• • • We do not use the following data for averages, fits, limits, etc. • • • 
$<0.03$ 

^{ 1} 

ASTR 
$<9.6 \times 10^{3}$ 
95 
^{ 2} 

ASTR 


^{ 3} 




^{ 4} 

NMR 
$<65$ 
95 
^{ 5} 

CNTR 
$<6.6$ 
90 
^{ 6} 

EDE3 
$<0.085$ 
90 
^{ 7} 

ASTR 
$<12.7$ 
95 
^{ 8} 

CNTR 
$<0.01$ 

^{ 9} 

ASTR 


^{ 10} 


$<93$ 
90 
^{ 11} 

HPGE 
$<4$ 
90 
^{ 12} 

PNDX 


^{ 13} 


$<177$ 
90 
^{ 14} 

CDEX 
$<0.079$ 
95 
^{ 15} 

ASTR 
$< 100$ 
95 
^{ 16} 

CNTR 


^{ 17} 




^{ 18} 




^{ 19} 




^{ 20} 




^{ 21} 




^{ 22} 

COSM 


^{ 23} 

ASTR 
$<250$ 
95 
^{ 24} 

CNTR 
$<155$ 
90 
^{ 25} 

EDEL 
$<8.6 \times 10^{3}$ 
90 
^{ 26} 

CNTR 
$<1.4 \times 10^{4}$ 
90 
^{ 27} 

BORX 
$<145$ 
95 
^{ 28} 

CNTR 


^{ 29} 

CNTR 


^{ 30} 


^{1}
LEINSON 2019 is analogous to BEZNOGOV 2018 , but estimating the axion luminosity based on the Tolman's analytic solution to the Einstein equations of spherical fluids in hydrostatic equilibrium. The dimensionless axionneutron coupling is constrained as $\mathit g_{Ann}$ $<$ $1.0 \times 10^{10}$.

^{2}
LLOYD 2019 is analogous to BERENJI 2016 . They highlight that the limit obtained with this technique strongly depends on the assumed NS core temperature.

^{3}
SMORRA 2019 look for spinprecession effects from ultralight axion dark matter in the ${{\overline{\mathit p}}}$ spinflip resonance data. Assuming ${{\mathit \rho}_{{A}}}$ = 0.4 GeV/cm${}^{3}$, they constrain the dimensionless axionantiproton coupling as $\mathit g_{ {{\mathit A}} {{\overline{\mathit p}}} {{\overline{\mathit p}}} }$ $<$ $2  9$ at 95$\%$ CL for ${\mathit m}_{{{\mathit A}^{0}}}$ = $2 \times 10^{23}  4 \times 10^{17}$ eV. See the right panel of their Fig. 3.

^{4}
WU 2019 look for axioninduced timeoscillating features of the NMR spectrum of acetonitrile2${}^{13}\mathrm {C}$. Assuming C$_{p}$ = C$_{n}$ and ${{\mathit \rho}_{{A}}}$ = 0.4 GeV/cm${}^{3}$, they constrain the dimensionless axionnucleon coupling as ${{\mathit g}_{{ANN}}}$ $<$ $6 \times 10^{5}$ for ${\mathit m}_{{{\mathit A}^{0}}}$ = $10^{21}  1.3 \times 10^{17}$ eV. Note that the limits for ${\mathit m}_{{{\mathit A}^{0}}}$ $<$ $10^{21}$ eV in their Fig. 3(a) should be weaker than those for heavier masses. See ADELBERGER 2019 and WU 2019C on this issue.

^{5}
AKHMATOV 2018 is an update of GAVRILYUK 2015 .

^{6}
ARMENGAUD 2018 is analogous to ALESSANDRIA 2013 . The quoted limit assumes the DFSZ axion model. See their Fig. 4 for the limit on product of axion couplings to electrons and nucleons.

^{7}
BEZNOGOV 2018 constrain the axionneutron coupling by assuming that thermal evolution of the hot neutron star HESS J1731347 is dominated by the lowest possible neutrino emission. The quoted limit assumes the KSVZ axion with the effective PecceiQuinn charge of the neutron C$_{n}$ = $0.02$. The dimensionless axionneutron couling is constrained as $\mathit g_{Ann}$ $<$ $2.8 \times 10^{10}$.

^{8}
GAVRILYUK 2018 look for the resonant excitation of ${}^{83}\mathrm {Kr}$ (9.4 keV) by solar axions produced via the Primakoff effect. The mass bound assumes ${\mathit m}_{{{\mathit u}}}/{\mathit m}_{{{\mathit d}}}$ = 0.56 and $\mathit S$ = 0.5.

^{9}
HAMAGUCHI 2018 studied the axion emission from the neutron star in Cassiopeia A based on the minimal cooling scenario which explains the observed rapid cooling rate. The quoted limit corresponds to $\mathit f_{A}$ $>$ $5 \times 10^{8}$ GeV obtained for the KSVZ axion with C$_{p}$ = $0.47$ and C$_{n}$ = $0.02$.

^{10}
ABEL 2017 look for a timeoscillating neutron EDM and an axionwind spinprecession effect respectively induced by axion dark matter couplings to gluons and nucleons. See their Fig. 4 for limits in the range of ${\mathit m}_{{{\mathit A}^{0}}}$ = $10^{24}  10^{17}$ eV.

^{11}
ABGRALL 2017 limit assumes the hadronic axion model used in ALESSANDRIA 2013 . See their Fig. 4 for the limit on product of axion couplings to electrons and nucleons.

^{12}
FU 2017A look for the 14.4 keV ${}^{57}\mathrm {Fe}$ solar axions. The limit assumes the DFSZ axion model. See their Fig. 3 for massdependent limits on the axionelectron coupling. Notice that in this figure the DFSZ and KSVZ lines should be interchanged.

^{13}
KLIMCHITSKAYA 2017A use the differential measurement of the Casimir force between a ${}^{}\mathrm {Ni}$coated sphere and ${}^{}\mathrm {Au}$ and ${}^{}\mathrm {Ni}$ sectors of the structured disc to constrain the axion coupling to nucleons for $2.61$ meV $<$ ${\mathit m}_{{{\mathit A}^{0}}}$ $<$ 0.9 eV. See their Figs. 1 and 2 for mass dependent limits.

^{14}
LIU 2017 is analogous to ALESSANDRIA 2013 . The limit assumes the hadronic axion model. See their Fig. 6(b) for the limit on product of axion couplings to electrons and nucleons.

^{15}
BERENJI 2016 used the Fermi LAT observations of neutron stars to look for photons from axion decay. They assume the effective PecceiQuinn charge of the neutron C$_{n}$ = $0.1$ and a neutronstar core temperature of 20 MeV.

^{16}
GAVRILYUK 2015 look for solar axions emitted by the M1 transition of ${}^{83}\mathrm {Kr}$ (9.4 keV). The mass bound assumes ${\mathit m}_{{{\mathit u}}}/{\mathit m}_{{{\mathit d}}}$ = 0.56 and $\mathit S$ = 0.5.

^{17}
KLIMCHITSKAYA 2015 use the measurement of differential forces between a test mass and rotating source masses of ${}^{}\mathrm {Au}$ and ${}^{}\mathrm {Si}$ to constrain the force due to twoaxion exchange for $1.7 \times 10^{3}$ $<$ ${\mathit m}_{{{\mathit A}^{0}}}$ $<$ 0.9 eV. See their Figs. 1 and 2 for mass dependent limits.

^{18}
BEZERRA 2014 use the measurement of the thermal CasimirPolder force between a BoseEinstein condensate of ${}^{87}\mathrm {Rb}$ atoms and a ${}^{}\mathrm {SiO}_{2}$ plate to constrain the force mediated by exchange of two pseudoscalars for 0.1 meV $<$ ${\mathit m}_{{{\mathit A}^{0}}}<$ 0.3 eV. See their Fig. 2 for the massdependent limit on pseudoscalar coupling to nucleons.

^{19}
BEZERRA 2014A is analogous to BEZERRA 2014 . They use the measurement of the Casimir pressure between two ${}^{}\mathrm {Au}$coated plates to constrain pseudoscalar coupling to nucleons for $1 \times 10^{3}$ eV $<$ ${\mathit m}_{{{\mathit A}^{0}}}<$ 15 eV. See their Figs. 1 and 2 for the massdependent limit.

^{20}
BEZERRA 2014B is analogous to BEZERRA 2014 . BEZERRA 2014B use the measurement of the normal and lateral Casimir forces between sinusoidally corrugated surfaces of a sphere and a plate to constrain pseudoscalar coupling to nucleons for 1 eV $<$ ${\mathit m}_{{{\mathit A}^{0}}}$ $<$ 20 eV. See their Figs. $1  3$ for massdependent limits.

^{21}
BEZERRA 2014C is analogous to BEZERRA 2014 . They use the measurement of the gradient of the Casimir force between ${}^{}\mathrm {Au}$ and ${}^{}\mathrm {Ni}$coated surfaces of a sphere and a plate to constrain pseudoscalar coupling to nucleons for $3 \times 10^{5}$ eV $<$ ${\mathit m}_{{{\mathit A}_{{0}}}}$ $<$ 1 eV. See their Figs. 1, 3, and 4 for the massdependent limits.

^{22}
BLUM 2014 studied effects of an oscillating strong $\mathit CP$ phase induced by axion dark matter on the primordial ${}^{4}\mathrm {He}$ abundance. See their Fig. 1 for massdependent limits.

^{23}
LEINSON 2014 attributes the excessive cooling rate of the neutron star in Cassiopeia A to axion emission from the superfluid core, and found C${}^{2}_{n}{{\mathit m}^{2}}_{{{\mathit A}^{0}}}$ $\simeq{}$ $5.7 \times 10^{6}$ eV${}^{2}$, where C$_{n}$ is the effective PecceiQuinn charge of the neutron.

^{24}
ALESSANDRIA 2013 used the CUORE experiment to look for 14.4 keV solar axions produced from the M1 transition of thermally excited ${}^{57}\mathrm {Fe}$ nuclei in the solar core, using the axioelectric effect. The limit assumes the hadronic axion model. See their Fig. 4 for the limit on product of axion couplings to electrons and nucleons.

^{25}
ARMENGAUD 2013 is analogous to ALESSANDRIA 2013 . The limit assumes the hadronic axion model. See their Fig. 8 for the limit on product of axion couplings to electrons and nucleons.

^{26}
BELLI 2012 looked for solar axions emitted by the M1 transition of ${}^{7}\mathrm {Li}{}^{*}$ (478 keV) after the electron capture of ${}^{7}\mathrm {Be}$, using the resonant excitation ${}^{7}\mathrm {Li}$ in the ${}^{}\mathrm {LiF}$ crystal. The mass bound assumes ${\mathit m}_{{{\mathit u}}}/{\mathit m}_{{{\mathit d}}}$ = 0.55, ${\mathit m}_{{{\mathit u}}}/{\mathit m}_{{{\mathit s}}}$ = 0.029, and the flavorsinglet axial vector matrix element $\mathit S$ = 0.4.

^{27}
BELLINI 2012B looked for 5.5 MeV solar axions produced in the ${{\mathit p}}$ ${{\mathit d}}$ $\rightarrow$ ${}^{3}\mathrm {He}{{\mathit A}^{0}}$.The limit assumes the hadronic axion model. See their Figs. 6 and 7 for massdependent limits on productsof axion couplings to photons, electrons, and nucleons.

^{28}
DERBIN 2011 looked for solar axions emitted by the M1 transition of thermally excited ${}^{57}\mathrm {Fe}$ nuclei in the Sun, using their possible resonant capture on ${}^{57}\mathrm {Fe}$ in the laboratory. The mass bound assumes ${\mathit m}_{{{\mathit u}}}/{\mathit m}_{{{\mathit d}}}$ = 0.56 and the flavorsinglet axial vector matrix element ${{\mathit S}}$ = 3${{\mathit F}}−{{\mathit D}}$ $\simeq{}$ 0.5.

^{29}
BELLINI 2008 consider solar axions emitted in the M1 transition of ${}^{7}\mathrm {Li}{}^{*}$ (478 keV) and look for a peak at 478 keV in the energy spectra of the Counting Test Facility (CTF), a Borexino prototype. For ${\mathit m}_{{{\mathit A}^{0}}}$ $<$ 450 keV they find massdependent limits on products of axion couplings to photons, electrons, and nucleons.

^{30}
ADELBERGER 2007 use precision tests of Newton's law to constrain a force contribution from the exchange of two pseudoscalars. See their Fig. 5 for limits on the pseudoscalar coupling to nucleons, relevant for ${\mathit m}_{{{\mathit A}^{0}}}$ below about 1 meV.
