Semiconductor package inspection with predictive model for wirebond radio frequency performance
Methods, systems and devices for use in semiconductor package manufacture of a package having a die and die substrate are provided. A method for use in semiconductor package manufacture includes steps of forming one or more wirebond interconnections between the die and the die substrate, capturing input data representative of wirebond interconnection features during inspection of the formed wirebond interconnections, and passing the captured input data to a machine learning (ML) engine. The method further includes processing the captured input data with the machine learning engine using a trained model to obtain an output array of data, evaluating the output array of data to determine a predicted radio-frequency (RF) performance rating, and outputting the predicted RF performance rating. Training data set processing may include applying image data and parameters to a multi-layer neural network to obtain a set of candidate ML models, and selecting an optimal ML model.
1. A method of inspection of a semiconductor package having a die and die substrate during semiconductor package manufacture, comprising:
forming one or more wirebond interconnections between the die and the die substrate;
capturing input data representative of one or more wirebond interconnection features during the inspection of the one or more formed wirebond interconnections, the input data comprising one or more digital images captured of the one or more formed wirebond interconnections;
processing the captured input data using a machine learning (ML) model to obtain an output array of data having a predicted value indicative of a radio-frequency (RF) performance of the semiconductor package with the one or more formed wirebond interconnections, wherein the ML model is trained based on wirebond parameters and RF performance parameters;
evaluating the output array of data to determine an RF (RF) performance rating for the semiconductor package;
outputting the RF performance rating; and
rejecting or passing a manufacture of the package according to the output RF performance rating.
2. The method of claim 1 , further comprising:
rejecting the manufacture of the package according to the output RF performance rating.
3. The method of claim 2 , wherein the rejecting of the manufacture of the package according to the output RF performance rating occurs during or after wirebonding assembly of the semiconductor package manufacture.
4. The method of claim 1 , further comprising:
generating an alert based on the RF performance rating.
5. The method of claim 4 , wherein the generating the alert occurs during wirebonding assembly in the semiconductor package manufacture.
6. The method of claim 1 , wherein the capturing the input data comprises capturing the one or more digital images with optical or infrared camera devices positioned relative to the package such that package regions having the one or more formed wirebond connections are within fields of view of the optical or infrared camera devices.
7. The method of claim 6 , wherein the capturing the input data comprises capturing distance data with a sensor system positioned relative to the package, the distance data including a set of data points representing distance values between the package regions having the one or more formed wirebond connections and the sensor system.
8. The method of claim 1 , further comprising:
storing a training dataset in computer-readable memory; and
training an ML algorithm based on the training dataset to obtain the trained model.
9. The method of claim 8 , wherein the ML algorithm is a multi-layer neural network algorithm, and the training the ML algorithm includes:
applying image data and parameters to the multi-layer neural network algorithm having feature extraction and classification to obtain a set of candidate ML models; and
selecting one of candidate ML model from the set of candidate ML models for use as the ML model.
10. The method of claim 9 , wherein the training dataset includes the wirebond parameters, the RF performance parameters, and RF rating labels generated from prior wirebond assemblies.
11. The method of claim 10 , wherein the wirebond parameters include the following parameters for each wirebond: wire bond planar length, wire bond curvilinear (3D) length, wire bond loop height, wire bond diameter, and angle of wirebond from a bird's eye view relative to a plane.
12. The method of claim 10 , wherein traces are coupled to the wirebonds in matching sections and the wirebond parameters include one or more of the following parameters: wire bond planar length, wire bond curvilinear (3D) length, wire bond loop height, wire bond diameter, number of wire bonds, rough spacing between neighboring wire bonds, angle of wirebond from a bird's eye view relative to a plane, trace widths of matching section, or trace length of matching section.
13. The method of claim 10 , wherein traces are coupled to the wirebonds in matching sections and the wirebond parameters include the following parameters: wire bond planar length, wire bond curvilinear (3D) length, wire bond loop height, wire bond diameter, number of wire bonds, rough spacing between neighboring wire bonds, angle of wirebond from a bird's eye view relative to a plane, trace widths of matching section, or trace length of matching section.
14. The method of claim 10 , wherein the training dataset further includes synthetic data relating to RF performance obtained from modeling software.
15. The method of claim 10 , wherein the ML model comprises a trained neural network model and the ML engine includes an inference stage, and the processing the captured input data includes classifying the input data in the inference stage with the trained neural network model to obtain the output array of data.
16. The method of claim 15 , wherein the RF performance rating identifies an RF performance level from among multiple performance levels.
17. An inspection system for use in manufacturing a semiconductor package having a die, a die substrate, and one or more wirebond interconnections formed between the die and the die substrate, comprising:
imaging sensors to capture input data representative of wirebond interconnection features during inspection of the one or more formed wirebond interconnections, the input data comprising one or more digital images of the one or more formed wirebond interconnections; and
a radio-frequency (RF) performance predictor tool operating on a computing device configured to:
process the captured input data with a machine learning (ML) engine having a ML model trained based on wirebond parameters and RF performance parameters to obtain an output array of data having a predicted value indicative of a radio-frequency (RF) performance of the semiconductor package with the one or or more formed wirebond interconnections;
evaluate the output array of data to determine an RF performance rating for the semiconductor package,
output the RF performance rating; and
reject or pass a manufacture of the semiconductor package based on the output RF performance rating.
18. The system of claim 17 , wherein the RF performance predictor tool is further configured to reject the manufacture of the semiconductor package based on the output RF performance rating.
19. The system of claim 17 , further comprising an alert generator operating on the computing device configured to generate an alert based on the RF performance rating.
20. The system of claim 17 , wherein the alert generator is configured to generate an alert based on the RF performance rating during wirebonding assembly of the semiconductor package manufacture.
21. The system of claim 17 , wherein the imaging sensors include one or more optical or infrared camera devices to capture the one or more digital images and positioned relative to the semiconductor package such that package regions having the one or more formed wirebond connections are within fields of view of the optical or infrared camera devices.
22. The system of claim 21 , wherein the imaging sensors are positioned relative to the semiconductor package and configured to capture distance data including a set of data points representing distance values between package regions having the one or more formed wirebond connections and the sensor system.
23. The system of claim 17 , further comprising:
a computer-readable memory configured to store a training dataset; and
at least one processor configured to process the training dataset with the machine learning engine to train an ML algorithm to obtain the ML model.
24. The system of claim 23 , wherein the ML algorithm is a multi-layer neural network algorithm and the at least one processor is configured to during the training to:
apply image data and parameters to the multi-layer neural network algorithm having feature extraction and classification to obtain a set of candidate ML models, and
select one of candidate ML model from the set of candidate ML models for use as the ML model.
25. The system of claim 23 , wherein the training dataset includes the wirebond parameters, the RF performance parameters, and RF rating labels generated from prior wirebond assemblies.
26. The system of claim 25 , wherein the wirebond parameters include the following parameters for each wirebond: wire bond planar length, wire bond curvilinear (3D) length, wire bond loop height, wire bond diameter, and angle of wirebond from a bird's eye view relative to a plane.
27. The system of claim 25 , wherein traces are coupled to the wirebonds in matching sections and the wirebond parameters include one or more of the following parameters: wire bond planar length, wire bond curvilinear (3D) length, wire bond loop height, wire bond diameter, number of wire bonds, rough spacing between neighboring wire bonds, angle of wirebond from a bird's eye view relative to a plane, trace widths of matching section, or trace length of matching section.
28. The system of claim 25 , wherein traces are coupled to the wirebonds in matching sections and the wirebond parameters include the following parameters: wire bond planar length, wire bond curvilinear (3D) length, wire bond loop height, wire bond diameter, number of wire bonds, rough spacing between neighboring wire bonds, angle of wirebond from a bird's eye view relative to a plane, trace widths of matching section, or trace length of matching section.
29. The system of claim 25 , wherein the training dataset further includes synthetic data relating to RF performance obtained from modeling software.
30. The system of claim 25 , wherein the ML model comprises a trained neural network model and the ML engine includes an inference stage, and the processing the captured input data includes classifying the input data in the inference stage with the trained neural network model to obtain the output array of data.
31. The system of claim 17 , wherein the RF performance rating identifies an RF performance level from among multiple performance levels.
32. The system of claim 17 , further comprising an ML engine coupled to the RF performance predictor.
33. The system of claim 32 , wherein the ML engine includes an inference stage, a training stage, and a multi-layer neural network, the multi-layer neural network corresponding to the ML model.
34. The system of claim 33 , wherein the multi-layer neural network includes a convolutional neural network (CNN).