A Fast Wafer-Level Spatial Variation Modeling Algorithm for Test Cost Reduction of Analog/RF Circuits Hugo Gonçalves 1,2 , Xin Li 1 , Miguel Correia 2 and Vitor Tavares 2 1 ECE Department, Carnegie Mellon University, USA 2 Faculdade de Engenharia, Universidade do Porto, Portugal 09/07/2014 Slide 1
Outline Motivation and background Virtual probe (VP) Proposed approach Dual Augmented Lagrangian method (DALM) Two-pass test flow Experimental results Conclusions Slide 2
Process Variation 45nm 32nm 22nm Small size Large variation V V V TH TH TH L L L Doping fluctuation Δ V TH Parametric variations Δ L Line edge roughness Slide 3
Wafer Probe Test Multiple test items must be measured for each die An industrial example of dual radio RF transceiver ~1 second testing time per die ~6500 dies per wafer ~ 2 hour testing time per wafer Measuring all test items is time-consuming ~2h per wafer Slide 4
Test Cost Reduction by Spatial Variation Modeling Measure a small number of dies at selected spatial locations Recover the full wafer map by statistical algorithm Measured dies 20 160 140 15 Y Axis ??? 120 10 100 5 80 5 10 15 X Axis Measured delay values (normalized) Recovered wafer from 282 industrial chips map [Chang11], [Kupp12], [Huang13], [Hsu13], etc. Slide 5
Virtual Probe (VP) List a set of linear equations based on measurement data 20 160 DCT basis function 140 15 Y Axis 120 10 100 5 Performance measurement 80 DCT coefficients 5 10 15 X Axis N Measured delay values (normalized) f ( x , y ) b x , y from 282 industrial chips i i i 1 Results in an under-determined linear equation, since we have less measurements than unknown DCT coefficients Slide 6
Virtual Probe (VP) Additional information is required to uniquely solve under- determined linear equation 20 160 DCT Coefficients (Magnitude) 300 140 15 200 Y Axis DCT 120 10 100 100 0 5 20 20 10 10 80 5 10 15 0 0 Y Axis X Axis X Axis Measured delay values (normalized) DCT coefficients (sparse) from 282 industrial chips If process variations are spatially correlated wafer maps show sparse patterns in frequency domain Slide 7
Virtual Probe (VP) Solve sparse DCT coefficients by L1-norm regularization DCT coefficients can be uniquely determined from a small number of measurements DCT basis function B Sum of absolute values of all elements 1 2 B α f α min 2 2 1 α Performance Regularization parameter measurement DCT coefficients f α (sparse) Slide 8
Virtual Probe (VP) 1 Linear regression problem: 2 B α f α min 2 2 1 α There is no closed-form solution A standard interior-point solver is not computationally efficient We aim to develop an application-specific solver to reduce computational time and, hence, testing cost Slide 9
Outline Motivation and background Virtual probe (VP) Proposed approach Dual Augmented Lagrangian method (DALM) Two-pass test flow Experimental results Conclusions Slide 10
Dual Problem 1 2 B α f α Primal problem: min 2 2 1 α Key idea: form a dual problem to reduce the number of unknowns Primal problem # of unknowns = # of DCT coefficients # of dies Dual problem # of unknowns = # of measurements Since we have substantially less measurements than unknowns, solving the dual problem is significantly more efficient Slide 11
Strong Duality Primal function Dual function 1 1 1 2 α Bα f α 2 2 P ( ) ( ) x x f f D 2 1 2 2 2 2 2 T . . B x S T D ( x ) α P ( ) D ( x ) α P ( ) Dual variable of P α x Size: Size: # of coefficients # of measurements Dual variable of D Slide 12
Dual Augmented Lagrangian Define an auxiliary variable z to form an equality constraint Dual problem Dual problem w/ equality constraint 1 1 2 2 1 1 max D ( ) x x f f 2 2 max D ( ) x x f f 2 2 2 2 x z , 2 2 2 2 x T S T . . z B x T S T . . B x z Solve the augmented Lagrangian of the dual problem 1 1 2 2 2 x z α x f f α z B x z B x z T T T max L , , A 2 2 2 2 2 2 x z , , z 0 ( z ) Primal variable , z size = # of DCT coefficients Slide 13
Alternating Direction Method Augmented Lagrangian 1 1 2 2 2 x z α x f f α z B x z B x z T T T max L , , A 2 2 2 2 2 2 x z , Solve optimization with alternating direction method [Yang10] Variable update Optimality conditions P ( z ) k z α k x , , L α ( k 1 ) ( k ) T ( k ) A z P / B x 0 z z k 1 α k L x z , , 1 A B α ( k 1 ) T ( k ) ( k 1 ) x I BB f Bz 0 x α k 1 α k B x k 1 z k 1 T AL step Slide 14
Fast Matrix Inverse 1 B α ( k 1 ) T ( k ) ( k 1 ) x I BB f Bz Since DCT basis functions are used, we have T BB I Hence, we do not need to explicitly calculate matrix inverse 1 B α ( k 1 ) ( k ) ( k 1 ) x f Bz 1 Slide 15
Fast Matrix-Vector Multiplication k 1 P k k α B x T z 1 B α ( k 1 ) ( k ) ( k 1 ) x f Bz 1 α k 1 α k B x k 1 z k 1 T Since DCT basis functions are used, we can calculate these matrix-vector multiplications by fast DCT or IDCT transform Slide 16
Two-Pass Test Flow Measure all dies on one wafer if its spatial pattern cannot be predicted by a number of pre-selected dies Pre-test analysis Test cost reduction Measure all dies on first Measure few dies on a wafer following wafer Extract spatial pattern Predict spatial pattern by by VP VP Y N Predictable? Error is small? Y N Determine pass/fail by Measure all dies for all Measure all dies VP following wafers Slide 17
Error Estimation Modeling error by VP must be sufficiently small to ensure small escape rate and yield loss Yield Loss f ~ Escape Rate pdf , f ub ~ f lb lb ub f lb ub f ~ ~ ~ ~ YL pdf f , f df d f ER pdf f , f df d f ~ lb f ub lb f ub ~ ~ f ub f lb f ub f lb ~ expected values from training f measured values from current wafer f Slide 18
Outline Motivation and background Virtual probe (VP) Proposed approach Dual Augmented Lagrangian method (DALM) Two-pass test flow Experimental results Conclusions Slide 19
Experimental Setup Production test data of an industrial dual radio RF transceiver 9 lots and 176 wafers in total 6766 dies per wafer and 51 test items per die – test items were selected by [Chang11] 1,089,120 good dies and 101,696 bad dies Lot ID 1 2 3 4 5 6 7 8 9 Wafer # 25 9 23 25 25 25 17 25 2 [Chang11]: H. Chang, K. Cheng, W. Zhang, X. Li and K. Butler, “Test cost reduction through performance prediction using virtual probe ,” ITC, 2011 Slide 20
Spatial Pattern Examples Spatial pattern is observed for a subset of test items, but not all test items 1 1 100 100 80 80 Y Axis Y Axis 60 60 0.5 0.5 40 40 20 20 0 0 20 40 60 80 20 40 60 80 X Axis X Axis 1 1 100 100 80 80 Y Axis Y Axis 60 60 0.5 0.5 40 40 20 20 0 0 20 40 60 80 20 40 60 80 X Axis X Axis Test item #1 Test item #48 Slide 21
Recommend
More recommend