*2.5. Analysis by GC–IMS*

In the experiment, the volatiles were concentrated and separated by headspace solidphase microextraction, with reference to Wang [11] and other methods and appropriate adjustments. Precisely-measured 50 μL of turmeric volatile oil sample was transferred into a 20 mL headspace bottle with Teflon spacer seal. The headspace bottle was heated at 80 ◦C and incubated for 10 min at 500 RPM. Then 100 μL of the sample was injected in non-shunt mode, and the temperature of the injection needle was kept at 85 ◦C.

The components of volatile compounds were identified by chromatography–ion mobility spectroscopy (GC–IMS; FlavourSpec®, G.A.S., Berlin, Germany). Gas chromatography (GC) was performed under the following conditions: carrier gas, nitrogen (99.99%); column, mxt-5 (15.0 m length × 0.53 mm ID × 1 μm thickness); running time, 50 min; flow rate, initial 2.0 mL/min, holding for 2 min, linearly increasing to 100 mL/min within 18 min, and holding for 20 min. Ion mobility spectroscopy (IMS) was carried out under the following conditions: drift gas, nitrogen (99.99%); flow rate, 150 mL/min; IMS detector temperature, 45 ◦C.

Three parallel samples are set for each irradiation intensity for volatile oil, and the difference in the spectrum of volatile organic compounds in the sample can be given after analysis. The NIST database and IMS database built into the software can conduct a qualitative analysis of substances.

### *2.6. Statistical Analysis*

The analysis software Vocal matched with the instrument is used to view the qualitative and quantitative analysis spectrum and data. The NIST database and IMS database built into the application software can be used for qualitative analysis of substances. The porter plug-in directly compares the spectrum differences between samples (threedimensional spectrum, two-dimensional top view, and difference spectrum). We compared the fingerprint of the gallery plot plug-in to intuitively and quantitatively compare the differences in volatile organic compounds between different samples. A dynamic PCA plug-in was used for dynamic principal component analysis, cluster analysis of samples, and rapid determination of unknown and unknown samples.

### **3. Results**
