Development of a high-power EUV irradiation tool in a hydrogen atmosphere

Extreme ultraviolet (EUV) lithography has recently been utilized as a high-volume manufacturing technology for advanced semiconductors. An EUV mirror can be easily contaminated in the existence of a residual hydrocarbon vapor gas inside an exposure chamber in a vacuum environment, which reduces the reflectance of the Mo/Si multilayer coating. To reduce this carbon contamination, hydrogen gas is introduced at a pressure of a few pascals in the EUV scanner. However, during this process, the multilayer may be damaged by hydrogen. In addition, the multilayer surface can become oxidized by residual water vapor in the vacuum chamber. Therefore, an EUV irradiation tool in hydrogen and water vapor atmospheres was developed and installed at BL-09 of the NewSUBARU synchrotron light facility to evaluate the cleaning effect and irradiation durability of the Mo/Si multilayer. The EUV irradiation intensity increased up to 6 W cm−2, and the hydrogen pressure reached 70 Pa.


Introduction
Since 2019, extreme ultraviolet (EUV) lithography has been used in high-volume manufacturing of advanced semiconductors. 1,2) The working wavelength of EUV lithography is 13.5 nm, which is near the soft X-ray region. [3][4][5][6] As absorption of the material is extremely high in the EUV region, a refractive lens cannot be employed for optics in EUV lithography. Thus, EUV optics are composed of reflective mirrors coated with Mo/Si multilayer. However, the reflectance of a single-layer coating at a normal angle of incidence is very low. For examples, the reflectance of a single-layer mirror of Pt and Ni at a normal angle of incidence is 0.4% and 0.2%, respectively. These metals are typically used as reflective coatings for grazing incidence X-ray optics. On the other hand, the Mo/Si multilayer has a reflectance of over 60% in the EUV region due to constructive interference of each layer. [7][8][9][10][11] Residual hydrocarbon gas in a vacuum environment is origin of carbon contamination on the mask and optics, which causes the reflectance drop. [12][13][14] To reduce this contamination, several pascals of hydrogen gas are introduced into the EUV scanner. Dolgov et al. reported the cleaning of carbon contamination by EUV-induced plasma. 15,16) In a hydrogen atmosphere, high-power EUV generates EUV-induced hydrogen plasma. However, hydrogen plasma would cause reflectance decrease of the Mo/ Si multilayer. Because hydrogen has a smallest atomic number, it easily diffuses into the Mo/Si interface in the Mo/Si multilayer film, which causes bubbles to form at the interface. This hydrogen damage is called a "blister". 17) Researchers have just begun analyzing the hydrogen plasma generation mechanism; 18,19) the mechanism of contamination cleaning using hydrogen gas and the blistering condition has not been clarified yet. Furthermore, the surface of the Mo/Si multilayer can be oxidized by residual water vapor in a vacuum chamber. 20) These reductions in the reflectance will worsen the throughput of the scanner.
Therefore, an EUV irradiation tool with a hydrogen and water vapor atmosphere is required to evaluate the blister and oxidization damage of the multilayer. Currently, the power of the EUV source is 250 W at the intermediate focus position, and the expected EUV power on the mask is 5 W cm −2 . 17) Thus, we developed a high-power EUV irradiation tool to evaluate the damage of the Mo/Si multilayer in a hydrogen and water vapor atmosphere at the BL-09 long-undulator beamline 21) of the NewSUBARU synchrotron light facility. Figure 1 shows a schematic drawing of the BL-09C end station, which is located downstream of the M8 mirror. A high-power-EUV irradiation tool was developed in an atmosphere of hydrogen gas and water vapor and was called "H 2 -exp tool." The H 2 -exp tool was installed 1050-mm upstream of the resist-outgas measurement chamber. [22][23][24][25][26][27][28] The EUV light in the horizontal and vertical directions was focused on the sample position of the H 2 -exp tool using spherical concave beamline mirrors, which were positioned upstream of M8 mirror. The M8 mirror shown in Fig. 1 is a flat mirror that changes the position of the beam focusing on the sample. The focused beam size was 2.4 mm and 0.5 mm in the horizontal and vertical directions, respectively. A 10.8 m long-undulator, which can provide a highly brilliant soft X-ray beam, was employed as the light source. This beamline is also used as an EUV interference lithographic tool, 29,30) as a interferometer of EUV optics, 31) for soft X-ray emission spectroscopy, 32) and as a resist-outgassing evaluation tool. [22][23][24][25][26][27][28] Figure 2 shows the spectrum generated from the 10.8 m long-undulator measured by a soft X-ray photodiode (AXUV100G, Optodiode Inc.). The undulator gap condition was set to 35 mm. This spectrum indicates the 1stharmonic-order of the light of the undulator. To obtain strong EUV power, we used the 0th-diffraction-order condition of the monochromator. This un-monochromatized specular light had a full width at half maximum (FWHM) bandwidth of 0.6 nm, which is mostly matched with the reflection bandwidth of the Mo/Si multilayer of 0.5 nm (FWHM).

Experimental methods
Two orifices were installed upstream of the H 2 -exp tool to allow for the differential exhaust of hydrogen gas as shown in Fig. 1. The vacuum pressure of the M8 mirror chamber should be maintained at 10 -5 Pa. By contrast, the hydrogen pressure of the H 2 -exp tool was several pascals. The orifices were chosen to be 3 mm in diameter to achieve low conductance. Thus, the maximum hydrogen pressure was 5 Pa, which was limited by the vacuum pressure of the M8 mirror chamber. In this study, the orifice condition was used for the irradiation experiments. Recently, we installed an additional orifice and pumping system, which improved the maximum pressure to 70 Pa. This additional orifice is not shown in Fig. 1. Figures 3 and 4 show the top and side views of the H 2 -exp tool. Hydrogen gas was introduced using a variable leak valve (59024-GE01, VAT Inc.). Water vapor was also introduced by a variable leak valve (951-7172, ANELVA). This leak valve was attached to the same position of the H 2 leak valve, which was not shown in Figs. 3 and 4. Deionized water was filled in a glass tube, which was connected to this leak valve. A capacitor manometer (CMR374, Pfeiffer Inc.) was installed to measure the absolute pressure up to 100 Pa. A quadruple mass spectrometer (M-200QA, Canon Anelva Inc.) was installed to measure the residual gas partial pressure, which can measure the mass number from 1 to 200. Figure 5 shows the results of the quadruple mass spectrometer of the residual gas in the H 2 -exp tool chamber. The acceleration voltage was set to 1400 V. The chamber vacuum pressure before hydrogen or water vapor introduction was 6 × 10 -6 Pa. The majority of residual gas was water.
The EUV photodiode PD Si/Zr coated with Si and Zr layers (SXUV-100 Si/Zr, Optodiode Inc.) was used upstream of the orifice to measure the EUV light intensity. A thick coating of Si and Zr was employed to eliminate the visible light component, which has a low EUV transmittance of less than 3%. Because a general CsI scintillator for X-rays is easily degraded by high-power EUV light exposure, a hexagonal boron nitride (hBN) sintered plate was employed as an EUV scintillator. hBN exhibited high durability to EUV light. A visible light camera was used to observe the EUV  The in situ reflectance of the sample during EUV irradiation was measured using two photodiodes, PD DB and PD R .
The PD DB , which was attached to a beam shutter cylinder, as shown in Fig. 3, can measure the direct-beam (DB) power. The PD R was attached to a rotary arm to measure the reflected (R) beam power. Both PD DB and PD R photodiodes  are SXUV-100 photodiodes (Optodiode Inc.), which is a standard EUV photodiode with high durability and high quantum efficiency. The photodiode surface was not coated with a thin film. The sample can be loaded into the irradiation chamber from the load-lock chamber using a transfer rod with linear and rotary motions. As this transfer rod is a magnet coupling type, the precision of the linear positions is not sufficiently high. The sample position can be monitored using a visible camera to accurately control the irradiation position, and the sample rotation angle can be adjusted to maximize the reflected-beam power. The maximum sample size was approximately 90 mm in diameter, which was limited by the gate valve size of GV1. Moreover, transmission samples such as EUV pellicles can be measured, where the in situ transmittance can be monitored by two photodiodes. In this study, the sample holder could hold six wafer samples and an hBN scintillator. Each wafer was 7.5 mm × 7.5 mm. The sample holder covered the sample edge area, and the covered area was not irradiated by EUV light.
In Fig. 4, the dashed line indicates the beam shutter cylinder, which has a diameter of 186 mm and a height of 82 mm. The direct-beam photodiode was attached to this cylinder, which had a trimmed entrance aperture (20 mm × 2 mm). The vertical position of the cylinder was controlled using a linear motion manipulator.    The EUV irradiation intensity can be calculated using Eq. (1) where I is the photodiode current of PD DB ; the sensitivity of the photodiode (p) is 0.19 A W −1 at a wavelength of 13.5 nm; a and b are the beam dimensions in the vertical and horizontal directions, respectively; and τ is the transmittance of the Si filter utilized for beam attenuation. The measured photodiode current was over 2 mA at the PD DB and PD R without the silicon attenuation filter. In this region, the photodiode current was not linear to the incident beam intensity. 33,34) The measured relationship between the photodiode currents of PD DB and PD SiZr is shown in Fig. 7. The current of PD DB is not linear to that of the PD SiZr over the 1.5 mA region of the PD DB . Thus, a Si filter with a transmittance (τ) of approximately 5% was installed to attenuate the beam intensity for photodiode measurements. Using this filter, the PD DB output current can be attenuated to be 0.7 mA, which exhibits a linear relationship with the incident EUV light intensity. The beam size was 0.5 mm (V ) × 2.4 mm (H) on the sample position. Thus, the EUV light intensity on the sample was ∼6 W cm −2 , which depended on the beam condition.

Irradiation sample and conditions
The irradiation sample was a Mo/Si multilayer coated on a silicon wafer, which had a silicon top layer and no metal capping layer. The sample was cut to a square of 7.5 mm × 7.5 mm. The period thickness was 6.9 nm. The Mo layer thickness ratio during this period was 0.4. The designed angle of incidence was 6°. Four samples were irradiated with a high-power EUV. The undulator gap was tuned to maximize the reflected EUV power because the irradiation wavelength should match the multilayer reflection wavelength. Table I shows the experimental conditions of the hydrogen gas and water vapor atmosphere, vacuum pressure, irradiation time, and EUV intensity. Samples #1 and #2 were irradiated in a hydrogen atmosphere of 5 Pa to evaluate the hydrogen damage. Sample #3 was irradiated as the reference condition in vacuum. Sample #4 was irradiated in a water vapor atmosphere of 1 × 10 -4 Pa.
The in situ reflectance was measured using the H 2 -exp tool every 30-60 min of irradiation. In this measurement, the EUV intensities were measured using PD DB and PD R photodiodes with an attenuation Si filter. After the EUV irradiation, the EUV reflectance distribution was measured around the irradiated area using an EUV reflectometer installed at the BL-10 beamline 27,35,36) at NewSUBARU. The angle of incidence was 6°, and the measured wavelength was tuned to the peak wavelength of each sample at approximately 13.5 nm. To estimate the carbon contamination and oxidization, the X-ray absorption spectra at the carbon K-edge and oxygen K-edge regions were measured at both the EUV-irradiated and unirradiated positions using the same reflectometer. Figure 8(a) shows the in situ reflectance results of the hydrogen atmosphere and the reference conditions for samples #1 to #3. The horizontal and vertical axes show the EUV irradiation dose and the normalized reflectance, respectively. The in situ reflectance was measured using an  1)]. For example, sample #3 was irradiated up to a dose of 130 kJ cm −2 for 9 h and 3.1 W cm −2 . The in situ reflectance was normalized by the reflectance of the un-irradiated reflectance (0 kJ cm −2 ). At the beginning of the irradiation, the reflectance of samples #1 and #2 increased by several percentage points. Because the sample was stored in an air atmosphere, the surface was contaminated. This reflectance rise might have been caused by cleaning with high-power EUV irradiation. The in situ reflectance behavior of samples #1 and #3 appeared to be the same, which dropped by 0.5% after an irradiation dose of 40 kJ cm −2 . Figure 9(a) shows the reflectance distribution measurement results of samples #1-#3, measured by the EUV reflectometer. The reflectance distributions were measured after the EUV irradiation experiment of Fig. 8. The total irradiation doses of samples #1, #2, and #3 were 80, 40, and 130 kJ cm −2 , respectively. The horizontal and vertical axes show the vertical position of the sample and the reflectance, respectively. The beam size of the EUV reflectometer was 0.1 mm (V ) × 0.8 mm (H) in FWHM. The sample reflectance measurement direction in the reflectometer was the same as the vertical direction of the sample stage in the H 2 -exp tool. The  EUV irradiation center is indicated by an arrow in Fig. 9. The outside position of >2 mm from the center was the un-irradiated position, which was covered by the sample holder. The irradiation centers of samples #1 and #3 had 0.7% and 0.2% lower reflectance than the un-irradiated positions, respectively. In samples #1, #2, and #3, reflectance drops (0.7%) were observed at the edge positions of the EUV irradiation area. However, these centers had almost the same reflectance. There were no differences with and without hydrogen gas. Figure 8(b) shows the in situ reflectance during the EUV irradiation of samples #3 and #4. Sample #4 was irradiated in a water vapor atmosphere. The in situ reflectance of #4 constantly decreased immediately after irradiation started. The reflectance was 8% lower than the initial value at an irradiation dose of 80 kJ cm −2 . Figure 9(b) shows the reflectance distribution of samples #3 and #4. The reflectance of the EUV irradiation center of sample #4 was 54%, which was caused by the oxidization of the multilayer. Two  The Japan Society of Applied Physics by IOP Publishing Ltd reflectance drops occurred at the edges of the EUV irradiation area, similar to the other samples. The reflectance at the 1.1-2.2 mm position (58.5%) was also lower than that of the unirradiated position (3.0 mm, 60%), which indicated that there was weak EUV scattering around the center. This scattering occurred at the orifice. The weak scattering also caused oxidization of the Si layer. Figure 10(a) shows the XAS results of sample #1 at the carbon K absorption edge of 275-310 eV. The horizontal axis represents the photon energy. The vertical axis represents sample absorption. These spectra were measured using the total electron yield method. A large absorption amount indicates significant carbon existence. The spectra were measured at the EUV-center-irradiated (0 mm), reflectance drop (0.5 mm), and non-irradiated positions (-2.0 mm). The measured positions were determined using the EUV reflectance distribution shown in Fig. 9. Figure 10(b) shows the absorption at 295 eV for the four samples. There was no difference in the contamination among the vacuum and  The Japan Society of Applied Physics by IOP Publishing Ltd hydrogen conditions of samples #1, #2, and #3. The carbon absorption of the un-irradiated position was low, and that of the center and edge positions were high. Notably, the edge position exhibited significant carbon absorption. Thus, the EUV reflectance drop at these positions was due to carbon contamination and not hydrogen damage under 80 kJ cm −2 EUV irradiation. The carbon contamination of sample #4 on the EUV irradiation position was cleaned by oxidization with a water vapor pressure of 1 × 10 -4 Pa. The residual water vapor pressure of the chamber was approximately 1 × 10 -5 Pa, as shown in Fig. 5. Even in sample #3 under vacuum conditions, the contamination of the center EUV region was less than the edge position. The center might be slightly cleaned by the residual water vapor pressure with high-power EUV irradiation. There was no cleaning effect of hydrogen plasma. Figure 11 shows the XAS results for sample #4 around the oxygen K absorption edge. Absorption at the center position was higher than that at the non-irradiated position. Thus, the EUV irradiation center was oxidized by water, which decreased the EUV reflectance.

Conclusions
A EUV irradiation tool in a hydrogen and water vapor atmosphere was developed and installed at the BL-09 long-undulator beamline of the NewSUBARU synchrotron light facility. The EUV intensity increased up to 6 W cm −2 , and the hydrogen pressure reached 70 Pa. The EUV irradiation durability of the Mo/Si multilayer under hydrogen gas and water vapor atmospheres was evaluated.
We found that the Mo/Si multilayer without the capping layer was not damaged in a hydrogen atmosphere (5 Pa) up to 80 kJ cm −2 EUV dose. However, carbon contamination was observed on the multilayer surfaces. Thus, no cleaning effect of the hydrogen plasma was observed, and no evidence of hydrogen damage was found in this experiment. The reflectance of the multilayer in the water vapor atmosphere (1 × 10 -4 Pa) was significantly reduced by oxidation of the Si surface. The Mo/Si multilayer without a capping layer was easily oxidized. Therefore, a capping layer is required to prevent surface oxidation.