How JMP is Used in the Semiconductor Industry
*Internal Use Only*
A semiconductor is a type of material that enables a transistor to be made. A transistor is a very small, low power electrical switch which can be used to construct semiconductor devices.
Semiconductors are used in the manufacturing of various kinds of electronic devices, including:
- Computers
- Cell Phones
- Digital Cameras
- Appliances
- Televisions
- Cars
- Bank ATM’s
- Anything that is computerized or uses radio waves most likely has a semiconductor
Who uses JMP in the Semiconductor Industry?
- Seagate
Use JMP as an integral part of wide-ranging data analysis techniques, statistical process control, reliability analysis, experimental design and modeling.
- Polar Semiconductor
Apply JMP to automate and optimize the collection, preparation, analysis, reporting and workflows of engineering and process data.
- SK Hynix
Use JMP to create machine learning algorithms that can expedite the use of data in manufacturing processes.
- Soitec
Apply JMP to improve yield and product quality by providing a standard, easy-to-use tool for visualizing data, analyzing the manufacturing process, and identifying potential opportunities.
- NXP
Use JMP to transform chip testing and manufacturing processes
- Texas Instrument
What Problems do JMP users have?
- Reporting on lots of test data from wafer/module/system test
- Correlating fab process data to test data (linking disparate datasets together)
- Identifying test excursions quickly and identifying root cause
- Correlating design simulation data to silicon characterization data
What do they use JMP for?
Visualization and analysis of semiconductor data pose unique challenges due to its size and volume. When dealing with hundreds or thousands of variables, and possibly millions of observations, an analysis package that is fast, flexible, extensive, and extendable. JMP data analysis software from SAS allows scientists and engineers in semiconductor industry to function more effectively by providing them with a tool that is easy to use, contains hundreds of statistical routines from simple to complex, and is customizable using a built-in scripting language.
Semiconductor Equipment Manufacturers and Foundries/IDMs
- Important: DOE!, Fit Model, GB, Distribution, Fit Y by X, Hierarchical Clustering, Partition, Screening, Quality/Process Tools, Multivariate Methods, Reliability and Survival
- Pro: Neural, Bootstrap Forest, Gen Reg, FDE
- JMP Live is a good fit as they have A LOT of standardized and ad hoc reporting needs but the difficulty is many have already gravitated to specialized semiconductor related data warehousing and reporting solutions
- Best possible JMP customers, they need it all!
Fabless Companies (they don’t have FABs!)
- Important: EDA, Standardized Reporting of large test data sets (JMP Live), Fit Model for simulation and characterization data, Reliability and Survival
- Not that important: DOE and Process Control Tools
- They don’t utilize DOE much and unlikely to be a compelling reason to acquire JMP
- A pitch could be made for designers to use DOE when performing simulations and product engineers to use DOE for characterization test plans to correlate with simulation conditions
- Pro: Not as much need as IDMs but JMP Pro can be pitched for predicting Final Test results from Wafersort data (prediction for speed/power binning)
- This use case landed JMP Pro deals at both Nvidia and Xilinx
- JMP Live situation similar as IDM case above
OSATs
- Important: DOE, Fit Model, GB, Distribution, Fit Y by X, Partition, Screening, Quality/Process Tools, Reliability and Survival
- Pro: Not much Pro play here
- Very strong need for automated reporting solution but already using other industry specific solutions
Contract Manufacturers
- Important: GB, Distribution, Fit Y by X, Screening, Quality/Process Tools
- Not as important: Reliability and Survival
- Pro: Not much Pro play here
- Very strong need for automated solution but already using other industry specific solutions
Do we have any good customer stories?
There are several success stories that may be found by going to the Customer Stories section of the JMP website and filtering by industry. One strong customer story is NXP Semiconductors.
NXP Semiconductors increasingly depends on analytics to optimize chipmaking and their success story can be read here.
Do we have any demo material?
Certainly，our customer always interested in whether we have any demo materials in the semiconductor industry.
I will introduce an easy example which is accepted by many semiconductor customer.
This graph was produced by Graph Builder in JMP which described the shape of wafer.
Using the interactive function in JMP Graph builder, We can use drill-through and tag functions to get the details of each chip.
The right plot shows the test variable information of a chip corresponding to the wafer which can help you quickly diagnose the quality control status in the production process.
Sources used:
JMP Product Management Library --> Semiconductor Industry Overview
