Discovery of the star that ionizes the North America and Pelican nebulae

Fernando Comerón (1) and Anna Pasquali (2)

(1) European Southern Observatory, Germany
(2) Institute of Astronomy, ETH Hoenggerberg, Switzerland


In July 2004 we used MAGIC and CAFOS at the CAHA 2.2m telescope to search for the ionizing star of the North America and Pelican nebulae. We focused our search in a circle of 1/2 degrees radius in the direction of the dark cloud L935 that separates both nebulae at visible wavelengths, where the ionizing star had long been expected to lie on the basis of geometrical considerations. The application of a near-infrared color-color criteria based on the 2MASS JHK photometry allowed us to narrow down our preliminary list of candidates to 19 stars, of which only 3 turned out to have early types according to the follow-up low-resolution infrared spectroscopy with MAGIC. Of these, one could be identified as a foreground F-type star and another has strong hydrogen lines indicative of an early A-type spectra. The third one is featureless at the resolution and signal-to-noise ratio of the spectrum obtained with MAGIC, but a spectrum in the visible obtained with CAFOS confirms its spectral type as O5, in excellent agreement with the prediction from the spectrum of the HII region. Its proximity to the expected position of the source near the geometrical center of the nebula, the lack of any other O-type stars in the large search region, and the excellent agreement between the spectroscopic and photometric characteristics of this star with those expected in a mid O-type star leads us to conclude that the long search for the ionizing star of these famous nebulae has finally succeeded. The results described in this report are presented in detail in Comerón and Pasquali (2005).


The complex formed by the North America and Pelican nebula is probably the most widely known HII region in the sky after the Orion nebula. Its distance to the Sun is only 600 pc (Laugalys et al. 2002), and its angular size of 3 degrees translates into a physical diameter of 32 pc. The North America/Pelican and the Orion regions differ in some fundamental ways: Orion contains dense molecular clouds with vigorous massive star forming activity, while the overall density of the North America/Pelican complex is much smaller (about 10 atoms per cubic centimeter), and its star forming activity, restricted to low masses, is confined to a few regions (Bally and Reipurth 2003).

Another surprising difference between both complexes is that, while the Orion nebula contains the most thoroughly studied young massive stars, the star or stars responsible for the ionization of the North America and Pelican nebulae have remained unidentified thus far. Hubble (1922) was apparently the first to propose that nearby Deneb may be the star responsible for the ionization of the complex. However, the relatively late type of Deneb (A2Ia) soon led to the realization that the ionizing star had to be found elsewhere, and in 1952 Sharpless and Osterbrock proposed the O6V((f)) star HD199579, which is projected on the Northern part of the North America nebula, as a much more suitable candidate. But HD199579 is not entirely convincing either: its position does not seem to be the one required to illuminate the bright rims seen scattered across the complex, and its spectral type is slightly too late to account for the nebular spectrum, which requires at least a O5 star. Recognizing this, Herbig (1958) proposed that the actual ionizing star had to be hidden by L935, the dark cloud in the foreground of the complex that separates the North America from the Pelican nebulae and causes among others the well-known 'Gulf of Mexico' feature in the former. Some subsequent searches in the infrared turned up some candidates: Bally and Scoville (1980) produced a list of 11 candidates that are bright in the infrared, but their observations did not provide conclusive evidence for any of them. Around the same time, Neckel et al. (1980) identified a heavily reddened near-infrared source as the likely ionizing star, but subsequent spectroscopy by Eiroa et al. (1983) showed it to be an unrelated cool giant instead.

Figure 1: A wide-field image of the North America and Pelican nebulae complex, where the yellow circle marks the area where we searched for its ionizing star. The image was obtained with the Schmidt Camera of Calar Alto, and was made available to us by courtesy of CAHA.

How to search for the ionizing star?

Even though the illumination geometry of the bright rims strongly suggests that the star responsible for the ionization of the North America and Pelican nebula must lie somewhere behind L935, the search area is still far too large for a star-by-star spectroscopic classification (Figure 1), and additional selection criteria other than the purely positional one are needed. We thus decided to apply a simple technique that we had developed and successfully applied to the identification of obscured massive members of the rich association Cygnus OB2 (Comerón et al. 2002). The technique is based on the fact that intrinsically bright, reddened background giants are easily separated from the also bright and reddened early-type association members in the 2MASS (J-H), (H-Ks) diagram, since both classes of objects cluster along two bands that are separated by about 0.4 mag in (J-H) for any given (H-Ks). In our study on Cygnus OB2, carried out in the summer of 2001 at the 1.23m CAHA telescope equipped with MAGIC (see Figure 2) we used this color-based criterion to select a sample of objects for near-infrared, low-resolution spectroscopic follow-up to determine the candidates whose spectroscopic characteristics were consistent with early-type photospheres. Subsequent studies in the visible of some of these objects (Hanson 2003) have confirmed that most if not all the objects that we classified as possible massive new members of Cygnus OB2 indeed have early spectral types.

Figure 2: Authors Comerón (left) and Pasquali (right) stand next to MAGIC at the 1.23m CAHA telescope at the beginning of the observing run in the summer of 2001 in which they confirmed a large number of new members of the Cygnus OB2 association.

It is straightforward to apply this technique to the search of the ionizing star of the North America and Pelican nebulae. To do this, we selected a search area of 0.5 degrees radius centered on RA(2000)= 20:55:17, Dec(2000)= +43:47:30, near the geometric center of the complex and close to the likeliest location of the ionizing star, as indicated in Figure 1. The radius of the search circle is large enough to include the entire area of the sky where the ionizing star could be located. Then, we extracted from the 2MASS catalog all the stars in this region brighter than Ks < 7.0 + 1.78 (H-Ks), ensuring the detection of any stars earlier than B0V out to a distance of 1 kpc. The dependency on (H-Ks) in our selection criteria is designed to correspond to a uniform detection limit in intrinsic magnitude regardless of the foreground extinction. We retained only stars with (H-Ks) < 2.0, which implies a visual extinction smaller than 32 mag in V for an intrinsically blue star. This is a safe threshold, higher than the highest extinctions measured in the direction of L935, and excludes highly reddened stars for which the J-band photometry becomes unreliable. We then used the reddening-free parameter Q = (J-H) - 1.70 (H-Ks), which measures the distance on the (J-H), (H-K) diagram to a vector that crosses the (J-H) = 0, (H-Ks) = 0 point. Stars with Q > 0.30 are closer to the reddening band defined by normal late-type giants, whereas Q < 0.30 characterizes stars with mid and early spectral types. We thus retained in our sample only those stars having Q < 0.30. The application of all these sample selection criteria left us with a sample of only 19 stars in the entire search area, a reasonable number for spectroscopic follow-up.

Although the Q < 0.30 criterion should select any early-type stars associated to the nebulae complex, including its ionizing stars, the selection may not be expected to be uncontaminated. The main source of contamination are cool stars such as AGB and carbon stars, as well as possible early-type, unrelated foreground or background stars. Of these, the cool stars should be easy to distinguish from early-type stars due to their prominent spectral features in the infrared (e.g. strong carbon absorption at 1.78 microns for carbon stars, or CO longwards of 2.29 microns for AGBs). Low resolution H- and K-band spectroscopy with MAGIC is sufficient to reliably separate such cool stars due to those features, whereas the low-resolution spectrum of O and early B stars should appear featureless.


Our low-resolution spectroscopy observations with MAGIC at the CAHA 2.2m telescope were obtained on the night of 30 July 2004. The grism used covers the entire 1.5-2.4 microns range at a resolution of 240 with the 1'' slit used. All 19 selected stars were observed with integration times ranging from 20 to 60 seconds, depending on the K magnitude of the source. Remarkably, 16 of the 19 objects were found to be cool stars: 5 were clearly carbon stars, 6 were AGB stars with prominent water vapor absorption, 3 were red giants with CO absorption but no water, and 2 had mid-to-late spectral types, with visible but moderate CO bands. Of the remaining 3, one had prominent Brackett lines of hydrogen indicating a spectral type A or late B. The second one had relatively blue colors inconsistent with it being located behind the L935 cloud, and we identified it as HD 199373, classified in the literature as F5V. In this way, we were left with only one candidate, J205551.3+435225, which was characterized by the MAGIC observations as a featureless object with a moderately reddened spectrum.

J205551.25+435224.6 has a faint, red visible counterpart listed in the USNO catalog as having B=15.5, R=11.7. A classification-quality spectrum in the 4000-5000 Angstrøm region is thus well within the reach of CAFOS at the 2.2m telescope. Although we had not initially foreseen the use of CAFOS, the discovery of a suitable candidate ionizing star and the realization that it was possible to nail down its accurate spectral classification with spectroscopy in the visible led us to request the use of that instrument to complete our observations. We are greatly thankful to the Calar Alto support staff for having accepted to install CAFOS at the telescope in advance of schedule so that we could still use it in the course of our observing run. In this way, on the night of 2 August 2004 we could integrate for 145 minutes on J205551.3+435225 and obtain the spectrum shown in Figure 3.

Figure 3: Visible spectrum of J205551.3+435225 obtained with CAFOS at the 2.2m telescope. The observed spectral features (see text) allow us to accurately classify it as O5V or possibly somewhat earlier, in excellent agreement with the spectral type expected from the emission spectrum of the North America and Pelican nebulae.

The ionizing star

Besides H Beta and H gamma, the spectrum of J205551.3+435225 displays prominent HeII Pickering and Fowler lines at 4200, 4542, and 4686 Angstrøm, leaving no doubt as to its classification as a O-type star. Also easily identifiable are the interstellar features at 4420, 4727, and 4762 Angstrøm. The prominence of the HeII lines is corresponded by a nearly total absence of HeI lines: the most intense HeI line in this interval, at 4471 Angstrøm, is barely visible. Its intensity ratio with respect to the nearby HeII line at 4542 Angstrøm, which is a sensitive spectral subtype indicator (Walborn and Fitzpatrick 1990), clearly indicates that J205551.3+435225 cannot be later than O5. On the other hand, the absence of NV lines and the ratio of the two HeII lines at 4686 and 4542 Angstrøm, both of which are luminosity class indicators in the mid-O range, rule out a luminosity class I or III. We thus classify J205551.3+435225 as a O5V star, noting that a somewhat earlier spectral subclass is not discarded.

Taking the tabulated average properties of O5V stars (Drilling and Landolt 2000; Tokunaga 2000) we obtain a visible extinction of 9.6 magnitudes and a distance of 610 pc, in excellent agreement with the distance to the North America and Pelican complex found in recent determinations (Laugalys and Straizys 2002) and with the typical extinction toward that region of L935 (Cambresy et al. 2002). Furthermore, the star is only 6'2 away from the estimated location suggested by Matthews and Goss (1980) based on the morphology of ridges seen in radiocontinuum maps. A X-ray source is detected in the Rosat All-Sky Survey close to the position of J205551.3+435225 with intensity and hardness properties similar to those expected from a mid- O star at that distance and with the amount of extinction derived from the visible and near-infrared data.

The properties of a O5V star also fit well with the bulk properties of the North America and Pelican complex derived from radio observations, for which a density near 10 electrons per cubic centimeter can be estimated using the measurements given by Wendker (1968) or Goudis (1976). The ionizing flux calculated for a O5V star (Schaerer and de Koter 1997) is only three times above that required to produce the observed radio continuum flux a spherical nebula of that density and 32 pc diameter as observed. Given the simplifying assumptions regarding the structure of the nebula that are needed to transform the observed radio flux into a ionization rate, the agreement can be considered as good. If the discrepancy between both determinations of the ionization flux turned out to be real it may indicate that the complex is density-bounded, rather than ionization-bounded, and that a fraction of the ionizing photons can escape the region.

Concluding remarks

J205551.3+435225 is a star of the right spectral type, at the right position, and at the right distance from the Sun to be the ionizing star of the North America/Pelican complex. Furthermore, it is the only star with these characteristics within a search radius that should include any other possible candidates. This makes us confident that the ionizing star of this famous complex has been finally found, after a long history of possible candidates starting with the pioneering work of Hubble over 80 years ago.

Our discovery raises a puzzling aspect, however. The O5V spectral type implies a mass of about 60 solar masses for J205551.3+435225, and such massive stars are well known to be gregarious, hardly ever being found in isolation. Our findings not only identify the ionizing star of the complex; they also rule out the possible existence of any other O or early B stars in the proximities of J205551.3+435225. The star is not a runaway, since proper motion measurements exist and are roughly those expected from the reflex solar motion at 600 pc in that direction. In view of this, and of the short lives of mid O-type stars, it rather appears that J205551.3+435225 may be a rare case of a massive star formed in isolation, or at least not in company of other similarly massive stars -yet another aspect in which the North America/Pelican complex differs from Orion, which hosts the best studied of the massive star clusters!