The astrophysics of the cosmic dawn, when star formation commenced in the first collapsed objects, is predicted to be revealed by spectral and spatial signatures in the cosmic radio background at long wavelengths. The sky-averaged redshifted 21 cm absorption line of neutral hydrogen is a probe of the cosmic dawn. The line profile is determined by the evolving thermal state of the gas, radiation background, Lyman α radiation from stars scattering off cold primordial gas, and relative populations of the hyperfine spin levels in neutral hydrogen atoms. We report a radiometer measurement of the spectrum of the radio sky in the 55–85 MHz band, which shows that the profile found by Bowman et al. in data taken with the Experiment to Detect the Global Epoch of Reionization Signature (EDGES) low-band instrument is not of astrophysical origin; their best-fitting profile is rejected with 95.3% confidence. The profile was interpreted to be a signature of the cosmic dawn; however, its amplitude was substantially higher than that predicted by standard cosmological models. Our non-detection bears out earlier concerns and suggests that the profile found by Bowman et al. is not evidence for new astrophysics or non-standard cosmology.
This is a preview of subscription content
Subscribe to Nature+
Get immediate online access to the entire Nature family of 50+ journals
Subscribe to Journal
Get full journal access for 1 year
only $8.25 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Tax calculation will be finalised during checkout.
Get time limited or full article access on ReadCube.
All prices are NET prices.
The processed spectrum that supports the findings of this study are available from the corresponding authors on reasonable request.
The dataset is stored and preprocessed in Miriad (https://www.atnf.csiro.au/computing/software/miriad/). Later stages of analysis are done in Python, extensively using publicly available routines in SciPy (https://scipy.org/), NumPy (https://numpy.org/) and matplotlib (https://matplotlib.org/). MCMC analysis was done using the emcee package (https://github.com/dfm/emcee), and Bayesian evidence was computed using PolyChord (https://github.com/PolyChord/PolyChordLite). Other processing and analysis steps were performed using custom code, which is available from the corresponding author upon reasonable request.
Pritchard, J. R. & Loeb, A. 21 cm cosmology in the 21st century. Rep. Prog. Phys. 75, 086901 (2012).
Fialkov, A., Barkana, R. & Visbal, E. The observable signature of late heating of the Universe during cosmic reionization. Nature 506, 197–199 (2014).
Furlanetto, S. R., Oh, S. P. & Briggs, F. H. Cosmology at low frequencies: the 21 cm transition and the high-redshift Universe. Phys. Rep. 433, 181–301 (2006).
Pritchard, J. R. & Loeb, A. Constraining the unexplored period between the dark ages and reionization with observations of the global 21 cm signal. Phys. Rev. D 82, 023006 (2010).
Bowman, J. D., Rogers, A. E. E., Monsalve, R. A., Mozdzen, T. J. & Mahesh, N. An absorption profile centred at 78 megahertz in the sky-averaged spectrum. Nature 555, 67–70 (2018).
Fialkov, A. & Barkana, R. Signature of excess radio background in the 21-cm global signal and power spectrum. Mon. Not. R. Astron. Soc. 486, 1763–1773 (2019).
McGaugh, S. S. Predictions for the sky-averaged depth of the 21 cm absorption signal at high redshift in cosmologies with and without nonbaryonic cold dark matter. Phys. Rev. Lett. 121, 081305 (2018).
Barkana, R. Possible interaction between baryons and dark-matter particles revealed by the first stars. Nature 555, 71–74 (2018).
Muñoz, J. B. & Loeb, A. A small amount of mini-charged dark matter could cool the baryons in the early universe. Nature 557, 684–686 (2018).
Barkana, R., Outmezguine, N. J., Redigolo, D. & Volansky, T. Strong constraints on light dark matter interpretation of the EDGES signal. Phys. Rev. D 98, 103005 (2018).
Ewall-Wice, A. et al. Modeling the radio background from the first black holes at cosmic dawn: implications for the 21 cm absorption amplitude. Astrophys. J. 868, 63 (2018).
Hills, R., Kulkarni, G., Meerburg, P. D. & Puchwein, E. Concerns about modelling of the EDGES data. Nature 564, E32–E34 (2018).
Singh, S. & Subrahmanyan, R. The redshifted 21 cm signal in the EDGES Low-band spectrum. Astrophys. J. 880, 26 (2019).
Bradley, R. F., Tauscher, K., Rapetti, D. & Burns, J. O. A ground plane artifact that induces an absorption profile in averaged spectra from global 21 cm measurements, with possible application to EDGES. Astrophys. J. 874, 153 (2019).
Sims, P. H. & Pober, J. C. Testing for calibration systematics in the EDGES low-band data using Bayesian model selection. Mon. Not. R. Astron. Soc. 492, 22–38 (2019).
Nambissan T., J. et al. SARAS 3 CD/EoR radiometer: design and performance of the receiver. Exp. Astron. 51, 193–234 (2021).
Girish, B. S. et al. SARAS CD/EoR radiometer: design and performance of the digital correlation spectrometer. J. Astron. Instrum. 9, 2050006 (2020).
Raghunathan, A. et al. A floating octave bandwidth cone–disk antenna for detection of cosmic dawn. IEEE Trans. Antennas Propag. 69, 6209–6217 (2021).
CODE ionosphere products. http://ftp.aiub.unibe.ch/CODE/
Thompson, A. R., Moran, J. M., Swenson, J. & George, W. Interferometry and Synthesis in Radio Astronomy 3rd edn (Springer, Cham, 2017).
Vedantham, H. K. et al. Chromatic effects in the 21 cm global signal from the cosmic dawn. Mon. Not. R. Astron. Soc. 437, 1056–1069 (2014).
Sathyanarayana Rao, M., Subrahmanyan, R., Udaya Shankar, N. & Chluba, J. Modeling the radio foreground for detection of CMB spectral distortions from the cosmic dawn and the epoch of reionization. Astrophys. J. 840, 33 (2017).
de Oliveira-Costa, A. et al. A model of diffuse Galactic radio emission from 10 MHz to 100 GHz. Mon. Not. R. Astron. Soc. 388, 247–260 (2008).
Zheng, H. et al. An improved model of diffuse galactic radio emission from 10 MHz to 5 THz. Mon. Not. R. Astron. Soc. 464, 3486–3497 (2016).
Foreman-Mackey, D., Hogg, D. W., Lang, D. & Goodman, J. emcee: the MCMC hammer. Publ. Astron. Soc. Pac. 125, 306–312 (2013).
Monsalve, R. A., Rogers, A. E. E., Bowman, J. D. & Mozdzen, T. J. Results from EDGES High-band. I. Constraints on phenomenological models for the global 21 cm signal. Astrophys. J. 847, 64 (2017).
Singh, S. et al. SARAS 2 constraints on global 21 cm signals from the epoch of reionization. Astrophys. J. 858, 54 (2018).
EDGES data releases. http://loco.lab.asu.edu/edges/edges-data-release/
Schwarz, G. Estimating the dimension of a model. Ann. Stat. 6, 461–464 (1978).
Spinelli, M., Bernardi, G. & Santos, M. G. On the contamination of the global 21-cm signal from polarized foregrounds. Mon. Not. R. Astron. Soc. 489, 4007–4015 (2019).
Burns, J. O. et al. A space-based observational strategy for characterizing the first stars and galaxies using the redshifted 21 cm global spectrum. Astrophys. J. 844, 33 (2017).
Patra, N., Subrahmanyan, R., Raghunathan, A. & Udaya Shankar, N. SARAS: a precision system for measurement of the cosmic radio background and signatures from the epoch of reionization. Exp. Astron. 36, 319–370 (2013).
Patra, N., Subrahmanyan, R., Sethi, S., Shankar, N. U. & Raghunathan, A. SARAS measurement of the radio background at long wavelengths. Astrophys. J. 801, 138 (2015).
Landecker, T. L. & Wielebinski, R. The galactic metre wave radiation: a two-frequency survey between declinations +25° and −25° and the preparation of a map of the whole sky. Aust. J. Phys. 23, 957–958 (1970).
Singh, S. et al. SARAS 2: a spectral radiometer for probing cosmic dawn and the epoch of reionization through detection of the global 21-cm signal. Exp. Astron. 45, 269–314 (2018).
Singh, S. et al. First results on the epoch of reionization from first light with SARAS 2. Astrophys. J. Lett. 845, L12 (2017).
Sathyanarayana Rao, M., Subrahmanyan, R., Udaya Shankar, N. & Chluba, J. GMOSS: all-sky model of spectral radio brightness based on physical components and associated radiative processes. Astron. J. 153, 26 (2017).
Bevins, H. T. J. et al. maxsmooth: rapid maximally smooth function fitting with applications in Global 21-cm cosmology. Mon. Not. R. Astron. Soc. 502, 4405–4425 (2021).
Harris, F. J. On the use of windows for harmonic analysis with the discrete Fourier transform. IEEE Proc. 66, 51–83 (1978).
Hampel, F. R. The influence curve and its role in robust estimation. J. Am. Stat. Assoc. 69, 383–393 (1974).
Narula, S. C. & Korhonen, P. J. Multivariate multiple linear regression based on the minimum sum of absolute errors criterion. Eur. J. Oper. Res. 73, 70–75 (1994).
Handley, W. J., Hobson, M. P. & Lasenby, A. N. POLYCHORD: next-generation nested sampling. Mon. Not. R. Astron. Soc. 453, 4384–4398 (2015).
Kass, R. E. & Raftery, A. E. Bayes factors. J. Am. Stat. Assoc. 90, 773–795 (1995).
We thank the staff at the Gauribidanur Field Station for assistance with system tests and measurements, and the Mechanical and Electronics Engineering Groups at the Raman Research Institute for building, assembling and deploying SARAS 3. S.S. thanks J. Mirocha and A. Liu for useful discussions on global signal modelling and acknowledges support from a McGill Astrophysics postdoctoral fellowship. J.N.T. acknowledges a PhD bursary from the Australian Research Council (ARC) Future Fellowship under grant FT180100196.
The authors declare no competing interests.
Peer review information
Nature Astronomy thanks the anonymous reviewers for their contribution to the peer review of this work.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The building blocks and signal flow within the instrument is shown from the antenna to the digital spectrometer.
Panel (a) shows the calibrated spectrum measured in the laboratory, in a long duration test, with the antenna replaced by an antenna simulator that approximates its reflection coefficient. A maximally smooth function fit to the spectrum is shown using a red dashed line; the data, with deviations from the fit magnified by factor 50, is shown as a blue continuous line. Panel (b) shows the residuals smoothed to successively lower resolutions, using a Hann function, from the native 61 kHz to 3.906 MHz. The FWHM of the smoothing functions and the r.m.s. values of the residuals on smoothing is in the legend. For clarity, the traces have been offset vertically and magnified by the ratio of r.m.s. value after smoothing to the r.m.s. value at native resolution.
Panel (a) shows the fit of a maximally smooth function to the reflection efficiency, as a red dashed line. Overlaid is the measurement, as a continuous blue line, with the difference between the fit and measurement magnified by factor 1000 for clarity. Panel (b) shows the difference between the measurement and MS fit.
Panel (a) shows the total and radiation efficiencies of the instrument as measured on water in the Sharavati backwaters, where most of the observations were made. Maximally smooth function fit to the total efficiency is overlaid on the measurement and the difference between the best-fit MS function and measurement is shown in panel (b).
Measured sky brightness temperature versus local sidereal time (LST), at a set of frequencies across the science band. The brightness temperature at any time is the global radio sky viewed by the antenna beam pattern of the instrument.
Panel (a) shows mock sky spectra constructed using erroneous calibrations for efficiencies, water thermal emission and receiver noise, with errors reflecting the uncertainties in their derivations. Panels (b), (c), (d), (e) and (f) show, in red, the systematic residuals that might result due to calibration errors and using 3rd, 4th, 5th, 6th or 7th order polynomials to represent the foreground. For comparison we also show, in black, the residual expected if the global sky spectrum contained the profile found by Bowman et al.
Panel (a) shows the logarithm of Bayesian evidence versus the order of the polynomial: in black symbols is shown the case where the sky spectrum measured by SARAS 3 is modelled as a polynomial plus the profile found by Bowman et al. with a free scaling factor, and in red symbols is shown the case where the model is a plain polynomial. Higher the evidence, more preferred is the model. Panel (b) shows Bayesian Information Criterion: here models with lower values are preferred. Panel (c) shows the r.m.s. value of the residuals smoothed to 1.4 MHz resolution: the spectrum is consistent with measurement noise for 6th and higher orders; panel (d) shows exclusively the r.m.s. values for order 6 and above.
Mock spectra of the global sky are generated using GSM and GMOSS, using the SARAS 3 beam, and processed to impress systematic errors in calibrations for total efficiency (panels a, b & c), water thermal emission (panels d, e & f) and receiver noise (panels g, h & i). Calibrations with errors that span the expected distributions are shown in the first column; our best-estimate calibrations are also shown using black dashed lines. Mock spectra with these calibration errors are shown in the second column. Residuals from fitting and subtracting out 6th-order polynomials from the mock spectra are shown in the last column.
Panel (a) shows flattened Gaussian profiles with parameters in the ranges charted by Bowman et al.5; the relative density in profile space displays their relative likelihood. The dashed orange line shows the best-fitting profile; the vertical lines and shaded domain demarcate the band used in our analysis. Lower panels show the result of MCMC analysis using the SARAS 3 data that determines the distribution of scale factors associated with each of these profiles: (b) shows distribution of scale factors for the whole class and (c) shows the distribution of the mean scale factor corresponding to each of the profiles. The shaded region in panel (b) spans 5th − 95th percentile values, and dash–dotted black line represents the mean. The dashed black lines represent 1σ confidence intervals. The profile set in panel (a) as a whole is rejected with 90.4% confidence.
Scatter plots corresponding to the different parameters are shown in separate panels, colour coded to show the significance with which the different parameter spaces are rejected by our MCMC analysis using the SARAS 3 data. The parameters describing the signal are the same as in Bowman et al.: amplitude (in mK), central frequency f0 (in MHz), full width at half maximum (FWHM) of the profile (in MHz) and the flattening parameter τ.
About this article
Cite this article
Singh, S., Nambissan T., J., Subrahmanyan, R. et al. On the detection of a cosmic dawn signal in the radio background. Nat Astron 6, 607–617 (2022). https://doi.org/10.1038/s41550-022-01610-5