SandBox Studio™ Suite
SandBox Studio™ accelerates product delivery by enabling engineers to rapidly predict process outcomes using physics based, statistical, and machine learning modelsSign Up for 15 day free trial
Simulate complex 2D and 3D stack profiles to predict process outcomes
Architect is an advanced 2D and 3D stack design tool which acts as the starting point for all your Sandbox Studio™ projects. It eliminates key bottlenecks in your recipe development process by making it effortless to define your recipe requirements and monitor all your recipe targets be it for an across-wafer or topography optimization.In Architect, you can easily import the incoming and target profile of your stack. Architect automatically detects the materials in your stack based on color by pulling from your material database allowing for convenient material specification.
Using your target profile, you can interact with your stack and directly tell SandBox Studio™ the mission critical features you need to optimize in your stack. In less than five minutes you can completely define all your process requirements.
- Visually define the incoming and target profiles of your stack
- Specify the materials in your stack quickly and conveniently
- Interactively identify your critical recipe targets
Optimize Incoming Profile
When optimizing critical dimensions for a stack in Architect, you begin by importing the profile of what the stack looks like before the etch or deposition process takes place. What is unique to the Architect is that you can easily define asymmetrical or rounding characteristics in your incoming profile. SandBox Studio™ will help you correct for any of the unwanted characteristics as you optimize your etch process. Architect also allows you to import a profile for as many “slices” or features that you are optimizing at any magnification that you wish. Using Architect’s built in scale bar tools, you can set the scale of your incoming profile by directly measuring the image.
Specify Material Properties
After importing your Incoming profile, Architect will automatically detect each color in the profile as a distinct material. Each colored material will show up in your material viewer. From this list, you can define each material in your stack. Architect will automatically identify materials already stored in your database.
Optimize Target Profile
For each slice you defined in Architect, you can import the target profile of the stack. The target profile is your “idealized” structure following your etch or deposition process. Using Architect’s built in measuring tools, you can interact with the profile and easily identify your recipe targets such as trench height, bow width, sidewall angle, and much more. You can then enter the target criteria and tolerance specifications for each recipe target that you defined. The Architect makes of complex stacks optimization simple and straightforward.
Streamline your process development workflow
Oculus forms the backbone for all of SandBox Studio’s analysis and visualization applications. Connecting to data on-premise or in the cloud, its easy-to-use interface keeps track of process recipes and experimental data and helps you to maintain your organization’s legacy knowledge. Oculus solves the problem of repetitive data input and output allowing you to focus on what you do best: the science.
- Efficiently design complex, multistep or cyclic processes
- Input experimental data quickly and conveniently
- Export detailed analysis and repots and present your work beautifully
- Securely share your data across your team in real-time
The Recipe Builder is your playground for recipe development. Here you can view all the reactors in your fabrication facility with material compatibility to your stack. After selecting your reactor of choice, you can define your recipe exploration space and design complex workflows that can include multistep or cyclic processes. Each reactor in the recipe builder has an up to date catalogue of the reactor’s process knobs and available gases. You can specify windows for each gas or process parameter to carve out your parameter space.
Design of Experiments
The design of experiments (DOE) tool in Oculus will help you lower your development costs and speed up your learning cycles for recipe development. By helping you to quantify the relationships of your process parameter inputs, a DOE will enable you to quickly improve your model development. SandBox Studio™ offers several experimental designs including full factorial, Box-Behnken, and Central Composite methods.
Quickly identify target process windows based on built-in models
Analytics is a data-analytics and model optimization tool that will allow you to understand what physical and/or statistical models best fit your data.
The bigger picture for most machine learning workflows requires setting an objective, inputting the data, and then fitting models to the data/deciding which performs best. The diagram shows this process step by step, and the model fitting step on experimental data is what SandBox Analytics™ solves. Before one can even use Envision or apply a model, the Analytics step is crucial to understanding how your process performs.
- Make data-driven decisions that reduce process development cost by 3x and have a 3x faster time to market
- State of the art tools to make this decision without having to rely solely on anyone else. You will have the key metrics to understand what hidden insights are underlying your data.
- Analytics is part of a larger workflow that easily integrates into your process and has improvements for other aspects like data visualizations and process optimization.
Sandbox Analytics™ employs a variety of proprietary physical and statistical models that scale to the needs and complexity of our customers. A crucial part that we take into consideration is creating robust models which fit to your complex data sets while making sure they generalize to newer observations.
Analytics also uses physical models that give us an advantage over typical machine learning solutions. Our process modeling expertise in fabrication steps includes plasma etching and deposition, among others. To build a physical model in Analytics, simply reach out to our experts about your process development needs.
|General Analytics/Machine Learning Software||Hiring data scientist||Sandbox Analytics™|
|Strong understanding of process conditions|
|Powerful computational and modeling skills|
|Long-term growth and scalability|
|Cost Effective Solution|
Summary Statistics and Key Metrics
Analytics provides key insights into the performance of a model not just by showing how well it fits to the data you provided. We also display industry leading and heavily tested metrics that show how robust the models would be to new observations. We empower you to make these decisions without having to trust a black-box model or having to rely on anyone else.
Model Performance Insights
Once you have selected a model that optimizes your recipe targets you can apply that model to better understand your process and how it would perform under different conditions. This will enable you to reduce uncertainty about your process outcome.
You can also plug the you generate in Analytics seamlessly to our proprietary software tools in Envision that allow you to visualize the process output space and perform design of experiments.
Visualize multi-dimensional process spaces
Envision helps process engineers visually interact with their process models and data. It is designed to reveal the fundamental mechanisms governing your experimental data and enables you to uncover hidden process trends. By providing a simple and useful representation of complex, multi-dimensional process spaces, Envision significantly reduces the time you spend performing data analysis and helps increase your process insights even with fairly limited experiment sets.
- Intuitive visualization of multi-dimensional process spaces and etch fronts.
- Identify optimal process parameters to quickly and accurately meet your targets in multi-dimensional process spaces.
- Co-optimize of multiple process objectives and recipe targets.
In Envision, you can quickly generate beautiful, presentation-ready charts such as scatter plots, bar charts, and topography simulations. The model generated bar charts and scatter plots will help you compare your experimental and simulated results while the topography simulator will enable you to visualize the evolution of the etched structure over multiple time steps.
The most distinguished feature of Envision is Quilt®. Quilt® process maps are 2D representations of complex, multi-dimensional process spaces. They will intuitively help you to follow the data and easily discover nonlinear process trends. With Quilt®, you can generate Pass-Fail process maps to help you co-optimize multiple process objectives.
Optimal Recipe Prediction
Quilt® is generated by running thousands of simulations across the user-defined process pace. To dissect this process space, you can use Optimal Recipe Prediction (ORP), to carve our the best process space based on your recipe targets. ORP finds your three largest process parameter “pass region.” By maximizing the optimal process window in the simulated parameter space, ORP will help you not only achieve your recipe targets, but also ensure your process stability.