The Secret Lives of Cepheids: δ Cep—The Prototype of a New Class of Pulsating X-Ray Variable Stars^*

Although Cepheids are luminous at optical wavelengths, their moderate levels of X-ray activity (${L}_{{\rm{X}}}\approx {10}^{29}$ erg s⁻¹) combined with their distances make them comparatively faint at X-ray wavelengths (${f}_{{\rm{X}}}\lt {10}^{-14}$ erg s⁻¹ cm⁻²). Relatively long exposures (and the need for multiple visits) are required to achieve sufficient data quality and phase sampling, and this has slowed the X-ray studies progress. Over the past decade (Polaris having been observed with Chandra in 2006 February), only two Cepheids have been observed with sufficient pulsation phase coverage for potential X-ray variability to be uncovered. The first is δ Cep, and the second star is β Dor  ($\langle V\rangle =3.78$; d = 318 pc; F4–G4 Ia–II; ${P}_{\mathrm{pulsation}}=9.843$ days), which has the more symmetric light curve common to Cepheids with pulsation periods near ∼10 days; it will be the subject of a future paper in preparation.

From 2010 to 2013, five independent pointings to δ Cep were carried out with XMM-Newton (Engle et al. 2014; Engle 2015). One of the program's early observations with XMM-Newton showed a much higher X-ray flux at the beginning of a long ∼73 ks exposure, which then declined to a lower flux level similar to that observed at other pulsation phases (see Engle et al. 2014). Although searching for X-ray variability was the initial goal, the phase at which it was found to occur was very surprising. All FUV–UV emission lines peak near phase $\phi \approx 0.90\mbox{--}0.96$, just before δ Cep reaches maximum visual brightness and just after the phase in the star's pulsation where it has begun to expand again. This expansion forms a shock front that then propagates through the Cepheid's outer atmosphere, compressing and heating atmospheric plasmas to produce the observed strong phase-dependent FUV–UV emissions. To our surprise, however, the phase of the enhanced X-ray emissions occurred ∼2.7 days later, at $\phi \approx 0.45$. This is just after the Cepheid reaches its maximum size and coolest temperature, and begins to shrink again. It was an unexpected phase for enhanced X-ray activity, and relied on a single sub-exposure, so confirmation was necessary to prove that the X-ray variability was pulsation-phase-specific, and not transient or arising from a companion star.

In 2015 May, a ∼42 ks visit of δ Cep was carried out with the Chandra Advanced CCD Imaging Spectrometer (ACIS-I ). The visit was constrained to occur at the phase of increased X-ray flux previously observed with XMM-Newton. The data were reprocessed with the Chandra Interactive Analysis of Observations (CIAO) v4.9 suite to ensure the latest calibrations were applied. Analyses of the data were made using the Sherpa modeling and fitting package (distributed as part of CIAO), and MEKAL models (Drake et al. 1996) were used for the final fitting and flux calculations.

To increase phase resolution, the observation was divided in half, creating two equal 21 ks sub-exposures that were analyzed separately. The distance and activity level of the target resulted in relatively low count rates, so single temperature plasma fits were applied to the resulting energy distributions, as two-temperature models showed no improvement in the fits. As shown in Table 1 and Figure 1, the Chandra data confirm the X-ray flux peak found previously with XMM-Newton. A steady decline in X-ray flux is observed over the course of the Chandra exposure, with the second subdivision displaying just over half the X-ray flux of the first subdivision. Figure 2 compares the X-ray energy distributions of the Chandra subdivisions. Both distributions peak around 1 keV, with peak plasma temperatures from 1.1 keV ($T\approx 13\,\mathrm{MK}=1.3\,\times {10}^{6}\,{\rm{K}}$) to 1.7 keV ($T\approx 20\,\mathrm{MK}$). But the first subdivision, in addition to higher overall activity, shows an increase in harder X-ray counts, in the 1.2–1.5 keV range, and potentially in the 2.4–2.9 keV range as well, although the signal is very weak at these energies.

Zoom In Zoom Out Reset image size

Table 1.  X-Ray Observations of δ Cep

ObsID	Start Time (UT)	End Time (UT)	Phase Range	kT	kT	${f}_{{\rm{X}}}$ (0.3–2.5 keV)	${f}_{{\rm{X}}}$ error	${L}_{{\rm{X}}}$	$\mathrm{log}\,{L}_{{\rm{X}}}$
	Start Time (JD)	End Time (JD)		(kev)	Ratio	(10−15 erg s−1 cm−2 )	(10−15 erg s−1 cm−2 )	(1028 erg s−1)
Previous XMM-Newton Results
552410401	2008 Jun 05 14:26	2008 Jun 05 21:53	0.33–0.39	1.00	1.00	5.14	1.40	4.56	28.66
	2454623.101	2454623.412		0.40	0.64
0603741001 I			0.43–0.48	0.63	0.57	19.08	1.80	16.92	29.23
				1.83	1.00
0603741001 II	2010 Jan 22 18:05	2010 Jan 23 14:17	0.48–0.54	0.33	1.00	10.36	1.40	9.19	28.96
	2455219.254	2455220.095		1.47	0.95
0603741001 III			0.54–0.59	0.79		5.87	0.40	5.21	28.72
603740901	2010 Jan 28 18:04	2010 Jan 21 12:37	0.05–0.12	0.25	1.00	6.07	3.00	5.38	28.73
	2455217.253	2455218.026		0.70	0.63
0723540301 I	2013 Jun 28 6:34	2013 Jun 29 13:49	0.84–0.96	0.32	0.97	4.34	1.00	3.85	28.59
	2456471.774	2456473.076		1.33	1.00
0723540301 II			0.96–0.08	0.41	1.00	4.35	2.20	3.86	28.59
				1.38	0.65
0723540401 I	2013 Jul 02 6:17	2013 Jul 03 7:51	0.58–0.68	0.61	1.00	3.71	0.70	3.29	28.52
	2456475.762	2456476.827		0.74	0.01
0723540401 II			0.68–0.78	0.36	1.00	5.94	1.50	5.27	28.72
				0.96	0.42
Chandra 2015 Visit
16684 I	2015 May 08 20:17	2015 May 09 07:53	0.48–0.52	1.6		17.38	3.70	16.72	29.22
16684 II	2457151.345	2457151.828	0.52–0.56	1.3		10.41	1.50	10.12	29.01

Download table as:  ASCIITypeset image

δ Cep is a relatively weak X-ray source, whether it is in a phase of high X-ray activity or not. Thus, the plasma temperature values returned by the model fits are less certain because of the low count rates. The X-ray luminosities (${L}_{{\rm{X}}}$) of δ Cep (as well as other Cepheids; see Tables 1 and 2) are approximately the same as observed for young, chromospherically active G–M stars, and about 100× larger than the mean X-ray luminosity of the Sun. However, despite the large radii and thus very large surface areas of Cepheids, their X-ray surface fluxes (${F}_{{\rm{X}}}$) are much smaller than even inactive coronal X-ray sources like the Sun.

Table 2.  Cepheid X-Ray Detections

Observed	Start Time (UT)	End Time (UT)	Start	End	kT	${f}_{{\rm{X}}}$ (0.3–2.5 keV)	${L}_{{\rm{X}}}$	Possible
Cepheid	Start Time (JD)	End Time (JD)	Phase	Phase	(keV)	(10−15 erg s−1 cm−2	(1028 erg s−1)	Companiona
V659 Cen	2013 Sep 07 20:11	2013 Sep 08 02:22	0.15	0.20	0.88	4.74	32.4	Yes
	2456543.341	2456543.599
R Cru	2014 Jan 04 19:57	2014 Jan 05 02:30	0.75	0.79	1.24	7.67	63.1	Yes
	2456662.331	2456662.604
V473 Lyr	2013 Sep 22 09:49	2013 Sep 22 12:03	0.53	0.59	0.66	20.5	75.9	No
	2456557.909	2456558.002
S Mus	2013 Jan 05 14:37	2013 Jan 05 21:48	0.01	0.04	0.93	34.6	288.4	Yes
	2456298.109	2456298.408
Polaris	Multiple				0.1–0.6	28–76	6.31–15.85	Yes
β Dor	Multiple				0.3–2.1	5.3–44.3	6.31–50.12	No
δ Cep	See Table 1				0.3–2.1	4.3–19.4	3.16–15.85	Yes

Note.

^aPossible Companion: This column indicates which Cepheids have companions that the currently available X-ray images cannot resolve. For these stars, at present the X-ray activity can be attributed to either the Cepheid or the companion, with the exception of δ Cep, where the phasing of the X-ray variations allows us to attribute them to the Cepheid.

Download table as:  ASCIITypeset image

Adopting the mean interferometric radius of $\langle R\rangle =43.0\,{R}_{\odot }$ (Mérand et al. 2015) and the observed minimum and maximum ${L}_{{\rm{X}}}$ values of $\sim 4.0\times {10}^{28}$ and $17.4\times {10}^{28}$ erg s⁻¹, respectively, for δ Cep (see Table 1) returns minimum and maximum surface X-ray fluxes of ${F}_{{\rm{X}}}\approx 360$ and 1550 erg s⁻¹ cm⁻², respectively. Compared to the Sun (assuming $\langle {L}_{{\rm{X}}}{\rangle }_{\odot }\approx 1\times {10}^{27}$ erg s⁻¹ from Ayres 2005), the surface X-ray flux is $\langle {F}_{{\rm{X}}}{\rangle }_{\odot }=1.6\times {10}^{4}$ erg s⁻¹ cm⁻² . So, even for the maximum X-ray activity level of δ Cep, the surface X-ray flux is ∼10 times less than the Sun's (while at minimum, it is 50 times less). To further illustrate the relative X-ray activity levels, the average bolometric luminosity (${L}_{\mathrm{bol}}$) of δ Cep is ∼1800 times that of the Sun. Taking this into account, and using δ Cep 's average ${L}_{{\rm{X}}}$ = 10.8 × 10²⁸ erg s⁻¹, the ratio ${L}_{{\rm{X}}}$/${L}_{\mathrm{bol}}=1.57\times {10}^{-8}$, which is ∼6% that of the Sun. When compared to younger, chromospherically active G–K stars with ${L}_{{\rm{X}}}\approx {10}^{29}$ erg s⁻¹, the X-ray surface fluxes (${F}_{{\rm{X}}}$) of δ Cep are nearly 10⁵ times weaker.

Further comparisons can be made to the other stellar residents of and around the Cepheid region of the instability strip. For several G and early-K supergiants, distances and X-ray fluxes were obtained from Ayres (2005) and Ayres et al. (2005), along with angular diameter measures from Bourges et al. (2017). The resulting surface fluxes show a large spread. The K-type supergiants π Her (K3 Iab) and γ Aql (K3 II) have very low surface X-ray fluxes of ${F}_{{\rm{X}}}\approx 50$ and 40 erg s⁻¹ cm⁻², respectively. By contrast, the seemingly hyperactive G-type supergiants β Cam (G1 Ib–II) and β Dra (G2 Ib–II) have surface X-ray fluxes of ${F}_{{\rm{X}}}\approx 4.7\times {10}^{4}$ and $5.8\,\times {10}^{4}$ erg s⁻¹ cm⁻², respectively. In this context, δ Cep falls near the quieter end of cool supergiant X-ray emissions. The X-ray detections of some supergiants carries important implications for the Cepheids and, specifically, their X-ray generating mechanisms, as discussed in Section 4.2.

The very low surface fluxes of δ Cep, coupled with moderately high kT values of 0.4–1.6 keV, imply that if the X-rays from δ Cep originate from quasi-uniform coronal structures, as happens for the Sun, the layer of emitting plasma is likely relatively shallow. However, another possibility is that the X-ray emissions of δ Cep originate in tightly confined, heated regions, covering a small fraction of the Cepheid's surface, as a result of pulsation-induced magnetic fields and structures. In this case the observed X-ray emissions could arise from the interactions of turbulent plasmas and the resulting magnetic fields via magnetic reconnection mechanisms that occur in solar flares (see Drake et al. 2006).

Because Cepheids are young (ages <200 Myr; Bono et al. 2005; Marsakov et al. 2013), cool companion stars would also be young and thus rapidly rotating. As such, they would be chromospherically active stars and likely coronal X-ray sources. As mentioned earlier, cool members of the ∼125 Myr old Pleiades cluster typically have ${L}_{{\rm{X}}}$ values of $\sim 0.5\mbox{--}10\times {10}^{29}$ erg s⁻¹ (Micela et al. 1996). Because ∼50% or more of Cepheids are members of binary or multiple systems (Evans et al. 2015a, and references therein), it is important to ascertain whether X-ray emissions (if detected from a Cepheid) indeed originate from the Cepheid itself or from a companion. Because Cepheids are luminous in visible light, any close main sequence G–K–M companions, if present, are very difficult to resolve, since they are at least ∼8 mag fainter than their corresponding bright Cepheid hosts and thus contribute insignificantly to the Cepheid's optical flux. However, at wavelengths below ∼1400 Å, including X-rays, the photospheric continuum fluxes of the Cepheids and any cool companions are no longer significant. At these high energies, the companion's X-ray emissions will not be overwhelmed by those of the Cepheid.

It is possible that an unresolved, active cool companion star (as proposed by Anderson et al. 2015) could contribute to the X-ray emission and variability. Cool main sequence stars (especially younger ones) are also X-ray variable (see Guinan et al. 2016; Stelzer 2016, and references therein) on timescales of hours (flares), days (stellar rotation bringing spots and active regions in and out of view), years (magnetic activity cycles), and even on timescales of millions to billions of years (weakening of the stellar magnetic dynamo as the star evolves and spins down). However, although the pulsation period of δ Cep is ∼5.37 days, as shown in Figure 3 the phase of enhanced X-ray activity is tightly confined, with the enhancement diminishing in a matter of several hours. This timescale excludes long-term variations such as magnetic activity cycles and evolutionary weakening of the magnetic field. This leaves only flares from, or the rotational X-ray variability of, the companion star as a possible explanation.

Zoom In Zoom Out Reset image size

Rotational X-ray variability has been observed in a small number of young, cool dwarfs. Flaccomio et al. (2005) used 10 days of Chandra integration time, spread over a time span of 13 days, to study X-ray variability of cool dwarfs in the Orion Nebula Cluster (ONC). At ∼1 Myr, this cluster is much younger than δ Cep, but this is one of the most thorough studies of rotational X-ray variability, which is why it is used as reference. Flaccomio et al. found several cool dwarfs with probable, periodic X-ray variability closely matching their optically determined rotation periods, with X-ray amplitudes of ∼2 times and even ∼3 times in some cases. At first, this would appear to be a promising alternative explanation for the observed X-ray variability of δ Cep. However, as shown in Figure 3, in ∼420 ks of X-ray observations obtained over 6 years, only the observations within  ±0.1 of 0.45ϕ show X-ray variability. The X-ray flux ranges are similar to the X-ray amplitudes of the known rotation-variable dwarfs. This would imply that the rotation period of the companion is ∼1 day, which is entirely possible for a young, cool dwarf. However, if the rapid rotation of a companion star was responsible, then observations at other phases would also show X-ray variability. This is not observed for δ Cep. Finally, if the rotation period of the companion was close to the pulsation period of δ Cep, then the X-ray curve would also show a smoother, quasi-sinusoidal variation, as opposed to the X-ray flux curve of δ Cep. This leaves flare activity from the companion as the only remaining alternative explanation.

For comprehensive studies of X-ray flare properties of cool dwarfs, we again reference the Chandra ONC data (Favata et al. 2005; Getman et al. 2008, and references therein). The Chandra ONC data is one of the best data sets for efficiently carrying out an X-ray flare study of multiple targets. Typically, compiling a statistically significant sample of cool, main sequence star flare characteristics is rather difficult at present, due to the high demand on X-ray satellite observing time and the large amounts of said time required to comprehensively study flares. However, the K dwarf AB Dor has a particularly rich X-ray data set, due to it lying in the foreground of the often observed Large Magellanic Cloud. Lalitha and Schmitt (2013) presented the light curves of 32 separate XMM-Newton observations of this star. With an age of ∼50 Myr, AB Dor is somewhat younger than δ Cep (or any companion stars), but nevertheless has an incredibly rich X-ray data set. Flares of various intensities and durations are present in nearly all of the 32 independent exposures. However, for δ Cep only, X-ray observations occurring in the range of $\sim 0.35\mbox{--}0.55\phi$ show significant variations. The X-ray exposures at other phases show no significant variations.

As a further, more precise check on the timing of the X-ray variations apparent near 0.45ϕ, we subdivided all exposures in the range of $0.3\mbox{--}0.8\phi$ into 10 ks bins (see Figure 3). This was the smallest bin size with a consistent signal to noise value above ∼3 for all data sets plotted. It also provides much better time and phase resolution, as shown in Figure 3. The XMM-Newton observations were phased according to the recent ephemeris used by Engle et al. (2014): ${T}_{\max }=2455479.905+5.366208(14)\times E$. In the interest of thoroughness, because the period of δ Cep is continually undergoing small changes (see Engle et al. 2014), recent AAVSO photometry was analyzed to ensure that the newly acquired Chandra data were correctly phased. The ephemeris measured from the AAVSO photometry analysis, and applied to the Chandra data, is ${T}_{\max }=2456859.039+5.366279(21)\times E$. This amounted to an essentially insignificant change of ∼0.001ϕ. As shown in Figure 3, when the recent Chandra X-ray data are combined with our previous XMM-Newton data, the data sets (taken more than 5 years apart) almost exactly match each other in flux levels as well as variability. The resulting X-ray flux curve leaves little doubt that the X-ray activity of δ Cep is indeed periodic and varies by up to a factor of four, peaking at $\phi \approx 0.45$. The precise phasing of the X-ray fluxes with the Cepheid's pulsation period indicates that the X-ray emission arises from the Cepheid itself and not from a companion.

As discussed in Engle et al. (2014), and previously by others (see Bohm-Vitense & Love 1994, and references therein), the phased FUV–UV emissions are best explained by pulsation-induced collisional shocks. These shocks originate near the He ii ionization boundary, within the star's interior, and are produced shortly after the star is most compressed and poised to rebound to rapid expansion. As recently measured by Anderson et al. (2015), δ Cep has a radial velocity amplitude of $K\approx 39$ km s⁻¹. Also, the interferometric study of Mérand et al. (2015) found the angular diameter of δ Cep to vary by ∼10%, with minimum radius occurring near ∼0.9ϕ and maximum occurring near ∼0.4ϕ (see Figure 1). The resulting kinetic energies associated with this pulsation could be sufficient to heat atmospheric plasmas and account for the FUV–UV emission lines with the observed plasma temperature from $\sim {10}^{4}\,\mathrm{to}\,3\times {10}^{5}$ K. The emitting plasma is heated by a fast moving shock front arising from the rapid expansion of the interior of the star after maximum compression during the "piston-phase" of the stellar expansion, beginning at phase ∼0.9ϕ. The observed phasing of the FUV–UV emissions prior to maximum brightness of the star is in general accord with this model. Moreover, emission lines show additional broadening at these phases for δ Cep (Engle et al. 2014) and for several other Cepheids (see Bohm-Vitense & Love 1994), which arises from increased turbulence of the gas as the shock front propagates outward through the less dense layer of the stellar atmosphere. From our study, for example, measures of the FUV Si iv 1393 Å line broadenings from HST-COS spectra indicate turbulent velocities of up to 225 km s⁻¹ near $0.9\pm 0.1\phi$, when these emissions reach maximum strength. The other FUV line emissions show similar behavior and broadenings. In addition, from the analysis of high dispersion International Ultraviolet Explorer (IUE) spectra of the Mg ii h+k emissions, Schmidt and Parsons (1984b) report radial velocity elements displaced up to  ±100 km s⁻¹ from the photospheric velocities. This may indicate the presence of fast moving, heated plasmas in the outer atmosphere of the star.

Identifying the X-ray mechanism operating in δ Cep is, at this time, unfortunately not straightforward. Hot plasmas with $T\gtrsim {10}^{5}$ K are indicated in δ Cep (and other Cepheids) by FUV emission lines such as N v 1240 Å. The presence of X-ray emission with plasma temperatures $T\gt 5$ MK is surprising, given that Cepheids were not known to be X-ray sources until recently. As shown in Figure 1, the FUV emission lines show a strong pulsation phase dependence, attaining peak emissions near $0.9\mbox{--}0.95\phi$ in the case of δ Cep . It was initially assumed that X-rays (if present) would also peak at or near these phases. As shown in this study, this is not the case. The X-rays are present at all phases but reach maximum strength near 0.45ϕ, i.e., nearly 3 days after the FUV lines peak. Though δ Cep is the first Cepheid to be identified so far as a definite X-ray source, we are attempting to confirm suspected X-ray variability in several other Cepheids to determine their X-ray properties. With additional observations of other Cepheids with different properties (masses, ages, pulsation periods, etc.), it should be possible to arrive at a better understanding of the X-ray mechanism. In what follows, we briefly discuss several of the most promising mechanisms and theories to explain the X-ray properties of δ Cep (and perhaps other Cepheids as well).

With our current understanding of the complex δ Cep atmosphere, it is difficult to successfully apply the shock model to explain the X-ray activity and variability. In addition (as shown in Figure 1), we have not yet obtained COS spectra of δ Cep near 0.45ϕ to search for broadening or increased FUV emissions from the X-ray emitting plasmas as they cool. IUE measures of the Mg ii 2800 Å h+k emission lines (Schmidt & Parsons 1984b) peak near 0.9ϕ (as do other FUV emissions) but show no apparent secondary enhancements near 0.45ϕ, where the X-ray emission peaks. Shock-heating the Cepheid's atmospheric plasmas to the point of generating the observed X-ray emissions (in the range of $\sim 3\times {10}^{6}$ to $20\times {10}^{6}$ K, as found from the best fitting plasma models; see Table 1 and Figure 2) would require large (>200 km s⁻¹) shock velocities. It is possible that the shock wave accelerates as it moves through the rarefying outer atmosphere of the Cepheid (Ruby et al. 2016). This acceleration, combined with the potential in-falling material expelled from the Cepheid during previous pulsations, could achieve such high relative velocities. However, as discussed later, the generation of magnetic fields in the outer convective atmospheres of Cepheids that commonly occurs throughout the cool half of the H–R diagram (Güdel & Nazé 2009) for main sequence stars may also be responsible (directly or indirectly) for the X-ray emissions.

As previously discussed, the measured large FUV emission line broadening values indicate that velocity fields are present and are much higher than the photospheric pulsation motions estimates of ∼39 km s⁻¹. The large velocities of >200 km s⁻¹ observed in the FUV emission lines suggest high Alfvenic velocities, and hence the presence of hot magnetized plasma (via $B=v\sqrt{4\pi \rho }$) that could subsequently generate X-ray emission. In this expression, B is magnetic field strength, v is velocity, and ρ is the density of the plasma.

The role of short-scale magnetoacoustic convective modes, entailing the generation of longitudinal tube waves, has also been explored in detail, as pursued in the framework of time-dependent simulations. By considering photospheric-level magnetic parameters (Fawzy & Cuntz 2011), as well as the detailed treatment of shock formation and dissipation (Fawzy et al. 2012), it was found that processes associated with those modes are insufficient to explain the X-ray observations.

Another possible mechanism for generating magnetic fields and the resulting X-ray emission in δ Cep (and perhaps other Cepheids) are via a convective zone, or through a combination of convective and pulsation-driven motions and turbulence, within the stellar interior (Narain & Ulmschneider 1996, and subsequent work). The X-ray variability of δ Cep requires a periodic amplification of the magnetic field, heating the atmospheric plasmas and increasing X-ray activity. There are a number of potential (still qualitative) theories for this magnetic mechanism, such as a post-shock increase of turbulence in the stellar interior and chromosphere that strengthens the magnetic dynamo effect, or magnetic reconnection events (Christensen et al. 2009). In the case of δ Cep (and maybe also for β Dor), the X-ray emissions peak near the maximum diameter of the star. Thus the enhancements in X-ray emission near this phase could be explained by a turbulent magnetic dynamo that strengthens in the expanding, cooling, and increasingly convective atmosphere of the stars.

We have also considered the hypothesis previously advanced by Ayres et al. (2003) to account for X-ray emissions in some cool giants and supergiants. Ayres et al. theorized that these stars could have magnetic features that scale in extent to those of the Sun and other cool main sequence stars, but their lower gravities would allow for much more extensive chromospheres. These chromospheres would act as X-ray absorbers, explaining why so few red giants were observed to be X-ray active. Cepheids are comparable to red giants in terms of gravity, but have larger masses, diameters, and higher surface temperatures. This could result in magnetic features large enough to extend beyond the Cepheids' bloated chromospheres, generating the persistent X-ray activity found at all phases observed. Further, several cool non-pulsating supergiants have also been detected in X-rays. Because these stars do not appear to pulsate, they should not be generating strong shocks. Their X-ray activity, therefore, appears likely to arise from solar-like dynamo-generated magnetic fields. This makes it possible that the Cepheids' pulsations are not responsible for generating the X-ray activity. Rather, the X-ray activity is persistent in a number of cool supergiants (certain Cepheids included), and the Cepheid pulsations simply serve to enhance the X-ray activity as a specific phase. FUV–UV emission lines and X-ray activity have also been observed in a small number of non-pulsating F and G supergiants (Ayres 2007). Although the number of detections is still small, the FUV-UV emission fluxes and X-ray luminosities of the Cepheids and non-pulsating supergiants are similar. This indicates that the same X-ray heating mechanism may be at work in both classes of supergiants, and the pulsations of the Cepheids then serve to periodically modulate or amplify this stellar dynamo driven mechanism.