Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Detailed Mass-Luminosity Relation for Main Sequence Stars

  1. Jun 7, 2004 #1


    User Avatar
    Science Advisor

    Does anybody know of a decent curvefit for the exponent in the main sequence mass-luminosity relationship?

    L/Lsun = (M/Msun)^a

    One constant fits all (e.g., 3.5) doesn't seem to be good enough. I don't like piecewise discontinuous approximations because you'd get two different answers for the luminosity at the masses where one piece ends and the next one begins.

    Somebody's probably worked out a curvefit for the parts of the main sequence for which the ML relation is known, but I can't seem to find it anywhere.

    Anybody know?

    Jerry Abbott
  2. jcsd
  3. Jun 9, 2004 #2


    User Avatar

    I would also like to know this. I have been searching the internet and have only found a few more detailed approximations. Even these few approximations do not agree.

    An example of disagreements are the following:

    one site said:

    While anothe rsite told me something like masses close to one sm (I can't remember the range(something like .5 to 5 sm)) are repersented by L = M^4, while mass above and below the range are repersented by L = M^3.3. The two above examples make me wonder which is closer.

    I believe they are only accurate for masses closer to the solar mass (if you consider around .5 to 10 sm close). This is annoying for me since I am trying to construct a pseudoscientific model of a galaxy for a game I am trying to create. Also I need to know a function that relates mass to the probability of the star occurrence. Finally, although I probably need to search on my own more, I need to know the average distance between stars (6 to 7 ly) including the standard deviation.
    Last edited: Jun 9, 2004
  4. Jun 9, 2004 #3


    User Avatar

    This link may help :uhh:


    Although I did not take a close look, I believe it suggests that, at least so far, there is no "accurate" mass-luminosity relation. The mass-luminosity varies greatly to some extent and is, therefore, only an approximation. Even this page (the link I provide) is, I believe, only looking at masses ranging form .5 to 2 solar masses.

    So the jumps in "piecewise discontinuous approximations" may not be as bad as you thought, or I. Still I would like a better approximation then L = M^n
    n being between 3 and 4. Importantly for me, I would like the relation to be accurate for mass from .008 to 100 solar masses (although I wonder whether the accuracy drops with the increase of mass).
  5. Jun 9, 2004 #4


    User Avatar
    Science Advisor

    I found something at
    http://www.phys.unm.edu/~duric/phy536/3/node2.html [Broken]

    For M/Msun < 0.43, b = 0.23 and a = 2.30

    For M/Msun > 0.43, b = 1.00 and a = 4.00

    where L/Lsun = b (M/Msun)^a

    That sounds plausible for M/Msun below about 3.0. Above there, the exponent (a) seems to trail off slowly to about 3.4.

    There's a graph of data points at
    http://www-astronomy.mps.ohio-state.edu/~pogge/Ast162/Unit2/mlrel.gif [Broken]
    (Bad color scheme though; the green used for the points is too light.)

    You could fit your own log-log lines by measuring with a rular, then convert to the exponent form.

    There might be some difficulty getting the ML relation for very low mass main sequence stars because their dimness makes it hard to observe many eclipsing binaries and also because the flares of some of them have a higher brightness in proportion to the usual luminosity of the star itself.

    Jerry Abbott
    Last edited by a moderator: May 1, 2017
  6. Jun 9, 2004 #5


    User Avatar
    Science Advisor

    Yes, the composition is different. Some stars have more metals than others do, and that might make the main sequence ML diagram fuzzy. But it's worth going after the detailed ML relation for the Population I star of average composition, like the sun is maybe.

    The star mass low bound is 0.08, I think. I forgot what it's called. The Chandrasakar mass maybe?

    Jerry Abbott
  7. Jun 10, 2004 #6


    User Avatar

    Yeah, .08 not .008. My mistake.

    On further a look I found that m^4 is bad for high mass stars. According to my information stars with solar masses around 6 sm correspond to close to 300 L. m^4 gives a luminosity of nearly 1300. This is a little over 4 times the luminosity that my sources give me. I will try approximating like you suggested and, perhaps, check my sources.
    Last edited: Jun 10, 2004
  8. Jun 10, 2004 #7


    User Avatar
    Science Advisor

    Mass Luminosity Relation (from data on graph)

    I eyeballed these points for drawing lines between.

    point 1: M=0.1, L=0.0012
    point 2: M=0.2, L=0.0052
    point 3: M=0.4, L=0.025
    point 4: M=1.0, L=1.0
    point 5: M=1.42, L=4.0
    point 6: M=2.35, L=49
    point 7: M=5.9, L=1200
    point 8: M=42, L=41000

    Taking logarithms:

    point 1: -1.000, -2.921
    point 2: -0.699, -2.284
    point 3: -0.398, -1.602
    point 4: 0, 0
    point 5: +0.152, +0.602
    point 6: +0.371, +1.690
    point 7: +0.771, +3.079
    point 8: +1.623, +4.613

    P = log M
    Q = log L

    point 1 to point 2: Q = 2.116 P - 0.805
    point 2 to point 3: Q = 2.266 P - 0.700
    point 3 to point 4: Q = 4.025 P + 0.000
    point 4 to point 5: Q = 3.960 P + 0.000
    point 5 to point 6: Q = 4.968 P - 0.153 :eek:
    point 6 to point 7: Q = 3.472 P + 0.402
    point 7 to point 8: Q = 1.800 P + 1.691 :grumpy:

    Linear fit mass-luminosity relations:

    [M < 0.2], b=0.157, a=2.116
    [0.2 < M < 0.4], b=, a=2.266
    [0.4 < M < 1.0], b=1, a=4.025
    [1.0 < M < 1.42], b=1, a=3.960
    [1.42 < M < 2.35], b=0.703, a=4.968
    [2.35 < M < 5.9], b=2.523, a=3.472
    [5.9 < M < 42], b=49.09, a=1.800

    L = b M^a

    The high coefficient (49.09) tells me that I probably aimed too low on the bright end of the brightest segment of the main sequence. I wonder whether the jump in the exponent (a) from M=1.42 to M=2.35 might have something to do with the C-N-O fusion cycle starting to dominate proton-proton.

    A quadratic log curvefit might be better. Smoother, anyway. More likely to please the critical people who go number-crunching after science-fiction authors.

    P = log M

    point 1 to point 3 (mass under 0.4 solar masses):
    Q = +0.24853 P^2 + 2.5385 P - 0.6310

    point 3 to point 5 (mass from 0.4 to 1.42 solar masses):
    Q = -0.11746 P^2 + 3.9784 P + 0.0000

    point 5 to point 7 (mass from 1.42 to 5.9 solar masses):
    Q = -2.4160 P^2 + 6.2316 P - 0.2894

    L = 10^Q

    I don't have any good guesses for the luminosity above ~6 solar masses yet.

    Here are some points from the quadratic log fit.

    M, L

    0.1, 0.0012
    0.2, 0.0052
    0.3, 0.0129
    0.4, 0.0250
    0.5, 0.0619
    0.6, 0.1293
    0.7, 0.2404
    0.8, 0.4105
    0.9, 0.6572
    1.0, 1.0000
    1.1, 1.4604
    1.2, 2.0620
    1.3, 2.8300
    1.4, 3.7918
    1.5, 5.4077
    1.6, 7.6196
    1.7, 10.432
    1.8, 13.929
    1.9, 18.195
    2.0, 23.311
    2.5, 64.241
    3.0, 136.09
    3.5, 243.14
    4.0, 386.06
    4.5, 562.71
    5.0, 769.02
    5.5, 999.87
    6.0, 1249.7

    Jerry Abbott
    Last edited: Jun 10, 2004
  9. Jun 10, 2004 #8


    User Avatar

    I just thought of something. One could try to find an best fit equation or equations that relates mass to radius. I believe luminosity is both strongly related to mass and radius. Using this information the complexity of the equation would increase and, perhaps be more accurate.

    Also, if you are interested *shrug* data points for mass, luminosity, and radius can be found at:

    http://www.phy.umist.ac.uk/Teaching/PastExams/MoreExams2/00-2/3/P611_99-00/node2.html [Broken]


    Last edited by a moderator: May 1, 2017
  10. Jun 10, 2004 #9


    User Avatar
    Science Advisor

    There might not be enough data for a bivariate model L(M,R). And R seems to be a fairly tight function of M, too. For most of the main sequence,

    R = M^0.72

    Or something like that.


    L = M^a


    R = M^b

    then the effective temperature is

    T = 5770 Kelvin M^(a/4 - b/2)

    You can develop equations for the star's average density and time-on-main-sequence, too.

    Jerry Abbott
  11. Jun 10, 2004 #10


    User Avatar
    Science Advisor

    Whoever made those websites used different data than I used. If I had an observatory, I'd probably go after definitive answers. There's too wide a variation in the M-L relation for it to be a matter of composition differences only.

    Jerry Abbott
    Last edited by a moderator: May 1, 2017
  12. Jun 10, 2004 #11


    User Avatar

    I am not sure if this is helpful but I came across the following from a Russian astronomy journal:
    http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=ATROES000042000006000793000001&idtype=cvips&gifs=yes&jsessionid=2345211086884141292 [Broken]

    I believe that in the above article Msub(bol) mean the bolometric magnitude.
    Last edited by a moderator: May 1, 2017
  13. Jun 10, 2004 #12


    User Avatar

    Perhaps there is not enough data like you said or it would not be any more accurate but doesn't luminosity equal 4piR^2T^4 ? Where pi is 3.14..., R is the radius, and T is temperature. If there was enough data one could relate mass and radius, and also relate mass and temperature using an approximation method. In relating them you would put them in the following forms: R(M) and T(M). Substituting R(M) for radius and T(M) for temperature in the above luminosity relation would give you L(M). Perhaps this new approximation would be more accurate. :confused:
  14. Jun 11, 2004 #13


    User Avatar
    Science Advisor

    The relations L(M) and R(M) are determined empirically from observations of eclipsing binary stars. Once they are known, you can use the Stephan-Boltzmann relation to get T(M) because it defines a star's effective temperature. This is why the effective temperature is distinct from color temperature, which is based on U-B and B-V color excess, although the two are usually fairly close.

    Jerry Abbott
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook