Atomic Force Probing (AFP) Lab

Atomic Force Probing (AFP) is a high-resolution electrical characterization technique that deploys multiple conductive AFM cantilevers or sharp tungsten nanoprobes to land directly onto the individual contact plugs (CA/M0) of a single failing transistor inside an integrated circuit. Once non-destructive fault isolation (such as OBIRCH or photon emission) localizes a defective cell, AFP measures the individual transistor's full IV characteristics (Ids-Vds, Ids-Vgs, gate leakage) to isolate anomalies like junction leakage, punchthrough, or contact voiding.

At room temperature (300 K), the fundamental thermionic emission limit for a field-effect transistor's subthreshold swing is SS = ln(10) * (kB * T / q) * (1 + Cdep / Cox) ≈ 59.6 mV/decade (commonly rounded to 60 mV/decade for an ideal body factor m = 1). Subthreshold swing values exceeding 85 to 110 mV/decade in sub-10nm FinFETs indicate severe short-channel degradation, sub-surface punchthrough, or interface trap density accumulation.

In sub-5nm technologies, the intrinsic resistance of contact plugs and probe tips can range from hundreds of ohms to tens of kilohms. A standard 2-point measurement forces current and measures voltage through the same lines, conflating probe wire resistance, tip-sample contact resistance, and transistor channel resistance. A 4-point Kelvin configuration routes high-impedance sense lines independently of current-carrying force lines, ensuring zero current flows through the measurement probes and eliminating parasitic contact voltage drops.

A 6T SRAM cell relies on two symmetric cross-coupled inverters to hold a binary bit. An asymmetric threshold voltage (Vth) shift on one pull-up PMOS or pull-down NMOS distorts the dual voltage transfer curves (VTCs), compressing the characteristic butterfly curve. This collapses the Static Noise Margin (SNM), making the cell vulnerable to random thermal noise, read disturb flips, or data retention failures at low operating voltages.

A premier multi-probe AFP or SEM nanoprober system represents a capital investment of $1.25M to $1.65M with annual probe tip and vacuum maintenance expenses of $65,000. By delivering conclusive root-cause electrical isolation on advanced-node test chips and qualification failures, the tool eliminates costly mask re-spins and accelerates yield ramps, delivering $2.8M to $4.2M in annual yield benefits, an internal rate of return exceeding 60%, and a discounted payback period of 1.3 to 1.6 years.