The redshifted ultraviolet light from early stars at z \sim 10 contributes to the cosmic near infrared background .
We present detailed calculations of its spectrum with various assumptions about metallicity and mass spectrum of early stars .
We show that if the near infrared background has a stellar origin , metal-free stars are not the only explanation of the excess near infrared background ; stars with metals ( e.g . Z = 1 / 50 Z _ { \sun } ) can produce the same amount of background intensity as the metal-free stars .
We quantitatively show that the predicted average intensity at 1–2 { \mu m } is essentially determined by the efficiency of nuclear burning in stars , which is not very sensitive to metallicity .
We predict \nu I _ { \nu } / \dot { \rho } _ { * } \simeq 4 - 8 ~ { } { nW~ { } m ^ { -2 } ~ { } sr ^ { -1 } } , where \dot { \rho } _ { * } is the mean star formation rate at z = 7 - 15 ( in units of { M _ { \sun } ~ { } yr ^ { -1 } ~ { } Mpc ^ { -3 } } ) for stars more massive than 5 ~ { } M _ { \sun } .
On the other hand , since we have very little knowledge about the form of mass spectrum of early stars , uncertainty in the average intensity due to the mass spectrum could be large .
An accurate determination of the near infrared background allows us to probe formation history of early stars , which is difficult to constrain by other means .
While the star formation rate at z = 7 - 15 inferred from the current data is significantly higher than the local rate at z < 5 , it does not rule out the stellar origin of the cosmic near infrared background .
In addition , we show that a reasonable initial mass function , coupled with this star formation rate , does not over-produce metals in the universe in most cases , and may produce as little as less than 1 % of the metals observed in the universe today .