Start with the observation
Growth B uses the same nominal recipe as Growth A, a reference with an acceptable measured outcome. Both have process logs and an in-situ thickness trace. The unexpectedly low final thickness in Growth B is the observation to explain. Open Analysis, select New Notebook, and enter:Investigate why Growth B reaches 32 nm while reference Growth A reaches 40 nm under the same nominal recipe. Compare the available process logs and thickness measurements. Check whether the runs and measurements are comparable, identify where the behavior first diverges, and show evidence for competing explanations.Click Create Notebook. Confirm the intended growths and measurements are included. See Creating a Notebook for the full instructions, or Choose a Reference Growth if you need to select a reference first.
Align the process before comparing the traces
Both runs include preparation before deposition begins. Set time zero to the recorded deposition start in each run, then keep actual elapsed minutes on the horizontal axis.
The same duration with a lower thickness gain points toward a difference during deposition or in its measurement. Matching setpoints alone do not establish that the process conditions were identical.
Ask the notebook to retain measurement units, model settings, and the original timestamps. Normalizing every run to a 0–1 time axis could hide a duration difference in another comparison; use the alignment that fits your question.
Find where the behavior first diverges
Fictional thickness gains aligned to deposition start. Growth B initially follows the reference, then gains thickness more slowly after minute 4.
Focus on minutes 3–6 after deposition starts. Overlay the recorded flow, pressure, temperature, and process events for both growths. Check clock alignment and measurement gaps before comparing event order.
Follow the difference back to its context
In this fictional example, the selected flow channel has a setpoint of 20 sccm in both runs. Its readback remains at 20 sccm in the reference but falls to 15 sccm after minute 4 in Growth B.
The flow signal is a lead, not a demonstrated root cause. The notebook should distinguish the commanded value from the measured readback, and a change in readback does not itself establish a change in material delivery.
Test the explanation against alternatives
Could a thickness-model change or a flow-sensor problem explain the same observations? Compare the measurement settings and event logs, and identify an independent measurement that would distinguish these possibilities.Check the measured temperature as well as its setpoint. If the relevant logs or measurement settings are unavailable, keep that gap in the conclusion. An independent final-thickness measurement can help test the thickness trace. An instrument check under your team’s procedures can help assess the flow readback. Neither requires treating the first correlation as the answer.
Leave a useful handoff
For this example, record:Growth B gains 32 nm over the same 20-minute deposition interval in which reference Growth A gains 40 nm. The thickness traces diverge after minute 4, near a decrease in the flow readback while its setpoint stays constant. Check independent final thickness and the flow measurement before attributing the lower growth rate to material delivery.Save the notebook with its alignment, source measurements, and unresolved checks. Record confirmed findings on the sample or project so colleagues can distinguish observations from hypotheses.
Adapt the investigation
If the difference is a short event within one run, use Investigate an Unexpected Change. To explore relationships across a campaign, see Relate Growth Conditions to Film Properties.