Chinese Journal of Astronomy and Astrophysis manusript no.(LATEX: henry_2005_01.tex; printed on May 27, 2006; 16:33)Bright Light from Dark Matter(Baryoni and Non-baryoni)Rihard Conn Henry ?Henry A. Rowland Department of Physis and AstronomyThe Johns Hopkins UniversityBaltimore, MD 21218 USAReeived 2005 July 31; a

epted 2006 Marh 01Abstrat Preliminary investigation of GALEX images of the di�use ultra-violet bakground radiation suggests that part of the di�use bakground rep-resents red-shifted Lyman � reombination radiation from the intergalatimedium, whih in turn would require, to maintain the observed high degreeof ionization of the intergalati medium, that the non-baryoni dark mat-ter deay with the emission of an ionizing photon. Tehniques for verifyingthis and for treating the data are desribed in some detail.Key words: tehniques: ultraviolet { osmology: di�use bakground1 INTRODUCTIONWith olleagues Jayant Murthy and Luiana Bianhi, and with the help of two students(P. Shalima and N. V. Sujatha) at the Indian Institute of Astrophysis, I am urrentlyinvestigating the di�use ultraviolet bakground radiation by means of a su

essful guestinvestigator proposal for NASA's GALEX mission.The purpose of the present paper is to lay out our planned methods of treatment ofour GALEX data by applying my software to publily available GALEX Deep ImagingSurvey (\Dis") images. These images may be downloaded by anyone from the MASTweb site at the Spae Telesope Siene Institute: http://galex.stsi.edu/GR1/The intent of our GALEX proposal was to test the idea (Henry 1991, 1999) that asubstantial fration of the di�use ultraviolet bakground that is seen at moderate and highgalati latitudes originates neither in starlight sattered from dust, nor in the integrated? E-mail: henry�jhu.edu

2 Rihard Conn Henry

85°

85°

90°

90°

35°

30°

dis.08

dis.09

dis.10

dis.11

Fig. 1 Dust distribution map from Shlegel et al. (1998) showing the loations of fourGALEX Deep Imaging Survey (\Dis") targets. In the present paper I present resultsfor Dis-08 (whih has a strong ar of dust in the �eld of view) and Dis-09, whih islargely free of dust. The irles represent the full �eld of view (1:26Æ) of the GALEXameras.light of faint galaxies, but instead is red-shifted Lyman � radiation from the intergalatimedium. If this should turn out to be true, it would of ourse be an extremely importantresult for osmology and for osmogony. In partiular, the red-shifted Lyman � wouldbe radiation from the baryoni dark matter in the universe, while if that is orret, thisorigin would require emission of ionizing radiation by the non-baryoni dark matter (tomaintain the observed high degree of ionization of the intergalati medium in the faeof repeated rapid reombinations). Hene the title of this paper.

Bright Light from Dark Matter 3Table 1 The Two of the 14 GALEX Deep Imaging Survey Targets that I Analyse.Dis Name RA & De ` & b � t E(B�V )08 SIRTFFL 03 258.33 58.87 87.64 35.55 5558 0.02909 SIRTFFL 00 259.13 59.91 88.86 35.04 23765 0.0220.018 0.028 0.038 0.048

800

900

1000

1100

1200

1300

1400

1500

1600

FU

V B

right

ness

��������E(B-V)

0.90

0.018 0.028 0.038 0.0481200

1300

1400

1500

1600

1700

NU

V B

right

ness

��������E(B-V)

0.85Fig. 2 For eah pixel in the Dis-08 GALEX images, I plot the ultraviolet brightnessagainst the value of EB�V from Shlegel et al. 1998. The NUV brightness has beenredued by 350 photons m�2 sr�1 s�1 �A�1 to allow for zodiaal light. The goodorrelations (0.90, 0.85) prove that there is dust-sattered starlight present in theimage, while the non-zero interept shows that there is a great deal of additional UVlight that is not dust-sattered starlight. The reader should also notie that for boththe NUV and the FUV images, the satter in brightness, as a funtion of EB�V , ismuh greater than the standard error in a single pixel (a typial error bar is shown).I present here my results for two of the fourteen GALEX Dis targets (Table 1). One ofthese, Dis-08, I have seleted beause the image turned out to ontain a very substantialontribution from dust-sattered starlight, while the seond, Dis-09, is hosen beause (aswe shall see) it ontains almost no dust-sattered starlight, despite having a very strongdi�use UV bakground present|far more than ould possibly be due to the integratedlight of unresolved galaxies.Thus, this preliminary investigation suggests that the projet of Henry, Murthy,Bianhi, Shalima, and Sujatha, might well have exiting results.2 THE GALEX IMAGES.There are two GALEX ameras: the \FUV" amera images a band from 1300 �A to 1700 �A(orresponding to redshifts of 0.07 to 0.4) while the \NUV" amera images a band from1700 �A to 3000 �A (redshifts of 0.4 to 1.5). The redshifts are those of hydrogen Lyman �on the hypothesis whih we expet eventually to test: that a substantial fration of thedeteted signal is due to reombination radiation from the intergalati medium.

4 Rihard Conn Henry0.018 0.019 0.020 0.021 0.022 0.023 0.024 0.025 0.026

900

1000

1100

1200

FU

V B

right

ness

��������E(B-V)

0.14

0.018 0.019 0.020 0.021 0.022 0.023 0.024 0.025 0.0261300

1350

1400

1450

NU

V B

right

ness

��������E(B-V)

0.17Fig. 3 The same as Figure 2, but now for Dis-09. In ontrast with the previous �gure,we see that there is essentially no orrelation of the Dis-09 ultraviolet signal withEB�V . We onlude that the signal is almost entirely due to something other thanstarlight sattered from dust. (In the ase of the NUV image, part of that exess ismerely zodiaal light, but from all NUV data I have removed 350 photons m�2 sr�1s�1 �A�1 to allow for that.)The MAST web site provides images from both ameras, and also provides the sameimages with all point soures removed. It is the latter images, of ourse, that are ofinterest for the present projet.The angular resolution of the GALEX ameras is six seonds of ar, but in orderto ahieve the signal-to-noise that is neessary for our purposes, we have redued thenumber of pixels by a fator of 900, the image being onverted from 3840 � 3840 pixelsto 128 � 128 pixels.3 REMOVAL OF DUST-SCATTERED STARLIGHTThe �eld of view of eah amera has diameter one degree (ignoring the outer edge ofthe �eld, where instrumental sattered light is learly present). I have onsulted theinterstellar dust maps of Shlegel et al. (1998) in order to determine what fration of thedi�use bakground signal in eah image is due to dust-sattered starlight. I have reateda map (Figure 1), using the data of Shlegel et al., whih shows the dust distribution inthe region that inludes Dis-08 and Dis-09, the two images that I analyse in the presentpaper. These two images were seleted as \typial," in that one (Dis-08) is found toontain a onsiderable amount of dust-sattered starlight (prinipally in the ar that islearly visible in Figure 1), while the other ontains almost no dust-sattered starlight.The result of my analysis appears in Figure 2, where for both the FUV and NUVimages of Dis-08, the ultraviolet brightness is plotted against the EB�V values fromShlegel et al. In Figure 3 is given the same information for Dis-09.The orrelations for FUV and for NUV for Dis-08 are very strong, and show on-lusively that there is a onsiderable signal in these images that must represent dust-sattered starlight. But, it is also apparent that there is, in addition, signal that annotbe due to dust-sattered starlight: �rst, there is a non-zero interept to the least-squares

Bright Light from Dark Matter 5 800 900 1000 1100 1200 1300 1400 1500 1600

1250

1300

1350

1400

1450

1500

1550

1600

1650

NU

V B

right

ness

FUV Brightness

0.94

300 400 500 600 700 8001000

1050

1100

1150

1200

1250

1300

1350

1400

NU

V B

right

ness

FUV Brightness

0.79Fig. 4 The pixel-by-pixel orrelation between the FUV and the NUV image forDis-08. There is a muh stronger orrelation (0.94) before the removal of dust-satteredstarlight, but there is ertainly still a strong orrelation (0.79) afterward. Is this beausethe removal proess is imperfet, or beause the signal that is not due to dust- satteredstarlight is orrelated between the two images? There is no way to be ertain. That iswhy the Dis-09 observation, whih is free of dust-sattered starlight, is so important.�t to the data, indiating that even were there no dust, there would still be a strong sig-nal; and seond, the satter about the least-squares straight-line �t is muh greater thanthe statistial unertainty in the UV brightness determination for eah pixel (a typialerror bar giving that statistial unertainty is shown in eah plot).In extreme ontrast, the orrelations for Dis-09 that appear in Figure 3 are veryweak: indeed, almost non-existent. It is lear that there is almost no signal in Dis-09that is due to dust-sattered starlight. And, again, the satter is very large omparedwith the statistial unertainty assoiated with eah point, indiating that our fator 900expansion of the pixel size has allowed us to detet signi�ant struture in the image;struture that, in the ase of Dis-09, is not at all orrelated with the amount of dust inthat diretion.4 CORRELATION BETWEEN THE FUV AND NUVIMAGES|DUSTY TARGETSIn Figure 4, I show the pixel-by-pixel orrelation between the FUV and NUV imagesof Dis-08 before and after the removal of dust-sattered starlight. If, in some GALEXimage, it should happen that almost all of the signal were due to dust-sattered starlight,we would expet an almost perfet orrelation between the two images before removalthe of dust-sattered starlight, and, of ourse, no residual. In fat (Figure 4) for the aseof Dis-08 we see a very strong orrelation (orrelation oeÆient 0.94) before the removalof dust-sattered starlight), but a still very strong orrelation (0.79) between the twoimages for the strong residual signal that remains after the removal of dust-satteredstarlight.

6 Rihard Conn Henry 800 900 1000 1100

1200

1300

1400

NU

V B

right

ness

FUV Brightness

0.14

Fig. 5 Here is the pixel-by-pixel orrelation for Dis-09, after removal of the verysmall amount of dust-sattered starlight. The orrelation oeÆient is 0.14 while beforeremoval it was 0.15. This establishes that the ultraviolet bakground radiation that isnot due to dust-sattered starlight is unorrelated between the NUV and FUV images,whih is diÆult to understand other than under the hypothesis that the signal is fromthe intergalati medium.How are we to interpret this residual signal? The problem is that the resolution ofthe Shlegel et al. dust maps is muh worse than the resolution of the GALEX images,and evidently the variation in dust aross a single Shlegel element is onsiderable. Thusmy removal of dust-sattered starlight is very far from perfet.So what is the onlusion regarding Dis-08? It is that these images (and similar imagesfor other GALEX targets that turn out to have a substantial dust-sattered starlightomponent to them) an only be of qualitative interest, in our main task of attemptingto asertain the nature of that part of the image that is not of dust-sattered-starlightorigin. Fortunately, suh dust-dominated images turn out to be only approximately 1/3of the Dis images.5 CORRELATION BETWEEN THE FUV AND NUVIMAGES|DUSTLESS TARGETSMy presentation of the GALEX Dis-08 dusty target in this paper was mainly for thepurpose of getting it out of the way: dust-sattered starlight is only of modest interest;a strong UV bakground that is not due to dust-sattered starlight is (potentially) ofvery muh greater interest. With Dis-08, my intention was simply to demonstrate that

Bright Light from Dark Matter 70.50.60.70.80.9

1

Fig. 6 This is the FUV image of Dis-08 before the removal of signal that is identi�ed(Figure 2) as starlight sattered from interstellar dust.I an reliably identify targets that are substantially ontaminated with dust-satteredstarlight, so that suh targets an subsequently, largely, be ignored.Of far greater interest are targets, suh as Dis-09, that turn out (Figure 3) to havelittle or no dust-sattered starlight. Almost all of the very strong UV bakgroud in Dis-09is not due to dust-sattered starlight. What an it be?Well, what ould it be? Apart from the idea of radiation from the intergalatimedium, almost the only soure that has been suggested is the integrated light of faintgalaxies. That signal has been estimated by Armand, Milliard, & Deharveng (1994) to bein the range 40 to 130 photons m�2 sr�1 s�1 �A�1, while Gardner, Brown, & Ferguson(2000) have atually measured the integrated bakground using HST, �nding a lowerlimit of 144 photons m�2 sr�1 s�1 �A�1 and and upper limit of 195 photons m�2 sr�1s�1 �A�1.Well, the observed di�use ultraviolet bakground in Dis-09 is (Table 2) very muhlarger than any of these predited or observed values for the integrated light of faint galax-ies, whih therefore may be onlusively eliminated as the soure the observed radiationfor Dis-09.

8 Rihard Conn Henry0.4

0.6

0.8

1

Fig. 7 This is the FUV image of Dis-08 after the removal of signal that is identi�ed(Figure 2) as starlight sattered from interstellar dust. Comparison of this image withthat of Figure 7 suggests, but does not prove, that the proess of removal of signaldue to dust-sattered starlight is very imperfet, as the main residual after removal isloated exatly where the main removed omponent was loated.Table 2 GALEX Target Ultraviolet Brightnesses (photons m�2 sr�1 s�1 �A�1).Dis FUV (mean) FUV (interept) NUV (mean) NUV (interept) Correlation08 1109 470 1391 1132 0.7909 990 876 1348 1305 0.14One other predited soure of di�use UV bakground should be mentioned, for om-pleteness, and that is two-photon emission from metastable hydrogen atoms in the in-terstellar medium. However, that soure has been estimated by Deharveng, Joubert, andBarge (1982) to ontribute only about 75 photons m�2 sr�1 s�1 �A�1 to the di�use ul-traviolet bakground, whih means that this soure, too, an be eliminated as a possiblesoure for the strong radiation that is observed from Dis-09.

Bright Light from Dark Matter 90.8

0.850.9

0.951

Fig. 8 This is the NUV image of Dis-08 before the removal of signal that is identi�ed(Figure 2) as starlight sattered from interstellar dust.6 THE FUV AND NUV IMAGES OF DIS-08In Figures 6 and 7 I show the atual FUV images (before and after removal of signalidenti�ed as being due to dust-sattered starlight) for the dusty Dis-8 target. Pleaseompare with Figure 1. The ared ridge that is apparent in Figure 1 is the pronounedrise that is apparent on the right in both Figure 6 and Figure 7. Keep in mind that inFigure 1 the full �eld of view of the GALEX ameras is shown, while in Figures 6 and 7,only the data from the entral one degree of the �eld is plotted, to avoid the substantialinstrumental sattered light that is near the edge of eah GALEX image.In Figures 8 and 9, similarly, I show the atual NUV images for the same dusty target,Dis-08. In these �gures, and in the preeding �gures, it is apparent that the removal ofdust-sattered starlight appears to be inomplete. But it annot be exluded that this isoinidene, and that the additional signal, whih is not due to dust-sattered starlight,just happens to be similarly distributed. For this reason the GALEX images of dustytargets are of limited value.

10 Rihard Conn Henry0.8

0.850.9

0.951

Fig. 9 This is the NUV image of Dis-08 after the removal of signal that is identi�ed(Figure 2) as starlight sattered from interstellar dust.7 THE FUV AND NUV IMAGES OF DIS-09Finally, in Figures 10 and 11, I present the FUV and NUV images, respetively, for thetarget Dis-09 that has a very large UV di�use bakground signal but that is almostentirely free of dust-sattered starlight, as has been demonstrated in Figure 3. No dust-sattered starlight has been removed from these images.This demonstrates the quality of data that will be available to us in attempting to de-termine whether this radiation is Lyman� reombination radiation from the intergalatimedium. Figure 5 shows essentially no orrelation at all between the di�use ultravioletbakground radiation in the FUV and NUV images of Dis-09, whih is exatly what ispredited on the reombination radiation hypothesis, beause essentially totally di�erentranges of redshift are observed by the two GALEX ameras. Before one an ome to a�rm onlusion regarding the intriguing FUV and NUV images seen in the GALEX Distargets, one must ask whether the struture that appears in the �gures is real, and if itis real, if it is of astronomial origin. The most powerful way to takle that question is towork with subsets of the data, to test for orrelation. Suh subsets will be available to meand my olleagues for our own guest investigator data, and we look forward to ompletingsuh analysis. For example, geophysial e�ets of some kind oneivably ould ontam-

Bright Light from Dark Matter 110.7

0.8

0.9

1

Fig. 10 This is the FUV image of Dis-09, whih is almost entirely free of dust-sattered starlight. Struture is present that, if real, is perhaps assoiated with theintergalati medium. Note the presene of small-sale struture, and also the preseneof a muh larger sale slope aross the �eld, suggestive of struture larger than the �eld.In all of the images, we only show the entral one degree, thus avoiding instrumentalsattered light whih is present at the edges of the FUV and NUV �elds of view.inate the data. Additionally, our guest investigator data involves targets hosen by usas partiularly suitable for testing our extragalati reombination radiation hypothesis.In partiular, the main basis for our suggestion that hydrogen Lyman � reombinationradiation might be the soure of the di�use UV bakground that is universally observedlongward of 1216 �A is the fat that Murthy et al. (1999) observed many loations where,shortward of Lyman �, only an upper limit of 30 photons m�2 sr�1 s�1 �A�1 is observed,whih is learly suggestive that the longer-wavelength radiation is red-shifted Lyman �.Some of these low-bakground targets are inluded in our guest investigator observingplan. We expet to omplete our guest investigator work over the next several months.Aknowledgements This work was funded by Maryland Spae Grant Consortium.ReferenesArmand, C., Milliard, B., & Deharveng, J. M., 1994, A&A, 284, 12Deharveng, J. M., Joubert, M., & Barge, P. 1982, A&A, 109, 179

12 Rihard Conn Henry0.9

0.920.940.960.98

1

Fig. 11 This is the NUV image of Dis-09, also almost entirely free of dust- satteredstarlight (Figure 3). Again interesting struture appears in the image whih if realperhaps an be attributed to the intergalati medium. Similar (but not idential)strutures appear in all 14 of the Dis FUV and NUV images. In the ase of theFUV images, but not the NUV images, there is typially in addition to smaller-salestruture, a general slope in di�use bakground intensity, with the diretion of slopevarying from target to target.Gardner, J. P., Brown, T. M., & Ferguson, H. C., 2000, ApJ, 542, L79Henry R. C., 1991, ARA&A, 29, 89Henry R. C., 1999, ApJ, 516, L49Murthy, J., Hall, D., Earl, M., Henry, R. C., & Holberg, J. B., 1999, ApJ, 522, 904Shlegel, D. J., Finkbeiner, D. P., & Davis, M., 1998, ApJ, 500, 525Zurita, A., Bekman, J., & Rozas, M., 2002, Mem.S.A.It., 73-1, 162D�ISCUSSIONJ�OHN BECKMAN: The ionization you require might not need radiative deay of thenon-baryoni dark matter: see Zurita, Bekman, and Rozas (2002).HENRY: Thank you; I will investigate that.This manusript was prepared with the ChJAA LATEX maro v1.0.