Transcription

1 1

2 To EUV or not to EUV? That is the question. Do we need EUV interferometry and EUV optical testing? 17 Things you need to know about perfecting EUV optics. 2

3 The main things you need to know #1 What does it take to reach diffraction-limited performance? Rayleigh criterion seeing Maréchal criterion Strehl λ/4 P-V 3.35 nm λ/14 RMS 0.96 nm Lithographic criterion CD variation λ/50 RMS 0.27 nm Bohr Radius 0.05 nm H 3

4 Numerous factors contribute to the EUV wavefront 4

5 The main things you need to know Properties that affect the wavefront Stability of the housing Alignment of the elements Figures of the surfaces Quality/Properties of the ML coatings #2 Alignment means rigid body motion In theory, each mirror has 6 degrees of freedom. So far, there are no adaptive optics in use for EUV. 5

6 The main things you need to know #3 The lens-maker s motto If you can measure it, you can make it. #3b The lens-maker s inside joke If you can t measure it, I d be happy to make it for you. 6

7 The main things you need to know #4 The accuracy of any test must be several times better than the target Precision is related to noise sources: Repeatability is easier to achieve Accuracy comes from systematic error calibration: Being Right is much harder! Systematic errors scale with NA 2, NA 3, NA 4,... measurement difficulty scales the same, or worse. Imagine trying to measure 0.05 nm after subtracting a 10 nm systematic term (and that s only 1 part in 200!) 7

8 The main things you need to know #5 In optical testing (interferometry) λ is the measuring stick λ EUV = 13.5 nm λ Visible = nm λ Visible / λ EUV = 39.5 testing... 1 nm = λ EUV / 13.5 routine λ Visible / difficult scale to 0.1 nm and you reach the state of the art 8

9 The main things you need to know #6 In interferometry, you count fringes. Interference fringes reveal the path length difference between two waves. Each fringe = 1 λ interferogram 9

10 The main things you need to know #7 EUV interferometry is only for assembled systems EUV Lenses are made with 2 8 aspherical mirrors. Measuring a single aspherical mirror is exceptionally difficult. A 10-µm aspheric departure = 19 λvis = 741 λeuv x2! sphere A complete optical imaging system produces a spherical wavefront. EUV can only be used on assembled optical systems with multilayer-coated elements. It can only be the final step in alignment. 10

11 The main things you need to know #8 You can test a lens from either side #9 You have to measure across the field lens lens lens Field dependence Distortion 11

12 Multilayer behavior depends on λ, d, θ R λ θ d/d 0 Φ R 3š/2 š š/ wavelength [nm] angle of incidence [deg] layer thickness Si: 4.18 nm Mo: 2.69 nm θ Mo/Si optimized for λ =13.4nmat -incidence Kenneth Goldberg, Center for X-Ray Optics, LBNL,

13 High sensitivity to multilayer properties 13

14 The main things you need to know #10 EUV and visible-light reflect differently from ML mirrors. There is a phase change on reflection that must be known and compensated. θ EUV visible path length difference Δs θ incident angle A systematic wavefront difference A B B 14

15 Φ(d) and R(d): Vis. and EUV vs. ML layer thickness #11 Only EUV can see the full wavefront phase of EUV optics.

16 The main things you need to know #12 Because of the short EUV coherence length, EUV interferometers must have a common-path design i.e. Test and reference beams must travel together 16

17 The main things you need to know #13 EUV interferometers are lensless collimated lens CCD re-imaging lens... But no re-imaging lenses are good enough for EUV optics! (True for EUV and visible-light interferometers) lens lensless CCD Mathematically propagate the field back to the exit pupil. 17

18 Interferometers for all occasions Kenneth Goldberg, Center for X-Ray Optics, LBNL,

19 EUV Interferometers used at LBNL Kenneth Goldberg, Center for X-Ray Optics, LBNL,

20 EUV Interferometers Everything starts with coherent, spherical-wave illumination The quality of the diffracted wave is very important Kenneth Goldberg, Center for X-Ray Optics, LBNL,

21 the illuminated MET pupil EUV Light transmitted light Kenneth Goldberg,, SPIE 2005,

22 EUV Interferometers Kenneth Goldberg, Center for X-Ray Optics, LBNL, Efficiency: HIGH Dynamic Range: Limited A great first test. Qualitative feedback Helps to identify the image plane. Self-calibrating.

23 EUV Light

24 EUV Interferometers Kenneth Goldberg, Center for X-Ray Optics, LBNL, Efficiency: HIGH Dynamic Range: Variable Very easy to align. Analysis is challenging Works up to ~0.3 NA Susceptible to higher-ordered aberrations

25 shearing interferogram EUV Light efficient measurement method Kenneth Goldberg,, SPIE 2005,

26 EUV Interferometers CGLSI cross-grating lateral shearing interferometer Efficiency: HIGH Dynamic Range: Variable Analysis is challenging Can go beyond ~0.3 NA Kenneth Goldberg, Center for X-Ray Optics, LBNL,

27 EUV Wavefront Sensor Efficiency: HIGH Dynamic Range: Variable high coherence Limited sensitivity not required Low-NA only Kenneth Goldberg, Center for X-Ray Optics, LBNL,

28 0.008-NA K-B system EUV Light

29 EUV Interferometers Efficiency: VERY LOW Dynamic Range: LOW Susceptible to defects/contamination Limited accuracy Kenneth Goldberg, Center for X-Ray Optics, LBNL,

30 Fresnel zoneplate

31 EUV Interferometers Efficiency: Medium Dynamic Range: LOW Ultra-high accuracy reference waves Requires < 30-nm pinholes for 0.3-NA Kenneth Goldberg, Center for X-Ray Optics, LBNL,

32 PS/PDI interferogram EUV Light ultra-high accuracy Kenneth Goldberg,, SPIE 2005,

33 EUV Wavefront Sensor TIS transmission image sensor Efficiency: LOW Dynamic Range: LOW Uses mask illumination. Feedback for low-frequency aberrations Kenneth Goldberg, Center for X-Ray Optics, LBNL,

34 A long track record of EUV Interferometry, alignment optimization and imaging at LBNL (since 93) Berkeley 10x 10xI 10xA 10xB 10xA 10xB2 10xB2 F2X ETS Set-1 ETS Set-2 MET 2-mirror, 10x Schwarzschild objectives NA 0.08 ƒ/ 6.3 higher resolution 4-mirror, 4x ETS projection optics NA = 0.1 ƒ/ mirror, 5x MET optic NA = 0.3 ƒ/ 1.67 time higher quality 34

35 ETS Projection Optic: off-axis, large field Work sponsored by the EUV LLC ~1.1-m mask-to-wafer M2 M4 M3 M1 35

36 36

37 37

38

39 MET at-wavelength interferometry and alignment MET Micro-Exposure Tool Wavefront measurement during alignment central field point astig coma sph ab trifoil h-o s nm 0.06 nm 0.04 nm 0.14 nm 0.37 nm RMS 0.55 nm λ/24.5 EUV interferometry & alignment sets astigmatism, coma, spherical aberration arbitrarily small. 39

40 What is the key to achieving high accuracy? Consulting the Oracle at Delphi

41 #14 γνῶθι σεαυτόν = Know Thyself Modern Translation Characterize the heck out of your interferometer x Pinhole-diffracted beams x Lensless Geometry x Cross-calibration Apollo

42 Developing state-of-the art pinholes for spherical reference-wave accuracy object pinholes image pinholes TEMPEST-3D Modeling vector E-M field simulations 100-nm Ni 150-nm Ni Nanofabrication (Nanowriter) object pinhole TEM SEM 100 nm image pinhole 25 nm Intensity Intensity Pinhole-array diffraction diffraction angle diffraction angle 42

43 The lensless geometry creates measurable systematic errors that are removed through calibration

44 grating null-test interferogram system calibration for high accuracy Kenneth Goldberg,, SPIE 2005,

45 two-pinhole null-test interferogram system calibration for high accuracy Kenneth Goldberg,, SPIE 2005,

46 What is the accuracy limit? Consulting the Oracle at Delphi

47 Sorry, The Oracle only allows one question per customer. Next! Apollo

48 48

49 Directly compare EUV and visible-light measurements... 49

50 Cross-calibrate the different techniques, wavelengths visible-light EUV PS/PDI EUV shearing Excellent work in this area by the EUVA team members! Murakami, Sugisaki, Ouchi, Otaki, Liu, Zhu, Hasegawa, Kato, Ishii, Saito, Niibe, Takeda, and others. EUVA, Univ. Hyogo, (Canon, Nikon) 50

51 Intercomparison Results, and a ray of hope #15 How good are the inter-comparisons? We always find: nm RMS disagreement Astigmatism is the main difference. Astigmatism Most sensitive to system misalignment. Hardest aberration to measure interferometrically * Easiest aberration to detect in printing. 51

52 But wait... #16 What about distortion? MISALIGNMENT SPACE Distortion Aberrations M 1 M 2 M 3 M 4... alignment 52

53 Inaccurate interferometry cost NASA $Billions Before After Hubble Space Telescope Kenneth Goldberg,, SPIE 2005, Hubble repair, 1993 COSTAR optic installed 53

54 Will we learn from history and invest in interferometry? Without EUVI? With EUVI? EUV Stepper Kenneth Goldberg, Performing EUV Interferometry? 54

55 #17 You better align it right, because at 10 7 Torr No one can hear you scream Kenneth Goldberg, 55