Metal resist for extreme ultraviolet lithography characterized by scanning transmission electron microscopy

We characterized the structures of metal resists used in EUV lithography by low-voltage aberration-corrected scanning transmission electron microscopy (STEM) combined with electron energy-loss spectroscopy (EELS). This study presents the first atomic-level observation of resist components in resist film. The structures of metal (zirconium or titanium) oxide cores are unambiguously identified, and the local elemental distribution in the resist film is obtained. The initial size of zirconium oxide cores is well maintained in the resist film. However, titanium oxide cores tend to aggregate to form an indefinite structure. The spatial distribution of metal cores may influence lithographic characteristics.

T he trend in optical lithography is that smaller is better. A demand for smaller patterning in semiconductor devices has led to the use of shorter-wavelength light sources in lithography for the fabrication of high-density and high-performance semiconductor devices. EUV lithography 1,2) has received attention as the next-generation method that uses wavelengths as small as 13.5 nm. However, the low power of EUV light sources limits EUV exposure tool technology [3][4][5] and necessitates the use of highly sensitive EUV resist materials. Therefore, the development of EUV resist materials continues to be a critical challenge for the implementation of EUV lithography. [6][7][8] Some important properties of resist materials that must meet certain requirements include resolution, line-edge roughness, and sensitivity. 9) Conventional, chemically amplified resists have been studied for use as EUV resists. 10,11) New metal resists [12][13][14][15][16][17][18][19] were reported to offer superior resolution, 12,13) sensitivity, [14][15][16] and etching resistance, 12) and have good shelf-life stability, 19) but none meets all the requirements simultaneously. Our group at EIDEC has developed a metal resist that comprises a metal oxide core and an organic molecule shell. 7) Figure 1 shows typical scanning electron microscopy (SEM) images of the line patterns of the EIDEC metal resist. This metal resist has good sensitivity of 9 mJ=cm 2 , which is a very important property for obtaining high throughput with the low-intensity EUV light sources currently available. The finest line pattern in Fig. 1(a) has a line width of 11 nm and a line pitch of 120 nm, but the resolution and line-edge roughness need to be further improved.
Transmission electron microscopy (TEM) as well as semiconductor device inspections 20) have been used to characterize resist materials for EUV lithography. Cross-sectional TEM images of metal resist line patterns provide information on device structure and material morphology. 21) It is also important to understand the spatial distribution of the shells of organic molecules that surround individual metal oxide cores in the resists, which conventional TEM techniques have been unable to show. Recently, the use of aberrationcorrected scanning TEM (STEM) at low electron-accelerating voltages has provided atomistic information such as atomic positions and electronic states in nanomaterials without seriously damaging their structures as occurs with electron beam irradiation. 22) The structures and reaction mechanisms of metal resists can be characterized in further detail using the latest techniques, enabling the development of new resist materials with improved resist properties. In this study, we demonstrate atomic-level imaging and elemental identification of metal resists using low-voltage aberrationcorrected STEM.
The metal resist materials evaluated in this study were composed of metal oxide nanoparticles and organic molecules forming a core-shell state. Zirconium oxide (ZrO x ) and titanium oxide (TiO x ) were selected as the metal oxide cores and methacrylic acid (MAA) was used as a component of the shell. The metal resist materials were synthesized using the sol-gel method. The resist solutions were prepared by adding appropriate amounts of the nanoparticle powder to propylene glycol monomethyl ether acetate (PGMEA).
Light scattering measurements with a Zetasizer Nano ZS (Malvern Instruments) evaluated the size of the metal resist particles in the PGMEA solution. The refractive index and viscosity of PGMEA at 20°C were 1.402 and 1.200 mPa·s, respectively. The refractive index of ZrO 2 was 2.17 and the refractive index and absorption coefficient of TiO 2 were 1.95 and 0.01, respectively. The metal element content in the metal resists was analyzed by inductively coupled plasma atomic emission spectroscopy using a Seiko Instruments SPS400. The carbon and hydrogen contents of the resist powder were determined simultaneously using a vario MICRO cube system (Elementar Analysensysteme).
STEM images of the metal resists were obtained using a JEOL JEM-2100F transmission electron microscope, equipped with DELTA spherical aberration correctors, 23) at an electron-accelerating voltage of 60 kV. Energy dispersive X-ray spectrometry (EDS) and electron energy-loss spectroscopy (EELS) analyses were performed using JEOL Centurio silicon drift detectors and a Gatan quantum spectrometer, respectively, installed on the microscope. For STEM analyses, each metal resist solution was dropped onto a molybdenum microgrid coated with an amorphous carbon film with holes and air-dried at 20°C. The microgrid was then heated at 80°C for 1 h in a vacuum to eliminate volatile species, such as residual PGMEA solvent, prior to STEM observation. Commercially available nanoparticles of ZrO 2 (Wako Pure Chemical Industries) and TiO 2 (Ishihara Sangyo) were used as references for EELS analysis.
First, we examined the line patterning of the ZrO x -MAA resist. Spin-coated films of the resist on silicon wafers were delineated using a High-NA small field exposure tool (HSFET; NA = 0.5, quadrupole illumination) at EIDEC and developed for 30 s with n-butyl acetate. Satisfactory patterning properties (Fig. 1) were achieved with the ZrO x -MAA resist. The post-coating delay of the ZrO x resist was comparatively better than that of the TiO x -MAA resist.
We then characterized these two metal resists using STEM to further understand the differences between their line-patterning properties with respect to their morphology.  21) the STEM observation at a reduced electronaccelerating voltage of 60 kV preserved the amorphous nature of ZrO x and TiO x cores, even after high-magnification imaging.
The difference in the distributions of the ZrO x and TiO x cores is attributed to their affinity to the surrounding MAA molecules, the PGMEA solvent, or both, and to the interactions between the cores. Since inhomogeneity of the components in the resist film is crucial in determining such properties as resolution, line-edge roughness, and sensitivity, 24) techniques to evaluate the homogeneity of metal resists are important for the development of improved resist materials. The images presented in Fig. 2 prove that low-voltage STEM    EDS and EELS analyses proved that the Ti=C atomic ratio in the TiO x -MAA resist layer was 1.4-1.5 times greater than the Zr=C ratio in the ZrO x -MAA layer. Assuming that all carbon atoms detected in the resist layers existed as MAA molecules, MAA=ZrO x and MAA=TiO x molar ratios were estimated to be 0.38-0.50 and 0.29-0.34, respectively. These molar ratios were equal to 1.5 in the original resist solutions, but two-thirds or more of the MAA molecules were eliminated together with the PGMEA during preparation of the dried resist layers prior to STEM analyses. The ZrO x -MAA resist layer was found to contain more MAA than the TiO x -MAA resist layer, suggesting that ZrO x cores exhibit a higher affinity to MAA.
Elemental distribution in the metal resist layers was further analyzed by STEM-EELS chemical mapping. Figures 4(a) and 5(a) show the ADF-STEM images of freestanding ZrO x -MAA and TiO x -MAA resist layers, respectively, that were selected for analysis. EELS images were acquired from the boxed areas indicated in the figures, and elemental distributions were determined on the basis of the intensities of carbon and oxygen K edges and zirconium M or titanium L edge at each measured point. Figure 4(b) shows the elemental maps for the selected area of the ZrO x -MAA resist layer in Fig. 4(a). Individual ZrO x cores that appear as bright regions in the ADF-STEM image are also clearly seen in the zirconium map. The carbon map profile is distinctly different from that of the zirconium map. The signal intensity of carbon is higher in the space between the ZrO x cores, suggesting that MAA-based components are concentrated there. Oxygen atoms exist in both the ZrO x cores and MAA molecules but are more concentrated around the cores as determined from the observed higher signal intensity in the oxygen map. Figure 5(b) shows the elemental maps for the selected area of the TiO x -MAA resist layer in Fig. 5(a). Individual TiO x cores are not separately identified in the titanium map because of aggregation that is suggested by the contrast profile in the ADF-STEM image. Examination of the carbon map clearly shows that MAA-based components exist mainly in the gap between the aggregated TiO x cores. Oxygen is detected in both the TiO x cores and the MAA region; its distribution is relatively uniform even though its signal intensity is lower than that of the other elements.
The STEM-EELS chemical mapping proves that both ZrO x -MAA and TiO x -MAA resist layers are actually composites of metal oxide cores and MAA molecules with the spaces between the metal oxide cores occupied by MAA. ZrO x cores were found as individual nanoparticles, while aggregation of TiO x cores was confirmed by the titanium map of the TiO x -MAA resist layer. In metal oxide-based resists, carboxyl groups of the surrounding organic molecules such as MAA are expected to be bound to a metal oxide core, resulting in a core-shell state. 14) The present study suggests that such a core-shell state is maintained in a more stable condition for the ZrO x -MAA resist than for the TiO x -MAA resist, even during the process of PGMEA solvent elimination, preventing ZrO x core aggregation. The excellent post-coating delay of ZrO x -MAA compared to that of TiO x -MAA is reasonably well explained by these models.
In conclusion, we have demonstrated atomic-level imaging and elemental identification of metal resists using lowvoltage aberration-corrected STEM. To the best of our knowledge, this study is the first atomic-level observation of resist components in resist films. STEM observed the morphology of the metal resist film, and the resist component of a single core of ZrO x was identified as an isolated nanoparticle, while TiO x cores were found to be aggregated. The morphologies of each component in the resist films may influence lithographic properties such as resolution, line-edge roughness, and sensitivity, in addition to the postcoating delay. The inhomogeneity of resist components, such as photoacid generators and quenchers, is being examined as it becomes more important for the delineation of single nanometer patterns. 24) STEM is a useful tool for the visualization and atomistic study of resist materials. Further study of the role of shell molecules in metal resists is underway.