Stress analysis of a high fill-factor micromachined bolometer for thermal imaging applications

Size: px

Start display at page:

Download "Stress analysis of a high fill-factor micromachined bolometer for thermal imaging applications"

Garry Hardy
6 years ago
Views:

1 Stress analysis of a high fill-factor micromachined bolometer for thermal imaging applications M.Safy, A.Hafz Zaky, H.Abdalla, Y. Elshaer Department of Electronics Military Technical College EGYPT Abstract: - The stress analysis of infrared radiation bolometer which is made of a thin film of asi:h has been performed using ANSYS (11) CFD program. This bolometer is realized in a multilevel electro thermal structure having a high fill factor. Using the multilevel structure, thermal isolation can be independently optimized without sacrificing the IR absorption area. The analysis showed that optimization of the arm dimensions can be done without failure of structure breakdown or buckling. The design shows that the thermal time constant is ms, the responsivity is of 4.65X10 3 V/W, and the detcetivity is of 7.97X10 8 cm Hz 1/2 w -1 which can be achieved in a 50µmx50µm michromachined structure Key-Words: - Bolometer-Infreared- Stress Analysis-Micromachined-Detectors-FEA 1 Introduction Merging the micromachined techniques with thermal bolometers has lead to the ability of fabrication of uncooled devices having comparable performance to its cooled photonic counterparts, this also provided the ability of integration of array readout circuits with bolometer arrays using the same IC post processing techniques hence lowering the cooling cost, overall fabrication cost. Surface micomachined uncooled detectors are mostly based on microbolometer approach, where infrared radiation heats the sensor material through an absorbing mechanism, changing its resistance according to its temperature coefficient of resistance (TCR) of its active material. This approach allows the implementation of small pixel sizes, such as 50µmx50µm.Vanadium oxide (VO x ), the most widely used material has a TCR of about 2-3% /K [1], however VO x is not a standard material in IC fabrication. Other materials such as amorphous silicon [2] and poly silicon SiGe [3] have also high TCR values of 3-6%/k and 2-3%, respectively. However these materials suffer special annealing problem in addition to limited design flexibility to thermal optimization. We believe that greater design flexibility is provided using amorphous silicon semiconductors utilizing λ/4 cavity absorber structure [4,5,6] as will be discussed later. The fill factor is the detection area to the detector area; its values depend on the structure and the shape of the supporting leg. The metal bolometer is realized in a multilevel electro thermal structure with a fill factor over 92% [7] compared to values of 51%and80% for normal λ/4 air cavity structures [6, 7].The objective of this paper to make a model of the mostly used bolometers, then optimizing the bolometer from compatibility, design flexibility, TCR, fill factor and thermal isolation points of view, paying high attention to the structure mechanical stability. The stability of the suspended structures depends mainly on the stress in the different layers of which it is composed. High overall compressive stress results in structure buckling and the structure might stick to the substrate loosing its thermal isolation. On the other band, high overall tensile stress might cause cracking of the suspended structure. Ideally the stress must be zero, or practically, it must have a very low value [8]. 2 Structure Optimization The proposed bolometer structure is in a sandwichgap form consists of two sandwich elements in series, with a gap component in parallel [9]. Figure.1 Sandwich Gap geometry For achieving high thermal isolation and high fill factor, this bolometer is to realized in a electrothermal structures [7]. ISSN: ISBN:

The proposed structure is shown in Figure.2.The bottom metal consists of a thermally evaporated titanium reflector having sheet resistance of 10 Ω /.

2 The proposed structure is shown in Figure.2.The bottom metal consists of a thermally evaporated titanium reflector having sheet resistance of 10 Ω /., forming a multilevel structure using 2 sacrificial layer steps, a quarter-wave length absorber which consists of active material for the bolometer and nickel chromium film which is impedance matched to free space having a sheet resistance of 377 Ω /. Titanium was chosen for the reflector film because it gives excellent adhesion, has low film stress and good mechanical properties. Note that the bottom metal also supplies the electrical connection to the bolometer and provides the mechanical support for the thermal isolation structure. This type of detector has the potential of yielding 90% absorption integrated over the 8-14μ.m wave band [10]. Figure.2 Proposed high fill-factor bolometer using surface micromachined multilevel structure a.si:h was chosen since by optimizing deposition parameters of hydrogenate amorphous silicon (a.si:h) thin film, superior thermistors with a high temperature coefficient of resistance and low electronic excess noise will be obtained [2]. The 377 Ω /.metal absorber layer is impedance matched to free space: hence 50% of the incident radiation is absorbed and 50% transmitted, with none reflected. The transmitted portion of the radiation travels through the quarter-wave semiconductor layer, is reflected by the lower metal film, and destructively interferes with the incident wave, giving zero reflection and almost 100% absorption. The bottom part of Figure.2 shows the multilevel electro thermal isolation structure of the proposed bolometer. The first level forms the suspended arms which are mechanically supporting the top IR absorbing area by connecting it to substrate. High fill factor, independent high thermal isolation can be achieved from optimizing the dimension and layout of the arms due to folding underneath the bolometer. How special attention should be paid to stress of the subsequent layers since it is the major factor of the structure stability. 3 Predicted performance Prediction for the performance in vacuum is obtained using simple analytical expressions for thermal conductance, time rise and time constant. Parameters used for performance calculation are contained in table.1 Table.1 Definition of parameters associated with microbolometer Symbol Description ρ (asi:h) The resistivity = Ω.m g The gap length = 6μ.m d The thickness = 0.8 μ.m w The detector side length = 50 μ.m l The length of the bottom layer = 22 μ.m K The thermal conductivity of the thermal loss path = 7.44 w. (m.k) - 1 A p The crossectional area of the thermal thermal path = 5X0.3X10-12 m 2 L the length of the thermal path = 50 μ m I The bias current = 2 μa R The detector resistance = 6.19X10 5 Ω α TCR = 0.022/K 0 ε The emissivity = 1 G The thermal conductivity = 8.928X10-7 W/K B The bridge factor = (R L / (R L +R) )= 0.5 R L Load resistance = R A The detector area [w(2d 2 +l 2 g)] V n The total detector noise = 2.9X10-8 v/hz 1/2 at 80 Hz [42] C The total thermal capacitance = 5.79 J/K The detector resistance is given by: The thermal conductance is given by: G=K (A p /L) = 4.45X10-7 W/K The responsivity is given by: at 80 Hz ISSN: ISBN:

3 The detectivity is given by: The thermal time constant: Table.2 Properties bolometer materials Thermal Specific Denisity conductivity (w/mk) Heat (J/Kg.K) (Kg/m 3 ) Ni-Cr asi:h Ti Stress Analysis The main reasons of stress affecting the structure are the weight of the bolometer and the residual stress in the deposited layers of the bolometer, which consists of three layers of NiCr, asi:h and Ti. The large difference in thermal expansion coefficients between the layers results in a considerable stress at the interface surfacces which can be detrimental for film adhesion. The residual stress in NiCr layer can be ignored due to its small thickness (.05µm), the residual stress is considered mainly for the other two layers. This model considers a.si:h layer deposited by Plasma enhanced chemical vapour deposition (PECVD). It is generally accepted that a large amount of hydrogen dilution into SiH4 plasma could result in the formation of nanocrystalline Si films, and hydrogen plays a dominant role by enhancing atom mobility, passivating the dangling bond defects, etching disordered phases, reconfiguring the subsurface bonds and modifying Si Si network [11 13]. There are some reports on the stress variation during film growth as a function of hydrogen dilution [14, 15]. However, because of the controversy about the growth mechanisms of crystallites in the nanocrystalline silicon [15], there are no systematic studies for the stress evolution. Ion bombardment, hydrogen content, defects and nanocrystal formation are some possible factors for the development of film stress. Few reports are available on the relationship of the stress evolution with surface morphology and microstructure evolution, especially for the transition region between a-si and nc-si films. In accordance published data of measured stress parameters for a.si:h layers [16] will be used for the modeling. To evaluate the stress in Ti layer, we use a new technique depend on the relation between the composite materials [17] modeled as follows: The average stress is given by P=σ A (1) Where: P The stress force σ The normal stress A The cross sectional area Since: P=P t +P a (2) Where: P t Titanium layer force= σ t A t P a Hydrogenate Amorphous silicon = σ a A a With: σ A= σ t A t + σ a A a (3) And σ =E ε (4) Where ε is the strain. E Modules of elasticity For equal strain then [17]: E=E t (A t /A)+E a (A a /A) (5) The modules of elasticity (E t ) of (Ti) can be obtained practically by stress-strain measurements. Titanium films of (0.3µm) thickness were grown using thermal evaporation technique on rubber substrate having 1.92Cm thickness, using Electron Beam Evaporator, under deposition parameters shown in table (3). Table. (3) Deposition parameters for Ti Sample thickness Power supply sample. Vacuum measurements 0.25μm 4KV,300 ma 10-7 Torr Figure (4.3) shows the stress-strain measurements for a blank rubber substrate, while the operation conditions for testing are shown in table (4). ISSN: ISBN:

4 Table. (4) Operation conditions for tensile testing. Load cell 10KN Test speed 500 mm /min Grip to grip separation 25 mm Where: E Total E 1 E 2 t 1 Total Young s modulus=2.72 MPa Ti Young s modulus= GPa Rubber Young s modulus=2.59 MPa Ti thickness=0.3μm T Total thickness= x10-3 m t 2 Rubber thickness=1.92x10-2 m Fig. (3) Strain stress curve for rubber substrate The stress-strain measurements for rubber substrate coated with 0.3 μm thick titanium layer at the same operating conditions is shown in figure (4). Simulated Stress Analysis: In this work the stress within the bolometer has been solved by using the computational fluid dynamics (CFD) model. This model has been solved to predict the behavior of the stress for the bolometer. ANSYS (11) CFD program has been used to solve the stress affects on the bolometer. Calculated parameters for the Ti and a.si:h layers are listed in table (5). Table. (5) Shows the parameters of materials for simulation Titanium(Ti) Amorphous silicon (asi:h) Density=4850 Kg/m 3 Density=4850 Kg/m 3 Fig. (4) Strain stress curve for rubber substrate with Ti coating. Finally the modules of elasticity can be calculated by the relation [17] E Total =(t 1 /t) E 1 +(t 2 /t) E 2 (6) Thermal Expansion =8.6X10-6 /C Tensile stress, Yield=140ΜPa Modules of elasticity= GPa Poisson ratio=0.34 Thermal Expansion =4.4X10-6 /C Tensile stress, Yield=120MPa Modules of elasticity=112.4 GPa Poisson ratio=0.280 ISSN: ISBN:

Results Finite Element Analysis (FEA) calculations were performed for the

Residual stress in layers can be extracted from these figures.

a- Deformation shape a- Deformation shape b- Equivalent Stress Fig.

5 Results Finite Element Analysis (FEA) calculations were performed for the design. Stress distribution for the structures is shown in figures (5), (6). Residual stress in layers can be extracted from these figures. 10 µm leg width. a- Deformation shape a- Deformation shape b- Equivalent Stress Fig. (6) Residual stress after deposition for 5 µm leg width. b- Equivalent Stress Fig. (5) Residual stress after deposition for ISSN: ISBN:

6 Table. (6) Demonstrate the X, Y and Z stress Stress direction for 10 and 5μm leg width 10 leg width (M Pa) 5m leg width (M Pa) X Y Z Concolusion Further decrease of supporting arms can't be proposed, because this will lead to increase in the thermal capacity which will tend to increase the time constant of the bolometer. From the simulation results, it s clear that for 10 µm leg and 5µm leg width, the proposed structure is affected by residual stress of the deposited layers which cause a deformation in the proposed structure shape. For 10µm leg width the maximum deformation at the bolometer center equal to 0.465µm and µm for 5µm leg width. This deformation has no effect on the bolometer stability or performance. On the other hand this deformation has a considerable affect for structures utilizing λ/4 air cavity underneath bolometer for infrared absorption. This shown the greater amount of design flexibility for λ/4 absorber amorphous semiconductor structures. References: [1]R.A.Wood, Uncooled Thermal Imaging Monolothic Silicon Focal Arrays, Infrared Technology XIX Proc.SPIE, 2020, 322 (1993). [2] J-L Tissot et al., "LETI/LIR's amorphous silicon uncooled microbolometer development," Proc. SPIE Vol. 3379, Infrared Detectors and Focal Plane Arrays 111(1994). [3] S.Sedky, Paolo Fiorini,Matty Caymax,Agnes Verbist,Chris Bart, IR bolometer made of polycrystalline silicon germanium, Sensor and actuators A66 (1998) [4] J.Brady, T.Schimert, D.Ratcliff, R.Gooch,B.Richet, P.McCardel,K.Rachel, Advanced in amorphous silicon uncooled IR systems, SPIE Vol.3698,April [5] K.C. Liddiard,M.H.Unewisse and O.Reinhold, Design and fabrication of thin film monolithic uncooled infrared detector arrays, SPIE Vol, [6] I.F. Brady, "Advanced in amorphous silicon uncooled IR systems," Proc. SPIE Vol. 3698, Infrared Technology and Applications XXV, p; 161 (1999). [7]Hyung-Kew Lee,Jun-Bo Yoon,Euisik Yoon,Sang-Baek ju,yoon-joong Yong,Wook Lee,and Sang-Gook Kim, A high fill factor infrared bolometer using micromachined multilevel electrothermal structures, IEEE Transactions on electron devices, vol.46. No.7, July [8]S. Sheng, X. Liao, G. Kong, APPL. Phus. Lett. 78, 2509, (2001). [9] K.C.Liddiard,M.H.Unewisse and O.Reinhold, Design and fabrication of thin film monolithic uncooled infrared detector arrays, SPIE Vol, [10]M.H.Unewisse,S.J.Passmore,K.C.Liddiard and R.J.Watson"Performance Of Uncooled Semiconductor Film Bolometer Infrared Detector",SPIE Vol.2269 Infrared technology XX (1999). [11] S. Miyazaki, N. Fukhara, M. Hirose, J. Noncryst. Solids (2000). [12] D. C. Marra, E. A. Edelberge. R. L. Nano, E. S. Aydil. Appl, Surf. Sci 133, , (1998). [13] K. S. Stevens, N. M. Johnson, J. Appl, Phys. 71 (1992) 2628 [14] U. Kroll, J. Meier, A. Shan, S. Mikhailov, J. Weber, J. Apple. Phys. 80, 4971(1996). [15] S. Hamman, P. Roca, l. Cabarrocas, J. Appl. Phys. 81, 7282, (1997). [16] B. Kalache, et al., J. Appl. Phys. 93, 1262, (2003). [17] I. Barsoum, K. S. Ravi Chandran," Stress Intensity Factor Solutions for Cracks in Finite- Width Three Layer Laminates With an Without Residual Stress Effects," Engineering Fracture Mechanics (2003). ISSN: ISBN:

Micro-Electro-Mechanical Systems (MEMS) Fabrication. Special Process Modules for MEMS. Principle of Sensing and Actuation

Micro-Electro-Mechanical Systems (MEMS) Fabrication Fabrication Considerations Stress-Strain, Thin-film Stress, Stiction Special Process Modules for MEMS Bonding, Cavity Sealing, Deep RIE, Spatial forming