Evidence for Solar-Cycle Modulation of the Alpha-to-Proton Temperature Ratio in Solar Wind

This study provides observational evidence that the alpha-to-proton temperature ratio in the solar wind strongly depends on solar wind speed, collisional age, and solar-cycle phase. The distributions of Tα /Tp reveal two dominant populations corresponding to near-equal temperatures (Tα /Tp ∼ 1) and mass-proportional temperatures (Tα /Tp ∼ 4), whose relative occurrence changes systematically with solar wind speed. The transition from an equal-temperature population in the slow solar wind to a mass-proportional population in the fast solar wind highlights the changing nature of ion heating and energy partitioning across different solar wind regimes.

In the slow solar wind (1, the distributions are dominated by the population centered near Tα /Tp ∼ 1, indicating that intervals with nearly equal proton and alpha-particle temperatures occur most frequently. As the solar wind speed increases, the relative contribution of the higher-temperature population progressively increases. In the fastest solar wind (> 500 km s−1), most intervals satisfy Ac < 1, and the distributions are instead dominated by the population centered near Tα /Tp ∼ 4, with a substantial fraction of intervals exhibiting mass-proportional or even super-mass-proportional temperature ratios. The intermediate-speed solar wind (400-500 km s−1) exhibits properties between these two limiting cases, reflecting a gradual transition rather than an abrupt change in plasma characteristics.

The close correspondence between solar wind speed, collisional age, and the relative occurrence of the two temperature populations is further reflected in the occurrence fractions of Ac 4. As shown in Figure 4, weakly collisional plasma (Ac < 1) and enhanced alpha-particle temperature ratios become increasingly common with increasing solar wind speed, while intervals characterized by Ac > 1 and Tα /Tp ∼ 1 dominate the slow solar wind. These observational trends indicate that the occurrence of mass-proportional temperature ratios is closely associated with weakly collisional plasma conditions, whereas near-equal proton and alpha-particle temperatures are preferentially observed in more collisionally processed solar wind.

The observed relationship between Tα /Tp and collisional age is broadly consistent with the established role of Coulomb collisions in regulating non-equilibrium ion properties during solar wind expansion. Since the collisional age scales approximately as Ac ∝ np /(vsw Tp)3/2 (J. C. Kasper et al. 2008), slow solar wind, which is generally denser and cooler, accumulates more Coulomb collisions than fast solar wind. Consequently, collisions progressively reduce temperature differences between ion species, leading to distributions dominated by Tα /Tp ∼ 1 in the slow solar wind. In contrast, the low collisional age of the fast solar wind allows elevated alpha-to-proton temperature ratios to persist over heliocentric distances, resulting in the predominance of the Tα /Tp ∼ 4 population at 1 AU. This interpretation is consistent with previous observational studies showing that alpha-particle differential heating and differential streaming decrease with increasing collisional processing (J. C. Kasper et al. 2008; B. A. Maruca et al. 2013; B. L. Alterman et al. 2018; P. J. Tracy et al. 2015; T. Ďurovcová et al. 2021).

However, the present observations alone cannot distinguish whether the elevated Tα /Tp values observed at 1 AU result from continued local heating during solar wind expansion or simply represent the preservation of heating signatures established closer to the Sun. Instead, our results demonstrate that high temperature ratios are preferentially associated with weakly collisional plasma, whereas near-equal proton and alpha-particle temperatures are preferentially observed in more collisionally processed plasma. This distinction is important because collisional age primarily governs the degree to which non-equilibrium signatures are preserved rather than directly identifying the physical mechanism responsible for generating them.

The preferential heating itself is widely believed to originate much closer to the Sun, where the plasma first becomes effectively collisionless (J. L. Kohl et al. 1998; S. R. Cranmer et al. 2009; E. Landi & S. R. Cranmer 2009; L. Ofman 2010; B. D. G. Chandran et al. 2013; P. Mostafavi et al. 2024; M. M. Martinovic et al. 2025; M. M. Martinović et al. 2026). Remote spectroscopic observations indicate that heavy ions become substantially hotter than protons only a few tenths of a solar radius above the photosphere as the Coulomb collision frequency decreases rapidly with height (J. L. Kohl et al. 1998; E. Landi & S. R. Cranmer 2009; S. R. Cranmer et al. 2009). Parker Solar Probe measurements have likewise revealed strong proton and alpha-particle temperature anisotropies, enhanced Alfvénic fluctuations, and weak collisional coupling in the near-Sun solar wind (P. Mostafavi et al. 2024; Yogesh et al. 2026). These observations are broadly consistent with theoretical models in which wave–particle interactions, turbulent dissipation, stochastic heating, ion-cyclotron resonance, and drift instabilities provide the free energy required for preferential ion heating (B. D. G. Chandran et al. 2010, 2013; S. R. Cranmer 2012; M. M. Martinović et al. 2020; M. M. Martinovic et al. 2025; T. A. Bowen et al. 2025; M. M. Martinović et al. 2026). While our measurements at 1 AU cannot directly determine which of these processes dominates, they are consistent with a scenario in which preferential heating is established in the near-Sun corona and inner heliosphere and subsequently modified by Coulomb collisions during outward expansion.

Our results therefore support a picture in which the observed alpha-to-proton temperature ratio at 1 AU reflects the combined effects of kinetic heating near the Sun and collisional relaxation during heliospheric transport. The relative importance of these two processes depends strongly on the collisional history of the plasma, naturally explaining why mass-proportional temperature ratios are most commonly observed in weakly collisional solar wind, whereas nearly equal ion temperatures dominate in more collisionally processed intervals.

In addition to the strong dependence on collisional age, our results reveal a clear solar-cycle modulation in the occurrence of both weakly collisional plasma (Ac 4). In the slow solar wind (< 300 and 300–400 km s−1), both quantities increase during solar maxima, whereas the intermediate (400–500 km s−1) and fast (> 500 km s−1) solar wind exhibit the opposite behavior, with larger occurrence fractions during solar minima. These opposite correlations indicate that the occurrence of mass-proportional temperature ratios depends not only on solar wind speed but also on the evolving distribution of solar wind plasma sampled throughout the solar cycle. It is worth noting that the interpretation of the slowest solar wind (< 300 km s−1) requires some caution. Previous studies have shown that this speed range may be affected by instrumental limitations or sampling biases, and is therefore often excluded or treated separately in statistical analyses (J. C. Kasper et al. 2007; B. L. Alterman & J. C. Kasper 2019; B. L. Alterman et al. 2025). Nevertheless, the systematic trends identified here persist across the broader slow-wind regime (300–400 km s−1), indicating that the observed solar-cycle dependence is not driven solely by the lowest-speed intervals.

The interpretation of these trends, however, requires caution because neither collisional age nor solar wind speed uniquely identifies the solar wind source region. Previous studies have shown that collisional age (Ac) and proton specific entropy (Sp ∝ log[Tp /np]) are closely related through their common dependence on the Coulomb collision frequency (νp ∝ np Tp −3/2), such that − log(νpp) scales with Sp up to an additive constant (B. L. Alterman 2019; V. Heidrich-Meisner et al. 2020; B. L. Alterman et al. 2018; F. Xu & J. E. Borovsky 2015). Consequently, plasma with low collisional age (Ac < 1) generally also exhibits high specific entropy, indicating that it has experienced relatively little collisional processing during its expansion from the Sun. Proton specific entropy has been identified as one of several useful classifiers of solar wind source regions rather than a unique discriminator (F. Xu & J. E. Borovsky 2015). Likewise, solar wind speed alone does not uniquely distinguish plasma originating from coronal holes, active regions, or streamer-belt sources, particularly in the intermediate-speed regime where multiple source populations coexist (R. d’Amicis et al. 2021; B. Alterman & R. D’Amicis 2025). Consequently, the present analysis does not attempt to assign a unique source region to individual intervals but instead examines the statistical evolution of plasma populations observed near the ecliptic over nearly three solar cycles.

A plausible interpretation is that the observed solar-cycle dependence reflects changes in the relative contributions of different solar wind source populations reaching 1 AU. During solar minima, the near-ecliptic heliosphere is more frequently occupied by long-lived high-speed streams originating from open magnetic-field regions associated with coronal holes (D. J. McComas et al. 2003, 2008; Y.-M. Wang et al. 2009). These streams typically exhibit lower densities, higher proton temperatures, stronger Alfvénicity, lower collisional ages, and enhanced non-equilibrium ion signatures, allowing preferential heating established closer to the Sun to remain visible at 1 AU (R. d’Amicis et al. 2021; Y. J. Rivera et al. 2024; B. Alterman & R. D’Amicis 2025). In contrast, during solar maxima the ecliptic is increasingly populated by slower and more variable solar wind associated with streamer belts, active regions, and the higher occurrence of transient coronal mass ejections (CMEs) (S. K. Antiochos et al. 2011; L. Abbo et al. 2016; S. L. Yardley et al. 2024; Y. J. Rivera et al. 2025a,b). Such plasma generally experiences greater collisional processing during its expansion and consequently exhibits a larger fraction of intervals with nearly equal proton and alpha-particle temperatures. The strongest solar-cycle variability is observed in the solar wind of speed from 300-500 km s−1, where plasma originating from coronal holes and closed-field regions is expected to coexist, suggesting that the observed trends primarily reflect changes in the relative occurrence of these source populations sampled by Wind near the ecliptic over the solar cycle. By contrast, the comparatively weak correlations in the slowest (500 km s−1) solar wind suggest that the observed long-term variations should be interpreted cautiously. In particular, the weak correlations in the fastest (> 500 km s−1) solar wind are consistent with this regime being dominated by relatively homogeneous coronal-hole wind, whose properties vary less over the solar cycle.

The intermediate-speed (400−500 km s−1) solar wind deserves particular attention because it exhibits the strongest reversal in the solar-cycle dependence. Rather than representing a distinct plasma state, this speed range is widely recognized as a mixture of solar wind originating from multiple coronal environments, including coronal-hole boundaries, active-region outflows, and streamer-associated plasma (R. d’Amicis et al. 2021; Y. J. Rivera et al. 2024; B. Alterman & R. D’Amicis 2025, 2026). The observed reversal near 400 − 500 km s−1 therefore likely reflects changes in the relative abundance of these source populations over the solar cycle, rather than a fundamental transition occurring at a single solar wind speed. This interpretation is consistent with the gradual evolution of the fitted population fractions presented in Figures 2 and 4, which show no evidence for an abrupt boundary between slow and fast solar wind.

Overall, our results suggest that the alpha-to-proton temperature ratio observed at 1 AU is controlled by the combined influence of preferential heating established in the inner heliosphere, subsequent collisional relaxation during solar wind expansion, and the changing mixture of solar wind source populations sampled throughout the solar cycle. Future studies combining in-situ plasma measurements with compositional diagnostics, magnetic connectivity, and remote observations of coronal source regions will be required to distinguish more directly between these contributing effects.

Using nearly three solar cycles of Wind observations, we investigated the dependence of the alpha-to-proton temperature ratio (Tα /Tp) on solar wind speed, collisional age, and solar-cycle phase. The observed distributions reveal two dominant populations corresponding to near-equal temperatures (Tα /Tp ∼ 1) and mass-proportional temperatures (Tα /Tp ∼ 4), whose relative occurrence varies systematically with solar wind speed. The slow solar wind is dominated by more collisionally processed plasma with larger collisional ages and temperature ratios close to unity, whereas the fast solar wind exhibits systematically lower collisional ages and persistent preferential alpha-particle heating with Tα /Tp ≳ 4. The occurrence of mass-proportional temperature ratios also shows a clear solar-cycle dependence that varies across different solar wind speed regimes. In the slow wind, mass-proportional temperature ratios become more common during solar maxima, whereas in the intermediate and high-speed wind they are enhanced during solar minima.

These results indicate that the observed alpha-to-proton temperature ratio at 1 AU reflects the combined influence of preferential ion heating established closer to the Sun, subsequent Coulomb collisional relaxation during solar wind expansion, and the evolving mixture of solar wind source populations sampled over the solar cycle. While the present analysis cannot uniquely distinguish between ongoing local heating and the preservation of near-Sun heating signatures, the observed trends are consistent with weaker collisional processing and stronger preferential heating in fast solar wind, and enhanced thermalization in slower, denser solar wind. Overall, this work provides new observational constraints on the interplay between preferential ion heating, collisional evolution, and solar-cycle variability, thereby improving our understanding of the thermodynamic evolution of heavy ions in the solar wind plasma.

Improvements to AI Systems:

1. Physics-Informed Predictive Models for Plasma State Estimation: Train AI models (e.g., neural networks or Gaussian processes) to predict the alpha-to-proton temperature ratio (Tα/Tp) as a function of solar wind speed, collisional age (Ac), and solar-cycle phase. The improved AI can now forecast whether a given solar wind interval will exhibit near-equal temperatures (Tα/Tp 1) or mass-proportional temperatures (Tα/Tp 4) with quantified uncertainty, enabling real-time classification of plasma regimes for space weather forecasting.

2. Collisional-Age-Aware Anomaly Detection: Develop AI systems that use the observed bimodal distribution (peaks at Tα/Tp 1 and 4) to detect anomalous plasma intervals that deviate from expected collisional relaxation trends. The improved AI can flag intervals where Tα/Tp exceeds 4 (super-mass-proportional) or where slow wind unexpectedly shows high Tα/Tp, potentially indicating undetected heating events or instrumental artifacts, thereby enhancing data quality control for heliospheric datasets.

3. Solar-Cycle-Adaptive Source Region Classification: Build a classifier that integrates solar wind speed, collisional age, and Tα/Tp to probabilistically assign plasma intervals to source regions (coronal holes, streamer belts, active regions) while accounting for solar-cycle-dependent mixing. The improved AI can now distinguish between source-region-driven variations and intrinsic heating signatures, reducing misclassification in intermediate-speed wind (400–500 km/s) where multiple sources coexist.

4. Causal Inference for Heating vs. Relaxation Disentanglement: Implement causal models (e.g., structural equation modeling or counterfactual reasoning) that use collisional age as a mediating variable to separate the effects of near-Sun preferential heating from in-transit Coulomb relaxation on Tα/Tp. The improved AI can estimate the relative contribution of each process for any given solar wind interval, providing a quantitative answer to the unresolved question of whether observed high Tα/Tp at 1 AU is due to local heating or preservation of near-Sun signatures.

5. Time-Series Forecasting of Non-Equilibrium Ion Properties: Train recurrent or transformer-based models on multi-decadal Wind data to predict future Tα/Tp distributions as a function of solar cycle phase and speed regime. The improved AI can generate probabilistic forecasts of the occurrence fractions of Ac4 for upcoming solar maxima/minima, aiding mission planning for Parker Solar Probe or Solar Orbiter observations.

6. Physics-Constrained Interpolation for Sparse Plasma Measurements: Develop generative models (e.g., variational autoencoders) that learn the joint distribution of Tα/Tp, Ac, and solar wind speed, constrained by the physical relationship Ac ∝ np/(vsw Tp 3/2). The improved AI can fill gaps in in-situ measurements (e.g., missing alpha-particle data) by generating plausible Tα/Tp values conditioned on proton parameters and collisional age, improving dataset completeness for heliospheric studies.

7. Automated Detection of Solar-Cycle-Dependent Regime Transitions: Create an AI system that monitors the relative occurrence of the two Tα/Tp populations in real time and detects shifts in the transition speed (e.g., 400–500 km/s) across solar cycles. The improved AI can alert researchers to changes in source population mixing, enabling early identification of solar cycle phase transitions from plasma data alone, without relying solely on sunspot numbers.

8. Multi-Scale Fusion for Coronal-Source Inference: Integrate AI models that combine in-situ Tα/Tp and Ac measurements with remote coronal observations (e.g., EUV images, magnetograms) to infer the coronal heating mechanisms (e.g., ion-cyclotron resonance vs. stochastic heating) responsible for preferential alpha-particle heating. The improved AI can map observed Tα/Tp signatures back to specific coronal conditions, providing a data-driven link between remote and in-situ observations.

9. Robust Statistical Clustering for Plasma Population Identification: Apply unsupervised learning (e.g., Gaussian mixture models with physics-informed priors) to identify and track the two dominant Tα/Tp populations across different solar wind speeds and solar cycle phases. The improved AI can automatically quantify the fraction of each population and detect subtle shifts in their centroids or widths, offering a more nuanced view of ion heating than fixed thresholds (e.g., Tα/Tp >4).

10. Uncertainty-Aware Space Weather Impact Assessment: Build an AI system that translates predicted Tα/Tp and collisional age into estimates of plasma instability potential (e.g., drift instabilities, temperature anisotropy-driven waves) for spacecraft safety and communication systems. The improved AI can provide probabilistic risk assessments for satellite operations by forecasting when weakly collisional, high-Tα/Tp intervals are likely to occur, which are associated with enhanced kinetic-scale turbulence and potential electromagnetic interference.

The influence of collisional age (A c) on the alpha-to-proton temperature ratio (T alpha/T p) has been explored in the past. However, the modulation of this ratio with respect to the solar cycle has remained unexplored so far. We show solar-cycle modulation of T alpha/T p and A c using nearly three decades of in-situ observations from Wind spacecraft across distinct solar wind speed regimes and solar activity phases. Our results reveal that in the slow solar wind with velocity <400 km s-1, where A c happens to be typically >1, the ratio T alpha/T p stays close to unity. This suggests frequent Coulomb collisions efficiently iron out temperature differences. In contrast, the fast wind with velocity >500 km s-1, where A c happens to be typically <1, mass-proportional heating is most pronounced, with T alpha/T p often exceeding 4. The intermediate speed regime (400 - 500 km s-1) represents a gradual transition between the slow and fast wind populations in terms of their solar-cycle dependence. This behavior reflects the changing dominance of high-speed streams from polar coronal holes during minima to denser slow wind during maxima. These results suggest that mass-proportional ion heating at 1 AU is not solely governed by local collisional physics but is significantly modulated by the solar cycle dependent variations in the solar wind sources.