KalphaTech
Case study Exsolidus® Screening Solid-form screening

Catching a trace polymorph across 12,000 screening patterns

In a simulated high-throughput screen, Exsolidus flagged every pattern with 1 wt% or more of an unwanted polymorph, reached down to 0.2 wt% with longer scans, and raised almost no false alarms.

Published · 4-minute read

At a glance

  • ChallengeIn polymorph screening experiments, detect new phases, or those that differ from a reference.
  • What we didProcessed about 12,000 simulated laboratory patterns with Exsolidus, masked the main form and searched what was left for impurity peaks.
  • Result100% of patterns flagged at 1 wt% and above. Detection limit down to 0.2 wt%. False positives at most 3%.

The challenge

A solid-form screen can produce hundreds or thousands of PXRD patterns. Most show the expected form. The ones that matter carry a little of something else: a second polymorph at a fraction of a percent, with peaks that sit among those of the main form.

Reviewing every pattern by eye does not scale, and on laboratory instruments the weakest signals are blurred into the background. The question was simple: how small an amount of an unwanted form can Exsolidus flag automatically, and how often does it raise a false alarm?

Data and conditions

Material
Model organic compound with two known polymorphs: 4-(tosylamino)benzoic acid
Dataset
Six simulated datasets of about 2,000 patterns each; second polymorph at 0, 0.1, 0.2, 0.3, 0.5, 1, 2, 3, 5 and 10 wt%
Measurement
Laboratory reflection profiles, 2θ 3–34°, six counting times from very short (scale 25) to long (scale 800)
Analysis
Exsolidus aberration removal, reference masking, peak search and profile-fit validation

Our approach

  1. Remove aberrations. Patent-pending technology turn each laboratory pattern into a sharp, near-noiseless one, so weak peaks stand clear of the background.
  2. Mask the main form. A reference pattern of the pure main form, supplied by you or inferred from the dataset, marks every region its peaks occupy.
  3. Search what is left. In the free regions, candidate impurity peaks are located in every pattern.
  4. Confirm before flagging. Each candidate is checked by profile fitting. A pattern is flagged only when at least one candidate passes.
Low-angle region of a simulated laboratory pattern with 0.3 wt% of the second polymorph. The raw input shows a sloping background; the Exsolidus output is sharp and reveals a small impurity peak at about 4.5 degrees 2-theta.
Figure 1. Scale 100, 0.3 wt% of the second polymorph, low-angle region. Blue: raw laboratory input. Red: Exsolidus output. The impurity peak at 2θ ≈ 4.5° is barely visible in the raw pattern and clear after processing.

Results

  • 100%of patterns flagged at 1 wt% and above, at every counting time
  • >90%flagged at 0.5 wt%, for all but the shortest scans
  • 0.2 wt%detection limit with the longest scans; 0.5 wt% with the shortest
  • ≤3%false positives on pure patterns; zero at the two longest counting times
Line chart of the share of patterns flagged against impurity concentration for six counting times. All counting times reach 100% at 1 wt%; longer counting times flag more patterns at 0.2 and 0.3 wt%.
Figure 2. Share of patterns flagged at each impurity level and counting time. The detection limit is the concentration at which half the patterns are flagged.

Sensitivity follows counting time. Longer scans move the detection limit from 0.5 wt% down to 0.2 wt%, so you can choose the scan length that matches the reporting level you need.

The detections are also consistent. Across hundreds of patterns and every impurity level, confirmed peaks cluster at the same positions, led by the impurity’s strongest low-angle peak at 2θ ≈ 4.5°. The few candidates near the main form’s peaks failed validation and were correctly rejected.

Bubble chart of confirmed impurity peak positions across all scale 100 patterns, drawn above the reference pattern. The largest cluster sits at about 4.5 degrees 2-theta; a few small gray bubbles mark rejected candidates.
Figure 3. Scale 100: confirmed impurity peaks across the whole dataset (green) and rejected candidates (gray), above the reference pattern. Bubble size shows how many patterns share each position.

What it means for your screen

Exsolidus Screening turns thousands of patterns into a short list. Patterns with a trace second form are flagged automatically and consistently, so your scientists spend their time on the wells that changed, not on the hundreds that did not.

It works on the laboratory data you already collect, with a reference form you choose, or one inferred from the screen itself. Flagged patterns can go straight on to identification, structure determination or quantification in other Exsolidus modules.

Exsolidus can save your time

If you perform screening experiments and Send us a set of screening patterns and your reference form. We will show you which ones Exsolidus Screening flags, and why.

Discuss your screening data

More case studies

All resources