in configuration text files (Fig.2, frame no.1). There arespecified the ambient temperature, support temperature andinner-device power dissipation source parameters.Configuration files also contain the specification of thedevice spatial extent and properties of used materials. Theinitial description uses only material data like heat capacity,thermal conductivity, specific heat, and density of thematerials used as particular device layers. Spatialconfiguration of the device covers the wafer shape and itsspatial extent (width, length and thickness). There are twokinds of layer. As wafers are real ones, inter-wafer spacesare treated as virtual layers. For automate model generationthe Hedoris system uses Matlab package (Fig.2, frame no.3).Automate model generation algorithms are implementedwithin Matlab environment as a set of standalone scriptsdriven from configuration text files. Model generation usesthe Matlab library of elements and modules used to form ane-Cubes device (Fig.2, frame no.2). The resulting thermalmodel of the device is implemented in Verilog-AMSstandard. The model structure is compatible with Cadencestructure of a project, so it can be imported into the Cadenceenvironment (Fig.2, frame no.5). To use the Cadence CIWbased(Command Interpreter Window) graphicalenvironment it can be directly copied into the project datastructures (Fig.2, frame no.4). A template project library hasto be prepared using Cadence tools. If there is no Cadencetemplate library prepared, the simulation has to be run fromthe Unix command-line (Fig.2, frame no.6).The visualization of the results is the last simulation stage(Fig.2, frame no.7). For transient simulations it is supportedby CIW graphical interface visualization tools. Steady-statetemperature distribution visualization is supported by Matlabenvironment.The Model ArchitectureThe device model architecture is hierarchical. It containsinstances of independent modules interpreted as modellayers. Inter-module layers and sub-area within each layerare also supported by the device model architecture. Alllayer types and instances are defined in the same way. Thesample-Cubes heterogeneous device formed as three-waferstack of modules is presented on the Fig.324-26 September 2008, Rome, ItalyThe definition of Wafer instances uses the wafer thickness,bottom gap thickness, and [X, Y] coordinates of shapecorner positions (NW_X, NE_X, SE_X, SW_X, NW_Y,NE_Y, SE_Y, SW_Y). The Via layer configuration specifies[X, Y] coordinates of all Via instances, and the type of Viageometry (rectangle, circle and ellipse). As Via instanceenumeration is automatic, the last parameter – Via number –is optional. The sample configuration file is presented on theFig.4.Wafer(0 , 15e-6, 0, 425e-6, 425e-6, 0, 425e-6, 425e-6, 0, 0);Wafer(12.5e-6, 15e-6, 0, 425e-6, 425e-6, 0, 425e-6, 425e-6, 0, 0);Wafer(12.5e-6, 15e-6, 0, 425e-6, 425e-6, 0, 425e-6, 425e-6, 0, 0);Via( 37.5e-6:50e-6:387.5e-6, 37.5e-6, 3, 1, 3, 1);Via( 37.5e-6:50e-6:387.5e-6, 387.5e-6, 3, 1, 3, 1);Via(112.5e-6:50e-6:212.5e-6, 212.5e-6, 3, 1, 3, 1);Via( 212.5e-6,262.5e-6:50e-6:312.5e-6, 3, 1, 3, 1);Via( 37.5e-6, 87.5e-6:50e-6:337.5e-6, 3, 1, 3, 1);Via( 387.5e-6, 87.5e-6:50e-6:337.5e-6, 3, 1, 3, 1);HeatSource( 100e-6,100e-6,25e-6,25e-6,3,1,2,2,10e-3);HeatSource( 325e-6,325e-6,25e-6,25e-6,3,2,3,2,10e-3);AreaOfInterest(100e-6, 325e-6, 10e-6, 0, 1, 2, 1 );AreaOfInterest(325e-6, 100e-6, 10e-6, 0, 1, 3, 2 );Fig.4. Sample initial configuration of the device model.The device configuration and model have a hierarchicalstructure. The e-Cubes wafer instances support the devicefunctionality (sensors, data processing etc.). They are placedon the top of the device model hierarchy. The devicearchitecture (Fig.3) contains instances of wafer (“Wafer”)and interconnection (“Via”) layers. Each layer containsvarious instances of elements supporting particular materialproperties for the defied area (Fig.5).Fig.3. Three-wafer e-Cubes device with 3 wafers (WAF), 2 heatsources (HS), 2 area of interest (AOI) and 33 inter- and throughchipvia (VIA).The configuration file specifies all modules of the device.Fig.5. The hierarchical architecture of the thermal modeldescription.Hence one can found through-chip via (VIA), inter-chipvia (ICV), silicon (WAF), and inter-wafer gap (FIL)elements within “Wafer” and “Via” layers. Each of them isformed on the basis of parameterized basic discretisationelement (BDE). The BDE parameterization is supported byinternal model factors defining thermal material parametersof modeled layer area. Also physical extent of BDE elementis used each time the BDE element is instantiated. The BDEelement is a 3D element and has a form of triangular prism.As presented on Fig.6 the BDE element has 6 thermal nodesand 6 thermal resistors (R TH_XX where XX index definethermally connected nodes). The presented element belongs©<strong>EDA</strong> <strong>Publishing</strong>/THERMINIC 2008 81ISBN: 978-2-35500-008-9
III. HEDORIS SIMULATIONSThermal simulations of various device configurationsconfirm the system usability. The set of device thermalsimulations defined by the configuration presented above(Fig.4) is presented on Fig.7. The temperature distribution ofthe bottom (Fig.7a), middle (Fig.7b), and top wafer (Fig.7c)is presented. As the ambient and support temperature is setto the level of 310K, the bottom device wafer is thermallybound to the support temperature. Gaps between wafers arefilled with air. According to the Fig.3 all wafers arethermally and mechanically connected by 33 via/wafer 33through chip via/wafer instances. The Hedoris output post-24-26 September 2008, Rome, Italyto the n’th layer of the device. Assuming the n’th layer to be simulation data shows the local inner-wafer heat-flow.the 1’st wafer, the adjacent n’th+1 layer would be the gap The most important advancement of the Hedoris system isbetween 1’st and 2’nd wafers, filled with air and 33 via that the simulation can be run within Cadence embeddedinstances. Then the n’th+1 layer could model some solid, Spectre universal simulator.liquid or gas filling the gap.aFig.6. The equivalent thermal structure of the BDE elementApart from the ambient and substrate temperature,standalone heat source elements (HS) are defined inconfiguration file by HeatSource statement (Fig.4). Heatsources impose Dirichlet, Neuman or Cauchy boundarythermal conditions to the structure. Heat sources applyparticular temperature (K) or define a power dissipationlevel (mW) for the selected area of the device or any of itscomponents.For transient simulation purposes instances of virtual areaof interest (AOI) can be defined in initial configuration fileby the statement AreaOfInterest (Fig.4). The AOI instancesmake it possible to find the final transient temperaturedistribution in particular regions of the device and tovisualize the time dependent local variation of temperature.Referring to the initial configuration of the device, spatialcoordinates of the layer/device (x, y), the layer/device extent(dx, dy, dz), type of the shape (circle, rectangle, polygon)and the layer/wafer consecutive number have to be specifiedas layer parameters. To define heat source parameters, theinitial configuration file also describe the HS-typedependent: temperature or power dissipation levels. Thedeclaration of the area of interest (AOI) is similar: first fourparameters define spatial coordinates and extent of the AOI(refer to the HS statement). Remaining parameters are theAOI shape and wafer number the particular AOI belongs to.The last parameter is optional. The AOI consecutive numbercan be automatically assigned. If specified manually, theAOI consecutive number is verified and (if necessary)automatically corrected by Hedoris system.Fig.7. Results of thermal simulation for the structure shown onthe Fig.3 for double 40mW heat-flow source.The same can be done in any other multidomain HDLsimulator.IV. COVENTOR SIMULATIONSFinite element modeling and simulation method (FEM) isan alternative to the high level modeling method presented inthe previous paragraph. One of the FEM methoddisadvantages is a need of commercial software dedicated tosupport a particular type of simulation. One of importantFEM-based software advantages is that it is possible to ispossible to create a physical model of almost any real siliconstructure or nanostructure. CoventorWare [7] system is agood example here. It supports defining new materials, realtechnological process list and mask sets. The systemsupports creation of a structure realistic model. ITE usesCoventor-Ware package for FEM based modeling andsimulation. Coupling between thermal, electrical andbc©<strong>EDA</strong> <strong>Publishing</strong>/THERMINIC 2008 82ISBN: 978-2-35500-008-9
- Page 1 and 2:
http://cmp.imag.fr/conferences/ther
- Page 3 and 4:
24-26 September 2008, Rome, Italy©
- Page 5 and 6:
24-26 September 2008, Rome, Italy©
- Page 7 and 8:
24-26 September 2008, Rome, Italy©
- Page 9 and 10:
24-26 September 2008, Rome, ItalyBL
- Page 11 and 12:
24-26 September 2008, Rome, ItalySe
- Page 13 and 14:
24-26 September 2008, Rome, ItalyED
- Page 15 and 16:
Staggered Cartesian meshes tolerate
- Page 17 and 18:
VIII. THE CASE FOR MCAD-EMBEDDED EC
- Page 19 and 20:
24-26 September 2008, Rome, ItalyTr
- Page 21 and 22:
solution to the eikonal equation(
- Page 23 and 24:
24-26 September 2008, Rome, ItalyS
- Page 25 and 26:
24-26 September 2008, Rome, ItalyTr
- Page 27 and 28:
24-26 September 2008, Rome, Italyda
- Page 29 and 30:
q (RthΣ, t)q (R th , t )10.80.60.4
- Page 31 and 32:
24-26 September 2008, Rome, ItalyEv
- Page 33 and 34:
24-26 September 2008, Rome, Italy*R
- Page 35 and 36:
24-26 September 2008, Rome, Italy[7
- Page 37 and 38:
24-26 September 2008, Rome, ItalyNe
- Page 39 and 40:
24-26 September 2008, Rome, ItalyS=
- Page 41 and 42: 24-26 September 2008, Rome, ItalyVI
- Page 43 and 44: 24-26 September 2008, Rome, Italyth
- Page 45 and 46: 24-26 September 2008, Rome, Italypr
- Page 47 and 48: 24-26 September 2008, Rome, ItalyFi
- Page 49 and 50: 24-26 September 2008, Rome, ItalyPo
- Page 51 and 52: 24-26 September 2008, Rome, ItalyWL
- Page 53 and 54: 24-26 September 2008, Rome, ItalyTa
- Page 55 and 56: 24-26 September 2008, Rome, ItalyII
- Page 57 and 58: Thermal resistance (K/W)0.250.200.1
- Page 59 and 60: 24-26 September 2008, Rome, ItalyB.
- Page 61 and 62: temperatrue rise [K]76543210Sim0,00
- Page 63 and 64: 24-26 September 2008, Rome, ItalyCo
- Page 65 and 66: 24-26 September 2008, Rome, ItalyPr
- Page 67 and 68: 24-26 September 2008, Rome, ItalyPr
- Page 69 and 70: 24-26 September 2008, Rome, ItalyBl
- Page 71 and 72: conclusion, every tile is contacted
- Page 73 and 74: and B10), three space blocks (W4, W
- Page 75 and 76: 24-26 September 2008, Rome, ItalyMu
- Page 77 and 78: After having determined the k eff v
- Page 79 and 80: 24-26 September 2008, Rome, ItalyVI
- Page 81 and 82: 24-26 September 2008, Rome, ItalyTh
- Page 83 and 84: PTjunc24-26 September 2008, Rome, I
- Page 85 and 86: tures - has been shown previously [
- Page 87 and 88: 24-26 September 2008, Rome, ItalyAu
- Page 89 and 90: IV.CONCLUSIONSMentor Graphics Exped
- Page 91: 24-26 September 2008, Rome, ItalyIn
- Page 95 and 96: 24-26 September 2008, Rome, Italytr
- Page 97 and 98: 24-26 September 2008, Rome, ItalyDe
- Page 99 and 100: 24-26 September 2008, Rome, ItalyII
- Page 101 and 102: datasheets, these values are usuall
- Page 103 and 104: 24-26 September 2008, Rome, Italyte
- Page 105 and 106: 24-26 September 2008, Rome, ItalyCo
- Page 107 and 108: 24-26 September 2008, Rome, Italy20
- Page 109 and 110: material to handle difficulties suc
- Page 111 and 112: 24-26 September 2008, Rome, ItalyAt
- Page 113 and 114: 24-26 September 2008, Rome, Italymo
- Page 115 and 116: heatspreaders deposited during the
- Page 117 and 118: 24-26 September 2008, Rome, ItalyTh
- Page 119 and 120: 24-26 September 2008, Rome, Italyla
- Page 121 and 122: Fig. 13 Transient of the output dio
- Page 123 and 124: 24-26 September 2008, Rome, ItalyIn
- Page 125 and 126: chamber to calibrate the thermal te
- Page 127 and 128: V.Table 2: FE-Simulation test-matri
- Page 129 and 130: 24-26 September 2008, Rome, ItalyCo
- Page 131 and 132: 24-26 September 2008, Rome, Italy(c
- Page 133 and 134: Case coverSupercapacitor n°2Electr
- Page 135 and 136: 24-26 September 2008, Rome, Italyfr
- Page 137 and 138: 24-26 September 2008, Rome, ItalyTA
- Page 139 and 140: 24-26 September 2008, Rome, ItalyTh
- Page 141 and 142: 24-26 September 2008, Rome, Italych
- Page 143 and 144:
24-26 September 2008, Rome, ItalyDe
- Page 145 and 146:
Peltier control units (PCUs). The t
- Page 147 and 148:
24-26 September 2008, Rome, ItalyTh
- Page 149 and 150:
every step: the fewer cells the mor
- Page 151 and 152:
TABLE IIEffect of multiplication-bl
- Page 153 and 154:
24-26 September 2008, Rome, ItalyTh
- Page 155 and 156:
24-26 September 2008, Rome, ItalyRe
- Page 157 and 158:
Drain source leakage current (A)Rev
- Page 159 and 160:
24-26 September 2008, Rome, ItalyFP
- Page 161 and 162:
V IN =V DD /2Fig. 5. Short circuit
- Page 163 and 164:
24-26 September 2008, Rome, ItalyTh
- Page 165 and 166:
The overall objective of the NANOPA
- Page 167 and 168:
24-26 September 2008, Rome, ItalyRe
- Page 169 and 170:
predictions show that carbon nanotu
- Page 171 and 172:
24-26 September 2008, Rome, ItalyFl
- Page 173 and 174:
55. Campbell, R.C., S.E. Smith, and
- Page 175 and 176:
24-26 September 2008, Rome, ItalyCo
- Page 177 and 178:
24-26 September 2008, Rome, ItalyTh
- Page 179 and 180:
24-26 September 2008, Rome, ItalyEX
- Page 181 and 182:
24-26 September 2008, Rome, Italyen
- Page 183 and 184:
24-26 September 2008, Rome, ItalyKE
- Page 185 and 186:
An original three-step etch process
- Page 187 and 188:
with a network of built-in sensors
- Page 189 and 190:
24-26 September 2008, Rome, Italyan
- Page 191 and 192:
24-26 September 2008, Rome, ItalyFu
- Page 193 and 194:
24-26 September 2008, Rome, ItalyFi
- Page 195 and 196:
24-26 September 2008, Rome, ItalyFi
- Page 197 and 198:
24-26 September 2008, Rome, ItalyWe
- Page 199 and 200:
semiconductors is of the order of 1
- Page 201 and 202:
24-26 September 2008, Rome, ItalyDu
- Page 203 and 204:
transistor, through the bulk and be
- Page 205 and 206:
impedance behaviour and on the othe
- Page 207 and 208:
Tungstenmicro-heaterGas SensingMate
- Page 209 and 210:
Tungsten micro-heater resistance (
- Page 211 and 212:
24-26 September 2008, Rome, ItalyPo
- Page 213 and 214:
24-26 September 2008, Rome, ItalyT3
- Page 215 and 216:
24-26 September 2008, Rome, ItalyEv
- Page 217 and 218:
24-26 September 2008, Rome, ItalyD.
- Page 219 and 220:
24-26 September 2008, Rome, ItalyOn
- Page 221 and 222:
24-26 September 2008, Rome, ItalyAc
- Page 223 and 224:
24-26 September 2008, Rome, Italy
- Page 225 and 226:
24-26 September 2008, Rome, ItalyLO
- Page 227 and 228:
24-26 September 2008, Rome, Italymu
- Page 229 and 230:
24-26 September 2008, Rome, Italy(R
- Page 231 and 232:
24-26 September 2008, Rome, ItalyPr
- Page 233 and 234:
24-26 September 2008, Rome, ItalyPa
- Page 235 and 236:
24-26 September 2008, Rome, ItalyEl
- Page 237 and 238:
Temperature rise (°C)1,81,61,41,21
- Page 239 and 240:
i résistances carbone (A)i résist
- Page 241 and 242:
calculates the dissipation distribu
- Page 243 and 244:
in the next. It can also be used to
- Page 245 and 246:
allows creating special methods of
- Page 247 and 248:
OLED device (see Fig. 2.) provided
- Page 249 and 250:
24-26 September 2008, Rome, ItalyFi
- Page 251 and 252:
24-26 September 2008, Rome, ItalyIn
- Page 253:
24-26 September 2008, Rome, ItalyWa