In many industries, the control of fine dust or other foreign matter plays a vital and decisive role in the performance and quality of the products produced. The areas involved include semiconductor production, LCD screen production, biochemical engineering, biotechnology, Precision instruments, OEM industries, pharmaceuticals, hospitals and other industries, especially Semiconductor wafer manufacturing processes have the highest requirements for dust-free levels , because today’s wafers have developed to the nanometer level, any particle of dust falling on the wafer may cause damage to the entire wafer circuit, or even cause the wafer to be scrapped. The previous efforts have paid off. The flow of water may affect the quality and capabilities of the overall product, and thus affect the external view and evaluation of the company. Although it sounds like a small thing, it has a wide impact, so we have to be careful.
With the advancement of the times, product production time and development period have also been greatly reduced. This also means that the industry must be more accurate in risk assessment and production line monitoring to detect loss of control of the production line early and reduce possible losses. . Under this trend, How to quickly discover abnormal products and try to catch defective products in the early stages of production to avoid producing products that do not meet specifications is also a top priority faced by production lines and quality management departments in various industries. .
JMP provides many dust analysis tools to help customers analyze dust data and find potential problem points. From preliminary data drawing, to integrating data for simple statistical analysis, or subsequent use of statistical tools for in-depth analysis of abnormal projects, JMP can provide perfect support. Today I will share how to use JMP to conduct semiconductor fallout analysis. .
Data drawing and preliminary graphic analysis
JMP's Graph builder platform can provide a fast observation platform to draw data into graphics, and can draw corresponding point sizes according to the particle size of dust points, and use gradient colors to distinguish particle sizes, allowing you to understand the actual situation more quickly and intuitively status.
Below we use the data of dust falling on the panel as an analysis case illustration.
Figure 1 uses Graph builder to create a scatter diagram of falling dust, and uses point size and color to indicate particle size.
JMP also has many interactive chart functions that allow you to observe the relationships between factors by adjusting the graphics, and even help you find key influencing factors. The Local data filter under each platform can filter the categories you want to discuss based on the variables you set. Using this function, we can confirm whether there are different allocations between different panels.
Figure 2 Graph builder cooperates with local data filter to filter data
Using the column switcher function, we can quickly switch the discussed parameters on the Graph builder platform, and quickly find out the factors with greater influence. In other words, this factor can better distinguish the data.
Figure 3 Graph builder cooperates with Column Switcher to switch variables
If you want to compare the proportion of dust particles in different lots or panels, you can also use a pie chart to compare. Take the following figure as an example. If we want to compare the proportion of dust particles in panel No. 07 to be closer to 06 or 12, you can Drawing a simple pie chart shows that the proportion of dust particles in Panel 07 is closer to that of Panel 12.
Figure 4: Pie chart comparing the similarities and differences in the proportion of dust
For a small amount of data, we can draw a pie chart to observe the dust situation and find panels with similar dust ratios. But when the amount of data is slightly larger, are there other ways to find problematic panels?
Group the data into groups with similar proportions
In order to solve this problem, JMP provides many simple platforms to help users analyze anomalies in larger data. First, we first try to use JMP to perform grouping according to the proportion of dust.
In order to simplify the calculation, we first do a simple grouping of particle size, and use JMP's distribution platform to find the overall distribution percentile data, as shown below, we based on the 75th percentile, 90th percentile, 97.5th percentile, 99.5th percentile The quantile is used as the dividing point to divide the material into fine particles, small particles, medium particles, large particles, and ultra-large particles.
Figure 5 uses Distribution for narrative statistical analysis
Next, use Hierarchical Clustering to divide the panel into According to the proportion of particles, we can find that JMP Group those with similar proportions into the same group , with the help of JMP, we classified the overall panel into three categories. We can find that category one seems to have a ring-shaped dust falling trend, while category three seems to have some special large particle points and clusters on the upper side and lower left corner. When agglomeration occurs, and more samples are collected and included in the follow-up, these data can be used as a panel group to judge which dust ratio the new material is closer to. Combined with the subsequent production line results, it can be pre-detected in the front-end process. Identify potential opportunities for defective products to avoid subsequent production of defective products.
Figure 6 uses Hierarchical Clustering to classify according to the proportion of dust.
After understanding how to group samples with the same trend, we further need to clarify which panels have special dust problems.
use JMP Analyze abnormal dust fallout data
In order to find those panels with special dust fallout conditions, we first simply divide abnormal dustfall into several items.
- Large Particles - Relative to the overall dust particle size data, the panel has a dust particle size that is much higher than the overall dust particle size data
- Large quantity - Compared with the overall dust quantity data, the panel has a much higher dust quantity than the overall one.
- Clustered dust - Compared with the overall dust data, the panel dust has a higher concentration in certain areas.
JMP Explore Outliers on The platform can quickly search for the outliers we set , based on the fallout case, we use the Quantile Range Outliers interface to find that Tail Quantile is 0.005, and Q is 4, and set the outlier point to be greater than [percentile 0.995+4*(percentile 0.995-percentile 0.005 )] or less than [percentile 0.005-4*(percentile 0.995-percentile 0.005)], it can be found that the upper limit is 541. We regard particles that exceed it as large particles and need to mark them.
Figure 7 Use Explore Outliers to filter outliers
Using Graph builder or Tabulate we can find that Panel 4~12 all have large particles.
Figure 8 uses Graph builder and Tabulate to display the number of large particle dust in each panel.
If you want to select panels with a large amount of dust particles, use Tabulate to sort out the number of dust particles on each panel, and then use the Explore Outliers platform to select panels with a larger amount of dust particles. In this example, we will think that the three panels 06, 07, and 02 have a lot of dust problems.
Figure 9 uses Tabulate to sort the amount of dust and uses Explore Outliers to filter panels with a large amount of dust.
Next we have to deal with the problem of clustered particles. JMP's Normal mixtures platform provides the ability to determine whether the data points are general random scattered points or special concentrated points. JMP divides the data into several categories based on the degree of point aggregation. It is somewhat similar to the method of memorizing pictures, using general random scattered points as the background, and then classifying the remaining points into groups. In this example, we divide all the data after excluding noise points into 5 groups, and then draw them with Graph builder. We can find that there are ring-shaped clustered particle points at 02, 06, 07, 09, and 12, especially at 02, 06 , 07 is particularly obvious on the three panels, while 05, 10, 11, and 08 have some clusters of points in the upper and lower left corners, 04 has two clusters on the left, and the clusters of 21~30 are not so Obviously, most of them are small-scale cluster points or only one cluster point. Such results also correspond to the previous results of Hierarchical Clustering.
Figure 10 uses Graph builder to draw dust classification after Normal mixtures classification (distinguished by color)
Dust analysis conclusion
Based on the previous analysis and using JMP to organize, the following results are obtained.
Figure 11 Analysis and finishing table of dust problem
Using statistical analysis and related interactive graphical tools on JMP, we can quickly and accurately understand the entire data. Graphical presentation also helps us find problem points more quickly and cooperate with JMP Convenient and highly adjustable tools can help users quickly identify key problem points. In this example, ring particle clustering seems to be the main problem. Using JMP, we narrow the problem to a few specific Panels. We can then do subsequent analysis to confirm whether the problem is concentrated on a certain equipment or personnel operation. etc. Oriented. By using JMP, you can be faster, save time and effort, and maintain the quality of your products with less money.
Recommended reading:
Learn how to draw wafer diagrams with JMP in 7 minutes
How to quickly identify wafer defects? Data analysis under high-tech semiconductors
How to perform reliability testing on IC products? JMP Reliability Acceleration Analysis Application
