An Equivalent Circuit Model for a Faraday Cage Substrate Crosstalk Isolation Structure Joyce H. Wu ( ) and Jesús A. del Alamo Massachusetts Institute of Technology Faraday cage TransmitterReceiver 70 µm Substrate via 200 µm Deep reactive ion etch (DRIE) Silicon nitride barrier liner Electroplated Cu fills via Substrate-Via Technology Key Features 12 µm Cu Si 12 µm x 100 µm vias before Cu CMP step (aspect ratio = 8) Faraday Cage Isolation Structure Technology Faraday Cage Test StructureReference Test Structures Measurements Substrate Via Model: RvRv LvLv Z 11 Substrate-Via Impedance Faraday Cage Substrate Noise Isolation 100 µm Ground Signal System-on-Chip Substrate crosstalk is considered one of the biggest problems in mixed-signal circuits Motivation Si Grounded via Cu ground plane Al Silicon nitride Noisy or sensitive devices/circuits A/D Digital Logic Analog/RF Noisy Substrate Fabrication (2) DRIE through the substrate (3) photoresist strip, silicon nitride deposition from front and backside (4) e-beam deposition of Ta-Ti-Cu seed on backside (5) Cu electro- plating to close bottom of via (7) Cu CMP frontside for a flush surface (8) Al e-beam deposition and patterning to form contact to via Si substrate Al Cu Silicon nitride Ta-Ti-Cu seed µm (1) photoresist mask patterning (6) Cu electro- plating to fill via 1 GHz: 41 dB improvement 10 GHz: 30 dB improvement 50 GHz: 16 dB improvement At 100 µm distance, on average: Substrate thickness=77 µm Separation dist.=100 µm Via separation=10 µm Via diameter=10 µm Faraday Cage Air Reference Re(Z 11 ) Im(Z 11 ) Simulation Measured Faraday cage

Equivalent circuit lumped-element values Imag S 21 Measured Simulation Real Equivalent Circuit Model Reference StructureReference Structure with center split Simulations Comparison of Measurement and Simulation Increase Tx-Rx separation distance Increase via spacing Developed a simple, lumped- element equivalent circuit model Model matches experimental data into mm-wave regime Model will be useful to evaluate substrate noise isolation schemes in actual circuits Simple model matches data well (including real and imaginary S 21 ) Range of R v and L v consistent with measured values Spread of R v and L v of substrate via accounts for spread in S 21 of Faraday cage Increasing pad separation reduces substrate noise  R 1 and  C 1 to account for greater pad separation Increasing via spacing reduces substrate noise isolation Effectiveness of R v -L v shunt is reduced due to fewer vias Only need to increase R v -L v values for larger via spacing Faraday Cage Structure 10-µm via spacing 70-µm via spacing Simulation Measured R2R2 R2R2 R1R1 C1C1 R3R3 C3C3 R3R3 C3C3 C pad R1R1 C1C1 100-µm transmitter-receiver separation RrRr CrCr R2R2 R2R2 R3R3 C3C3 R3R3 C3C3 C pad R r = 5 k  C r = 3 fF R 1 = 2.5 k  C 1 = 6 fF R v =25 Ω L v =50 pH R v =1 kΩ L v =500 pH R v =250 Ω L v =200 pH R v =70 Ω L v =70 pH Reference Faraday Cage Simulation Measured Change only R v and L v to evolve from reference to Faraday cage structure R2R2 R2R2 R1R1 C1C1 R3R3 C3C3 R3R3 C3C3 C pad R1R1 C1C1 RvRv LvLv 100-µm transmitter-receiver separation Reference Faraday Cage Simulation Measured 100-µm pad separation Simulation Measured Faraday Cage Reference 100-µm 200-µm 100-µm 200-µm Conclusions C 1 = 2C r R 1 = RrRr 2 Add series R v and L v of via Model unchanged by split R 1 = 2.5 k  C 1 = 6 fF R v =45 Ω L v =30 pH R v =20 Ω L v =30 pH