Search for photon oscillations into massive particles

Recently, axionlike particle search has received renewed interest, and several groups have started experiments. In this paper, we present the final results of our experiment on photon-axion oscillations in the presence of a magnetic field, which took place at the Laboratoire pour l’Utilisation des Lasers Intenses, Palaiseau, France. Our null measurement allowed us to exclude the existence of axions with inverse coupling constant $M>9.\times {10}^5\mathrm{GeV}$ for low axion masses and to improve the preceding Brookhaven-Fermilab-Rochester-Trieste (BFRT) Collaboration limits by a factor of 3 or more for axion masses $1.1<m_a<2.6\mathrm{meV}$. We also show that our experimental results improve the existing limits on the parameters of a low mass hidden-sector boson usually dubbed “paraphoton” because of its similarity with the usual photon. We detail our apparatus which is based on the “light shining through the wall” technique. We compare our results to other existing ones.

Figure 1

Sketch of the apparatus. The wall and the blind flanges are removable for fiber alignment.

Figure 2

Focal spot without correction (a) and with wave-front correction (b). This correction allows one to maintain a spot of one or two diffraction limits despite the amplifiers’ not being in thermal equilibrium.

Figure 3

Number of high energy pulses versus laser energy during the four weeks of data acquisition.

Figure 4

Scheme of XCoil. Magnetic fields ${\overrightarrow{B}}_1$ and ${\overrightarrow{B}}_2$ are created by each of the racetrack shaped windings. This yields a high transverse magnetic field $\overrightarrow{B}$ while allowing the necessary optical access for the laser photons $\gamma$.

Figure 5

Transverse magnetic field inside the magnet along the laser direction. At the center of the magnet we have a mean maximum magnetic field $B_0=12\mathrm{T}$. Integrating $B$ along the optical path yields 4.38 Tm.

Figure 6

Magnetic field $B_0$ at the center of the magnet as a function of time. The maximum is reached within 1.75 ms and can be considered as constant ($\pm 0.3\%$) during ${\tau }_B=150\mu \mathrm{s}$. The 3–5 ns laser pulse is applied during this interval. Inset: temporal profile of a 4 ns laser pulse.

Figure 7

Amplified APD output (upper curve) and logic signal (lower curve) of the detector as a function of time. The capacitive transients on the APD output signals are due to the gated polarization of the photodiode in Geiger mode. (a) Signals with no incident photon. (b) Signals when a photon is detected.

Figure 8

Detection efficiency (●) and dark count per 5 ns bias pulse (△) as a function of the discriminator threshold (a) ($V_{\mathrm{APD}}$ fixed to 78.4 V) or APD bias voltage (b) ($V_d$ fixed to 0.760 V). The APD temperature is fixed to the lowest achievable value 221.5 K. Dashed lines indicate the chosen working point.

Figure 9

Experimental setup to measure the detection efficiency of the single-photon detector. The detector is illuminated with a laser intensity lower than 0.1 photon per 5 ns. This intensity is calculated through the measurement of the supermirrors transmission and the laser power before those supermirrors. A half wave plate and a polarizer are used to change the number of incident photons.

Figure 10

Monitoring of the optical path followed by the high energy beam. Losses due to misalignment are estimated by comparing the center of the beam to the center of the black cross. The upper image corresponding to an uncorrected laser beam pointing exhibits 30% injection losses, while the lower one is perfectly corrected.

Figure 11

Coincidence rate between the arrival of photons on the detector and its 5 ns detection gate as a function of an arbitrary delay time. The dashed line indicates our working point, chosen in order to maximize the coincidence rate.

Figure 12

$3\sigma$ limits for the axionlike particle—two-photon inverse coupling constant $M$, as a function of the axionlike particle mass $m_a$, obtained from our null result. The area below our curve is excluded.

Figure 13

Limits on the axionlike particle—two-photon inverse coupling constant $M$ as a function of the axionlike particle mass $m_a$ obtained by experimental searches. Our exclusion region is first compared to other purely laboratory experiments such as the BFRT photon regeneration experiment [3], the GammeV experiment [33], and the PVLAS Collaboration [34] with a $3\sigma$ confidence level. Those curves are finally compared to the 95% confidence level exclusion region obtained on CAST [5] and the more than 90% confidence level on microwave cavity experiments [4, 35, 36]. Model predictions are also shown as a dotted stripe between the predictions of the KSVZ model (lower line, $E/N=0$) [44] and of the DFSZ model (upper line, $E/N=8/3$) [45].

Figure 14

95% confidence level limits on photon-paraphoton mixing parameter as a function of the paraphoton mass obtained to our null result (deep gray area). Shaded regions are excluded. This is compared to excluded regions obtained by the BFRT photon regeneration experiment [3] (light gray area), to searches for deviations of the Coulomb law [37] (points) and to comparisons of the Rydberg constant for different atomic transitions [38] (stripes).