DIRECT OBSERVATIONAL EVIDENCE OF FILAMENT MATERIAL WITHIN INTERPLANETARY CORONAL MASS EJECTIONS

Coronal mass ejections (CMEs) explosively arise from the 1 MK solar corona, hurtling enhanced magnetic fields, mass, and energy into the heliosphere. CMEs are driven by the explosive release of free energy gained by the evolution of the solar magnetic field. This process likely involves reconnection (i.e., Klimchuk 2001). Thus far, thousands of CMEs have been observed using coronagraphs (Hundhausen 1993; Gopalswamy et al. 2004, 2010), and recently even in three dimensions using STEREO (e.g., Mierla et al. 2009). There is a rich and varied set of observational signatures within these ejections, yet there is a basic property that should be found within most CMEs.

More than 70% of CMEs are associated with filament eruptions (e.g., Gopalswamy et al. 2003; Webb & Hundhausen 1987; Munro et al. 1979). Filaments, also known as prominences if observed on the solar limb, are cool (T ∼ 10⁴ MK) quasi-stationary structures in the corona. They are localized over magnetic neutral lines, and are generated through run-away radiative cooling in locations of enhanced densities (Karpen & Antiochos 2008). The cooling process is sufficiently effective to allow filament material to reach T ∼ 10⁴ MK (e.g., Gopalswamy et al. 1999). Plasma in such an environment yields a significant fraction of neutrals and low-charge ions with cool ionization temperatures (He⁺, O⁺, O²⁺, C²⁺, etc.), but, filament plasma is far away from local thermal equilibrium (e.g., Labrosse et al. 2010).

Most CMEs are generated in conjunction with a filament eruption: despite their large mass density, a filament is lifted out of the Sun's gravitational well and becomes part of the CME eruption (e.g., Webb & Hundhausen 1987, and references therein). Together with the interaction signatures of these CMEs, the filament is a crucial part of the classic three-part structure of CMEs observed in coronagraphs. Figure 1 depicts an eruption of a CME observed by the Solar and Heliospheric Observatory (SOHO) Large Angle Solar Coronagraph (LASCO) with a classic three-part structure. The bright rim is likely a feature of the plasma interaction of CMEs with the ambient coronal plasma. Behind this bright front is a so-called cavity, with significantly lower density, presumably associated with an enhanced magnetic field and evolving toward its Taylor state, or minimum energy state for a given helicity, or possibly ejected as such from the low corona (Lynch et al. 2004; Low 1994). Toward the back of the CME structure is the core, consisting of dense and comparatively cool filament plasma. While filament material is relatively denser in the corona and often visible in UV, it may only occupy a small fraction of the ejected CME. The filament's distinct spectroscopic signatures have been observed up to the heliospheric distance of 3.5 solar radii (Raymond 2002; Akmal et al. 2001).

Zoom In Zoom Out Reset image size

Based on these remote observations, one might expect to be able to resolve a similar three-part structure in the interplanetary counterparts of the CMEs, interplanetary CMEs (ICMEs). This is not typically the case, resulting in one of the biggest observational puzzles related to ICMEs (Zurbuchen & Richardson 2006, and references therein).

ICMEs routinely exhibit two of the three signatures. They often show dynamic density enhancements ahead of magnetically dominated ejecta, often near a Taylor state, with a magnetic field in a near flux–rope configuration. However, there are no reports of consistent in situ composition signatures of the cool material associated with the erupting filament.

The most obvious signature to look for is the presence of low-charge ions, indicative of low coronal electron temperatures near the Sun and observed by SOHO Ultra-Violet Coronal Spectrograph (UVCS; e.g., Raymond 2002). It is clear that such a signature only persists into the heliosphere if plasma heating near the Sun, a proven consequence of the eruption process, is unsuccessful at further ionizing these ions.

More than 80% of all ICMEs exhibit elevated ionic charge states or hot ionization temperatures, hotter than the ∼1 MK corona (i.e., O⁸⁺, O⁷⁺, C⁶⁺, Fe¹⁶⁺, etc.; Lepri et al. 2001; Henke et al. 2001; Richardson & Cane 2004). The underlying physical processes that cause this heating are thought to relate to magnetic reconnection and the subsequent formation of current sheets that are also at the heart of the generation of flares (Lepri & Zurbuchen 2004; Reinard 2005; Rakowski et al. 2007).

Even though relatively sparse, UVCS spectroscopic observations reveal the presence of both hot and cold plasma as part of ICME eruptions. Based on these observations, the presence of cool ions should be routinely expected in the heliosphere. Yet, there are only a handful of reported observations of cool ions within ICMEs, leading to a major observational puzzle at the heart of solar-terrestrial relations and space weather predictions.

Schwenn et al. (1980) and Zwickl et al. (1982) analyzed data from 1972 to 1980 using various spacecraft and found three ICMEs with enhancements of He⁺. Zwickl et al. also showed observations that indicated the presence of Fe⁵⁺, C⁴⁺, and O²⁺ inside these ICMEs. This composition could not be unambiguously determined at the time, as it was derived using a simple electrostatic analyzer, assuming equal speeds of all species. There are two events with substantial He⁺ reported during the past solar cycle (Burlaga et al. 1998; Skoug et al. 1999). The 1998 May 1 and 2 event described by Skoug was the first to be analyzed with a modern solar wind composition instrument, the Solar Wind Ion Composition Spectrometer (SWICS) on the Advanced Composition Explorer (ACE), and showed significant additions of cold ions (Gloeckler et al. 1998). But this remains the only such event described to date in the literature.

This Letter presents the first systematic search for cold material in ICMEs, as identified by a comprehensive set of heavy-ion charge states at 1 AU. This analysis uses a novel technique, taking advantage of the full power of ACE SWICS (Gloeckler et al. 1998; Stone et al. 1998). Our study focuses on oxygen ions at charges as low as O²⁺, carbon as low as C²⁺, and iron as low as Fe⁴⁺. Section 2 will describe the ACE SWICS sensor and discuss the methods used to search for these cold events. Section 3 will discuss filament events and analyze their rate in the context of solar observations of filaments in CMEs. Section 4 will summarize and discuss these results.

The ACE spacecraft, launched in 1997, is currently in orbit about the L1 point, 1.5 million km sunward of Earth (Stone et al. 1998). Of particular importance to this study is SWICS (Gloeckler et al. 1998). SWICS is a time-of-flight (TOF) mass spectrometer paired with energy-resolving solid-state detectors (SSDs) and an electrostatic analyzer (ESA) that measures the ionic composition of the solar wind. Ions with the appropriate energy per charge are selected in the ESA. Ion speed is determined in the TOF telescope and the residual energy measured by the SSDs enables particle identification. These measurements allow the independent determination of mass, M, charge, Q, and energy, E, and are virtually free of background contamination. These triple-coincidence measurements allow nearly unambiguous identification of a wide range of charge states of He, C, O, Fe, etc. in the energy per charge range of 0.49–60 keV e⁻¹, which is in the thermal energy range of the solar wind and ICMEs.

From these measurements, we calculate the charge state distributions of C, O, and Fe with a time resolution of 2 hr, typically sufficient to achieve <20% accuracy for C, O, and Fe charge states. We have recently proven that SWICS is able to resolve charge states in triple mode (with successful TOF and SSD measurements) down to C²⁺, O²⁺, and Fe⁴⁺. However, the data have to be corrected for instrument efficiencies, measurement duty cycle, and event priorities that affect the probability of these ion measurements to be telemetered to the ground (von Steiger et al. 2000; Lepri et al. 2001). The Lepri et al. process has been extended to these cool ions to enable the work presented here. In routine operation, we can now perform analysis of C^2–6+, O^4–8+, and Fe^4–24+, substantially extending the charge range, thus far available. In some special cases, we are able to extend the range even farther.

Very low ionic charge states, such as singly charged He⁺, C⁺, and O⁺ or O²⁺, are most likely observed in double coincidence (with successful TOF, without successful SSD measurements due to the energy detection threshold). These double-coincidence measurements have substantially different signal-to-noise characteristics and event priorities within the SWICS data stream. Here, we limit ourselves to triple-coincidence measurements.

Figure 2 provides three representative examples of charge distributions for C (panel (a)), O (panel (b)), and Fe (panel (c)) during 2 hr of integration time. The "Cold CME" charge state distribution, starting on 1998 May 2 (DOY 122) at 20:11 UT, is shown as blue bars in panels (a)–(c). "Normal Solar Wind," starting on 1998 May 11 (DOY 131) at 14:12 UT, is shown as green bars in panels (a)–(c). The "Hot ICME," starting on 1998 September 25 (DOY 268) at 18:08 UT, is shown as red bars in panels (a)–(c).

Zoom In Zoom Out Reset image size

Typical solar wind charge state distributions peak at C⁵⁺, O⁶⁺, and Fe¹⁰⁺, indicating coronal electron temperatures near 1 MK. The hot ICME charge state distributions are as previously described in Lepri et al. (2001) and Lepri & Zurbuchen (2004). These charge state distributions contain significant contributions from C⁶⁺, O⁷⁺, and Fe¹⁶⁺ (the Ne-like ion of the Fe charge state series) and higher. The cold ICME example shown here reveals strong peaks at the lowest charge states and enhancements at the highest charge states for C, O, and Fe. In general, the observations of cold ICMEs extend to some of the lowest charge states analyzed (C²⁺, O²⁺, Fe⁴⁺), and some cold ICMEs may extend to lower charge states not covered by this analysis method. In all of the cold ICME cases, we identified that there are contemporaneous measurements of elevated charge states, such as C^4–6+, O^6–7+, and Fe^12+–17+ (as shown in Figure 2). Such clearly nonthermal behavior is a common feature of many cold ICME intervals down to the SWICS time resolution of 12 minutes. In fact, the hot components observed as part of the cool time periods are very much comparable to the charge distribution of hot ICMEs, as shown in Figure 2.

We now discuss the application of this methodology for the analysis of ICMEs, as identified by Cane and Richardson (CR) from 1998–2007. ICMEs are identified based on a multitude of signatures (Cane & Richardson 2003; Richardson & Cane 2010). The details of the ICME list can be found in the aforementioned papers.

To reliably identify cold filament material, we used a set of highly restrictive conditions designed to eliminate spurious events caused by distributed in situ sources with planetary origin (Gloeckler & Geiss 2004) and events with low statistics.

For each 2 hr data period, we calculate the relative abundances of cold ions of C, O, and Fe as defined below:

$$\begin{equation} {\rm C}_{\rm c} = \frac{{{\rm C}^{2 + - 3 + } }}{{{\rm C}^{2 + - 6 + } }},\quad {\rm O}_{\rm c} = \frac{{{\rm O}^{4 + } }}{{{\rm O}^{4 + - 8 + } }},\quad {\rm Fe}_{\rm c} = \frac{{{\rm Fe}^{4 + - 7 + } }}{{{\rm Fe}^{4 + - 24 + } }}. \end{equation} \tag{ 1 }$$

This selection of charge states represents the lowest charge states we can identify with sufficient accuracy in SWICS' triple-coincidence measurements throughout the entire ACE mission, without being impacted by SSD efficiency effects as well as significant biasing of O^2+–3+, which strongly depends on the prioritization scheme utilized for telemetry download. As mentioned previously, these charge states are generally not present in nominal solar wind flows and their presence indicates an LTE source temperature as low as 60,000–200,000 K, much lower than typical coronal temperatures of 1 MK.

Since the total abundance of low-charge-state ions varies with the solar wind density and flux, we developed an identification criterion that focused on the relative abundance to avoid biases in the analysis. For each element, thresholds were determined using the following process: we examined the probability distributions for the logarithm of the ratios C_c, O_c, and Fe_c over the entire ACE mission. The probability distributions for these ions resembled a Poisson distribution, hence a mean value could be calculated. The mean values from these probability distributions represent the nominal solar wind values of C_c, O_c, and Fe_c. Significant enhancements of low-charge-state ions were identified by looking at time periods when the relative abundance of those ions were simultaneously elevated more than one standard deviation above their mean relative abundance (C_c > 0.034; O_c > 0.009; Fe_c > 0.121).

Furthermore, we require the speed of the low-charge-state ions to be within 15% of the velocity of He²⁺. Hefti et al. (1998) showed that different solar wind species typically have bulk speeds well within 10%. Our range of 15% allows for larger velocity changes that could be expected because of reduced Coulomb drag (von Steiger & Zurbuchen 2006). In summary, any 2 hr period is considered to contain cold plasma only if all four conditions—three compositional conditions and one speed criterion—are fulfilled simultaneously.

Figure 3 shows ACE observations of cold ejecta in 2005 May as defined by the above criteria. The vertical solid red line indicates the front boundary of the ICME, while the vertical dashed red line indicates the rear boundary. Figures 3(a)–(c) show the proton density, temperature, and speed from SWEPAM. Figure 3(d) shows the magnetic field magnitude from MAG. Figures 3(e)–(g) show the time evolution of charge state distributions of Fe, O, and C, respectively. The color bars in these panels indicate the relative density in each of the charge states. The green track in the Q_Fe panel shows the average charge state of Fe. Outside of the ICME, C⁵⁺ and C⁶⁺, O⁶⁺, Fe¹⁰⁺ are most prevalent. Inside the ejecta, the charge states of Fe increase to near 16+ before abruptly dropping along with C and O to extremely low charge states for a short period before returning to elevated states through much of the ejecta. The rear half of the ejecta is characterized by normal solar wind-type charge states. The ejecta contain both hot and cold plasma, as shown in panels (e)–(g). Cold plasma is confined to a short period shaded in yellow.

Zoom In Zoom Out Reset image size

Table 1 summarizes the cold events identified according to the criteria defined in Section 3. We find a total of 11 anomalously cold events in the ACE SWICS data from 1998 to 2008. These events are listed in Table 1 with their ICME associations based on the CR list. First, and most importantly, all cold events are associated with ICMEs. Out of the 11 events, 10 exhibited measurable cold plasma within the boundaries of the ICME. One event, DOY 226, 1999 (August 14), has signatures of cold plasma < 2 hr beyond the rear boundary of the ICME.

To further investigate the linkage with filament eruptions, we conducted a cursory analysis of LASCO Extreme-ultraviolet Imaging Telescope (EIT) movies (195 Å) and examined data from the Mauna Loa Solar Observatory (MLSO) to look for filament eruptions near in time and space to the initial CME. We identified filament eruptions in EIT movies as the presence of dark filamentary material which appeared to rise up abruptly and expand outward with the CME. In the MLSO observations, we look for the disappearance of filaments between frames. For a few of the ambiguous cases, we received assistance from H. Gilbert (2010, private communication). Of the nine events for which a LASCO CME was identified (Richardson & Cane 2010) and EIT data were available, eight of them exhibited probable filament eruptions. The one that did not clearly exhibit such an eruption preceded the Halloween Storm of 2003, and was linked to multiple CMEs at the Sun, thereby making it difficult to identify a filament eruption. Hence, nearly 90% of these cold events have likely associations with filament eruptions, further supporting the idea that the origin of the cold material is a filament that erupts in association with the CME.

Out 283 ICMEs identified by CR during the study period, approximately 4% exhibit the presence of cold plasma using our restrictive criteria. This rate is substantially (>10 times) higher than indicated by previous in situ studies, and may be even higher if less restrictive criteria are used for analysis. The determination of this enhanced rate can be entirely attributed to improvements in the signal-to-noise ratio of low-charge-state ions enabled by our new measurement and analysis technologies. Note also that cold ICMEs are observed during all phases of the solar cycle.

The observations of cold plasma occur contemporaneous with observations of other plasma and composition signatures. First, ∼82% of events (9 out of 11) with cold plasma occur inside ICMEs with magnetic cloud (MC) signatures (Zurbuchen & Richardson 2006, and references therein). Furthermore, all of the ICMEs exhibiting signatures of cold plasma show mixtures of cold and hot ionic charge states, as described above. This indicates that these ICMEs at the Sun may be associated with "eruptive flares," a term coined by Zhou et al. (2003) to represent an eruption that contains both a flare and filament. Examining the spatial distribution of cold plasma within the ICME, we found that in 35% of the cases, cold plasma is detected in the leading third of the ICME, in 20% of the cases, cold plasma appears in the middle of the ICME, and the remaining 45% of the cases exhibit cold plasma in the trailing third of the observed ICME periods. Thus, there is no strong spatial ordering of the cold material inside of the ICMEs.

We report the results of the first comprehensive search for cold plasma in the heliosphere, using the full power of the SWICS measurement and analysis techniques. We analyzed 10 years of data and, using highly restrictive criteria, identified 11 cold events, all of which are associated with ICMEs. We find that while these events exhibit unusually cold plasma, they tend to simultaneously exhibit charge states indicative of hot plasma.

In ∼90% of the cold ICME events associated with CMEs observed by LASCO in white light, a probable connection to a filament could be established. For 10% of the cases, such a connection could not be established because the event was linked to multiple ICMEs and therefore no one event could be identified as associated and/or because the detection of filament eruptions also vary strongly with respect to their signal-to-noise ratio. Hence, these numbers support our conclusion that the cold plasma observed in situ and identified by our process originated in a filament near the Sun.

Overall, we find that more than 4% of all ICMEs observed at ACE exhibit signatures of filament plasma. If filament material is only expected to be found in ICMEs that are MCs, then this represents approximately 12% of all MCs. Gilbert et al. (2006) suggest that the average filament mass is ∼20% of the average ICME mass. Since at least some of the filament material is likely to be ionized during the eruption, our findings are reasonably in line with these estimates. The findings of our study increases the known rate of such observations by over one order of magnitude, and may increase it further if less restrictive signatures are used.

However, this fraction is much lower than the 70% of ICMEs observed near the Sun that are associated with filament eruptions. This remaining discrepancy may reflect our highly restrictive selection criteria, or may be a direct measure of the filling ratio of filaments within ICMEs. However, it could be a result of heating processes near the Sun that would tend to erase low—charge-state signatures in the filament plasmas observed near Earth.

The authors acknowledge conversations with H. Gilbert, N. Gopalswamy, and the SHRG at the University of Michigan. This work has been supported by NNH08ZDA001N-LWSTRT, NNX08AM64G, and the ACE mission 44A-1085637. T.H.Z. acknowledges the hospitality of the International Space Science Institute in Bern, Switzerland, where much of his contribution to this work was performed.