Axion astronomy with microwave cavity experiments

Terrestrial searches for the conversion of dark matter axions or axionlike particles into photons inside magnetic fields are sensitive to the phase space structure of the local Milky Way halo. We simulate signals in a hypothetical future experiment based on the Axion Dark Matter Experiment that could be performed once the axion has been detected and a frequency range containing the axion mass has been identified. We develop a statistical analysis to extract astrophysical parameters, such as the halo velocity dispersion and laboratory velocity, from such data and find that with only a few days integration time a level of precision can be reached, matching that of astronomical observations. For longer experiments lasting up to a year in duration, we find that exploiting the modulation of the power spectrum in time allows accurate measurements of the Solar peculiar velocity with an accuracy that would improve upon astronomical observations. We also simulate signals based on results from N-body simulations and find that finer substructure in the form of tidal streams would show up prominently in future data, even if only a subdominant contribution to the local dark matter distribution. In these cases, it would be possible to reconstruct all the properties of a dark matter stream using the time and frequency dependence of the signal. Finally, we consider the detection prospects for a network of streams from tidally disrupted axion miniclusters. These features appear much more prominently in the resolved spectrum than suggested by calculations based on a scan over a range of resonant frequencies, making the detection of axion minicluster streams more viable than previously thought. These results confirm that haloscope experiments in a postdiscovery era are able to perform “axion astronomy.”

Figure 1

Example simulated power spectra as a function of time. Each line is the average power spectrum observed over a 10 day period. The top panel shows the spectra for a smooth Maxwellian halo, and the bottom panel shows the spectra for a pure tidal stream with parameter values displayed in Table 1. The purple line in the frequency-time plane shows the evolution of the frequency at which the power is maximized, $2\pi {\nu }_{\max }=m_a(1+v_{\mathrm{lab}}^2/2)$ and $2\pi {\nu }_{\max }=m_a(1+|{\mathbf{v}}_{\mathrm{lab}}-{\mathbf{v}}_{\mathrm{str}}{|}^2/2)$ for the Maxwellian halo and stream, respectively.

Figure 2

Reconstructed axion mass and coupling as well astrophysical parameters, $v_{\mathrm{lab}}$ and ${\sigma }_v$, for a smooth Maxwellian halo model. We show sets of 68% and 95% confidence level contours in the $m_a-g_{a\gamma \gamma }$ and $|{\mathbf{v}}_{\mathrm{lab}}|-{\sigma }_v$ planes (left and right panels, respectively). We express the axion mass as $\mathrm{\Delta }{m}_a$, which has the true (input) value subtracted. The blue, green, and red sets of contours correspond to the estimates with experiments of different durations: 10 days, 0.5 yr, and 1 yr, respectively. The maximum likelihood values are indicated by triangles, and the input values for the parameters are indicated by dashed lines and a yellow star.

Figure 3

Reconstructed lab velocity components $(v_{\mathrm{pec}}^x,v_0+v_{\mathrm{pec}}^y,v_{\mathrm{pec}}^z)$ at 68% and 95% confidence for three data sets of length 10 days, 0.5 yr, and 1 yr, indicated by the blue, green, and red sets of contours, respectively. The maximum likelihood values are indicated by triangles, and the true (input) values are indicated by dashed lines with a yellow star.

Figure 4

Reconstructed parameters for multiple stochastic realisations the data. The 1 and 2 sigma error bars are shown for five parameters, from top to bottom, $m_a$, $g_{a\gamma \gamma }$, $|{\mathbf{v}}_{\mathrm{lab}}|$, ${\sigma }_v$, and the noise (which we express as $P_N/k_B\mathrm{\Delta }\nu$). There are 30 sets of repeated measurements for three different experimental durations ${\tau }_{\mathrm{tot}}=1\mathrm{day}$, 10 days, and 1 yr (from left to right).

Figure 5

Set of laboratory frame speed distributions of the 200 samples chosen from the VL2 simulation. The shaded regions indicate the range of $f(v)$ values for a given $v$. The solid purple lines indicate the maximum and minimum values of $f(v)$ and the dashed line is the mean distribution over all samples. The black line is the SHM Maxwellian with parameters from Table 1. We label particular samples which contain prominent streams with arrows and the sample number.

Figure 6

Left: Axion conversion power spectra for a selection of four Earth-radius dark matter velocity distributions from the VL2 simulation. In each of the four examples, the power spectrum has the amplitude of the noise power ($P_N$) subtracted and is displayed as a function of time (running over two years from 2016–2018). The frequency dependence is shown as the difference between the photon frequency and the axion mass. Right: The same set of power spectra after performing the various cuts detailed in the text. The remaining points show fluctuations in the axion power spectra after the time independent components have been subtracted. The best fit to Eq. (40) is shown as a red line; the power spectrum lying along this best fit line is then extracted to measure the properties of the stream. For clarity, in the left-hand power spectra, we have divided the noise amplitude by 4 so that the substructure is clearer; however, the right-hand data (used to do the reconstruction) retain the full noise amplitude with $T_S=4\mathrm{K}$.

Figure 7

Measurements of stream velocity (vertical black lines) and intervals at the 68% and 95% confidence levels (dotted and dashed lines, respectively) for each of the four sample VL2 velocity distributions. The one-dimensional speed distributions in each Galactic coordinate $(v_x,v_y,v_z)$ correspond to the first three columns. Each row corresponds to the four sample distributions chosen. The final column is the reconstruction in the plane of the stream density fraction and dispersion.

Figure 8

Simulated power spectra observed over a ten year period for a halo model consisting of a smooth population of axion dark matter with an additional component from a network of tidal streams stripped from orbiting miniclusters. The abundance is based on the calculation of Ref. [75]. The signal from minicluster streams appears as short-lived enhancements which are modulated in frequency due to the orbit of the Earth. The power spectra are displayed as a function of time from Jan 2016 to Jan 2021 and frequency shifted by the axion mass.