Anda di halaman 1dari 22

Chen A Calibration of Spectral Absorption Line Strengths to the Magnesium and Iron Content of Red Giant Stars Executive

Summary In recent decades, many new and powerful astronomical instruments, such as the HIRES spectrograph at the Keck observatory, have been constructed. Even so, the data acquired from these instruments can and should still be improved to increase their precision. The strengths of irons and magnesiums absorption lines were initially measured by analyzing how far each elements absorption line dipped. These values (called equivalent width values) were then compared to the metallicity and luminosity values of the stars. A mathematical equation, relating these three variables, was derived and calibrated to high resolution data points. The final equations can facilitate astronomical research by providing higher quality data. Precise values for metallicity, which is a ratio of elemental abundances, are especially important. Specific metallicities such as [Fe/H] can be compared to the magnesium to iron abundance of a star in order to discover more about its history. Because the [Mg/Fe] value changed following periods of supernovae explosions, the evolutionary history of a star is imprinted within its elemental abundance. In particular, these metallicity values can be applied to the stars of the Dwarf Spheroidal galaxies, which may have played a major role in the creation of the Milky Way galaxy. Moreover, the elements that make the different astronomical bodies are the very same elements that allow life to function on Earth. Ultimately, attaining more information about the origins of the different astronomical bodies can uncover secrets about the universe abroad as well as those at home.

Chen A Calibration of Spectral Absorption Line Strengths to the Magnesium and Iron Content of Red Giant Stars Abstract Advances in astronomy can be facilitated with precise data; however, some of the existing spectra, such as those for the Dwarf Spheroidal galaxies, have lower signal to noise ratios compared to the spectra of the globular clusters. This paper aims to calibrate a quantitative relationship between the luminosity, metallicity, and elemental equivalent widths to high resolution data from the Red Giant stars. The final equation acts as a proxy, allowing the derivation of these astronomical values from alternative sources(e.g. deriving the metallicity will only require the luminosity and equivalent width values), mitigating imprecision issues associated with low resolution data.1 In the end, two mathematical relationships were created for both magnesium and iron.

where

is the

and

is the equivalent width of the absorption line.

where

is

and

is the value for the equivalent width of the absorption feature.

The final calibrated equation for magnesium revealed a root mean squared error (RMSE) of 1.20 dex and a standard deviation of 0.181 for the high resolution [Mg/H] data and a RMSE of 1.74 dex and a standard deviation of 0.193 for the medium resolution data. Moreover, the average RMSE between the calibrated equation for iron and the actual [Fe/H] metallicities was
1

High resolution denotes spectra with R >15,000, where R= (Wavelength of Interest)/ (Width of the unresolved line). 2

Chen 1.19 dex with an average standard deviation of 0.224. There was a constant offset in the medium resolution magnesium metallicities, whose data came mostly from the Dwarf Spheroidal stars. Despite this offset, the low standard deviation of 0.193 for these medium resolution points still signifies a precise final equation. Nevertheless, further work should investigate the cause of this deviation. The calibrated equations can be applied to stars with low resolution data in order to facilitate important research such as extracting star formation histories.

Chen A Calibration of Spectral Absorption Line Strengths to the Magnesium and Iron Content of Red Giant Stars Introduction Clues about the origins of the Milky Way are imprinted within the chemical compositions of dwarf galaxies and globular clusters, which are dense collections of stars (ranging from 10 thousand to 10 million stars) bound together by their mutual gravity. Precise data is thereby necessary to facilitate such research and discover more about the universe. Even with the advent of advanced scientific instruments in recent decades, the precision of current astronomical data can and should still be improved. This paper aims to calibrate a quantitative relationship between the V magnitude, metallicity, and elemental equivalent width (EW) to high resolution data from the red giant stars of the Milky Way galaxy in order to develop a method for obtaining these same values for stars with lower resolution data.2 Metallicity is the proportion of the objects composition that includes elements other than hydrogen and helium, the V magnitude is a measure of a stars luminosity, and the equivalent width is a numerical measure of the strength of an elements absorption feature. This calibrated relationship can facilitate analysis of globular clusters, dwarf spheroidal galaxies (dSphs), or any other collection of stars with properties similar to those of the Milky Way globular clusters. The elements magnesium and iron were analyzed for several reasons. First, the magnesium and iron lines at 8689 and 8807 respectively were studied because they are the next strongest lines after those of the calcium triplet (the calcium absorption lines at 8498 , 8542 , and 8662 ), facilitating precise equivalent width extraction. Second, these elements are applicable to an expansive range of astronomical research. In particular, the calibrated
2

In this paper, high resolution denotes spectra with resolving power R >15,000, where R= (Wavelength of Interest)/ (Width of the unresolved line). 4

Chen measurements can be used as star formation indicators by analyzing the magnesium to iron ratio in relation to the [Fe/H] metallicity for globular clusters or the Dwarf Spheroidal galaxies. The latter may have very well been active players in the formation of the Milky Way (White & Reese 1978; Arimoto, Ikuta, & Jablonka, 2004). Star Formation Histories The aforementioned elements, magnesium and iron, can also facilitate the extraction of star formation histories (SFH) from globular clusters and dSph galaxies. The element oxygen is ejected by Type II supernovae while iron is created by Type Ia and Type II supernovae. The first star formation burst in the universe involved primarily Type II supernovae, and in this initial event, matter with high levels of [O/Fe] was ejected into the interstellar medium (ISM). However, when Type Ia supernovae later ejected a different mixture of elements into the ISM, this elemental ratio decreased due to a larger iron abundance (Tinsley, 1968). Even though Gilmore and Wyse (1990) examine star formation in relation to oxygen, the above sequence of events can also be applied to magnesium. Magnesium is created in a nuclear fusion reaction called the Alpha process, wherein the element oxygen is bombarded twice consecutively by an alpha particle to create neon and then magnesium. Because the abundance of magnesium is approximately correlated to that of oxygen, Gilmore and Wyses (1990) description of the star formation process can similarly be tracked through the magnesium to iron ratios. Furthermore, because [Fe/H] is approximately time dependent, where [Fe/H] increases roughly monotonically with time, analyzing [Mg/Fe] in relation to [Fe/H] allows one to discern star formation histories through a chemical evolution model. A chemical evolution model relating the relevant elements, iron and magnesium, will facilitate the SFH characterization.

Chen Iron At a constant iron abundance and with increasing V magnitudes, the strength of iron absorption lines in red giants declines. A fainter red giant is hotter, translating into a weaker iron absorption line (assuming a constant composition). With a constant V magnitude or temperature, the strengths of the absorption lines increase with increasing iron abundance. By looking at trends in the available data, the relationships between the metallicities, equivalent widths of the absorption features, and V magnitudes can be quantitatively examined. Magnesium The magnesium absorption line was analyzed in order to calibrate a relationship between the equivalent width, the , and the [Mg/H] metallicity. The current spectra for Dwarf

Spheroidal galaxies have lower signal to noise ratios than those of globular clusters. By calibrating the relationship between the equivalent width of the magnesium absorption line, the V magnitude and the metallicity, [Mg/H], to higher quality spectra, this same relationship can be applied to the spectra of dSph galaxies to attain more precise metallicity values, facilitating SFH extraction.

Theory and Methods Measuring the Absorption Features The spectra used in the research were created by the medium resolution Deep Extragalactic Imaging Multi-Object Spectrograph (DEIMOS) instrument from the Keck Observatory.3 In order to analyze the spectra, Espec2, a program coded in the Interactive Data

During the course of the research, the DEIMOS instrument was not used to acquire the spectra as analysis was conducted on already created spectra. For more information on acquiring spectral data, go to http://www2.keck.hawaii.edu/inst/deimos/. 6

Chen Language (IDL), was used to obtain equivalent width measurements. This program is capable of extracting EWs with three different methods: Lorentzian, Numerical, and Gaussian. The equivalent width is given by (Equation 1) where is the flux in the absorption line and is the flux in the continuum. Extracting the

equivalent widths was possible because iron and magnesium absorption features are relatively strong, making them distinguishable from the rest of the spectrum. Stronger lines have lower noise to absorption strength ratios, which allow the program to the fit the absorption features more precisely. The theoretical shape of an absorption feature can be fit with a Moffat function, which is a convolution of a Gaussian and Lorentzian function. However, for magnesium and iron, resolving the Moffat shape would require a spectrograph with the ability to obtain extremely high quality spectra and be exposed on any one star for an extremely long period of time. In fact, even Keck Observatorys highest resolution spectrograph, HIRES, which has a resolving power of up to 84,000 (Vogt 1994), is unable to resolve the Moffat profile for these elements absorption features. The absorption lines created by the medium resolution DEIMOS spectrograph are smeared, which means that the smoothing function that was used on the spectral data was Gaussian in nature; thus, the lower the resolution of the data, the more Gaussian the resulting absorption line. Moreover, the magnesium and iron absorption lines studied were narrow compared to the broader calcium triplet (the calcium absorption lines at 8498 , 8542 , and 8662 ), which further resulted in more Gaussian absorption features. The other possible option was numerical fitting, which computes the total area covered by the absorption line and derives an equivalent width value from this measurement. However, such a fitting method would have been more sensitive to the random fluctuations (Poisson noise) when 7

Chen deriving an equivalent width measurement. In the end, a Gaussian function was chosen to fit the absorption features as the spectral data were characteristically Gaussian and the other methods were imprecise. During the fitting process, the spectra with signal to noise ratios (S/N) lower than 10 per pixel were discarded. For example, Figure 1 depicts a noisy spectrum focused on the 8807 magnesium line that would have been eliminated from the data set. On the other hand, Figure 2 depicts a clean spectrum that would have been kept. Eliminating noisy spectra would not skew the results because the amount of noise depends on the stellar magnitude, which is independent of iron or magnesium content. The EWs were measured for 14 different globular clusters and 10 Dwarf Spheroidal galaxies across a total of 62 DEIMOS slit masks.

Figure 1: This spectrum depicts a noisy 8807 magnesium absorption line. Data obtained from spectra such as this were discarded.

Chen

Figure 2: This spectrum depicts the same 8807 magnesium absorption line as in the previous figure. Data obtained from spectra such as this were kept as this spectrum is quite clean. Data Stars with too large measurement uncertainties were eliminated. Those with s greater than 0 were taken out of the final data set. Moreover, [Mg/Fe] and [Fe/H] values with uncertainties greater than 0.3 were also eliminated. This data trimming is effectively a cut on the signal-to-noise ratio of the spectra. Such a cut does not bias the results of the calibration, as there is no systematic relationship regarding the size of the uncertainty.

Results and Discussion Iron There are several possible factors that can influence the strength of an elements absorption line. The effective temperature of the star and the abundance of the element have the largest effect on the line strength (Gray 2005). Initially, these two variables were analyzed independent of each other, meaning the absorption line strength was analyzed in relation to the effective temperature, without accounting for elemental abundance. Because the [Fe/H] 9

Chen metallicity for a star is constant across each globular cluster, the process of analyzing each variabless effect on line strength was facilitated as data analysis could then be conducted separately on each globular cluster without concern for a differential metallicity value.4 The homogenous nature of the [Fe/H] values within a globular cluster is theoretically believed to be due to the uniform elemental composition of the cloud from which the globular cluster came from. The equivalent width measurements were plotted against the values, which

are the distance-independent extinction-corrected luminosities of stars. Moreover, (Equation 2) where is the absolute extinction given by (Equation 3) where is the interstellar reddening or color excess. was used as a proxy

for the effective temperature due to the precision of its component measurements and its monotonic variance with the temperature, allowing it to serve as an adequate replacement. Polynomial functions of varying degrees were fit through the points and, in the end, a second degree function fit the data most precisely without potential over fitting complications. For each individual globular cluster, a second degree polynomial function was fit to the iron line strengths vs. . The resulting graph can be seen in Figure 3.

[Fe/H] Calibration A relationship between the [Fe/H] metallicity, the equivalent width values, and the was derived from the data. First, a second degree equation was fit for the iron equivalent width vs. graph for each of the globular clusters. The coefficients of the

where

and

are the numbers of each elements

atoms in a given volume.

follows the same structure. 10

Chen individual fit equations were averaged and the resulting aggregate equation is seen in Equation 4. (Equation 4) where is the equivalent width of iron, is , and is the y intercept of the and

function. The second degree sloping of the graphs (as measured by the coefficients for

of Equation 4) should theoretically be the same for all of the clusters. Equation 4 reveals that the first coefficient is consistent with 0, while the second coefficient has an uncertainty of 4.3. As such, it was initially assumed that the only difference between the fit equations of the different globular clusters were the y intercept values for each cluster. This assumption is proven later in this section to be statistically valid by comparing the theoretical [Fe/H] values to the actual [Fe/H] values. The [Fe/H] metallicity was further graphed versus the y intercept values, and another second degree function, located below, was fit through the points. (Equation 5) where is the same y intercept seen in Equation 4. Equation 5 and Equation 4 were further

combined to yield Equation 8, which relates all three variables: [Fe/H], iron EW, and .

(Equation 6) where denotes the and is the equivalent width of the absorption line.

Figure 3 reveals the expected negative correlation between these two values. Even though some of the clusters have larger spreads, this negative correlation can still be seen quite systematically in all of the globular clusters. Table 1 reveals that the iron calibrated equation acts as an adequate proxy as the average RMSE and standard deviation between the calculated [Fe/H] values and the actual [Fe/H] values are 1.19 and 0.224 respectively. The precision of the final 11

Chen [Fe/H] equation further confirms that the second degree fitting of the EW vs. (Equation 4) is statistically adequate. Figure 3 does not display all 14 globular clusters as there were either not enough points to conduct a second degree fit or the spread in the points was too large to meaningfully fit such a function. Moreover, the data for the Dwarf Spheroidal galaxies were not included in the [Fe/H] calibration as each galaxy does not have a constant metallicity, which is a prerequisite for this method of calibration. However, Equation 6 can still be applied to dSphs galaxies despite this characteristic difference as only the method of calibration, not the equation itself, requires a homogenous metallicity within a cluster. Globular Cluster RMSE(between Eq. 6 and
[Fe/H] Actual)

data

Standard Deviation(between Eq. 6


and [Fe/H] Actual)

N288 0.289 0.126 N1904 1.05 0.120 Ng5904 0.653 0.158 N6205 0.695 0.168 N6838 0.301 0.082 N7078 3.39 0.365 N7492 1.14 0.309 N7089b 0.788 0.174 N2419 2.41 0.514 Table 1: The RMSEs between Equation 6 and the actual [Fe/H] values were calculated. The average RMSE is 1.19 and the average standard deviation is 0.224.

12

Chen

Figure 3: This graph depicts iron equivalent width vs. . A second degree polynomial function is also fit through each globular clusters points. The globular clusters located in this figure are NGC 7492, M13, M5, NGC 2419, NGC 288, M79, M2, and M15. Not all of the globular clusters analyzed are located in this figure as (a) there were not enough points to fit a second degree function or (b) the EW spread was too large, indicating problems with the quality of the spectra. Unnatural kinks in the graphs can be attributed to minor graphics rendering issues, which have no effect on the precision of the final calibrated [Fe/H] equation.

13

Chen Magnesium Whereas [Fe/H] is constant for the stars in each globular cluster, this is not the case for the magnesium abundance. Therefore, a different approach was necessary to analyze the nature of the magnesium absorption features. , the EW values for magnesium, and the

[Mg/H] metallicity values were analyzed together instead of separately by graphing these values in three dimensions. A two dimensional surface was fit through the points to examine the relationship between these three variables. A function in IDL called Sfit facilitated this process. The function fits a 2 dimensional planar equation. (Equation 6) where is the coefficient of the th and th x and y variables, is and is the

equivalent width measurement, with both to their respective th and th powers. During the fitting process, potentially over fitting the data was an issue, so careful steps were taken to prevent using an equation with too high of a degree, which can factor in minor fluctuations in the data caused by noise instead of a potential underlying relationship. In order to prevent this problem, the max degree option of the function was also set so that could be no

higher than the user inputted degree, limiting the number of parameters in the resulting fit equation to 10. The resulting derived equation is located below.

(Equation 7) where is and is the value for the equivalent width of the absorption feature.

The Precision of the Data 14

Chen Figure 3, Figure 5, Figure 6, and Figure 7 reveal varying degrees of data spread. Poisson noise and operator error are a couple sources of random data fluctuation. The first accounts for much of the imprecision and it is caused by the inherent limitations in photon counting statistics. In other words, spectral extraction is inherently imprecise in that data must come from a limited number of photons. Operator error refers to human error in measuring the equivalent widths. Because some of the EWs were measured by the operator imposing wavelength boundaries (the operator visually selects a minimum and maximum bound on the spectrum in order to measure the equivalent width between the bounds), this type of error may also account for much of the data fluctuation.5 The iron absorption line at 8805 , which could contaminate and skew the 8807 magnesium line, is another potential source of error. This phenomenon can be seen in Figure 4. The iron line can be seen partially melding with the larger magnesium absorption line. However, this effect is not significant as the contamination is likely to be less than 10%, which translates into an error of about 0.1 on the [Mg/Fe] range (conversation with Kirby 2010). Moreover, because this melding systematically affects all of the EW measurements, the effect of the contamination can be further mitigated as it does not preferentially influence certain spectra.

To clarify, the majority of the EWs were measured by the program algorithm imposing wavelength boundaries. However, if the program failed, then the operator had to rectify the situation by visually carrying out this process. 15

Chen

Figure 4: The Mg 8807 absorption feature is contaminated by the 8805 iron absorption feature. Magnesium To evaluate the precision of the resulting surface equation for magnesium, a graph was created of the [Mg/H] metallicity values from the fit equation vs. the actual [Mg/H] values for the high resolution stars.6 These two data sets were graphed against each other and a 1:1 line was overlaid across the points. The points were also color coded by globular cluster to examine whether there was any systematic deviation due to a particular globular cluster. This graph can be seen in Figure 5. In Figure 5, there is a larger spread around -1.3; however, this aberration is not caused by a particular globular cluster preferentially giving a higher or lower value of [Mg/H] as there is a random spread in the color across the 1:1 line, denoting a non systematic spread in the globular clusters metallicity values. The [Mg/H] values from the fit equation were then graphed against the actual [Mg/H] values for the medium resolution stars to further ensure the surface equations
6

Due to the logarithmic nature of metallicity and because [Mg/H] metallicity values were not directly located in the data, [Mg/H] values were acquired by adding together [Fe/H] and [Mg/Fe] values. 16

Chen precision; this graph is seen in Figure 6. For this second graph, there is a constant offset of the points from the 1:1 line. The majority of the medium resolution points also came from the Dwarf Spheroidal galaxies, so the offset can probably be attributed to a unique characteristic of dSphs that can cause a systematic [Mg/H] deviation. However, the exact underlying cause is unknown, and further research should investigate this phenomenon. Finally, a three dimensional graph was created of the derived [Mg/H] metallicity from the fit equation vs. the magnesium equivalent width vs. the . Figure 7 is the side view of this graph.

Figure 5: The [Mg/H] Fit vs. [Mg/H] Actual graph for the high resolution data is plotted and a 1:1 line is overlaid across the points. The points are further color coded by cluster. None of the clusters preferentially give higher or lower [Mg/H] values as there is a random spread in the color across the line, denoting a non systematic spread in the globular clusters metallicity values.

17

Chen

Figure 6: [Mg/H] values from the fit equation are plotted against the actual [Mg/H] values for the medium resolution dSph stars. There is a constant data offset from the 1:1 line.

Figure 7: This is the graph of vs. Magnesium Equivalent Width vs. [Mg/H] with the opacity set to 0.5. The fainter blue points represent points behind the plane of the graph while the darker blue points represent points in front of the plane. There are 23 points below the plane and 20 points above (above and below refer to changes along the [Mg/H] axis), demonstrating a relatively equal distribution across the plane. 18

Chen

Further error analysis was conducted by finding the root mean squared error and standard deviation. The final calibrated equation for magnesium has a RMSE of 1.20 and a standard deviation of 0.181 for the high resolution [Mg/H] data and a RMSE of 1.74 and a standard deviation of 0.193 for the medium resolution data. Moreover, the average RMSE between the derived [Fe/H] metallicity values from the fit equation and the actual [Fe/H] metallicities is 1.19 with an average standard deviation of 0.224. Even with the aforementioned offset in the [Mg/H] metallicity data for the medium resolution points, the low standard deviation of 0.193 still signifies a precise final equation. Nevertheless, further investigation should be conducted to discern the cause of this offset. Moreover, the visually apparent spread at -1.3 for the high resolution [Mg/H] data in Figure 5 is not a statistically significant problem as the standard deviation is still 0.181.

Conclusion and Future Work In this paper, magnesium and iron equivalent widths were mathematically analyzed in relation to the and the metallicity. The relationships were further calibrated to the

available high quality data points in order to formulate an equation that can be applied to other clusters and galaxies. The inverse relationship between the equivalent width of iron and the luminosity confirmed established knowledge, and a second degree function produced a standard deviation of 0.224 and RMSE of 1.19 for the final calibrated equation. The ability to derive numerical data from alternative sources is crucial in astronomy or any science. In the field, high enough quality spectra and data do not exist for every globular cluster and galaxy, which impedes progress in the field and slows further discoveries about the universe. A calibrated mathematical equation allows for data verification as well as a proxy 19

Chen method to access more precise data, which increases the confidence level of results and conclusions. One particular application of the calibrated equations is deriving star formation histories, which necessitates precise [Mg/Fe] and [Fe/H] values to create chemical evolution models. The Dwarf Spheroidal galaxies are especially nebulous in that the current data for these galaxies are of lower precision than those of globular clusters. The calibrated relationships for magnesium and iron can be applied to these galaxies in order to discover more about their origin and nature of their creation. The elements that make the different astronomical bodies are the very same elements that allow life to function on Earth. Ultimately, obtaining more information about the origins of the different astronomical bodies can uncover secrets about the universe.

Future Work Future work should consider investigating the cause of the constant offset in the medium resolution [Mg/H] data. Even though, statistically, this offset should not have a significant effect on the final calibrated equation, as evidenced by the low standard deviation of 0.193, this phenomenon is particularly interesting in that these points are mainly composed of Dwarf Spheroidal galaxy stars, not those of globular clusters. Finding the underlying cause of this offset can potentially reveal key differences in the composition and attributes of these two types of astral entities. Moreover, the calibrated relationships can be applied to the Dwarf Spheroidal galaxies or globular clusters to extract star formation histories. Analyzing the former can potentially reveal the processes that occurred during the birth of the Milky Way galaxy (White & Reese, 1978; Arimoto, Ikuta, & Jablonka, 2004).Mathematical relationships can also be derived for the alpha

20

Chen elements (O, Ne, Si, S, Ar, Ca, Ti) other than magnesium. These other elements can be used in place of magnesium in extracting star formation histories as their activity mimics that of oxygen in the star formation process outlined by Gilbert and Wyse (1990).

21

Chen References Arimoto, N., Ikuta, C., & Jablonka, P. (2004, February). Star Formation Histories of Dwarf Galaxies in the Local Group. Gilmore, G., & Wise, R. F. (1990, November). Chemical Evolution with Bursts of Star Formation: Element Ratios in Dwarf Galaxies. Chicago, Illinois: Astrophysical Journal Letters. Gray, D. F. (2005). The Observation and Analysis of Stellar Photospheres (3rd ed.). New York: Cambridge University Press. Reese, S. D. M., & White, M. J. (1978, May). Core condensation in heavy halos - A two-stage theory for galaxy formation and clustering. London: Royal Astronomical Society. Rutledge, G. A., Hesser, J. E., & Stetson, P. B. (1997, May). Galactic Globular-Cluster Metallicity Scale from the Ca II Triplet. II. Rankings,Comparisons, and Puzzles (Paper No. II). Chicago, Illinois: Astronomical Society of the Pacific. Rutledge, G. A., Hesser, J. E., Stetson, P. B., Mateo, M., Simard, L., Bolte, M., . . . Copin, Y. (1997, May). Galactic Globular-Cluster Metallicity Scale from the Ca II Triplet. I. Catalog (Paper No. I). Chicago, Illinois: Astronomical Society of the Pacific. Starkenburg, E., Hill, V., Tolstoy, E., Gonzalez Hernandez, J. I., Irwin, M., Helmi, A., . . . Jablonka, P. (2010, February). The NIR Ca II Triplet at Low Metallicity Searching for Extremely Low-Metallicity Stars in Classical Dwarf Galaxies. Les Ulis, France: Astronomy and Astrophysics. Tinsley, B. M. (1968). Evolution of the Stars and Gas in Galaxies. Vogt, S. S. (1994, May 4). HIRES Users Manual [Brochure].

22

Anda mungkin juga menyukai