Does the solar oxygen abundance change over the solar cycle? An investigation into activity-induced variations of the O I infrared triplet

The determination of the solar oxygen abundance remains a central problem in astrophysics, as its accuracy is limited not only by models but also by systematics. While many of these factors have been thoroughly characterized, the effect of the solar activity cycle has so far remained unexplored. Due to its relative strength and accessibility, the O I infrared triplet is typically the primary choice for abundance studies. However, previous investigations have shown that abundances inferred from this triplet tend to be higher than expected on active stars, whereas such an overabundance effect is not observed for the much weaker forbidden O I 6300 Å line. This raises the question of whether a similar trend can be found for the Sun. To address this question, we analyze two decades’ worth of synoptic disk-integrated Sun-as-a-star datasets from the FEROS, HARPS-N, PEPSI, and NEID spectrographs, focusing on the infrared triplet (7772, 7774, 7775 Å) and the forbidden O I 6300 Å line. The excellent signal-to-noise ratio of the PEPSI observations allows us to detect a weak but significant variation in the equivalent widths of the infrared triplet, corresponding to about 0.01 dex difference in abundance between activity minimum and maximum. This value is significantly smaller than the typical uncertainties on the solar oxygen abundance. Due to higher scatter, no comparable trend is found in the other data sets. Based on these results, we conclude that within the typical uncertainties presented in other works, we can assume the inferred solar oxygen abundance to be stable across the solar cycle, but that this effect may be significant for other, more active stars.

💡 Research Summary

This paper addresses the long‑standing question of whether the solar oxygen abundance, A(O), varies over the 11‑year solar activity cycle. While the O I infrared triplet (7772, 7774, 7775 Å) is the workhorse for solar and stellar oxygen determinations, previous studies of active stars have reported systematically higher abundances from this triplet compared to the much weaker forbidden line at 6300 Å. The authors therefore set out to test whether a similar activity‑related bias exists for the Sun itself.

Four long‑term Sun‑as‑a‑star data sets were assembled: FEROS (2003‑2011, R≈48 000, S/N≈200), HARPS‑N (since 2015, R≈120 000, S/N≈400), PEPSI (since 2015, R up to 250 000, S/N≈1000) and NEID (since 2020, R≈110 000, S/N≈320). For each instrument the highest‑quality spectrum per month (within a three‑day window around the 15th) was selected, yielding a time series that spans a substantial fraction of a solar cycle. Equivalent widths (EWs) of the three triplet lines and the forbidden line were measured automatically with ARES v2, which fits Gaussian profiles and normalises the local continuum.

Two statistical approaches were applied to each line‑instrument combination. First, the Pearson correlation coefficient (r) between EW and the daily sunspot number (the activity proxy) was computed, together with its p‑value. Second, the data were split into an “active” group (sunspot number ≥ 50) and a “quiet” group (< 50); the median EW of each group was calculated and the difference ΔEW was converted into an abundance difference ΔA(O) using the 3‑D NLTE curve‑of‑growth from Steffen et al. (2015).

The results are strikingly heterogeneous. FEROS and NEID show negligible correlations (r≈0.14 and 0.05, p > 0.3) and ΔEW ≤ 0.5 mÅ, corresponding to ΔA(O) ≤ 0.008 dex – well within the typical systematic error budget of solar oxygen studies. In contrast, the PEPSI data exhibit a strong positive correlation (r≈0.78–0.79, p < 10⁻⁴) for all three triplet lines. The active‑quiet EW difference is about 0.8–1.0 mÅ, which translates into a modest abundance increase of ≈0.01 dex at activity maximum. This signal is only detectable because of PEPSI’s exceptionally high signal‑to‑noise ratio (≈1000); the same effect is lost in the noisier FEROS, HARPS‑N and NEID data.

For the forbidden 6300 Å line, neither PEPSI nor HARPS‑N show any significant trend (r≈−0.06 to +0.16, p ≈ 0.5), and ΔEW is only ±0.03 mÅ – far below the detection threshold. This confirms earlier findings that the forbidden line is essentially insensitive to solar activity, likely because it forms deeper in the photosphere and is intrinsically weak.

The authors conclude that, within the precision of current solar abundance analyses, the Sun’s oxygen abundance can be regarded as constant over the activity cycle. The 0.01 dex variation detected in PEPSI is smaller than the typical systematic uncertainties (≈0.04–0.06 dex) that dominate the solar oxygen problem. However, the same mechanism could become a non‑negligible source of bias for more active solar analogues or young stars, where activity levels (e.g., log R′_HK > −4.3) are an order of magnitude higher. In such cases, the infrared triplet could overestimate A(O) by several hundredths of a dex, potentially affecting studies of Galactic chemical evolution, stellar interior modelling, and exoplanet host‑star characterization.

The paper also highlights the importance of ultra‑high S/N observations for detecting subtle activity‑related line changes and suggests future work using synthetic Sun‑as‑a‑star spectra generated with the Numerical Empirical Sun‑as‑Star Integrator (NESSI). By varying the filling factor of active regions in the models, one can quantify the expected EW modulation for different activity levels and develop empirical correction formulas. Such corrections will be essential when applying the O I triplet to stars that are significantly more active than the Sun.

In summary, the study provides the first systematic, multi‑instrument assessment of solar oxygen line variability over the activity cycle, demonstrating that the effect is real but minute for the Sun, and emphasizing its potential impact on abundance determinations for more active stars.