... Enhancements for Backside FA Immersion lens technology has proven to be an effective solution There are two types of immersion lens technology to date namely, Liquid Immersion Lens (LIL) [45] and Solid Immersion. .. concentrations and (b) p-Si doped at 1019 cm-3 for different Si thickness Quantum Efficiency of typical Si-CCD, MCT and InGaAs FPAs Liquid Immersion Lens Schematics of (a) Refractive Solid Immersion Lens and. .. liquid and solid immersion Solid immersion Lens (SIL) technology can be divided into Diffractive SIL (DSIL) and Refractive SIL (RSIL) Liquid immersion can enhance resolution by a factor of the refractive
DEVELOPMENT AND CHARACTERIZATION OF REFRACTIVE SOLID IMMERSION LENS TECHNOLOGY FOR FAR- FIELD INTEGRATED CIRCUIT FAILURE ANALYSIS USING LASER INDUCED TECHNIQUES GOH SZU HUAT A THESIS SUBMITTED FOR THE DEGREE OF DOCTOR OF PHILOSOPHY DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING NATIONAL UNIVERSITY OF SINGAPORE 2009 Abstract Abstract Backside failure analysis methods have become more important because of the increasing number of metallization levels within an integrated circuit which makes it difficult to locate defects in the active region from the frontside The transition to new packages like flip-chip and lead-on-chip also contribute to the need for backside analysis techniques However, the layer of silicon substrate poses tradeoffs in terms of degradations in image resolution This is due to the high absorption of low wavelength light in silicon as well as additional spherical aberration caused by imaging through a mismatched medium These give rise to a large diffraction limited resolution of approximately μm using 1340nm wavelength laser Refractive Solid Immersion Lens technology for spatial resolution enhancements is studied in this work An analytical model is devised to investigate the theoretical limits Characterization of resolution based on key design parameters identified is performed These theoretical findings are verified experimentally on a scanning optical microscope to a correlation accuracy of less than 10% A resolution enhancement of close to 70% is achieved This work also optimizes RSIL for backside laser induced fault localization techniques An enhanced laser induced signal of approximately 12 times is achieved The last part in this work demonstrates the effectiveness of combining RSIL and pulsed laser technique for 65nm technology node and beyond fault localization The findings will be useful to provide academia with a theoretical perspective to understand Refractive Solid Immersion Lens This work also provides useful i Abstract information to industry with the intention to incorporate and optimize Refractive Solid Immersion into their routine Failure Analysis flow to enhance the current backside fault localization capability ii ACKNOWLEDGEMENTS First and foremost, I would like to thank my main supervisor, Prof Jacob Phang for his guidance for the last years, beginning from my undergraduate final year In spite of his busy schedule, he never fails to allocate time to review my progress and gives timely advice It is also under his leadership that I am given the opportunities in project collaborations, both local and overseas He has trained me to be an independent and confident researcher Another mentor who plays a significant role in my project is my co-supervisor, Prof Colin Sheppard He is always approachable and I had benefited greatly from the numerous theoretical discussions with him I am inspired by his passion in research and strong technical ability despite his multiple commitments in other academia appointments I would also like to thank Prof Ludwig Josef Balk for his invaluable technical discussions whenever he visits my university A large contributing factor to the success of this project lies in the equipment and facilities support from Semicaps, Singapore Besides providing the resources, the CEO, Mr Chua Choon meng and his managers also provide guidance and advice in the effective use of their equipments in the course of our collaboration Through the numerous communications on the requirements of their industrial partners, I am able to identify the issues that are critical to industries currently and in the near future For these, I would like to take this opportunity to thank Semicaps and the staff iii Another key collaborator in this project is Dr Frank Zachariasse from NXP Semiconductors, formerly known as Philips Semiconductors I would like to thank him for his guidance in the fabrication of Diffractive Solid Immersion lens during my attachment to Philips on the Diffractive Solid Immersion Lens project I would also like to thank the staff of the Centre for Integrated Circuit Failure Analysis (CICFAR) for having provided excellent administrations and logistics support throughout my postgraduate In addition, I would like to thank Chartered Semiconductor Manufacturing for providing a part of my grant to support my postgraduate as well as the numerous opportunities to interact and learn from students under the same program Finally, not forgetting the fellow students of Prof Jacob Phang, this is an important group of people who is the main source of encouragement and morale support when the going gets tough We also had many valuable discussions and collaborations together I really appreciate them Thank you iv LIST OF TABLES Table 1.1 Table 3.1 Critical defect size (extracted from ITRS 2008) Comparison between Liquid and Solid Immersion technology 60 LIST OF FIGURES Fig 1.1 Fig 1.2 Fig 1.3 Fig 1.4 Fig 1.5 Fig 1.6 Fig 2.1 Fig 2.2 Fig 2.3 Fig 2.4 Fig 2.5 Fig 2.6 Fig 2.7 Fig 2.8 Fig 2.9 Fig 2.10 Fig 2.11 Fig 2.12 Fig 2.13 Fig 2.14 Fig 2.15 (a) SEM cross section of an IC with layer metallization and (b) layer metallization (a) Frontside accessible wire bond and (b) backside accessible flip chip die packages Light transmittance of (a) 500 μm p-Si for different doping concentrations and (b) p-Si doped at 1019 cm-3 for different Si thickness Quantum Efficiency of typical Si-CCD, MCT and InGaAs FPAs Liquid Immersion Lens Schematics of (a) Refractive Solid Immersion Lens and (b) Diffractive Solid Immersion Lens Technology Focusing of a collimated light beam to (a) a single point according to geometrical optics and (b) a finite spot with diameter of the order of the light wavelength according to diffraction theory Huygens’ Principle applied to (a) plane wave and (b) spherical wave Model of Huygens- Fresnel Principle Contribution from spherical wavelets from P1 from an aperture at P2 in terms of Rayleigh-Sommerfeld Diffraction Integral Wave propagation using diffraction formula Diffraction of incident plane wave on a slit Diffraction by a thin lens Airy disc at the focus of a lens Focusing by an objective lens Focusing of a spherical wave Resolution Criteria Relationship between a Gaussian and an Error complementary function (a) Spherical aberration from a high NA lens and (b) using an aperture to reduce spherical aberration (a) Spherical Aberration due to focusing light through a mismatched medium and (b) light propagating from a mismatched medium into air Diagram showing light focused by a lens into two media separated by a planar interface 10 12 17 18 19 20 21 22 23 25 25 26 29 30 31 32 33 v Fig 2.16 Fig 3.1 Fig 3.2 Fig 3.3 Fig 3.4 Fig.3.5 Fig 3.6 Fig 3.7 Fig 3.8 Fig 3.9 Fig 3.10 Fig 3.11 Fig 3.12 Fig 3.13 Fig 3.14 Fig 4.1 Fig 4.2 Fig 4.3 Fig 4.4 Fig 4.5 Fig 4.6 Fig 4.7 Fig 4.8 Fig 4.9 (a) Chromatic aberration in a converging objective lens and (b) using an additional diverging lens to reduce Chromatic aberration Aplanatic conjugate points on a sphere Laser focusing at (a) centric and (b) aplanatic point Schematic diagram of (a) conventional imaging, (b) backside imaging, (c) imaging using CRSIL and (d) imaging using an ARSIL Relationship between the field radius (horizontal) and the defocus distance using computer model at different ASIL thickness NAeff as a function of air gaps Time averaged electric energy density for focusing a laser beam polarized in azimuthally angle φ=0 direction and collection angle α (a) close to zero (small NA) and (b) 75° Angular spectra of several detection geometries: (a) aperture only and (b) aperture + SIL Notice that the combination of an aperture and a SIL gives larger overlap area and contrast Schematic of a confocal setup (a) Planar silicon-air interface and (b) continuous saw tooth profile interface (a) Plan view and (b) side profile for an approximated binary lens Difference between Fresnel Zone plate and Fresnel Phase plate DSIL fabrication by Gallium implantation and differential etching Standard SNOM setup consisting of a) an illumination unit, (b) collection and redistribution unit and c) a detection module Aplanatic sphere Geometry for point source at PA Wave aberration functions, as functions of the angle θ in the medium, for ZA= -130 mm Time-averaged electric energy density distributions in the x –z meridional plane for a probe depth of μm and for numerical apertures of (a) 0.3, (b) 0.6, and (c) 0.9 The calculations are for air (n1 = 1.0) and glass (n2 = 1.5) and a wavelength of λ = 0.6328 μm (He–Ne laser) 2.1E-3 is 2.1 X 10-3, and so on, in this and subsequent figures A truncated plane tilted along the (a) x-axis and (b) y-axis Spherical aberration model in light focusing through a sphere Spherical aberration function plot Contours of constant intensity in the x-y plane for an object plane Δu = μm for semi-aperture angle of 30° with (a) 15, (b) 35, (c) 65, (d) 105, (e) 125 and (f) 155 plot points At least 125 plot points is required to illustrate a continuous intensity distribution (Light wavelength, λ: 1340nm, spherical radius, r: 500μm) Contours of constant intensity in the x-y plane for an object plane Δu = μm for semi-aperture angle of (a) 1° limiting case, (b) 10°, (c) 30° and (c) 50° (Light wavelength, λ: 1340nm, spherical radius, r: 500μm) Normalized line profile in the x direction at Δu = μm for semiaperture angle of (a) 1°, (b) 10°, (c) 30° and (c) 50° (Light wavelength, λ: 1340nm, spherical radius, r: 500μm) 35 40 42 44 49 50 52 53 53 56 57 57 58 62 64 67 69 70 71 73 75 78 81 81 vi Fig 4.10 Fig 4.11 Fig 4.12 Fig 4.13 Fig 4.14 Fig 4.15 Fig 4.16 Fig 4.17 Fig 4.18 Fig 4.19 Fig 4.20 Fig 5.1 Fig 5.2 Fig 5.3 Fig 5.4 Fig 5.5 Fig 5.6 Fig 5.7 Fig 5.8 Fig 5.9 Fig 5.10 Fig 5.11 Fig 5.12 Fig 5.13 Fig 5.14 Fig 5.15 Change in azimuthally 90o spatial resolution with objective NA at different focusing planes Contours of constant intensity in the x-z plane for an object plane Δu = 100μm for the cases of objective NA (a) 0.15, (b) 0.4 and (c) 0.5 (Light wavelength, λ: 1340nm, spherical radius, r: 500μm) Contours of constant intensity in the x-z plane for an object plane Δu = 148μm for the cases of objective NA (a) 0.1, (b) 0.16 and (c) 0.19 (Light wavelength, λ: 1340nm, spherical radius, r: 500μm) Normalized axial intensity plot at focusing plane, Δu = 100μm, for objective NA (a) 0.29, (b) 0.4 and (c) 0.5 Imaging of an isotropic point receiver through a sphere Illumination through a sphere Change in spatial resolution with semi aperture angle, ψ at different focusing depths for a mm diameter spherical lens Change in semi collection angle θ with imaging depths for a mm diameter spherical interface under different aperture size and 0.5 objective NA constraint Change in spatial resolution with distance from centric location using 20X backing objective and variable clear aperture transparency for mm diameter RSIL Change in spatial resolution with distance from centric location using 50X backing objective and variable clear aperture transparency for mm diameter RSIL Additional transverse magnification from RSIL Experimental setup Reflected laser image using 1340 nm laser and 20X backing objective with (a) poor optical coupling and (b) improved optical coupling (a) Photograph and (b) cross section of decoupled RSIL (a) Photograph and (b) cross section of integrated RSIL Reflected laser image of pitch structures on a subsurface resolution target The smallest structure is the 0.4 μm pitch structure (a) Schematic of ray trace and (b) Reflected laser images at the different focal planes when laser is focused through a mm RSIL with height less than a hemisphere (a) Schematic of ray trace and (b) Reflected laser images at the different focal planes when laser is focused through a mm RSIL with height larger than a hemisphere (a) Reflected laser image from the top and (b) vicinity of the centre of a ball bearing of mm diameter Sample tilting for (a) a small die and (b) a large die Schematic of Coupled RSIL with (a), (b) poor optical coupling and (c), (d) good optical coupling between RSIL and interface Tilt compensation by flexible clear aperture on a decoupled RSIL Deflection of primary beam at the tilted interface Schematic of backside resolution target Reflected laser image of pitch structures on a subsurface resolution target 1340 nm laser reflected image (a) without RSIL and (b) with RSIL 82 84 85 86 89 91 92 93 95 96 98 102 103 104 105 106 109 110 111 113 114 115 115 117 117 118 vii Fig 5.16 Fig 5.17 Fig 5.18 Fig 5.19 Fig 5.20 Fig 5.21 Fig 5.22 Fig 5.23 Fig 5.24 Fig 5.25 Fig 5.26 Fig 6.1 Fig 6.2 Fig 6.3 Fig 6.4 Fig 6.5 Fig 6.6 over the marked area as (a) FWHM resolution calculated from an edge response when focal plane is at μm off centric (a) and (b) are captured with a clear aperture of 680 μm (c) and (d) are captured with a clear aperture of 950 μm Reflected laser image of subsurface resolution target features at μm below centric (a) - (d) are captured using 1340 nm laser wavelength (e) - (h) are captured with using 1064 nm laser wavelength Clear aperture size and backing objective NA are labeled in the respective figures FWHM resolution calculated from an edge response when imaging plane is at 125 μm below centric (a) and (b) are captured with a clear aperture of 680 μm (c) and (d) are captured with a clear aperture of 950 μm Reflected laser image of subsurface resolution target features Imaging plane is at 155 μm off centric (a) and (b) are captured using 1340 nm laser wavelength (c) and (d) are captured with using 1064 nm laser wavelength The clear aperture and backing objectives are labeled in the respective figures Change in spatial resolution with distance from centric location using variable backing objectives and 68% clear aperture Change in spatial resolution with distance from centric location using variable backing objectives and 95% clear aperture NA requirement and collection angle at various focus planes Reflected subsurface laser image with RSIL application using 1340 nm laser on IC features through silicon substrate In (a), 20X objective is used and FOV is limited by RSIL In (b), 50 X objective is used and FOV is limited by screen FOV Focusing depth is 13 μm below aplanatic location Additional magnification from RSIL is around 14 times Reflected subsurface laser image of 0.4 μm pitch structures in the middle at (a) digital zoom and (b) digital zoom 3, and 0.4 μm pitch structures at the edge of RSIL field of view at (c) digital zoom and (d) digital zoom 3, under RSIL application using 1340 nm wavelength laser and 20X objective Focusing depth is μm above aplanatic location Change in FOV with focusing depths for mm diameter RSIL Change in FOV with focusing depths for mm diameter RSIL Reflected- TIVA overlay image of 1.4 μm pitch polysilicon serpentine line structures with 20 X objective (a) without RSIL at digital zoom and (b) with RSIL at digital zoom TIVA image Comparison of (a) TIVA signal intensity and (b) normalized TIVA signal intensity with RSIL and without RSIL (a) Line profile and (b) normalized intensity plot across TIVA signal spot TIVA line profiles from RSIL with different backing objectives to show overfill effect at the back pupil Laser beam geometrical cone, (a) within mechanical aperture and 120 121 122 123 126 127 127 129 130 131 131 134 136 136 138 139 140 viii Fig 6.7 Fig 6.8 Fig 6.9 Fig 6.10 Fig 6.11 Fig 6.12 Fig 6.13 Fig 6.14 Fig 6.15 Fig 6.16 Fig 6.17 Fig 7.1 Fig 7.2 Fig 7.3 Fig 7.4 Fig 7.5 Fig 7.6 Fig 7.7 Fig 7.8 (b) larger than mechanical aperture Change in TIVA signal enhancements with varying laser focusing planes and objective NA “tail” artifacts in a TIVA image Horizontal line scan signal profiles of ac-coupled detection systems taken from a TIVA signal micrograph Line profile across a TIVA signal from an ac- coupled laser induced response of a laser beam pulsed at 100 Hz Backside reflected-TIVA overlay image using (a) 50X Mitutoyo objective (0.42 NA) and (b) 100X Mitutoyo objective (0.5 NA) 1340 nm wavelength laser is used Backside reflected-TIVA overlay image with (a) RSIL application and 20X Mitutoyo backing objective (0.24 effective NA) and (b) SEM image across AA’ (a) Backside reflected-TIVA overlay image and (b) Backside reflected Pulsed TIVA overlay image without RSIL application and using 20X Mitutoyo objective (0.24 effective NA) 1340 nm wavelength laser is used (a) Backside reflected-TIVA overlay image and (b) Backside reflected Pulsed- TIVA overlay image with RSIL application and 20X backing objective (0.24 effective NA) 1340 nm wavelength laser is used Backside reflected-TIVA overlay images (a) using 100X Mitutoyo objective (0.5 NA) and without RSIL application and (b) 20X backing objective with RSIL application 1340 nm wavelength laser is used [134] (a) Backside reflected-TIVA overlay image using 20X Mitutoyo objective (0.24 effective NA) with RSIL application, (b) GDS layout of area of interest and (c)SEM image of BB’ 1340 nm wavelength laser is used Backside reflected Pulsed-TIVA overlay images using (a) 50X Mitutoyo objective (0.5 NA) without RSIL application, (b) 50X Mitutoyo objective (0.42 NA) with RSIL application, (c) GDS layout of the area of interest and (d) TEM cross section along CC’ 1340 nm wavelength laser is used Dependence of measured resolution values on polarization direction Schematic design of a pair of (a) aplanatic RSIL and (b) analogous DSIL, with increasing size Schematic design of (a) an aplanatic RSIL and (b) analogous DSIL Schematic design of a DSIL (a) with diameter D1 and (b) D2 where D2 > D1 Schematic design of a DSIL with designed focal point at (a) k1d and (b) k2d where k2 > k1 and d is the thickness of silicon substrate Schematic design of a mobile DSIL Reflected laser image of subsurface features captured using 1340 nm laser and (a) 20X objective and (b) 50X objective Regions A and B denote the areas where the mobile DSIL will be positioned Reflected laser image of subsurface features captured using 20X 142 145 145 146 148 149 151 152 152 154 156 159 162 163 164 165 167 169 169 ix Chapter Literature Review It was also calculated that the FOV 2r for CRSIL can be expressed as: ⎡ 2aλ ⎤ 2r < ⎢ ⎥ ⎣ n(n − 1) ⎦ 0.5 (3.13) where r is the field radius 3.4.3 RSIL with Laser Polarization and Aperture Engineering RSIL was further improved by Serrels et al by introducing polarization effects to the excitation source via a half-wave plate [76] It is well known that spatial resolution is best when the edge of feature is parallel to polarization direction Fig 3.6 illustrates the case of focusing a laser beam polarized in azimuthally angle φ=0 direction In Fig 3.6(a), when objective lens NA is infinitely small, the focused spot size of the laser beam is symmetrical in all directions However, at high NA as shown in Fig 3.6(b), the focused laser spot size is elliptical The contour spacing along φ=π/2 is smaller than that along φ=0 The result is a better image spatial resolution in the φ=π/2 direction With an objective of 0.55, it was reported that a spatial resolution ranging between 120 nm to 250 nm was achieved depending on polarization state of 1530 nm laser Theoretical calculations could only explain the results partially Some possible reasons proposed were the imperfect interface and the effect of spatial irradiance distribution of incidence An imperfect interface near the focal plane could increase ellipticity of the focal spot which was predicted to be in excess of 1.7 at an air-glass interface [97, 98] This could lead to a decrease in the FWHM of the diameter The focal spot size and shape were also affected by the incident spatial irradiance distribution which was capable of a reduction factor of 0.7 from diffraction limit [99] Page | 51 Chapter Literature Review (a) (b) Fig 3.6 Time averaged electric energy density for focusing a laser beam polarized in azimuthally angle φ=0 direction and collection angle α (a) close to zero (small NA) and (b) 75° [62] RSIL combined with aperture engineering has also been used in optical storage to enhance collection efficiency [100] Conventionally when only the aperture is used as shown by Fig 3.7(a), the fiber optic illumination system with a small hole in the end of the fiber produces modulation over a wide angular range but they are not collected in the reflected path, limited by the NA of the fiber Although resolution is still determined by the aperture size, SIL enhances the collection efficiency in the angular spectrum as shown in Fig 3.7(b) A higher signal contrast is achieved with RSIL but aperture engineering is required for further resolution enhancement It has been shown that using a light wavelength of 690 nm, 200 – 300 nm circular and rectangular apertures fabricated on a mm diameter CRSIL with refractive index of 1.85 and effective NA of 0.95 could lead to an improvement of 20 % in resolution for the SIL aperture configuration over the case of using only SIL Page | 52 Chapter Literature Review (a) (b) Fig 3.7 Angular spectra of several detection geometries: (a) aperture only and (b) aperture + SIL Notice that the combination of an aperture and a SIL gives larger overlap area and contrast [100] 3.4.4 RSIL with Confocal Technique Another common optical technique to enhance resolution which can be applied on RSIL is the confocal technique [73, 92, 93] The confocal technique can be incorporated easily into most optical systems although there is a tradeoff in throughput The introduction of confocal pinholes to a conventional laser scanning microscopy is capable of enhancing both the lateral and axial resolution Fig 3.8 illustrates the location of pinholes A and A’ at the illumination path and in front of the imaging detector respectively in a typical confocal setup Fig 3.8 Schematic of a confocal setup The light intensity I collected by the point detector through an infinitely small pinhole in an ideal situation is given by [58, 101]: I (ν ) = [2πJ1 (v ) / v ] (3.14) Page | 53 Chapter Literature Review where v is the optical coordinates [58, 101] and J1 is the Bessel function of the first kind of the first order The intensity for the conventional optical imaging without a confocal pinhole is given by [58, 101]: I (ν ) = [2πJ1 (v ) / v ] (3.15) The Bessel function J1 can be used to estimate the diffraction pattern in the image plane A higher degree of Bessel function will translate to a narrower FWHM, hence a higher resolution enhancement factor By comparing the power in Eq 3.14 and 3.15, the enhancement factor from the confocal technique is close to 40 % However, the confocal pinhole needs to be approximately 10 μm to be fully effective In generally, confocal effects can be felt for pinholes less than 30 μm 3.4.5 RSIL Effect on Laser Induced Technique RSIL has been applied to enhance laser induced technique For FA fault localization, NAIL was used with lock-in amplifier on OBIRCH on sub 90 nm technology devices to achieve a 30 % reduction in signal spot size, to reduce the heat spread [36] This improves localization precision 230 nm line- space structures were also resolvable This is an improvement from 530 nm when observed without NAIL using a 50X objective lens Two photon optical beam induced current (TOBIC) is a multiphoton technique [102] which is commonly used in fluorescence microscopy As a brief introduction to fluorescence microscopy, a fluorescent molecule is excited by absorbing the energy of a photon and subsequently irradiating another photon as it relaxes to a lower energy Page | 54 Chapter Literature Review state The wavelength of emission for Stoke’s fluorescence is longer than that of the excitation (emission photon energy is lower than excitation) In multiphoton absorption, the energy of multiple photons are absorbed simultaneously by the fluorescent molecule and irradiated as a single photon However, the wavelength of emission is now 1/N of the illumination wavelength Therefore, when more photons are absorbed, the lower is the limit of excitation wavelength This is capable of improvement in lateral and axial FWHM of 20 – 30 % for and 20 photon processes Recently, an ARSIL or Supersolid Immersion Lens (SSIL) is applied to TOBIC [103] A 1530nm modelocked Erbium fibre laser was used as excitation source The laser generated 400 fs pulses at a repetition frequency of 30 MHz with an average output power of 75 mW This output power was focused by an objective lens of NA of 0.42 through a SSIL into the device under test An experimental resolution of 166 nm was achieved and essentially matched the calculated value of 168 nm In general, a two photon excitation is capable of a reduction of or 30 % improvement in spatial resolution 3.5 Diffractive Solid Immersion Lens DSIL designed for subsurface imaging involves fabricating concentric rings that look like Fresnel Zone plates [53, 54] on the Si substrate Diffractive lens are so called because diffraction of light occurs at the slits between the rings The widths of the slits are designed such that the first order diffraction light is collected or made to focus to a spot The first order light is used due to its high efficiency [54] Focal shifts which lead to SA not exist at the focal point Light rays within the geometrical cone of collection interfere constructively at the focus spot Page | 55 Chapter Literature Review For perfect constructive interference, the profile of a diffractive lens is continuous This can be understood by a simple model as shown in Fig 3.9 Fig 3.9(a) shows the rays and the wavefronts, emanating from a point source in Si Conventionally, when refraction occurs at the Si- air interface, focal shifts result One way to remove the aberrations is to create a saw tooth profile on the Si surface so that the rays incident normally on the modified interface This is shown in Fig 3.9(b) The regions in black are the areas to be milled (a) (b) Fig 3.9 (a) Planar silicon-air interface and (b) continuous saw tooth profile interface By maintaining a constant path difference of a wavelength from the point source, the rays have the same phase at different points at the interface In this way, they will focus constructively at the focus Due to fabrication simplicity, a DSIL with binary profile lens as shown in Fig 3.10 is usually created In this respect, DSIL works in a similar manner to a Fresnel Phase plate For a Fresnel zone plate, rays that will cause destructive interference are blocked so as to ensure only constructive interference at the focus However, the efficiency is reduced by half This can be overcome by using a Fresnel phase plate instead The difference between a Fresnel Zone plate and a Fresnel Phase plate is in Page | 56 Chapter Literature Review the introduction of a π phase change to the destructive rays This is shown in Fig 3.11 Fig 3.10 (a) Plan view and (b) side profile for an approximated binary lens [56] Fig 3.11 Difference between Fresnel Zone plate and Fresnel Phase plate Page | 57 Chapter Literature Review For Si backside application, a fabrication process which makes use of the concept of differential etch rate between bulk Si and Si implanted with Gallium (Ga) ions was developed [56] Under plasma etching, Si etches faster without Ga ions Prior to implantation step, a software is used to create a pattern for the locations of slits Subsequently, Ga ions are implanted using a Focused Ion Beam (FIB –fei Dual beam 860) on areas where the slits are not to be created The sample is etched in CF4 /O2 plasma (Nextral) in the final step It should be noted that the etch time varies with etch recipes and a calibration is necessary Fig 3.12 summarizes the fabrication process Fig 3.12 DSIL fabrication by Gallium implantation and differential etching [56] By fabricating a DSIL on Si backside, a pitch of 0.4 μm has been resolved [52] Other ongoing research in this field include reducing the fabrication time by using wet etching and improving the diffractive lens efficiency [104] By using wet etching, the Ga implant dose required for selective etching will be lower Hence, this will cut down on the time for implantation which is currently about 30 nm for a lens 150 μm in diameter However, the tradeoff is the isotropic etch characteristics which will affect the profile of the lens There are also other ways to fabricate DSIL like using electron beam lithography for instance [105, 106, 107] Page | 58 Chapter Literature Review For further resolution enhancement, it has been demonstrated that a large number of pinholes fabricated and distributed appropriately over the DSIL was capable of focusing X-Rays to spot sizes smaller than the diameter of the smallest pinhole [108, 109] These pinholes are known as photon sieves This arrangement ensures that a portion of the light diffracted from the pinholes interferes constructively at a focal point Photon sieves are capable of sharpening the focal spot with first order diffraction light and suppressing higher diffraction orders light By varying the pinhole locations, diameter and distribution, the optical performance can be optimized In general, the enhancement factor is d/ w where d is the diameter of the pinhole and w is the Fresnel zone width A resolution of 200 nm resolution was achieved Currently, the resolution of diffractive lens is also limited in the order of the smallest width of the outermost zone, in other words, the smallest slit structure (20 40 nm) [110] 3.6 Comparison of LIL, DSIL and RSIL LIL is easy to implement and is able to accommodate variable sample thickness However, the enhancement factor is significantly limited by the refractive index of most common commercial liquids available In this perspective, SIL, though complex, is more superior It retains most of the advantages of LIL and yet offers much better performance A SIL can be made of Si which yields a high enhancement factor of 3.5 It is also capable of enhancements to the factor of the square of refractive index of Si depending on design Although both RSIL and DSIL offer similar performance, DSIL suffers a few tradeoffs First, it is highly process dependent which means that it require certain Page | 59 Chapter Literature Review competence in operator skill and the stability of process in order to have consistent performance The performance is not easily reproducible for general systems More importantly, it is immobile It becomes impractical to create many lenses for different sites for analysis in view of the time factor As a guideline, about 25 – 30 minutes is required for each DSIL and as long as an hour can be taken to fabricate a high performance lens This is not taking into consideration errors in fabrication Based on these comparisons, RSIL is more practical to be adopted for industrial applications Table 3.1 compares the difference between LIL and SIL Table 3.1 Comparison between Liquid and Solid Immersion technology Sample Preparation Light Collection efficiency Critical Angle in light collection Spherical Aberration Light Wavelength Sample contamination Working Distance Mobility Imaging mode Implementation Liquid Immersion Variable sample thickness allowed Less sensitive to sample roughness Enhance light gathering power Restricted by critical angle Spherical Aberrations present at high NA RSIL Specific sample thickness Sensitive to sample roughness Enhance light gathering power Not restricted by critical angle Spherical Aberrations absent at ideal focal point Detection window is not restricted to monochromatic light Detection window is not restricted to monochromatic light Risk of sample contamination Low working distance No risk of sample contamination Variable working distance according to sample thickness High mobility Surface/subsurface High mobility Surface/ Subsurface Require liquid immersion compatible objective lens Simple to implement on existing optical system DSIL Variable sample thickness allowed Less sensitive to sample roughness Enhance light gathering power Not restricted by critical angle Spherical Aberrations absent at designed focal point Detection window limited to monochromatic light(designed wavelength) No risk of sample contamination Variable working distance according to lens design Not mobile Subsurface Not easy to create DSIL on location desired Page | 60 Chapter Literature Review 3.7 Near- Field Resolution Enhancement Techniques The near- field techniques of resolution enhancement took place in the early 20th century when a near-field detection of transmitted planar waves through a nanometer aperture (smaller than wavelength) [8, 111, 112] was proposed When the aperture size is much smaller than the light wavelength, at a small distance of λ/ to λ/ 3, the transmitted field is not diffracted as shown in Fig 2.6 Fig 3.13 shows a system configuration of a typical SNOM system [113] It consists of main sub-units, the illumination of the device through the aperture as shown in Fig 3.13 (a), the detection of the light after interaction with the sample by a highly efficient objective as shown in Fig 3.13(b) and finally the detection unit as shown by Fig 3.13(c) This configuration can give rise to a spatial resolution that is determined by the size of the detection aperture This technique is known as Scanning Near-Field Optical Microscopy (SNOM) and optical spatial resolution of ≤ 50 nm can be achieved typically [114, 115, 116] However, this gain in resolution comes with a tradeoff in detection sensitivity which is one of the main disadvantages of this technique The optical throughput T can be expressed as [117]: ⎛d ⎞ T =⎜ ⎟ ⎝λ ⎠ (3.16) where d is the diameter of the probe aperture and λ is the light wavelength The optical throughput is largely affected by the d since the variation of λ is relatively small over the light spectrum [117] For instance, the optical signal, due to a reduction factor of 10 to the aperture diameter, will reduce by 10,000 Page | 61 Chapter Literature Review Fig 3.13 Standard SNOM setup consisting of a) an illumination unit, b) collection and redistribution unit, and c) a detection module [113] This low throughput has been addressed by several groups over the past decades by improving probe design Frey et al used a back illuminated full body glass tips coated with a thin metal as local apertureless probetip [116] He optimized the thickness of the metal coating, angular illumination direction and polarization state for efficient coupling of light to surface plasmons excitation at the metal layer These surface plasmons are excited by the angular illumination at the metal layer and they propagate along the probe to the tip to produce a high light intensity confinement at the apex Alternatively, Kim et al enhanced light throughput by replacing conventional metal film coating on an oxide aperture with a double metal layer of Al/ Ti to control uniformity and shape of the aperture [118] The use of Ti layer resulted in a reduction of surface roughness This reduces light scattering and enhances throughput significantly Page | 62 Chapter Literature Review From SNOM, other near-field techniques have evolved Other relevant examples to this project include Scanning Thermal [119] and Thermal Conductivity Microscopy [10] and Scanning Near-Field Photon Emission Microscopy [11] Although there have been excellent results in overcoming the sensitivity issues and improving spatial resolution, there exists practical disadvantages which make it unattractive for routine FA fault localization The near- field condition imposes a very short working distance that is incompatible with backside analysis and together with the small depth of field, are some challenges that need to be overcome The long scan times resulting in low throughput must also be addressed before near-field techniques can be widely adopted by industrial FA analysis 3.8 RSIL Challenges for Backside Analysis The achievements of RSIL applications have been discussed in earlier sections However, in practice, there are still both hardware and theoretical challenges which need to be addressed before RSIL can be implemented and characterized for effective backside analysis One of the main issues arises from the limitation in the mechanical polishing of Si substrate Firstly, it is not possible to polish the substrate to a precise thickness to match the RSIL dimension perfectly for subsurface analysis Secondly, a nonuniformity of ± 5-10 μm in polishing is common Both factors will result in thickness errors which deviate from the ideal focal plane The knowledge of the effects of this parameter on RSIL performance and the tolerance to the deviations are essential prior to optimization of the lens for application in a non-ideal situation Page | 63 Chapter Literature Review In subsurface imaging through a spherical structure, it should be noted that off-axis effects exist The aplanatic locations are restricted to an aplanatic sphere as shown in Fig 3.14 and bounded by the following: Δu + Δr = 2 n2 r n1 (3.17) where Δu and Δr are the vertical and lateral displacements from the RSIL optical centre For an extended source or wide field imaging where Δr is non zero and μm < Δu < rn2/n1 μm, spatial resolution is expected to be enhanced in an annular manner In other words, lateral SA exists This is also an important parameter to characterize Fig 3.14 Aplanatic sphere Furthermore, although it is well known that when the focal plane is at the aplanatic location where light refraction at the spherical surface exists and the backing objective NA required for imaging is much smaller than operating at the centric locations, the performance of RSIL under varying backing objective NA need to be understood better Page | 64 Chapter Literature Review In a conventional scanning microscope system, Field of View (FOV) is dependent on the magnification of the objective lens Lens magnification is inversely proportional to FOV For an RSIL, there is additional magnification due to light refraction at the spherical interface As a result, FOV has another limiting factor It is necessary to characterize the relationship between RSIL dimension and its additional magnification during imaging so that FOV can be optimized This is particularly useful for fault localization where maximum details are preferred in a single scan image For RSIL implementation, one of the major issues is the coupling at the lens and sample interface A contact between two imperfect interfaces produces air gaps In order for evanescent wave coupling at the interface to occur, the air gap must be approximately equivalent to the wavelength of light inside the semiconductor [120] Some of the other challenges include aligning the centre of RSIL to the principal axis so that it can be navigated to the features of interest with precision for quantitative analysis Eventually, for commercial application, the RSIL must also be aligned so that it can be par focal with the other objective lens Tilt control is also important because in practice, Si thickness is never uniformly flat A good tilt control ensures that the light rays not get reflected off the principal axis after incidence on the tilted features The effect of lens spherical quality is also unknown These challenges have to be controlled if not overcome in order for RSIL performance to be characterized Page | 65