also serves to protect the reflector film. The mask is reflective at the quartz/reflector interface so the substrate is part of the projection optical path and so acts as the pellicle. We have investigated chromium, silicon and aluminum for the reflector material; their reflection coefficients at 248nm at the quartz- reflector interface are 30, 55, and 90 per cent respectively. Silicon has been chosen because it has a practical combination of reflectivity and ease of deposition and etching. Moreover films as thin as 30?m provide the full (bulk-value) reflection and so precise etching is further facilitated. Among possible absorber materials, novolak photoresist is a practical choice having a quartz/film reflectivity of 1%. Features down to 0.25?m are regularly patterned for these masks with a MEBES I using Shipley SAL-601 or PMMA electron beam resist.

Soft X-ray Project Lithography (SXPL) may be used to fabricate high resolution structures for future integrated circuits. This technique will use a reflection mask which is a substrate coated with an x-ray multilayer mirror and patterned with a thin ({approximately}50 nm) layer of x-ray absorber. Mask patterning processes must not degrade the reflectivity of the x-ray mirror and mask repair techniques must be developed. The technical challenges of conventional reflecting optical imaging system designs are severe and mask technology can have a significant impact on this issue. Specifically, innovative mask designs can reduce the complexity of the optical system by decreasing the number of mirrors and replacing aspheric optical surfaces with spherical surfaces. We have developed a technique, called Encoded Mask Lithography, with which we have designed an optical system which uses only two (spherical) imaging mirrors and has 30 mm diameter field of view.

The inhibition mechanism is novolak/diazonaphthoquinone resists is widely studied. We are interested in evaluating the effect of sensitizer distribution in the resist on the dissolution properties. Langmuir- Blodgett film deposition allows us to control the spatial distribution of sensitizer and thus evaluate its 'sphere of influence' on the dissolution inhibition of the novolak resin. We have performed both dissolution studies and lithographic studies on films with sensitizer deposition only on the surface as well as films with an embedded layer of sensitizer between two layers of novolak polymer. The surface layer films with one monolayer of sensitizer do not demonstrate any inhibitive properties unless the films are thermally treated. However, for the analogous embedded layer films, inhibition does occur without any baking of the samples. Also, in the lithographic studies, pattern contrast is improved considerably upon baking the samples; indicating that intimate interactions between resist components which may be induced by the thermal treatment is necessary for dissolution inhibition to take place.© (1991) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

To shift the onset of skin resistance in off-chip interconnects novel microstructures are proposed as a method to maximize conductor surface area for a given pitch. We consider lines of Scm length operating at 1GHz with ~6dB attenuation. Assuming thin but wide lines (aspect ratio 10:1) for low distortion then conventional microstrip lines would require a 5Oim pitch theoretical study shows that a vertical coplanar line essentially a tall double-strip coplanar line could be packed twice as densely. Circuit theory and field theory which accounts for fringing are used to predict a reduction in high frequency resistance in laminated microstrip lines as well. Reduction in resistance was observed in a large-scale model of a transposed multiple-conductor microstrip line. Microfabrication of phase-velocity matched laminated lines demonstrated the difficulty of matching material constants emphasizing the relative strength of transposed laminated lines which do not require matching.© (1991) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Central obscuration of the pupil is a prominent feature of many high performance reflective designs being considered for sub-200nm lithography. The performance of centrally-obscured designs were investigated using computer simulations of projected image intensity and major features from simulation were experimentally confirmed. The effect of partially-coherent illumination on centrally-obscured systems was studied and an optimized annular illumination system is proposed. 1.

In this paper we present a class of nonlinear adaptive filtering schemes to detect edges to the nearest pixel in digital images. These one- or two-dimensional filters are adapted by training to a subset of image data to produce peaked output at user-specified edge locations within the image. A nonlinear adaptive algorithm has been developed and has shown improved performance over standard cross correlation schemes in binary classification situations. The resulting filters are then applied non-adaptively to the entire image set, and signal peaks within the image are detected to produce a binary edge map. A short theoretical development of the algorithm is given, and results for images representative of harsh alignment conditions are presented.

Langmuir-Blodgett (LB) techniques can be used to prepare very thin organic films, and we have investigated their use in lithographic applications. Resist films were prepared usmg novolac and poly(p-hydroxystyrene) resins and both near and deep UV sensitive photoactive compounds (PAC). LB resist films formed from mixtures of polymer and PAC behaved lithographically in a similar fashion to spin cast films. LB films as thin as 300 A were sufficient to protect 500 A of chromium during wet chemical etching. Initial attempts to use LB films as the top layer in a bilayer resist system indicated that silylation of the LB films may yield etch rate ratios of up to 10: 1 vs hard baked novolac. Investigation of the "sphere of influence" of PAC molecules by controlling their distribution throughout the film indicates that strong interactions may exist between consecutive LB layers of PAC. LB films may also be useful for modifying the surface concentration of PAC.

The distortions that arise from thermal stresses in X-ray masks which employ the embedded absorber structure are modelled and analyzed. By using a quasi two-dimensional model, both in-plane and out-of-plane distortions were characterized and their dependence on the fractional absorber coverage was calculated. These distortions were found to be large when the absorber was initially deposited at a high temperature; however they can be greatly reduced by adding a buffer layer between the absorber and membrane. The Young's modulus and the linear expansion coefficient of this buffer layer are chosen such that the mask distortions are compensated for. Without the buffer layer, the shear and peeling stresses at the absorber-membrane interface were found to increase exponentially with distance near the absorber edges and may cause fatigue. These results were found in agreement with computer simulations.

A four fold improvement in photoyield has been observed in GaN films coated with CsBr films (CsBr∕GaN) relative to CsBr∕Cr photocathodes reported in previous studies. A model is presented involving photoemission from intraband states in the CsBr film and direct electron injection through the CsBr film from the GaN substrate. The lifetime of the films at high current density >90A∕cm2 is limited by the temperature rise in the GaN films caused by the high photon absorption at 257nm. The lifetime can be improved by utilizing a cooled substrate or a high thermal conductivity substrate under the GaN films.

Controlling the crystallization process of semiconductor materials is essential for many applications, including monolithic three-dimensional integration of devices. In particular, crystallization must be done while keeping to the thermal budget of underlying device layers. Using a 532nm laser pulse of 10ms duration, we demonstrate partial melting of silicon, which yields crystallites several microns in size upon cooling. A ⟨100⟩ preferential orientation was observed in these crystallites, which may be a consequence of partial melting. By observing the melt evolution, we show that a partial melt can form on the millisecond time scale and does not require thermal equilibrium to exist. By reducing the laser pulse duration further, such a process could be compatible with the thermal constraints of three-dimensional integration.

Experimental results on a new CsBr∕GaN heterojunction photocathode structure are presented. The results indicate a fourfold improvement in photoyield relative to CsBr∕Cr photocathodes. A model is presented based on intraband states in CsBr and electron injection from the GaN (with 1% addition of indium) substrate to explain the observed photoyield enhancement. The photocathode lifetime at high current density (>40A∕cm2) is limited by laser heating of the small illuminated area. Calculations are presented for sapphire and diamond substrates, indicating a factor of 20 reduction in temperature for the latter. The results are encouraging for the realization of a high photoyield photocathode operating at high current density with long lifetime.

The attractiveness of electron beam systems would be greatly enhanced if the throughput could be improved. One approach, described previously by the authors employs a uniform axial magnetic field to focus thousands of electron beams simultaneously [D. S. Pickard et al., J. Vac. Sci. Technol. B 21, 2709 (2003); T. R. Groves and R. A. Kendall, ibid., 16, 3168 (1998)]. The beamlets never combine to form a common crossover, thereby avoiding the throughput limitations due to space charge blurring. With this approach, one challenge was to fashion a detection scheme that maintains a tight beamlet packing density (250μm pitch) while minimizing cross-talk between adjacent secondary electron signals, either by crossing trajectories or within the detector. A pin-diode-based detector was investigated as a potential component of the multielement detection scheme for the authors’ system. The detector features a two-dimensional array of elements on high resistivity float-zone silicon. The detector attributes that were a...

There is a continuing need for high intensity electron sources that will operate in demountable vacuum and can be externally modulated. Materials with wide bandgap, e.g. diamond, are rugged but need photon energies exceeding the bandgap to emit efficiently and this rules out the use of CW lasers. We have found that a photocathode of CsBr is both adequately intense(>150A/cm{sup 2}) and rugged and can be excited with photons of energy of 4.8eV(257nm). This is below the energy gap of CsBr(7.3eV) but such operation can be explained by the presence of intraband states about 4eV below the conduction band minimum.

Shot noise in electron-beam and extreme ultraviolet (EUV) exposure fundamentally limits the useful sensitivity of resists. Here the exposure, amplification, and deprotection of chemically amplified resists are treated as a sequence of statistical events to determine the effect of shot noise. Noise among processes is additive, and a sequential Poisson process is used to illustrate the “acid bottleneck” that occurs when less than one acid electron is generated per input electron. Lateral mixing, due to effects such as electron scattering and acid diffusion, is accounted for in a binomial sorting model that shows that the effect is simply a function of the total number of quanta reaching the voxel irrespective of their path. Electron exposure is estimated to generate on average one acid every 3.8, 21, and 36nm along the trajectories of 5, 50, and 100keV electrons respectively. Acid generation appears well correlated with energy deposition, at about 22eV/acid required for EUV and electron exposure or three ti...

We have simulated and measured 60–120fs time jitter of photoelectron pulses emitted by a nitride photocathode at 100GHz rate as in order to evaluate the resolution performance of a previously proposed photonic analog to digital converter. Recently, there has been an increasing demand for high speed analog-to-digital converters (ADCs) for microwave bandwidth signals. State of the art electronic ADCs have reached 10Gigasamples∕second (GS/s), 6–12bit performance [P. W. Juodawlkis, J. C. Twichell, G. E. Betts, J. J. Hargreaves, R. D. Younger, J. L. Wasserman, F. J. O’Donnell, K. G. Ray, and R. C. Williamson, IEEE Microwave Theory Tech. 49, 1840 (2001)]. We have previously introduced a photoelectronic ADC implementation with measured performance of 3bits at 100GS∕s [K. Ioakeimidi, R. Leheny, S. Gradinaru, K. Ma, R. Aldana, J. Clendenin, J. S Harris, R. F. W. Pease, IEEE Trans. Microwave Theory Tech. (to be published); Conference on Lasers and Electro-Optics, Baltimore MD, 1–6 June 2003]. The basic operating pr...

The need for a high performance (low energy spread 3months per spot, emission stability

Achieving clean surfaces is a major and challenging requirement for the study of surfaces and surface reactions. Nondestructive cleaning is a crucial step in semiconductor manufacturing, growth of materials, and processing. We use photoemission spectroscopy (PES) to systematically study the core and valence band electronic structure of various chemical treatments of InP(100), GaAs(100), and GaN(0001). These surfaces undergo wet chemical cleaning of H2SO2/H2O2/H2O followed by thermal heating. In order to achieve the necessary surface sensitivity and spectral resolution, synchrotron radiation in the energy range of 60–1000 eV is used for PES. In tuning the sulfuric acid based chemistry, we achieve oxygen free GaAs and InP surfaces, as shown in our valence band and core level PES analysis. Furthermore, core level PES shows oxygen coverage of the GaN surface is reduced to less than 0.1 monolayer (ML). The carbon coverage is also reduced dramatically for the III–V surfaces

Negative electron affinity (NEA) photocathode electron sources operating within sealed off tubes (closed vacuum) and demountable vacuum systems (open vacuum), suffer from different factors that affect the initial high performance and longevity. To aid in the understanding of the factors affecting an open vacuum system, the chemistry of a cesium and oxygen (Cs/O) adlayer was investigated, focusing on the role of oxygen during the NEA co-deposition process (activation) and operation of a GaN (0001) emitter. Synchrotron radiation photoemission spectroscopy and a focused mercury arc discharge lamp are used for analyzing surface chemical species and total quantum yield, respectively, as the photocathode ages. The results for the activation within a demountable vacuum system indicated: (1) oxygen concentration increased over time on the Cs/O activated surface independent of the starting substrate as observed in related studies of InP (100) and GaAs (100), (2) the dominant initial oxygen species is assigned to an ion of molecular oxygen (di-oxygen) in the Cs/O activation layer having −2 charge state that changes to the −1 charge state as the quantum yield decays, and (3) the chemical changes of the Cs/O adlayer are not accompanied by a significant loss of cesium from the surface, but instead an increase in oxygen from the open vacuum system occurred. We present a chemical model for the activation layers and their transformation over time consistent with observations (1) through (3). The quantum efficiency (QE) from the GaN (0001) photocathode remained constant within a few percent over a 10 h period at ⩾20% QE and decayed by less than a factor of 2 over the subsequent 7 h as the chemical changes at the surface took place. The slowly decreasing monotonic yield gives preliminary evidence that GaN is a prospect for a more robust emitter. A long lived photocathode of this type offers near constant quantum yield over several hours when operating within open vacuum conditions by simply adjusting laser power with a feedback circuit. On the other hand, this decay can most certainly be avoided in sealed tubes by maintaining a pristine closed off environment with the proper over pressure of Cs. A detailed account of the effects of the chemical changes in the Cs/O adlayer on the resulting emission properties for GaAs and GaN (open vacuum) are discussed in a separate report [F. Machuca et al., J. Vac. Sci. Technol. B 20, 2721 (2002)].

A distributed-axis system can be used to avoid the space-charge limits of conventional electron-beam lithography systems. Our approach is to use a uniform magnetic field to simultaneously focus an array of beams. We have built a test bed and report the first experimental results. Sub-100 nm resolution has been obtained and is presently limited by aperture size and interference. Sub-10 nm resolution should be possible given a suitably small source.

Source contamination, photovoltage effects, and stimulated electron desorption of cesium are factors that determine the initial high performance and longevity of negative electron affinity electron sources. As part of the study of these factors, we investigate the effect of oxygen contamination during the activation process and operation of a traditional GaAs (100) and a novel GaN (0001) emitter. We use synchrotron radiation photoemission spectroscopy, a focused mercury arc discharge lamp, and a helium neon (HeNe) laser to obtain simultaneously elemental analyses of the emitting surface and the corresponding total quantum yield at various stages, respectively, for GaN and GaAs. Our results indicate: (1) carbon uptake does not occur in our carbon free system over time, (2) oxygen uptake is observed for both GaN (0001) and GaAs (100) activated surfaces, a property common to the chemistry of the Cs/O adlayer, (3) the oxygen species appears to change over time and the initial species is assigned to an ion of ...

The brightness of a multicusp-plasma ion source was characterized to determine its suitability for a proposed maskless ion-beam lithography system. In order to be competitive with similar systems utilizing electron sources, the brightness requirement must be met to satisfy throughput demands. For this requirement, 1 nA into each beam diameter of 50 nm is needed at the target plane; this implies a brightness of about 1000 A/cm2/sr at the entrance to the objective lens. The brightness of the source has been characterized using 5-μm-diam extraction apertures under two different extraction configurations: (1) with the apertures directly illuminated by the source and (2) with the apertures flooded using an extraction system. The maximum brightness achieved for either configuration was 15 A/cm2/sr at 3 keV beam energy, He+ ions, and continuous wave operation. Although this configuration falls short of the requirement for high-throughput exposures, it is also possible to operate the multicusp-plasma ion source i...