# Maximized Yield with 62% Fewer Experiments—Electrochemical R&D, Reinvented

#### Quick summary
Explore how AI is helping optimize freeze drying experiments and process conditions in pharmaceutical research.

##### Industry
Industrial Chemical R&D

#### **Client**  
A U.S. pilot chemical plant conducting advanced electrochemical process development. An innovation-focused leader in industrial-scale R&D, operating at the intersection of sustainability and precision chemistry.

#### **Challenge**  
Traditional optimization was slow, resource-heavy, and unable to consistently reach optimal yield. With 24 experiments over 6 weeks, the best achievable performance plateaued at 74%. Time and materials were being consumed at a rate that hindered project velocity.

#### **Goal**  
To reach the maximum possible yield in the anodic oxidation of DMP with fewer experiments, faster cycle times, and improved prediction accuracy to accelerate overall R&D timelines.

#### **Approach & Solution**  
SuntheticsML deployed its small-data ML modeling framework using Bayesian Optimization. With just 9 experiments (down from 24), the model rapidly converged toward the global maximum, leveraging pattern recognition to guide decisions and narrow down ideal process conditions.

#### **Results & Metrics**  
- **Yield increased from 74% to 93%** — a 19 percentage point improvement  
- **Number of experiments reduced from 24 to 9** — a **62.5% reduction**  
- **Time to results shortened from 6 weeks to 1.5 weeks** — **75% faster**  
- Identified ideal operating ranges for:  
  - **Current density**  
  - **DMP concentration**  
  - **Flow rate**  
- ML-guided predictions honed in on global maximum with minimal experimentation

Additional insight: SuntheticsML recommended optimal ranges for current density, DMP concentration, and flow rate, identifying the sweet spot that would otherwise have been prohibitively time-consuming to find through traditional methods.

#### **The Sunthetics Edge**  
_SuntheticsML allowed to bypass months of experimentation. The model's suggestions led directly to a 93% yield—faster and with fewer resources than ever before._

#### **Key Takeaways**  
- **62% fewer experiments** to reach optimal conditions  
- **75% time reduction** from problem to solution  
- **Bayesian Optimization** efficiently navigates complex chemical landscapes  
- Enables **lean, agile R&D**—ideal for constrained or high-cost environments