3.1.1 ASML’s technical route

In the EUVL system of ASML, the reticle is protected from being contaminated by covering a pellicle on it that has a higher reflectivity to DUV radiation than the BB, resulting in a greater impact on CD errors. To reduce its influence, a free-standing transmissive SPF, which is called a dynamic gas lock membrane (DGLm), is adopted by ASML in NXE:3400. Figure 7 depicts the details of the DGLm. The DGLm is mainly composed of four different layers^[ Reference van de Kerkhof, Jasper, Levasier, Peeters, van Es, Bosker, Zdravkov, Lenderink, Evangelista, Broman, Bilski and Last ^{32 ]}, of which DUV and IR active layers are the core of the whole membrane to filter out the OoB radiations. Different from most transmissive SPFs’ locations, the DGLm is located between the projection optics and the wafer. With this design, the DGLm acts as a physical barrier to prevent the contamination of projection optics caused by resistance outgassing^[ Reference Es, van de Kerkhof, Minnaert, Fisser, Klerk, Smits, Moors, Verhoeven, Levasier, Peeters, Pieters and Meiling ^{33 ]}. The DGLm is a free-standing device, and thus it can be replaced once damaged. What is more, as the membrane is placed further away from the IF, it is less sensitive to non-uniform transmission and particles^[ Reference van de Kerkhof, Jasper, Levasier, Peeters, van Es, Bosker, Zdravkov, Lenderink, Evangelista, Broman, Bilski and Last ^{32 ]}.

Figure 7 DGL's location and its influence on outgassing suppression. Reprinted from Ref. [Reference Fomenkov31].

With the design of the DGLm, more than 99% DUV and nearly 80% IR radiations can be suppressed^[ Reference van de Kerkhof, Jasper, Levasier, Peeters, van Es, Bosker, Zdravkov, Lenderink, Evangelista, Broman, Bilski and Last ^{32 ]}, as depicted in Figure 8. Since the total thickness of the membrane is below 50 nm and the absorption coefficient of the material is low, the EUV transmittance can reach 85%^[ Reference van de Kerkhof, Jasper, Levasier, Peeters, van Es, Bosker, Zdravkov, Lenderink, Evangelista, Broman, Bilski and Last ^{32 ]}. Without the DGLm, the CDs at the corners and edges are all less than 0.1 nm, which has already met the requirements of the N5 node, while with the DGLm, the CD reaches approximately 0.04 nm^[ Reference van de Kerkhof, Liu, Meeuwissen, Zhang, de Kruif, Davydova, Schiffelers, Waelisch, van Setten, Varenkamp, Ricken, de Winter, McNamara, Bayraktar, Felix and Lio ^{16 ]}, which represents an improvement of DUV suppression. It should be noted that the excellent performance without the DGLm is based on the suitable solutions of other EUVL components, such as the specific designs of the BB and the sensitive enough resistance, but the DGLm can improve the DUV suppression under conditions that are not good enough, such as with less sensitive resistance. However, a potential problem is that the OoB light is not filtered until it enters the wafer stage; hence, the mirrors in the optical path will be damaged by the high-power radiation.

Figure 8 OoB suppression performance with DGLm: (a) complete suppression of DUV radiation (<0.1% transmitted) as measured by PTB; (b) 78% IR suppression as measured off-line by Fourier transform infrared spectroscopy. Adapted from Ref. [Reference Fomenkov31].

4.1.1 Bragg diffraction

The EUV collector is also known as a Bragg reflector^[ Reference van den Boogaard, van Goor, Louis and Bijkerk ^{39 ]} for the design of coating Mo/Si multilayers on the surface of the collector to maximize the EUV reflectivity. The Bragg reflector is an optical device that uses the constructive interference of reflected light at different interfaces to enhance the reflection of specific wavelengths^[ Reference Kittel and McEuen ^{40 ,} Reference Haase ^{41 ]}. According to the Bragg’s law, when the optical path difference of the reflected light at two adjacent interfaces is half a wavelength, the reflected light at the interface will incur constructive interference and get a strong reflection^[ Reference Migura ^{42 ]}. The formula formation of Bragg’s law in the case of periodic multilayers^[ Reference Guen, André, Wu, Ilakovac, Delmotte, de Rossi, Bridou, Meltchakov, Giglia and Nannarone ^{43 ,} Reference Moseley ^{44 ]} is as follows:

(3) $$\begin{align}2d\sin \theta = n\lambda,\\[-20pt] \nonumber\end{align}$$

where $d$ is the multilayer period, $\theta$ is the incident angle, $n$ is diffraction order and $\lambda$ is the incident wavelength. The reflectivity is determined by the number of layers and the refractive index difference between materials.

4.1.2 Grating diffraction

Grating diffraction is generally based on the Fraunhofer multi-slit diffraction effect. Once the optical path difference from the two adjacent slits to the interference point is an integral multiple of the wavelength of the incident light, the two beams have the same phase and the interference will be enhanced^[ Reference Kittel and McEuen ^{40 ]}. The grating equation summarizes the phenomenon:

(4) $$\begin{align}p\left(\sin \delta +\sin \theta \right)= m\lambda, \end{align}$$

where $p$ is the grating period, $\delta$ is the diffraction angle, $\theta$ is the incident angle, $m$ is the diffraction order and $\lambda$ is the wavelength of the incident light. With proper grating designs according to the grating equation, the needed radiation can be concentrated around one specific diffraction order, and most of the OoB radiation is diffracted to other orders^[ Reference Feigl, Perske, Pauer, Fiedler, Zeitner, Leitel, Eckstein, Schleicher, Schröder, Trost, Risse, Steinkopf, Scholze and Laubis ^{36 ,} Reference Medvedev, van den Boogaard, van der Meer, Yakshin, Louis, Krivtsun and Bijkerk ^{45 ,} Reference Miles, McCoy, McEntaffer, Eichfeld, Lavallee, Labella, Drawl, Liu, DeRoo and Steiner ^{46 ]}.

Figure 10 shows the schematic of the OoB suppression design with the built-in SPF system, in which the blue line is the beam path of EUV radiation, while the red line is OoB radiation. After the laser radiations hit the Sn droplets, the EUV radiation is generated and the OoB radiations travel to the collector firstly. With the effect of the grating coated with Mo/Si multilayers, the EUV radiation concentrated at the zero order is reflected to the IF for the subsequent propagation, while the IR radiation is diffracted to higher diffraction orders and stopped by the beam stop.

Figure 10 Schematic of the IR suppression design with the collector integrated with the rectangular substrate grating. Adapted from Ref. [Reference Mizoguchi, Nakarai, Abe, Nowak, Kawasuji, Tanaka, Watanabe, Hori, Kodama, Shiraishi, Yanagida, Soumagne, Yamada, Yamazaki, Okazaki and Saitou47].

4.3.1 Design principles

Assume that the OoB radiation that needs to be suppressed is 10.6 μm and the grating is in the shape of a 1D rectangle. As Figure 14 shows, the grating is characterized by the period $p$ , width $d$ and depth $h$ of the grooves. When the 13.5 nm EUV radiation combined with the 10.6 μm is incident onto the grating, if the diffraction order $m=0$ , both radiations’ reflection angles are the same, while with higher diffraction orders, the characteristic values of the diffraction angles differ by about three orders of magnitude^[ Reference Medvedev, van den Boogaard, van der Meer, Yakshin, Louis, Krivtsun and Bijkerk ^{45 ]}. Therefore, the 10.6 μm radiation at m = 0 is what we need to suppress. Assuming that the grating is illuminated by the normal-incidence plane wave, the following equation shows the mth-order DE of the 1D rectangular phase grating:

(5) $$\begin{align}{R}_n={R}_{\mathrm{tot}}{\left|\sin \mathrm{c}\left( m\pi \right)+A\Gamma \sin \mathrm{c}\left( m\pi \right)\right|}^2, \\[-24pt] \nonumber\end{align}$$

where ${R}_{\mathrm{tot}}$ is the total reflected intensity, $\Gamma =\frac{d}{p}$ and $A=\exp \left(\frac{i\cdot 4\pi h}{\lambda}\right)-1$ , and

(6) $$\begin{align}{R}_0={R}_{\mathrm{tot}}\left\{1+2\Gamma \left(\Gamma -1\right)\left[1-\cos \left(\frac{4\pi h}{\lambda}\right)\right]\right\}\end{align}$$

is the particular case of zero diffraction order, which both can be derived from Equation (2) ^[ Reference Medvedev, van den Boogaard, van der Meer, Yakshin, Louis, Krivtsun and Bijkerk ^{45 ]}. To suppress the 10.6 μm IR radiation at zero order, Equation (5) should be equal to zero. It can be calculated that the fill factor $\Gamma =0.5$ and the minimum depth of the grooves is $h=\frac{\lambda }{4}=\sim 2.65$ μm. The period $p$ is more difficult to determine because it is a trade-off between IR reflectivity at the zero order ( $\sim p/\lambda$ ) and the angular separation of diffraction ( $\sim \arcsin \left(\lambda /p\right)$ )^[ Reference Medvedev, van den Boogaard, van der Meer, Yakshin, Louis, Krivtsun and Bijkerk ^{45 ,} Reference Johnson ^{67 ]}. What is more, the limited spatial and temporal coherence of the radiation should also be considered^[ Reference Medvedev, van den Boogaard, van der Meer, Yakshin, Louis, Krivtsun and Bijkerk ^{45 ,} Reference Huang, de Boer, Barreaux, van der Meer, Louis and Bijkerk ^{60 ]}.

Figure 14 Design of the rectangular substrate grating: (a) schematic of 1D rectangular substrate grating; (b) schematic of 2D rectangular substrate grating by IOF. Adapted from Refs. [Reference Huang, Medvedev, van de Kruijs, Yakshin, Louis and Bijkerk53,Reference Feigl, Perske, Pauer, Fiedler, Zeitner, Leitel, Eckstein, Schleicher, Schröder, Trost, Risse, Steinkopf, Laubis and Scholze62].

To suppress radiations of two wavelengths, in particular 10.6 μm and 1064 nm, this method can be used to determine the dual-layer grating. Since the difference between the two wavelengths is 10 times, the parameters of the grating are also 10 times different. The Fraunhofer Institute for Applied Optics and Precision Engineering (IOF)^[ Reference Feigl, Perske, Pauer, Fiedler, Zeitner, Leitel, Eckstein, Schleicher, Schröder, Trost, Risse, Steinkopf, Scholze and Laubis ^{36 ,} Reference Feigl, Perske, Pauer, Fiedler, Zeitner, Leitel, Eckstein, Schleicher, Schröder, Trost, Risse, Steinkopf, Laubis and Scholze ^{62 ]} designed and fabricated a dual-wavelength rectangular grating with optimized geometric parameters, as shown in Figure 14. At the IF, the measured reflectivity of 10.6 μm is 0.32%, the reflectivity of 1064 nm is only 0.08% and the average reflectivity of the EUV radiation at 13.5 nm is 64% which represents only 4.5% loss compared to the unstructured multilayers.

4.3.2 Fabrication process

At present, there are two main processes to fabricate the substrate gratings deposited with multilayers. The main difference between them is the preparation of the substrate grating, which involves the diamond turning process and ion beam etching.

The process of manufacturing the substrate grating used by Rigaku is the diamond turning^[ Reference Platonov, Kriese, Crucet, Li, Martynov, Jiang, Rodriguez, Mueller, Daniel, Khatri, Magruder, Grantham, Tarrio and Lucatorto ^{15 ]}. Firstly, Ni is injected into the substrate made of Al alloy and the surface of it is polished. Then the grating grooves are made on the Ni surface by diamond turning, which will result in turning patterns, as shown in Figure 15(a), and high spatial frequency roughness (HSFR) on the cut surface. Therefore, a glass smoothing layer is then needed to smooth the substrate grating. The other method, the ion beam etching process, is used by IOF^[ Reference Feigl, Perske, Pauer, Fiedler, Zeitner, Leitel, Eckstein, Schleicher, Schröder, Trost, Risse, Steinkopf, Scholze and Laubis ^{36 ,} Reference Feigl, Perske, Pauer, Fiedler, Zeitner, Leitel, Eckstein, Schleicher, Schröder, Trost, Risse, Steinkopf, Laubis and Scholze ^{62 ]} to prepare a dual-layer rectangular grating on the collector’s substrate, which is made of AlSi alloy. With the method, the substrate is firstly diamond turned and polished to reduce the HSFR to less than 0.2 nm rms, and then, the dual-layer grating is directly constructed into the NiP layer of the AlSi substrate by ion beam etching.

Figure 15 (a) AFM image of diamond-turned patterns and (b) the mechanical polished surface of (a). Reprinted from Ref. [Reference Trost, Schroder, Duparre, Risse, Feigl, Zeitner and Tunnermann49].

When manufacturing the substrate grating, the HSFR is a critical factor to achieve a smooth enough surface. To reduce the HSFR, some methods are proposed. Mechanical polishing^[ Reference Trost, Schroder, Duparre, Risse, Feigl, Zeitner and Tunnermann ^{49 ]} is the simplest smoothing method, which is mainly used to remove typical diamond turning structures. As shown in the Figure 15, it reduced the HSFR by more than 10 times. Ion beam polishing technology^[ Reference Soer, Gawlitza, van Herpen, Jak, Braun, Muys and Banine ^{68 ,} Reference Spiller, Baker, Mirkarimi, Sperry, Gullikson and Stearns ^{69 ]}, which is commonly used on collectors without gratings, uses the neutral ion beam to bombard the workpiece, removing atoms or molecules in a certain area of the surface, and achieves ultra-smooth polishing. The method of adding a smoothing layer^[ Reference Zhang ^{70 ]} refers to manufacturing a film on the substrate with a smaller HSFR to smooth the roughness, and the performance can be seen in Figures 16(a) and 16(b). Stock et al. ^[ Reference Stock, Hamelmann, Kleineberg, Menke, Schmiedeskamp, Osterried, Heidemann and Heinzmann ^{71 ]} used the electron beam evaporation method to deposit a single layer of carbon on the substrate of BK7 and Zerodur to smooth the roughness. Salmassi et al. ^[ Reference Salmassi, Naulleau and Gullikson ^{72 ,} Reference Salmassi, Anderson, Gullikson and Naulleau ^{73 ]} proposed a spin-on-glass resistance process based on HSQ to smooth a substrate with diamond-turned structures and presented a roughness reduction from 3.7 nm to 0.39 nm rms on the Al substrate. Rigaku^[ Reference Kriese, Platonova, Ehlersa, Jianga, Rodrigueza, Muellerb, Danielb, Khatrib, Magruderb, Granthamc, Tarrioc and Lucatortoc ^{8 ]} used a glassy smoothing layer on the substrate grating to reduce the HSFR of the surface, achieving the results presented in Figures 16(c) and 16(d), which is sufficient for high performance. Ulmer et al. ^[ Reference Ulmer, Dugard, Quispe, Buchholz, Stagon, Chung, Cao, Kritikos, Guerra, Stahl, Shiri, Vaidyanathan, den Herder, Nakazawa and Nikzad ^{74 ]} designed a mirror with a diameter of 1 m and smoothed the NiP substrate by covering the CN _x layer.

Figure 16 HSFR results (AFM) of adding a smoothing layer by Rigaku: (a) diamond-turned surface sample; (b) smoothed diamond-turned surface sample; (c) 0.14–0.29 nm rms over 2.2 μm scans; (d) 0.29–0.39 nm rms over 8.7 μm scans of the grating surface. Adapted from Refs. [Reference Kriese, Platonova, Ehlersa, Jianga, Rodrigueza, Muellerb, Danielb, Khatrib, Magruderb, Granthamc, Tarrioc and Lucatortoc8,Reference Platonov, Kriese, Crucet, Li, Martynov, Jiang, Rodriguez, Mueller, Daniel, Khatri, Magruder, Grantham, Tarrio and Lucatorto15].

4.3.3 Metrology

During the preparation of the substrate grating, some measurements are needed to measure the machining accuracy and the overall device performance. The three main aspects to be measured are the roughness, reflectivity for different wavelengths and uniformity of reflectivity.

Roughness. The thickness of the multilayers deposited on the substrate grating is at the level of nanometres, and thus the morphology and roughness of the substrate will be translated to the top layers^[ Reference Naulleau, Liddle, Anderson, Gullikson, Mirkarimi, Salmassi and Spiller ^{55 ,} Reference Zhang ^{70 ,} Reference Louis, Yakshin, Tsarfati and Bijkerk ^{75 ]}. It is clear that the prerequisite for reducing the roughness of the multilayers is to reduce the roughness of the substrate grating. The evaluation indicators for roughness include the average roughness, root mean square (RMS) roughness, peak-to-valley (PV) value, power spectral density (PSD), etc. The first three are simple statistical parameters and only the vertical component of surface roughness is calculated^[ Reference Zhang ^{70 ]}. The PSD measures the roughness from the perspective of Fourier spectra, thereby obtaining the spatial frequency distribution of the surface error, which means it also contains the transverse and longitudinal information of the surface^[ Reference Chen ^{76 ,} Reference Li ^{77 ]}.

Roughness in different spatial frequency ranges should be measured by different methods, as shown in Figure 17. According to the scattering theory, the scattering caused by roughness in different spatial frequencies can be generally divided into the following three types^[Reference Louis, Yakshin, Tsarfati and Bijkerk^75, Reference Yang, Zheng and Chen^78–Reference Schuster, Merkel, Metalidis and Kierey^80]:

1. (1) minimum angle scattering caused by low spatial frequency roughness (LSFR), which introduces basic aberration into the system, will blur the image and reduce the resolution of the optical system;

2. (2) small-angle scattering caused by medium spatial frequency roughness (MSFR), also known as stray light, will reduce the contrast of spatial imaging;

3. (3) large-angle scattering caused by HSFR will reduce the light flux reaching the image plane, thus reducing the energy transmission efficiency of the optical system.

Figure 17 Different measurements for roughness at different spatial frequencies. Adapted from Ref. [Reference Zhang70].

Taking Rigaku’s work as a reference^[ Reference Kriese, Platonova, Ehlersa, Jianga, Rodrigueza, Muellerb, Danielb, Khatrib, Magruderb, Granthamc, Tarrioc and Lucatortoc ^{8 ]}, the different spatial frequencies of roughness, corresponding to the above influences, are defined as follows: LSFR, more than or equal to 1 mm; MSFR, 1 mm–10 μm; HSFR, less than or equal to 10 μm.

The HSFR is a critical indicator to be measured because among the roughness in all frequency bands, the roughness with the spatial wavelength equivalent to the incident light wavelength has the strongest scattering ability^[ Reference Zhang ^{70 ]}. The HSFR of the grating surface used for the collector needs to be less than 0.3 nm rms^[ Reference Trost, Schroder, Duparre, Risse, Feigl, Zeitner and Tunnermann ^{49 ,} Reference Feigl, Perske, Pauer, Fiedler, Zeitner, Leitel, Eckstein, Schleicher, Schröder, Trost, Risse, Steinkopf, Laubis and Scholze ^{62 ]}, which is also the main challenge of the fabrication. At present, the main and most direct method used in HSFR for measuring surfaces with less than or equal to 0.1 nm rms is atomic force microscope (AFM) measurement, which determines the distance between the probe and the detection surface by the contact force. Its detection area is usually less than tens of micrometres, the longitudinal resolution is approximately 0.01 nm^[ Reference Li ^{77 ]} and the transverse resolution can generally be up to 0.1 nm^[ Reference Zhang ^{70 ,} Reference Chen ^{76 ]}, which depends on the radius of curvature of the probe and the depth from the top of the probe to the detection surface^[ Reference Zhang ^{70 ,} Reference Eastman and Bausmeister ^{81 ]}. In the IOF’s experiments on the dual-layer rectangular grating^[ Reference Feigl, Perske, Pauer, Fiedler, Zeitner, Leitel, Eckstein, Schleicher, Schröder, Trost, Risse, Steinkopf, Scholze and Laubis ^{36 ,} Reference Feigl, Perske, Pauer and Fiedler ^{82 ]}, with a scanning range of 10 μm $\times$ 10 μm, the roughnesses of the four facets of the grating were all below 0.25 nm rms. It should be noted that if PSD is used for characterization, it is necessary to measure the same surface at different locations and meet the sampling theorem.

The LSFR is primarily measured by optical profilometry, including white light interferometry (WLI), double-beam interferometry, etc., among which WLI is the main method currently used. WLI is a non-contact metrology, and has the advantages of a large measuring range, high vertical accuracy and fast measuring speed. Based on the principle of optical interference, WLI can analyse the relative height variation of the sample surface according to the contrast and position of the light interference fringe, thus generating a 3D image and obtaining the 3D morphology and surface roughness of the measured sample. However, due to the influence of dispersion, its lateral accuracy can only reach the scale of hundreds of nanometres^[ Reference Chen ^{76 ,} Reference Li ^{77 ]}. The IOF^[ Reference Feigl, Perske, Pauer, Fiedler, Zeitner, Leitel, Eckstein, Schleicher, Schröder, Trost, Risse, Steinkopf, Scholze and Laubis ^{36 ]} used the WLI to characterize the profile of its dual-layer substrate grating, as Figure 18 shows, and controlled the maximum error of the grating period, fill factor and groove depth within less than 1%. Rigaku^[ Reference Kriese, Platonova, Ehlersa, Jianga, Rodrigueza, Muellerb, Danielb, Khatrib, Magruderb, Granthamc, Tarrioc and Lucatortoc ^{8 ]} also used an interference microscope to measure the roughness of the substrate grating, and achieved an LSFR of 0.577 μm rms.

Figure 18 WLI analysis of a dual-layer rectangular substrate grating structure by the IOF. Reprinted from Ref. [Reference Feigl, Perske, Pauer, Fiedler, Zeitner, Leitel, Eckstein, Schleicher, Schröder, Trost, Risse, Steinkopf, Laubis and Scholze62].

The MSFR can be measured by laser speckle interferometry and light scattering measurement, both of which are indirect measurement. The former is primarily used in the detection range where the roughness is greater than the illumination wavelength^[ Reference Zhang ^{70 ]}, and has the advantages of simple operation and a large measurement range. The brightness, shape and contrast of the speckles generated by the coherent wave can effectively reflect the distribution of surface roughness. Light scattering measurement mainly includes the angle resolved scattering (ARS) measurement and total integrated scattering (TIS) measurement^[ Reference Mazule, Liukaityte, Eckardt, Melninkaitis, Balachninaite and Sirutkaitis ^{83 ]}. ARS determines the roughness according to the distribution of the scattered light intensity in the plane^[ Reference Böhm, Jech and Vellekoop ^{84 ]}, while TIS determines the roughness according to the ratio of the light intensity in the hemisphere to the light intensity reflected on the surface of the sample^[ Reference Wei ^{85 ]}. Compared with TIS, ARS is more complex and expensive, but it can correctly measure the spatial distribution of scattered light and obtain more roughness information^[ Reference Mazule, Liukaityte, Eckardt, Melninkaitis, Balachninaite and Sirutkaitis ^{83 ]}. Trost et al. ^[ Reference Trost, Schroder, Duparre, Risse, Feigl, Zeitner and Tunnermann ^{49 ]} used ALBATROSS (Figure 19), developed by the IOF, to measure the roughness of the multilayer grating. Hilpert et al. ^[ Reference Hilpert, Hartung, von Lukowicz, Herffurth and Heidler ^{86 ]} measured the roughness of the reflector in the spaceborne optical system by ARS and obtained a roughness value of 0.1 nm rms. Herrero et al. ^[ Reference Herrero, Mentzel, Soltwisch, Jaroslawzew, Laubis and Scholze ^{87 ]} of Physikalisch-Technische Bundesanstalt (PTB) exposed a new tool (EUV-ARS) based on the grazing incidence small-angle X-ray scattering instrument (Figure 20), which allows one to study the structural surface by analysing the scattered light and fluorescence signals, and to achieve the roughness measurement at the nanoscale.

Figure 19 ARS instrument ALBATROSS for scattering measurements in the UV-VIS-IR range. Components include laser sources (1), mechanical chopper for lock-in amplification (2), attenuation filters (3), beam preparation optics (4), polarizer (5), sample (6) and detector (7). Adapted from Ref. [Reference Feigl, Perske, Pauer, Fiedler, Zeitner, Leitel, Eckstein, Schleicher, Schröder, Trost, Risse, Steinkopf, Laubis and Scholze62].

Figure 20 EUV-ARS for the characterization of nanometre structures exposed by PTB. Reprinted from Ref. [Reference Herrero, Mentzel, Soltwisch, Jaroslawzew, Laubis and Scholze87].

OoB suppression and EUV reflectivity. The collector integrated with rectangular gratings is designed to suppress 10.6 μm and 1064 nm IR radiation without reducing EUV radiation, so it is necessary to detect the intensity of EUV and IR radiations at the focal plane. Compared with the collector without the grating structure, the intensity of IR radiation at the IF generally needs to be suppressed by more than 99%^[ Reference Bakshi ^{12 ]}, while the minimum loss of EUV radiation achieved nowadays is approximately 4.81%^[ Reference Sun, Jin, Yu, Guo, Yao, Deng and Li ^{88 ]}.

For the EUV detection, the equipment used includes the EUV reflectometry at BESSY II developed by PTB in Germany^[ Reference Scholze, Laubis, Dersch, Pomplun, Burgerc and Schmidt ^{89 –} Reference Laubis, Barboutis, Buchholz, Fischer, Haase, Knorr, Mentzel, Puls, Schönstedt, Sintschuk, Soltwisch, Stadelhoff and Scholze ^{92 ]} (Figure 21), the EUV reflectometry by the National Institute of Standards and Technology (NIST) in the USA^[ Reference Grantham, Tarrio, Lucatorto, Kriese, Platonov, Rodriguez and Jiang ^{93 ]} and the system MERLIN of the IOF^[ Reference Schröder, Herffurth, Trost and Duparré ^{94 ]}.

Figure 21 Mechanics of the EUV reflectometer by PTB. Reprinted from Ref. [Reference Scholze, Laubis, Dersch, Pomplun, Burgerc and Schmidt89].

For the detection of IR suppression, ALBATROSS^[ Reference Schröder, Herffurth, Blaschke and Duparré ^{95 ]} can also be used. In fact, ALBATROSS is a device that can measure the reflected light, transmitted light and scattered light distribution with a high dynamic range, high sensitivity and low noise. What is more, the measurement device can also be custom-built for the actual measuring needs, as shown in Figure 22. The laser firstly passes through the beam expander, chopper, etc., and is then reflected by the virtual light source sphere before being incident on the full-size collector. After being deflected by a folded plane mirror, it is focused on the HgCdTe detector for the measurement^[ Reference Kriese, Platonova, Ehlersa, Jianga, Rodrigueza, Muellerb, Danielb, Khatrib, Magruderb, Granthamc, Tarrioc and Lucatortoc ^{8 ]}.

Figure 22 Schematic of the IR suppression test stand. Reprinted from Ref. [Reference Kriese, Platonova, Ehlersa, Jianga, Rodrigueza, Muellerb, Danielb, Khatrib, Magruderb, Granthamc, Tarrioc and Lucatortoc8].

Reflectivity uniformity. The final step in the preparation of the substrate grating is to deposit Mo/Si multilayers, which aims to improve the EUV reflection performance, so measuring the EUV reflectivity uniformity is very important. The EUV reflectometry mentioned above is used to symmetrically select multiple positions on the collector’s surface along the x- and y-axes and measure their reflectivity for comparison. The IOF^[ Reference Feigl, Perske, Pauer, Fiedler, Zeitner, Leitel, Eckstein, Schleicher, Schröder, Trost, Risse, Steinkopf, Scholze and Laubis ^{36 ]} used the system of PTB to compare the average EUV reflectivity of the collector integrated with the dual-layer grating, and the change of EUV reflectivity is approximately 0.5%. Rigaku^[ Reference Kriese, Platonova, Ehlersa, Jianga, Rodrigueza, Muellerb, Danielb, Khatrib, Magruderb, Granthamc, Tarrioc and Lucatortoc ^{8 ]} used the same system as the IOF to measure the peak EUV reflectivity of the collector surface, and showed a variation of less than 0.0025 nm at the opposite positions, referring to 0.02% PV variation.

4.3.4 Gratings with other shapes

According to the existing technical schemes, the collector substrate of the rectangular grating structure cannot effectively filter the DUV radiation, so it is proposed to use a blazed grating to replace the rectangular grating since the structure was theoretically demonstrated to be effective at filtering out full-OoB radiations by van den Boogaard et al. ^[ Reference van den Boogaard, Louis, van Goor and Bijkerk ^{20 ]} when studying the multilayer grating.

Johnson^[ Reference Johnson ^{67 ,} Reference Johnson and Vac ^{96 ]} designed a collector integrated with the blazed grating structure that aims to divert the EUV radiation into the IF rather than diffracting IR radiation out of the IF, as shown in Figure 23(a), and two fabrication methods are given, which are the conformal-multilayer grating and patterned-multilayer grating. What is more, a power recycling component was designed surrounding the IF aperture to retroreflect IR radiation for a higher source conversion efficiency (CE). Following that, Hassanein et al. ^[ Reference Hassanein, Sizyuk, Sizyuk and Johnson ^{97 ]} continued the design of the blazed grating and furthered the idea of power recycling, which showed a potential 60% gain of EUV-CE with the preliminary plasma simulation, as shown in Figure 23(b). However, these relevant researches based on the substrate blazed grating are mainly theoretical analyses, and no experiments have been carried out to establish the feasibility of this scheme.

Figure 23 (a) Schematic of the collector integrated with the blazed substrate grating and (b) schematic of the collector with power recycling mirrors. Reprinted from Refs. [Reference Johnson and Vac96,Reference Hassanein, Sizyuk, Sizyuk and Johnson97].

Theoretically, the blazed grating can indeed suppress the radiation of the IR and UV bands, but the problem lies in the limitations of the manufacturing process. For example, it is difficult to control the blazed angle of the multilayers within ${3}^{\circ }$ ^[ Reference Zhang ^{35 ]}. Therefore, its actual effect is far from reaching the theoretical effect, so it is still at the theoretical stage.