DOE Master Guide: other issues

From LabAutopedia

Jump to: navigation, search
<< Back to Pre-experimental planning
Invited-icon.jpgA LabAutopedia invited article

Authored by: James N. Cawse, Ph.D.;Cawse and Effect LLC


Interaction and Curvature

In experimental design, an interaction is a change in the effect of one factor on a response that is caused by a change in a second factor.

No Interaction Interaction between X1 and X2
Although increasing X2 affects the response, the effect of changing X1 remains the same The response to X1 is steeper as  X2 is increased.

This concept is not self-explanatory and is often confused or overlooked.  HT experimentation is all about interactions, and usually high-order interactions. A search for main effects does not need HT equipment! New and substantially enhanced properties will be found in higher-order interactions. A sense of the degree and type of interactions to be searched is important in deciding on the experimental strategy and designs to be used.


If some factors are hard to vary, it is common for them to be held constant for relatively large fractions of the design, rather than randomly change them at every run. This changes the error structure of the experiment and can cause major errors in determining the significance of factor effects. Typical symptoms are “batches” of materials, “furnace loads” in processing, and “plates” in high throughput designs. A common nested or “split-plot” situation is a formulation x processing experiment, where a master plate of different formulations is split into daughter plates, which are then subjected to different processing conditions.

Restrictions on the Experiment

The team should be quick to put all known limitations and pitfalls on the table.  
  1. Are there combinations of highs and lows of the factors that cannot be studied for safety or operability reasons?
  2. Are some control factors harder to change than others?  Temperatures are often very slow to move.
  3. What are the limitations on the data acquisition equipment or the ability of the lab to handle samples?
  4. Are some combinations known to give bad results?

Automation Issues

Typical automation issues that can have a strong influence on the conduct and outcome of the experiment include:

  1. The standard size of the experimental array, such as a 96-well plate.
  2. The sources, masks, and shutters of a system used for producing landscape libraries
  3. The capabilities of the robot used for charging the ingredients, including:
  • Minimum and maximum aliquot size
  • Accuracy of addition (as a function of aliquot size, viscosity, etc.)
  • Robot speed for completion of the array (which may affect sample stability, solvent evaporation, kinetics, etc.)
  • Ramp up, ramp down, and uniformity of any processing steps (temperature, pressure...)
  • Speed, precision, and resolution of the analysis system
  • Capability of the analysis system to perform in situ measurements.

Scale of Experiments

The typical scale of high throughput experiments in the materials arena ranges from plate-based (typically 96-well) to microreactor-based (typically 8 reactors as a unit). Plate-based experiments are typically used for wide-ranging screening while microreactors are more for optimization. The distinctions between the two are summarized in the following table[1].

Questions Plate-Based Screening Microreactor-based screening
Primary objective Discovery Optimization
Starting points (number and quality) Limited number or not well developed Numerous or well developed
Experimental Space Very large Large
Quality of data required Qualitative or semi-quantitative Precise or quantitative
Level of detail (process and product information) More limited Extensive
Throughput Very high High
Cost per experiment (after amortization of HT equipment cost!) Very low Low

Trial Runs

Most experiments involve people  (and sometimes machines) doing things that they have never done before.   Often some practice helps.  Learn and refine experimental procedures before starting a big job.


  1. Adapted from Keith A. Hall, "Multistage Screening in Catalyst Discovery and Optimization: Integration and Iteration", Symyx Symposium 2009, Philadelphia, May 12-14 2009.
Click [+] for other articles on 
The Market Place for Lab Automation & Screening  Automation Software