Infrared transmission spectroscopy of direct bonded Si compound optics

Immersion gratings are made from monolithic hockey-puck sized Silicon boules. The massive size presents challenges when using conventional semiconductor equipment designed for DVD-sized Si wafers. An alternative to monolithic samples is to bond Si wafers directly to larger substrates. This presentation gives the history of Si direct bonding, and concludes with the introduction of a new optical technique to evaluate the quality of the bond interface.

Transcript

1. bonded Si optics
   gully
   monday, november 11, 2013
   

   View Slide

2. Key ideas
   History of Si bonding Literature
   How to directly bond Si-Si
   How to measure bonded Si optics
   – IR imaging
   – Cary5000
   – Matrix method
   

   View Slide

3. Shimbo+ 1986
   Stengl+ 1988
   Lehmann+ 1989
   Mitani+ 1990
   Mitani & Göselle
   1992
   Feijoó+ 1994
   Himi+ 1994
   Reiche+ 1995
   Göselle+ 1995
   Göselle+ 1995
   Reading the literature offers guidance on how to bond silicon pucks together.
   

   View Slide

4. Takagi+ 1996
   Reiche+ 1996
   Takagi+ 1998
   Han+ 2000
   Gracias+ 2001
   Greco+ 2001
   Litton+ 2001
   Haisma+ 2007
   Göselle & Tong
   1992
   Reading the literature offers guidance on how to bond silicon pucks together.
   

   View Slide

5. Bäcklund, Ljungberg, Söderbärg
   1992
   Journal of Micromechanical Engineering
   A suggested mechanism for silicon direct bonding from studying hydrophilic and
   hydrophobic surfaces
   They measure the bond energy in Joules per square meter
   (J/m2).
   The bond energy measurement is from “crack
   propagation”, W.P Maszara et al. 1998, Journal of
   Applied Physics.
   They find bond strength is:
   greater for hydrophilic Si for annealing 21 < T (°C) < 400
   greater for hydrophobic Si for annealing T (°C) > 400
   Hydrophilic bonds at room temperature are “reversible”
   The authors explain the physical processes going on.
   

   View Slide

6. Mitani and Gösele
   1992
   Journal of Electronic Materials
   Wafer BondingTechnology for Silicon-on-Insulator Applications: A Review
   An unconventional use of wafer bonding is the reversible room temperature bonding
   for the protection of polished wafer surfaces against organic and particle
   contamination.
   Annealing at T (°C) > 800.
   

   View Slide

7. For effective particle prevention, the initial bonding has to be performed in a clean room of
   class one or better quality right after chemical cleaning. Trapping of air can be
   prevented by propagating the bonding area radially from just one chosen contact
   point…
   …the most problems in wafer bonding is … particle-free bonding.
   This set-up allows wafer bonding without bubbles in a normal non-cleanroom laboratory
   environment by removing local particles between two silicon wafers, facing each other
   at a distance of about 1 mm, by deionized (DI) water flushing. The procedure of wafer
   bonding with using this micro-cleanroom setup completely eliminates particles and
   maintains particle-free surfaces once the wafers are cleaned and free from particles. So
   far, particle-free wafer bonding with up to 8 inch wafers has been realized in this
   micro-cleanroom setup.
   Mitani and Gösele
   1992
   cont…
   

   View Slide

8. 1 erg/cm2 = 0.001 J/m2
   Mitani and Gösele
   1992
   cont…
   

   View Slide

9. The bubbles tend to disappear if the annealing temperature is high enough.
   Lots of discussion on the origin of bubbles:
   water?
   hydrocarbons?
   Mitani and Gösele
   1992
   cont…
   

   View Slide

10. Mitani and Gösele
    1992
    cont…
    FZ wafers are better than CZ
    wafers, because oxygen
    diffuses into the Si in the FZ
    wafers
    

    View Slide

11. Mitani, Lehmann, Gosele
    1990
    IEEE Solid-state sensor workshop
    Bubble formation during silicon wafer bonding: causes and remedies
    

    View Slide

12. Mitani, Lehmann, Gosele
    1990
    

    View Slide

13. View Slide

14. Hydrophilic bond chemistry Hydrophobic bond chemistry
    

    View Slide

15. High Resolution Electron
    Microscopy
    Of the bond interface
    It’s about 3 nm of Si-O-Si bonds
    Recall, this is after 1100 °C annealing
    The hydrophobic case has
    Si – F H – Si bonds,
    But occasionally has ‘oxide islands’
    Anyways, 3 nm of bond is negligible
    on our optical performance.
    Their MIRS technique raises the
    question of whether in double pass
    the column is enough to introduce
    subtle spectral features. I think ‘no’.
    (they have ~ 18 passes)
    

    View Slide

16. View Slide

17. UT-JPL bonded Si optics 2012-2013
    

    View Slide

18. Experiments so far
    

    View Slide

19. Ash
    4.0 µm
    0.0
    1.0
    2.0
    3.0
    4.0
    VG03
    1.
    2.
    VG03
    

    View Slide

20. View Slide

21. Experiments so far
    

    View Slide

22. View Slide

23. View Slide

24. View Slide

25. View Slide

26. View Slide

27. View Slide

28. View Slide

29. View Slide

30. View Slide

31. View Slide

32. View Slide

33. View Slide

34. View Slide

35. View Slide

36. Metrology for small gaps
    Cary5000 and Matrix method
    

    View Slide

37. Example 1: Double Side Polished (DSP), thick Si
    wafer at normal incidence
    •  Zeroth order assumption:
    •  A few subtle effects to consider:
    1)  Reflection is a function of nSi
    , which is a modest function of
    λ and temp: nSi
    =n(λ, T)
    2)  There will be multiple reflections within the substrate, since
    reflection loss is multiply retroreflected in the propagation
    direction
    3)  λ << d Incoherent
    4)  Absorption per unit length
    n
    Si
    ~ 3.48 at room temp
    R =
    (n −1)2
    (n +1)2
    ≈ 0.31
    €
    T
    net
    = T2 = (1− R) × (1− R) ≈ 0.48
    Si
    Air Air
    

    View Slide

38. 1000 1200 1400 1600 1800 2000 2200 2400
    h (nm)
    0.0
    0.2
    0.4
    0.6
    0.8
    1.0
    Relative Transmission
    Example 1: Double Side Polished (DSP), thick Si
    wafer at normal incidence
    Absorption sets in
    at λ = 1150 nm
    The red line is the measurement. It agrees very well with the prediction for
    multiple reflections within the substrate, shown in black - - - -
    Zeroth order
    only: Tnet
    ~ T2
    Multiple reflections are a
    ~10% effect!
    Must include up to 3rd
    order terms!
    

    View Slide

39. 1)  The etalon is characterized by the
    Fresnel reflectivity of its sidewalls, and
    the gap size
    2)  The reflectivity is encapsulated in the
    coefficient of finesse, FSi-Air
    ~2.5
    3)  Multiple reflections within the substrate
    will interfere to make a λdependent
    Transmission spectrum Te
    (λ)
    4)  d ~ λ Coherent
    5)  Lossless: Re
    (λ) = 1 - Te
    (λ)
    Example 2: Fabry Pèrot etalon
    Air gap immersed in Si, normal incidence
    Re
    (λ)
    Air
    Si Si
    Te
    (λ)
    d
    

    View Slide

40. 1)  There will be dozens of permutations of
    reflections and transmission
    2)  The intermediate interface can be
    treated as a black box with a etalon
    reflection and transmission: Re
    (λ) and
    Te
    (λ)
    3)  The interface walls can be treated as a
    black box with Fresnel reflection and
    transmission: Rn
    (λ) and Tn
    (λ)
    4)  Then apply wave transfer matrix
    technique
    Example 3: Fabry Pèrot etalon
    Air gap immersed in Si, normal incidence, with
    incoherent multiple reflections from Si puck walls
    Re
    (λ) Te
    (λ)
    Rn
    (λ) Tn
    (λ)
    

    View Slide

41. Example 3: Fabry Pèrot etalon
    Air gap immersed in Si, normal incidence, with
    incoherent multiple reflections from Si puck walls
    2007 Saleh and Teich Fundamentals of Photonics
    

    View Slide

42. Example 3: Fabry Pèrot etalon
    Air gap immersed in Si, normal incidence, with
    incoherent multiple reflections from Si puck walls
    Re
    (λ) Te
    (λ)
    Rn
    (λ) Tn
    (λ) Rn
    (λ) Tn
    (λ)
    Rn
    (λ) Tn
    (λ) Rn
    (λ) Tn
    (λ)
    Example 1 revisited: DSP thick Si puck at normal incidence
    ** n = nSi
    = n(λ)
    

    View Slide

43. 1000 1200 1400 1600 1800 2000 2200 2400
    Λ nm
    0.1
    0.2
    0.3
    0.4
    0.5
    0.6
    Trans
    Model Fabry Perot Transmission; 3960 nm Air gap
    Part VG03 measured
    Transfer Matrix d3960 nm
    DSP Si measured
    DSP Si prediction
    

    View Slide

44. VG15 9/16/2013
    1200 1400 1600 1800 2000 2200 2400
    h (nm)
    0.80
    0.85
    0.90
    0.95
    1.00
    Transmission
    Measurement
    +/ï 0.3%
    Prediction
    VG09 9/16/2013
    1200 1400 1600 1800 2000 2200 2400
    h (nm)
    0.80
    0.85
    0.90
    0.95
    1.00
    Transmission
    Measurement
    +/ï 0.3%
    Prediction
    

    View Slide

45. VG08 9/16/2013
    1200 1400 1600 1800 2000 2200 2400
    h (nm)
    0.80
    0.85
    0.90
    0.95
    1.00
    Transmission
    Measurement
    +/ï 0.3%
    Prediction
    VG07 9/16/2013
    1200 1400 1600 1800 2000 2200 2400
    h (nm)
    0.80
    0.85
    0.90
    0.95
    1.00
    Transmission
    Measurement
    +/ï 0.3%
    Prediction
    

    View Slide