Structure is clear: a single workbook `Book1` where each column (5µW → 100µW) is already a merged spectrum at that power. Energy axis (eV), `suggested_x_range = [1.23, 1.45]`. No Si/reference folder exists, so no background merge is needed.

Per protocol, I must despike before any fit. Doing that now:



Despike complete — 130 cosmic-ray spikes removed across all 15 power columns, data written back. The main PL peak sits near ~1.32 eV, well inside the suggested range [1.23, 1.45].

Now running the Lorentzian power fit — this fits every spectrum, extracts intensity + xc per power, builds both panels, and runs the allometric (I = a·P^b) fit on each:



![Power Fit]([file link removed])

## Power-dependent PL analysis — `powerdep_manuscript.opj`

All 15 merged spectra (5 → 100 µW) were despiked (130 cosmic-ray spikes removed), then each was fitted with a **Lorentzian + slope** profile over 1.23–1.45 eV. Every column converged in the **HIGH** tier (R² = 0.887–0.984, SNR 7.6–24.0).

**Intensity vs power — allometric fit I = a·Pᵇ**
- a = 16.28, **b = 1.0048 ± 0.0261**, R² = 0.995
- Interpretation: essentially **linear (excitonic)** — no saturation or bimolecular recombination over this range.

**Peak position (xc) vs power**
- Blueshifts monotonically from **1.3206 eV (5 µW)** to **1.3356 eV (100 µW)** — a total shift of ~15 meV (~5.0 meV/decade).
- FWHM narrows from ~62 meV (5 µW) to ~54 meV (100 µV), consistent with reduced thermal/broadening contribution at higher excitation.

Full per-power table (intensity, xc, FWHM, R²): [power_fit_results.csv]([file link removed])

RESULT: b = 1.0048 ± 0.0261