Chip Failure Analysis Method Flow

Chip Failure Analysis Method Flow

This paper introduces the methods and processes of chip failure analysis, gives examples of typical failure case processes, summarizes the challenges and countermeasures of key technologies of chip failure analysis, and summarizes the precautions for chip failure analysis.

Chip failure analysis is a systematic project, which needs to be combined with various means such as electrical testing, physical analysis, and material characterization to gradually narrow down the scope of the problem and finally locate the root cause of the failure. The following is a detailed description of the typical analysis process and key methods:

Preliminary Information Collection and Failure Phenomenon Confirmation

1. Lapse background checks

Collect chip model, application scenario, failure mode (such as short circuit, leakage, abnormal function, etc.), failure ratio, and use environment (temperature, humidity, voltage), etc. Confirm whether failures are reproducible and distinguish between design flaws, process issues, or improper applications (e.g., overvoltage, ESD).
2. Verification of electrical performance
Replicate failures using an automatic test device (ATE) or probe station (Probe Station) to record critical parameters (e.g., I-V curves, leakage current, threshold voltage offset). Compare the differences in electrical characteristics between good products and failed chips to reduce the failure area (e.g., specific functional modules).

Non-destructive testing(NDA)

Objective: Initially locate the problem and avoid destructive operations interfering with subsequent analysis.

X-ray Imaging: Inspect the package for defects such as wire bonding, solder ball connection, delamination, etc. 3D X-ray tomography (CT): 3D reconstruction of the internal structure of the chip to identify microcracks and voids (as shown in Figure 1).

Thermal Imaging scans the temperature distribution on the surface of the chip after power-on and locates abnormal hot spots (such as short circuit areas).

3. Acoustic microscopy (SAM) uses ultrasound to detect interface defects such as delamination and cracks inside the package (especially effective for molded devices).

0021-02983 TXZ Inner Shield

Destructive physical analysis(DPA)

Goal: Penetrate deep into the chip to observe microstructural defects.

Decapsulation: Use acid (such as fuming nitric acid) to dissolve the epoxy resin package and expose the surface of the chip (the corrosion time needs to be controlled to avoid damaging the metal layer). Laser decapsulation: Localized precise removal of high-density packages (e.g., Flip-Chip).

Cross-Section: Prepare a cross-section of a specific area using a focused ion beam (FIB) or mechanical grinding to cut the chip. Scanning electron microscopy (SEM) was used to observe the profile, and the metal layer fracture, through hole void, gate oxygen breakdown, etc. (e.g., fracture caused by electromigration of metal wires) were detected.

3. Energy Dispersive Spectroscopy (EDS) for Material Composition Analysis: Analyze elemental composition at the point of failure and identify contamination (e.g., corrosion caused by Cl⁻ ions). Secondary Ion Mass Spectrometry (SIMS): Detection of trace impurities (e.g., leakage due to Na⁺ migration).

Circuit-level Failure Location

Goal: Locate faults at the transistor or circuit node level.

Photon emission microscopy (EMMI) detects the weak photon emission in the failure area when energized, and locates the precise location of leakage or short circuit.

Laser-Induced Voltage Change (OBIRCH) The laser scans the surface of the chip, monitors the resistance change, and locates the high impedance or break point.

3. Electron beam flaw detection (EBT) uses electron beams to excite changes in the internal potential of the chip and analyze circuit node anomalies.

0021-35749 Ring, Isolator TXZ, 200 Mm, REFURBISHED

Comprehensive diagnosis and root cause analysis

1. Data correlation integrates the results of electrical testing, physical analysis, and material characterization to verify the consistency of the failure mechanism (e.g., electrical migration leads to an increase in resistance, and SEM confirms that the metal wire becomes thinner).

Failure Mechanism ModelConstruct a failure model based on phenomena, e.g., Hot Carrier Injection (HCI): Gate oxygen damage leads to threshold voltage drift. Electrochemical Migration (ECM): Migration of metal ions in the presence of humidity to form conductive filaments.

3. Suggestions for improvement include design optimization (e.g., adding ESD protection circuitry), process improvement (e.g., optimizing metal deposition temperature), or application condition adjustment (e.g., reducing operating voltage).

An example of a typical failure case process

Case: A power management chip fails in batches at high temperatures

Electrical test: The leakage current increases abnormally at high temperature, and it is locked to an LDO module.

X-ray CT: Microcracks were found in the solder balls inside the package.

FIB/SEM profile: Confirm that the crack causes poor contact of the power line, and the thermal stress increases at high temperatures.

EDS analysis: sulfur contamination at the solder ball interface (from the molding material).

Conclusion: The sulfur element of the packaging material causes the corrosion of solder joints, and the problem is solved after improving the packaging process.

Key technical challenges and countermeasures

Challenge	Solution
Nanoscale defect detection is difficult	High resolution SEM/TEM (TEM)
Multi-layer stacked chips are complex to analyze	Combining FIB layer-by-layer etching and 3D reconstruction technology
Soft failures (intermittent failures) are difficult to reproduce	Use dynamic signal analysis(DSA)

Precautions analysis sequence:

Strictly follow the principle of "non-destruction before destruction" to avoid the loss of key information. Sample protection: After opening, the surface passivation treatment (such as gold plating) should be carried out in time to prevent oxidation from affecting the observation. Data cross-validation: A single method may have misjudgments, and multi-technology joint verification is required. Chip failure analysis is like "solving a case", which requires rigorous logic and diverse means, combined with the progressive strategies of "macro→ micro", "electrical → physical properties", and finally realizes closed-loop failure management.