Figure 1: Photograph of the interferometer in our laboratory in Vienna after recording the gratings.

To generate the gratings we are employing the photo-neutron refractive effect of deuterated polymethylmethacrylat doped with a photosensitizer [3-5]. A standard holographic two-wave mixing set-up was used to induce refractive-index changes for neutrons. As the number density of the polymer and the monomer differs, an irradiation of the samples with a sinuousidal light-interference pattern leads to density changes which act as gratings.

This process is enabled by a decomposition of the photosensitive component into free radicals which start a polymerisation process in the bright region.
Advantages of this technique are that the grating spacing can be varied between 180 nm and several μm, that the magnitude of higher diffraction orders and their diffraction efficiency can be tuned by the exposure, intensity and thickness of the samples.

Figure 2: Sketch of the interferometer for cold neutrons based on holographically generated density gratings in d-PMMA.

The incident beam is split at each grating, resulting in a total of 8 rays (R₁-R₄ and S₁-S₄) behind the third grating. Only two pairs (R₂+R₃ and S₂+S₃) contribute to interference. After the third grating we obtain a coherent superposition of the wavefunctions Ψ^I+Ψ^II. To record an interferogram a phase-flag of sapphire was inserted in both beam-paths between the first and second grating. A subsequent rotation Φ of the sapphire around an axis perpendicular to the plane of incidence generates a phase-difference between the beams and results in correlated intensity modulations behind the third grating. Those intensity changes were monitored on the detector matrix.

The observed decay of the interference fringes with increasing phase-difference gives direct experimental access to the absolute value of the normalised coherence function via the visibility [7,8], and as a by-product to the coherence length 1_c. The cross-correlation or van Cittert-Zernike theorem gives an interrelation between the coherence function and the spectral distribution of the neutron beam via a Fourier transform. For a mean neutron wavelength of λ = 2.0 nm and a spectral distribution of Δλ/λ = 10% we obtained a coherence length of l_c = 10.3±0.5 nm. The maximum visibility of the inference fringes is 21%. Those measurements were performed at the SANS instrument D22 at the ILL.

Figure 3: Interferogram obtained by rotating a 4.1 mm thick phase-flag from sapphire around an axis perpendicular to the plane of incidence. The decay of the visibility reflects the coherence function. The solid line represents a fit to a Gaussian decay function.