Chapter 2 Principle
2.5 Summary
As discussed above, the design principles of beam-shaping components of fiberlenses are mentioned. The recognition related to gratings is necessary for designing and fabricating gratings suitable for the pickup-head. We adopted symmetric transmission microgratings as the beam-splitter in our proposed optical pickup-head.
The grating designed in this thesis is according to Rigorous Coupled Wave Analysis (RCWA), which is a more accurate solution of Maxwell’s equations [11] [12].
Since the package software, GAOLVER, based on RCWA is already available to the analyses of diffraction, there is no further introduction about RCWA in this thesis. We used the software, GAOLVER, to perform the simulation which will be discussed in detail in chapter 3.
Chapter 3
Simulated Results
3.1 Introduction
In the pickup-head system, the gratings have been widely used as beam-splitters to divide into three beams for tracking error detection. We will discuss the design of microgratings in this chapter. First, the simulation tool, GSOLVER software, will be briefly introduced. Then, we use GSOLVER to calculate the optical performance of our micrograting. The parameters considered in our design, includes the material, period, thickness, and filling factor of the micrograting. Finally, a brief summary about the micrograting will be given.
3.2 Simulation Tool
GSOLVER software is an analytic tool available to diffraction simulation of gratings. It calculates diffracted fields, phases, and diffraction efficiency of plane wave illuminated to arbitrarily complex grating structures, while the illumination may be at any incidence (conical mounts) and any polarization (TE, TM, circular, or elliptical). The grating structure is defined by a piecewise constant approximation, so that it permits the analyses of simple classical grating profiles (blaze, sinusoid, holographic, binary, 3-point plotline, etc), and complicated structures (multi-layers, coatings, interweaving of material, shadows, etc).
GSOLVER employs full 3D rigorous coupled wave algorithm (RCWA), which can be applied to explaining the diffraction phenomenon of gratings. We will not discuss further theory of RCWA here, since the derivations of these equations are too complicated to be introduced here. The references of RCWA theory will be listed in the Appendix 2. Thus, for the pickup head, by calculating the energy distribution of all retained diffraction orders beams, the features of microgratings will be optimized in the following sections.
3.3 Material and dimension of microgratings
To achieve simultaneously high light-efficiency, expected energy distribution and realization of fabrication, the structure of microgratings, including their period, filling factor, and thickness, must be optimized. Thus, these parameters and material of microgratings will be determined in the following sections.
3.3.1 Material of microgratings
While most transmission gratings are used in the visible light spectrum, it is possible to extend their performance into the near UV (250nm) and near IR (2.5um), with choices of appropriate materials. Generally, maximum value of diffraction efficiency depends somewhat on the refractive index of the grating, in that lower index leads to greater efficiency, due to reduced backward diffraction that results from lowered reflectivity. But decreasing the index also calls for increased grating depth, while thick layers tend to have rough surface and even cracks or corrugation.
In the surface-micromachining processes, generally, SiO layers are applied to
the sacrificial layers, and Si3N4 layers are used as the structures or electric insulation.
Contrast to the poly-silicon, Si3N4 has higher transmission efficiency in the blue-laser spectrum. Compared with the photoresist, the thickness of Si3N4 layers can be controlled more accurately by LPCVD than the photoresist by the spinner, since the thickness of microgratings mainly determines the optical property. Although isotropic O2 plasma is used to trim the profiles and thickness of the grating ridges, unstable etching will reduce both the width and the height of the remaining photoresist ridges, and hence affect the filling factor of the final grating. Besides, due to the natural photoresist behavior, the groove form can not be easily manipulated, so that results differ from sample to sample. Compared with main kinds of materials applied to MOEMS process, as shown in Table 3-1, the nitride has high transmission light-efficiency in the blue-laser spectrum and is well-suitable for the surface-micromachining processes. Therefore, Si3N4 is a preferred candidate as the material of our micro-grating.
Table 3-1. Contrast between the materials applied to the surface-micromachining process [1].
Material SiO2(glass) poly-silicon photo-resist Si3N4
Transparent region (um) 0.16~8 1.1~14 broad 0.25~9 Control of layer thickness PECVD LPCVD Spinner LPCVD&
H3PO4
The Si3N4 layers are deposited in NFC (Nano Facility Center). Generally, the Si3N4 layers applied to structures are deposited by LPCVD, since ones achieved by PECVD (Plasma Enhanced Chemical Vapor Deposition) have considerable tensile or compressive stress. Generally, it is compressive stress when the deposition temperature is of less 350 ℃, while tensile stress exists at high deposition temperature, as shown in Fig. 3-1. The Si3N4 layers by LPCVD are belong to the Si-rich-Si3N4, and they have higher refractive index (n=1.96~2.2) and lower residual stress than N-rich-Si3N4, whilelarge residual stress will result in degrading optical property of components. Besides, the thickness of Si3N4 layers can be trimmed by the wet-etching of 86% H3PO4 ( the etching rate is 8.1nm/min at 180℃) or BOE (the etching rate is 57nm/min at 25 ℃ ). Therefore, the material of our micrograting is low-stress Si-rich-Si3N4 deposited by LPCVD.
Fig. 3-1 Compressive stress and tensile stress
3.3.2 Period of microgratings
For the tracking error detection, the grating is one of key components used to divide into three beams in the pickup-head system. As a rule, the zeroth order beams read the data pitches of compact disk, and the two first order beams read adjacent pitches to keep the head both centered and focused, as shown in Fig.3-2. Therefore, since diffraction efficiency of +1 and -1 orders must be equal, the profile of transmission gratings is symmetric, necessarily.
In the next-generation optical storage system, the distance between adjacent pitches is 5um, and the focal length of objective lens must be 100*n um at least, where n is the refractive index of the substrate. For the next-generation optical storage disks (Philips and Sony proposed), diffraction angle of gratings between the zeroth order and +1 or -1 order beams is 5*10-2 radians or 2.86∘(commercial specifications) at least, as shown in Fig. 3-3. The following formula is the grating equation:
n1sinθ1 = n2sinθ2 + mλ/d , m=0, 1, 2...,-1,-2..., Eq. 3-1
where n1 the n2 are the refractive indices of grating and ambience, θ1 and θ2 are the angles between the incident (and the diffracted) wave directions and the normal to the grating surface, λ is the wavelength, d is the grating period measured in um, and m is the order of diffraction.
According to Equation 3-1, the period of rectangular gratings is about 8 um. We set the diffraction angle between the zeroth order and the +1 orders beams to be 2.9∘+ 0.1∘. Thus, the period is 8um + 0.25um, where the grating period in our mask is 8um and the fabrication tolerance is 0.25um.
3.3.3 Thickness and filling factor of microgratings
Since the thickness and filling factor of gratings mainly determine the diffraction efficiency, and they are affected greatly by the quality of fabrication, they are chosen as the main variables of GSOLVER simulation. The relationship between the thickness and diffraction efficiency was simulated at different filling factor ranging from 0.3 to 0.7, respectively. For evaluating diffraction efficiency and these two
variables simultaneously, the results of GSOLVER simulation were plotted as the contour diagrams by the Sigma-Plot software according to the distribution of diffraction efficiency ratio values, as shown in Figs. 3-5, 3-6, 3-7, 3-8, 3-9, 3-10. The simulated results of diffraction efficiency distribution in the blue-laser spectrum were shown in Figs. 3-5, 3-6, 3-7, while Figs. 3-8, 3-9, 3-10, were in the red-laser spectrum.
In those figures, the green region represents these ratio-values ranging from 6 to 12 for the pickup head, while the blue and red areas represent ratio-values smaller than 6 and larger than 12, respectively. We further contoured the green region by the ratio-values difference of 1 to hold elaborately the distribution of diffraction efficiency ratio.
According to the thin-grating theory, the distribution of maximal diffraction efficiency is described as the following equation:
mλ Eq. 3-2 n - 1
t =
, where t is the thickness of thin gratings with maximal diffraction efficiency, λ is the wavelength of incident beams, n is the refractive index of thin gratings. We adopted that the wavelength of incident beams was 405nm, and refractive index of Si3N4 was 2.1, so the maximal diffraction efficiency happened when the thickness of thin gratings was 0.37um, 0.75um, 1.11um, etc. But the diffraction efficiency ratio was from 6 to 12 only when the thickness of thin gratings was about 0.75um.
For improving the feasibility of expected micro-gratings, these parameters in the green region having larger area, were used to realize the micrograting. From these simulated results, the micro-grating is the most suitable for the pickup head when its thickness is 0.75um, since filling factor ranging from 0.4 to 0.6 still serves the
objective at that thickness, while larger range of filling factor represents larger process tolerance. Therefore, our micrograting was designed to have filling factor of 0.5 and grating depth of 0.75um. Besides, the micrograting is still suitable for the pickup-head in the red-laser spectrum. Therefore, this micrograting can be broadly applied to pickup-head systems, whether their light-source is blue-laser or red-laser.
Because the free-space micrograting may not stand vertically to the substrate due to the fabrication error, the incident beam does not propagate normally to the grating.
Therefore, the GSOLVER simulations at different angles were required to explore the incident-angle tolerance. From the simulation, we concluded that the incident angle tolerance was as high as 15 degrees.
3-Beam Tracking
+30% off-track on track -30% off-track -1 order
Tracking Error Signal = E-F
Fig. 3-2 Three beams method for the tracking error detection.
Fig. 3-3 Schematic of the optical path of the next-generation optical storage system,
where θ represents the diffraction angle between the zeroth order and the +1 orders beams.
3.3.3.1 Thickness of microgratings
The proper thickness of gratings will enhance the diffraction efficiency and split beams at the optimal ratio suitable for the pickup heads. For the pickup head, the diffraction efficiency ratio of gratings is 6~12 (commercial specifications) between the zeroth order and the +1 orders beams, as shown in Fig. 3-4.
Fig. 3-4 Target of energy distribution of the three-beamed micrograting
0 order
According to the contour diagrams (Figs. 3-5, 3-6, 3-7, 3-8, 3-9, 3-10), the thickness of Si3N4 layers should be 0.75um + 0.05um, where the filling factor of microgratings is 50%. Namely, the target of film thickness can be set to be 0.75um, and the fabrication tolerance is 0.05um. The thickness of Si3N4 layers can be easily controlled by selecting suitable LPCVD deposition time. Once the Si3N4 film is thicker than 0.75um, H3PO4 or BOE is used to trim the thickness.
3.3.3.2 Filling factor of microgratings
In the fabrication flow, the linewidth of components tends to vary in photolithography and RIE steps. Therefore, the tolerance of the filling factor must be detected. According to the contour diagrams (Figs. 3-5, 3-6, 3-7), when the filling factor is from 40% to 60%, the diffraction efficiency ratio still meets the objective.
Namely, this range indicates that the linewidth of grating ridges can be varied from 3.2um to 4.8um without huge deviation of energy distribution, when the ridges length of microgratings is 4um. Generally speaking, the tolerance of 0.8um is achievable.
3.4 Summary
For the pickup-head, we designed a micrograting made of Si3N4 to achieve the expected diffraction efficiency ratio (6~12) between the zeroth and the +1 orders beams. According to the distribution of diffraction efficiency ratio, the thickness of 0.75um is preferred in the blue-laser or red-laser spectrum, where the period is 8um + 0.25um, and the incident angle of blue-laser beam is of less than 15∘Considering the process tolerance in advance, the feasibility of expected energy distribution of diffracted beams can be realized.
.
Fig. 3-5 Distribution of diffraction efficiency ratio between the zeroth and two first orders beams. GSOLVER simulation condition：λ= 405nm，period=8um，material is Si3N4，diffraction angle is 2.9∘, and the incidence angle is 0°.
≦6 6~12 ≧12 The thickness of micro-grating
The filling factor of micro-grating
≦6 6~12 ≧12 The thickness of micro-grating
The filling factor of micro-grating
Fig. 3-6 Distribution of diffraction efficiency ratio between the zeroth and two first orders beams. GSOLVER simulation condition：λ= 405nm，period=8um，material is Si N ，diffraction angle is 2.9∘, and the incidence angle is 5°
≦6 6~12 ≧12 The thickness of micro-grating
The filling factor of micro-grating
Fig. 3-7 Distribution of diffraction efficiency ratio between the zeroth and two first orders beams. GSOLVER simulation condition：λ= 405nm，period=8um，material is Si3N4， diffraction angle is 2.9∘, and the incidence angle is 10°
The filling factor of micro-grating
≦6 6~12 ≧12 The thickness of micro-grating
Fig. 3-8 Distribution of diffraction efficiency ratio between the zeroth and two first orders beams. GSOLVER simulation condition：λ= 633nm，period=8um，material is Si3N4， diffraction angle is 4∘, and the incidence angle is 0°.
Fig. 3-9 Distribution of diffraction efficiency ratio between the zeroth and two first orders beams. GSOLVER simulation condition：λ= 633nm，period=8um，material is Si3N4， diffraction angle is 4∘, and the incidence angle is 5°.
≦6 6~12 ≧12 The thickness of micro-grating
The filling factor of micro-grating
≦6 6~12 ≧12 The thickness of micro-grating
The filling factor of micro-grating
Fig. 3-10 Distribution of diffraction efficiency ratio between the zeroth and two first orders beams. GSOLVER simulation condition：λ= 633nm，period=8um，material is Si3N4， diffraction angle is 4∘, and the incidence angle is 10°
Chapter 4
Micro Fabrication Processes
4.1 Introduction
The fabrication technologies for the fiber-alignment structures and free-space microgratings will be presented in this chapter. The rhombus-shaped channels were fabricated for inserting the fiberlens. The out-of-plane micrograting made of Si3N4
was realized by the MOEMS process and the probe system. Finally, the future work and a brief summary were described.
4.2 Fabrication of the fiberlens [1]
The fabrication of the hemispherical-shaped fiberlens was mentioned in detail in [1]. We will directly adopt the fiberlens without further introduction of its fabrication process.
4.3 Fabrication of the rhombus-shaped channel
Anisotropically etched V-grooves in silicon or precision-machined glass blocks are widely used for this purpose of placing the fibers. These drawbacks of V-grooves are that the vulnerable bare fibers placed on top of the wafer, and shear stress
produced by the dried resins. Although alternative technologies [1][2][3] such as adding plate on top of the groove were developed to fix the fibers mechanically within
the V-grooves, their fabrication is complex and these additional mechanisms can neither hold the fibers accurately.
A fiber alignment structure, with precise rhombus-shaped channels [4] formed by {111}-equivalent planes within {100}-silicon, is proposed in our novel optical pickup head module shown in Fig.1-4. Compared with the V-groove channels, the fibers are totally buried and thereby protected. Therefore, the rhombus -shaped channels are suitable for the hybrid integration of other components on their top surface, and allow a simplified assembly of fiber ribbon arrays due to integrated funnels for fiber
insertion. Besides, adding resins on the fiber in the V-groove will result in reliability problem due to shear stress. However, the upper parts of rhombus provide the support force and precise alignment to the fiber, so the position of the fiber is more precise as shown in Fig. 4-1(a). Therefore, the rhombus-shaped channels are adopted for the fiber -alignment structures in our pickup head system.
Rhombic channels were first demonstrated in silicon using laser melting [5].
Alternative, deep silicon etching as well as wafer dicing can be used for the pre- structuring step. The vertical-anisotropic pre-structuring step causes a deep crystal damage that allows the wet etching to attack deeper areas of the wafer. A rectangular damage such as a trench in {100}-silicon aligned along the (110)-flat directly defines the shape and size of the rhombus. A small but deep trench results in a nearly closed rhombic channel, while a broad but shallow trench ends up in a V-groove. The final size of the rhombus is defined by the width w of the opening at the wafer and the depth d of the trench. The minimal depth has to be
Eq. 4-1
If the trench width is equal to the mask opening, a standard single-mode fiber is completely buried below the wafer surface for d = 2r =125um and w = 2r*tan(α/2) = 64.5um. A comparable V-groove would be as wide as 241 um, a U-groove requires 125um at least. If the trench width is smaller than the mask opening mainly due to the issues of lithography and etching, the depth has to be increased properly. The mask opening w precisely defines the two upper {111}-planes of the rhombus that provide the necessary precision for the alignment of fibers. If w varies by △w, the fiber core position is vertically shifted by △y = √2*△w, as shown in Fig 4-1(b). The achievable accuracy of rhombus -shaped channels can directly be examined by measuring the width w of the slit after KOH etching.Therefore, the axis height of optical systems can be controlled precisely.
Fig. 4-1 (a) The position of the fiber in the V-groove has huge deviation due to shear stress, and (b) the accuracy of rhombus-shaped channels can directly be examined by measuring the width of the slit after KOH etching.
The rhombus-shaped channels are fabricated with the bulk-micromachining technique. The fabrication process is illustrated in Fig. 4-2. First, a 0.5um layer of
nitride (Si3N4) deposited by LPCVD was used to resist the anisotropic-etching of KOH. Second, the width of trenches was defined with the photolighography. We adopted a layer of thick-photoresist AZP4620 spun up to 6um onto the chip as the mask layer of the Deep-RIE and ICP. The Deep-RIE (reactive ion etching) opened the silicon nitride mask, and this step determined the overall accuracy of rhombus -shaped channels. After that, the expectant trench depth was achieved using the ICP
(Inductively Coupled Plasma). Another approach to achieve the necessary trench was that milling with a wafer-dicing saw, which was more expensive but inaccurate than ICP. Finally, according the simulation of Intellisuite software, the rhombus-shaped channel was structured optimally by the anisotropic wet etching of 20% KOH solution at 60℃ in about 2 hours. The etching stops stably at {111}-planes, which are arranged under an angle of 54.7∘against the surface forming a rhombus.
The fibers were inserted from one end face and threaded easily without dedicated tools. Once introduced into the channels, the fibers were hold in the rhombus-shaped channels. After adding glue or resins, the fibers were slightly pressed into the upper part of rhombus due to the shear stress from the dried glue or resins. The excess glue or resins can escape through the slit of the top surface. A variety of glues can be used to fix the fibers in the rhombus, especially epoxy resins and UV-curing glues. We adopted epoxy resins called Acryfix that are hardened in 15 minutes at 25℃.
After fiber assembly, the quality of the mounting process is proven by cutting the glued channels. No residual film between the fiber and the upper part of rhombus is found and the cavity is completely filled with hardened resins. The deviation of the
After fiber assembly, the quality of the mounting process is proven by cutting the glued channels. No residual film between the fiber and the upper part of rhombus is found and the cavity is completely filled with hardened resins. The deviation of the