MATERIALES, MANO DE OBRA, EQUIPO, MAQUINARIA Y HERRAMIENTAS

We have performed a comparison of SFR estimates using Hα, bolometric infrared

and 2800 ˚A luminosities using the 38 common sources found between Pascual’s

sample [250] and SWIRE N1 catalogue. In order to compare the infrared and Hα star formation rates we used the relations:

SF R (Myear−1) = 4.5 × 10−44LIR (erg sec−1) (1.11)

SF R (Myear−1) = 7.9 × 10−42L(Hα) (erg sec−1). (1.12)

Equation 1.11 depends on the total infrared luminosity and assumes that young stars dominate the radiation field throughout the UV–visible regimes and that the optical depth is large. The total infrared luminosity is then a good representation of the bolometric luminosity of the galaxy. Although this situation may hold for LIRGs, it does not apply for normal galaxies where the emission may be dominated by a cirrus component [211]. To estimate the SFR(IR) we used the bolometric infrared luminosities (1–1000 µm) given in the SWIRE catalogue, which was compared with Pascual’s catalogue.

Equation 1.12 comes from evolutionary synthesis models, assuming solar metal- licity and no dust. In this case the total integrated stellar luminosity shiftward of the Lyman limit is reemitted in the nebular emission lines. In order to measure the Hα luminosity and correct the Hα flux, we follow Pascual’s procedure [250] which was based on a given magnitude scale [115]:

mHα = − 17.0 − 2.5 log fHα (1.13)

Hα flux was corrected for the presence of [NII]λλ6548, 6584 lines. In addition a

mean internal extinction correction was applied to the objects and an 8 % statistical correction to the measured flux. The corrected flux is given by:

f0(Hα) = f (Hα) × 100.4AHα × (1 + (Hα/[NII]λλ6548, 6584)−1)−1 × 1.08 (1.14)

where Hα/[NII]λλ6548, 6584 = 2.33 and AHα = 1. The contamination from other

emission lines such as [OII]λ3727, Hβ and [OIII]λλ4959, 5007 found to be negligible

Figure 1.13: Comparison of the SFRs derived using the 2800 ˚A and the Hα luminosities. The straight line is the best fit for the data.

Hα luminosity is derived from the corrected flux using:

L(Hα) = 4π D_L2 f0(Hα) (1.15)

where DLis the luminosity distance related to the transverse comoving distance by:

DL = (1 + z) DM. (1.16)

For a universe with a cosmological constant the transverse comoving distance DM

does not have an analytical solution. In order to calculate the luminosity distance

we used the numerical recipe LUMDIST routine in IDL. This routine calculates DL

of an object given its redshift, the Hubble parameter and the deceleration parame- ter. In this case each object’s redshift is retrieved from the SWIRE N1 catalogue. The Hubble parameter was set to 72 km/sec/M pc and the deceleration parame-

ter was set to default. All sources that exhibit SF R(IR) < 10−9, mainly due to

misidentifications, were removed from our sample. Our calculation agreed totally with the results produced by Pascual [250] since the faintest object in our sample had exactly the same Hα flux and luminosity as those measured by Pascual, namely

7.22 × 10−16 erg sec−1 cm−2 and 3.78 × 1041 erg sec−1.

Figure 1.12 shows the SFRs derived using the infrared and Hα luminosities. The straight line was fitted by using the numerical recipe LINFIT routine in IDL. This

function fits the paired data to the simple linear model

y = A + Bx (1.17)

by minimising the chi–square error statistic. In this case y is the SFR(IR) and x is the SFR(Hα). A, B are estimated automatically by LINFIT. As a result it produces a least squares fit to the data with a straight line. Finally we performed a Spearman rank correlation test for both SFRs which produced a correlation coefficient of 0.95 confirming the strong correlation between the infrared and Hα SFR estimators.

A similar procedure was performed in order to compare SFRs derived from Hα

and 2800 ˚A luminosities. The 2800 ˚A SFR is derived from the relation:

SF R (M year−1) = 1.4 × 10−28Lv(ergs sec−1Hz−1). (1.18)

The 2800 ˚A luminosity is used as an SFR tracer in objects at redshifts z > 2 where

all strong emission lines except Lyα are shifted outside the optical range. As men- tioned before this equation is derived from evolutionary synthesis models and is

valid between 1500 and 2800 ˚A , a region not affected by the Lyα forest and the

contribution from old populations.

To calculate the L(2800 ˚A) we used the absolute magnitudes at 2800 ˚A given at

our SWIRE catalogue which was compared with Pascual’s catalogue. The relation

between the flux and the absolute magnitudes at 2800 ˚A was taken from Hayes &

Latham [142]:

m(2800 ˚A) = − 48.6 − 2.5 log f (2800 ˚A). (1.19)

L(2800 ˚A) is derived from the f (2800 ˚A) using:

L(2800 ˚A) = 4π D2_L f (2800 ˚A) (1.20)

where DL in this case is equal to 10pc since these are absolute magnitudes. The

comparison of the 2800 ˚A and the Hα SFRs is given in Figure 1.13. The straight

line was fitted by using the same numerical recipe LINFIT routine in IDL. In this

case y is the SFR(Hα) and x is the SFR(2800 ˚A). The Spearman rank correlation

test for these two SFRs produced a correlation coefficient of 0.92. The agreement of the different star formation rate estimators implies that the infrared emission results from the same young stellar population that produces the Hα emission and suggests that the dust is heated in close proximity to the young star-forming regions.

Figure 1.12 demonstrates that there is good agreement between the SFR(Hα) and SFR (IR) for sources at z < 0.3 while for the higher redshifts, and as a result higher luminosities, the agreement becomes poorer. Assuming that both are valid mea- sures of the true star formation rate it is obvious that the SFR(Hα) traces almost all the SFR as derived from the infrared luminosity. However, this is not true for the more luminous objects which becomes obvious for the 3 brightest infrared sources in this sample where the SFR(IR) estimate exceeds the one from SFR(Hα) by a factor of > 2. This is indicative of the presence of highly obscured dust-enshrouded star-forming regions within these galaxies which means that more precise extinction correction has to be applied to high-z sources and that there should be a reddening correction taken into account. The comparison between SFR(UV) and SFR(Hα) (Fig. 1.13) shows that for most of the galaxies the SFR(Hα) values are above the 1 : 1 relation. All these sources lie at z > 0.3 while the remaining sources, below the 1 : 1 relation are sources at z < 0.242. The latter implies that the UV extinction correction might be underestimated and the actual UV flux be slightly higher. These results imply that UV/Hα decreases with increasing SFR for galaxies at 0 < z < 0.5 whereas for low star formation rate objects the UV luminosities lead to higher star formation rates than Hα. In the case of the high redshift objects it is obvious that the UV flux underestimates the SFR which can be either attributed to variations in the star formation history due to starburst events or an increasing attenuation of the UV emission relative to Hα for higher luminosity galaxies.