We demonstrate direct flare measurements of the 4-mirror projection optics in the Engineering Test Strand (ETS). By using a conventional resist clearing method, also called the Kirk method, flare was directly measured in the ETS. Although flare has been verified to be constant across the field in our study, different flare values have been observed in different directions. The flare anisotropic variation makes flare compensation more complicated, and it may even result in non-uniform CD changes in different directions. In order to fully identify the directional flare variations of the projection optics in the ETS, a special flare measurement feature, a spoke pattern, has been created. The spoke pattern has been used to measure flare in azimuthal directions (every 15 degrees), and rigorously measured flare in all directions is presented in this study. Measured flare is highest in horizontal direction and lowest in vertical direction. The azimuthal flare variation is believed to be due to anisotropic mirror polishing. Measured flare in all directions qualitatively agrees with the polishing signatures of individual mirrors in the measured optics.
The source of flare in EUVL systems is mostly from the mid-spatial frequency roughness (1 /μm - 1 /mm spatial periods) of mirrors. Due to the challenges in polishing mirrors to a small fraction of the wavelength, flare in EUV lithography tools is expected to be greater than flare in current DUV tools. Even though EUV flare is constant across the field, there can be within-die flare variations due to variations in layout density. Hence, it is expected that to meet the CD control requirements for the 32 nm node, Flare Variation Compensation (FVC), akin to Optical Proximity Correction (OPC) would be required. FVC needs the within-die flare level estimated by convolving the Point Spread Function due to scatter (PSFsc) with the mask layout. Thus, accurate knowledge of the system PSFsc is essential for FVC. Experimental results of the Modulation Transfer Function (MTF) technique to estimate flare and the PSFsc of the Engineering Test Stand (ETS) are presented. It was also determined that due to the nature of the PSFsc in EUVL tools a more accurate measure for flare would be to use the 0.5 μm line as opposed to the current 2 μm line standard for measuring flare on DUVL tools.
The minimum gate CD for the 32 nm node is 15 nm and the CD control requirement on the gate CD is < 2.5 nm 3σ. One of the major concerns for meeting these targets using EUV lithography is flare. Flare degrades the aerial image contrast which decreases the process window, and within-die chrome density variation results in local flare variation which worsens the CD control. Since mirror roughness contributes to flare, mirror polishing needs to be improved so that the Mid Spatial Frequency Roughness (MSFR) will be reduced to < 0.14 nm/mirror for 6 mirror imaging systems. In this paper, we will determine the minimum acceptable flare for the 32 nm node to meet the CD target and control requirements using modeling and present methods to meet them as demonstrated by experiments run on the Engineering Test Stand (ETS). Effectiveness of flare mitigation methods using chrome dummification and negative tone resists are quantified, and the capability of Flare Variation Compensation (FVC) to meet CD control targets are verified experimentally.
We demonstrate direct flare measurements of 4-mirror projection optics in the Engineering Test Stand (ETS) using a conventional resist clearing method (the Kirk method). Two extreme UV lithographic projection optics, one with higher flare than the other, have been characterized and the results compared. The measured results have also been compared to analytical calculations based on measured mirror roughness and the extended point spread function. Full-field flare across the 24 mm field width has been measured, and we have verified that flare is constant across the field for EUV lithography as predicted. Horizontal (H) and vertical (V) flare bias has been observed and the cause of the H-V flare bias has been investigated. The main cause has been identified to be anisotropic mirror polishing. Simulations with the 2D Power Spectral density function have confirmed the experimental results.
The defect printability for two types of patterns isolated lines and contact holes was studied through a finite-difference time-domain (FDTD) computer simulation approach at the EUV wavelength of 13.5 nm. A three-dimensional FDTD solver was employed for aerial image analysis to assess the overall acceptable defect size for both opaque as well as clear phase defects for the 32 nm technology node.
Early production EUV exposure tools may have difficulty achieving flare requirements in the 5-6% range for the 32nm technology node. In this case, flare compensation may be needed to achieve the necessary CD control budget for production. This paper explores both experimentally as well as computationally wafer CD compensation though mask CD resizing so that proper CD control across the exposure field can be maintained. Experimental resist data collected on POB#2 of the Engineering Test Stand (ETS) suggest that even a simple linear CD compensation model can signifantly improve CD contorl in the presence of flare variation. Extending a similar concpet to a hypothetical full-field 25×33 mm2 mask area containgin 20 nm gate CDs shwos taht CD compensation, while computationally demanding, can be realized through a convolution approach of a 1×1 mm2 mask area using a non-uniform adaptive grid.
The Power Spectral Density (PSD) function for a large-field EUV exposure system is used to compute the impact of flare on critical dimension (CD) control for masks exhibiting Cr density changes that result in cross-field flare variation. It is shown that open field flare must be controlled to 11 percent for 30 nm isolated features and 6 percent for 20 nm isolated features for an NA equals 0.25 system assuming a +/- 3 percent CD control budget allocated to flare. Based on these results individual mirror surface roughness specifications for the mid-spatial frequencies of about 0.14 and 0.11 nm RMS, respectively, are needed for future chip production. This paper also discusses that either mask 'dummification' or CD resizing can be applied to minimize the effects of flare on CD control in case mirror fabrication targets may be difficult to achieve.
According to the International Technology Roadmap (ITRS) EUV exposure tools are expected to support both logic and memory manufacturing for the 70 nm node and beyond. In order to meet the CD control and overlay requirements at the expected tolerance levels for high volume manufacturing, tool performance from ideal need to be assessed and quantified, the impact of lens aberrations and flare to the aerial image at the system level were studied for three different technology nodes of interest: 70, 50, and 35 nm. This was done by computing the electric field at the mask level by making use of a rigorous mask simulator based on the differential method. The subsequent aerial image was then computed by suing a standard scalar imaging model. The effects on critical dimensions (CDs) through focus as well as overlay were investigated in order to arrive at a tolerance level for lens aberration and flare contribution at the system level. It is shown that isolated line CD control requirements set an upper limit on the overall imaging performance while for overlay both feature types seem to play a similar role that defines the tolerance limit. Mandating a 10% CD control range leads to a 30 milli Wave (RMS) aberration requirement. This conclusion was reached by analyzing the response of each of 37 Zernike coefficients separately for both isolated and nested lines. Employing two separate sets of KrF PMI data that were scaled to a range of RMS values seemed to further reinforce the previous aberration tolerancing conclusion. A separate but similar flare analysis indicates that cross field flare variations need o be controlled to within 2% to meet CD control requirements that are consistent with ITRS roadmap.
EUV masks generally mandate rigorous scattering models as the thickness to wavelength ratio is typically on the order of 30:1. In addition, boundary conditions at the absorber/air interface lead to non-zero electric fields even in the absorber region while the phase itself generally experiences a complex cross mask behavior that is pattern type as well as size dependent. Using two different types of rigorous simulators that are based on the differential method (LithoLand) and the time-domain finite-element method (EMFLEX) we explore and quantify the subtleties associated with the EUV mask when compared to a thin mask obeying the Kirchoff approximation. Both rigorous simulators predict that 30 nm isolated pattern, especially for NA >0.20 experience a focus shift of about 50 nm relative to best focus for the thin mask case. This effect occurs even when the illumination is normally incident to the mask. Furthermore, the Bossung curve for isolated patterns show an asymmetry through focus that is absent for the thin mask case and is sensitive to the partial coherence condition of the illuminator. Nested features seem to be far more immune to this unexpected anomaly. The origin of the predicated focus shift and asymmetry is explained through the complex phase behavior for the EUV mask which is fundamentally different from the thin mask case.
While condenser aberrations under Koehler illumination were previously treated in the literature their mathematical derivation did not take conservation of radiance into consideration. Here we make use of a more rigorous derivation of the mutual intensity where the source deformation term is treated in the context of radiance conservation. The derivation predicts that condenser aberrations lead to radiance invariance while aberrations have a direct bearing on illumination uniformity and the angular extent of the local effective source. This result significantly contrasts with the previously established conclusion in the literature that condenser aberrations lead to a modification of the source radiance but preserves irradiance in the reticle plane. Source aberrations of first and third order are derived and then systematically explored both analytically and numerically. Aberration impact on linewidth control are further considered and quantified from the aerial image perspective. It is shown that third order coma has the most significant impact on CD control as a result of the asymmetry in the deformation of the source shape. Similarly coma also significantly impacts overall mask illumination uniformity.
It is known that the wafer shot maps generated by the Nikon stepper software are not truly optimum, and if one can understand the uniqueness of the Nikon software, more die may be placed on the wafer. We have developed two software applications: WAFER OPT finds an optimum wafer shot layout, and GRAPH DRAW draws an existing wafer shot map generated by the Nikon stepper software for comparison to the optimized version. A `cleaner' shot map, with fewer clusters yet significantly more die on the wafer, can be achieved after assessing the validity of changing an existing wafer map to an optimum one. Hence, the two computer applications can result in greatly enhanced stepper throughput and reduced defectivity from the partial clusters often necessary in a nonoptimum wafer shot map. Given a set of parameters such as exposure field dimensions, die layout within a reticle, partial cluster criterion, wafer size, wafer flat length, wafer edge exclusion width, and wafer identification character size, the WAFER OPT program mathematically finds a shot map having the maximum number of complete die using the minimum number of shots and having the least sensitivity to map placement errors of the stepper stage. In addition, WAFER OPT calculates x and y map offsets, uniquely defined in the Nikon software, that can be directly entered in the stepper data file. Another software application called GRAPH DRAW can be used to draw an existing wafer shot map for comparison to the optimized version. A case that resulted in a significant increase in the number of die per wafer and a decrease in both the number of exposure fields and the number of partial die on the wafer is presented.
It is shown that second-harmonic generation efficiency in a photovoltaic model depends on the relationship of illuminated size to grating period. Model test experiments are proposed and realized. To explain the photo-induced second-harmonic generation (SHG) in glass optical fibers, two groups of models have been suggested, some based on separation of charges and appearance of a strong electrostatic field the others based on orientation of defects. However, a strong electrostatic field (approximately equals 104 V/cm) appears also in the last models. That is why neither the experiment on the discovery of such a field nor the experiment on the measurement of the component (chi)(2) tensor ratio gives an answer to the preference of either model. This paper shows that the efficiency of SHG by a (chi)(2) grating, resulting from a coherent photocurrent, depends strongly (approximately equals (r0/(Lambda) )-4) on the ratio of the transverse size r0 of the light beam, used for the grating preparation, and the grating period (Lambda). So we think that an experiment on SHG in a bulk sample to check this point will make it possible to clear up the mechanism of (chi)(2) grating formation. Another possibility is an experiment with the (chi)(2) grating writing by pump and second-harmonic waves, propagating in opposite directions in a fiber. In this case the grating period will decrease to (lambda)p/4 approximately equals 0.25 micrometers.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.