LiNbO3 is a promising photorefractive material for many applications [1]. Phase holograms formed in strip and planar waveguides may be used for phase conjugation, narrow-band holographic filters, or light amplifiers [2]. In most cases Fe2=3 serves as the photorefractive center. On the one hand, the maximum value of the light-induced refractive index changes has been found to be proportional to the concentration of Fe3 [3]. On the other hand, there are several hints that iron doping also has a significant influence on the dark conductivity of the material [4], thereby limiting the storage time. In waveguides the Fe concentration can be easily varied by indiffusion. In this letter we will show that the dark conductivity of heavily iron-doped planar waveguides in LiNbO3 increases exponentially with the iron concentration, and this correlation may also be applicable to samples with a much lower doping. Planar photorefractive waveguides are fabricated by two subsequent surface diffusions using commercially available optical grade y-cut LiNbO3 wafers of congruently melting composition. At first, an 80 nm-thick evaporated titanium layer is indiffused at 1000 C for 24 h in wet argon atmosphere to increase the surface refractive indices of LiNbO3. In a second step, evaporated iron layers of different thicknesses (30, 40, 60, 80, 100, and 120 nm) are indiffused at 1000 C for 18 h in wet oxygen to increase the photorefractive sensitivity of the samples. For comparison, we investigate a titanium-diffused waveguide (80 nm-thick Ti layer, indiffused at 1000 C for 18 h in wet oxygen) on a homogeneously doped LiNbO3 : Fe substrate [5], too. The concentration profiles of both, titanium and iron, can be well described by Gaussian functions, ci ci exp yy2=2r2, with penetration depths rTi 3:6 mm and rFe 10:7 mm, respectively [6]. All waveguides are multimode with typical depths of the nearly Gaussian refractive index profiles of rwg 5 mm. The relative iron concentration at the surface is measured with the help of an electron microprobe, where we demonstrate that cFe is proportional to the thickness of the evaporated iron layer. The absolute value of the surface concentration of the indiffused iron, cFe y 0, is determined by X-ray photoelectron spectroscopy (XPS) [6]. Only Fe3 can be detected, while the concentration of Fe2 is below the measuring accuracy. Thus we estimate a concentration ratio cFe2=cFe3 well below 0.01. The dark conductivity in the waveguide is measured with a holographic two-wave mixing technique. Both, a TE and a TM mode are excited with the help of a rutile prism to propagate collinearly along the x-axis of the sample, and the transmitted powers are measured. Modulated photovoltaic currents are generated by these orthogonally polarized waves, and a periodic space charge field builds up which leads to a perturbation of the dielectric constant via the electrooptic effect. Details of the formation of the gratings can be found in [7,8]. When the phase gratings are written to saturation, the two writing beams are switched off, and the decay of the grating in the dark is monitored by subsequently measuring the diffraction efficiency for short time intervalls (50 ms) with a low-intense reading beam. We define the diffraction efficiency as the ratio of diffracted to total read-out light intensity. The dark conductivity sd is calculated from the relaxation time td of the gratings via sd ee0=td, where e0 is the vacuum permittivity and e is the static dielectric constant of LiNbO3. In Fig. 1 the measured value of dark conductivity is shown as a function of the iron concentration cFe cFe3 at the surface of different waveguides. Samples with iron contents larger than 2 1020 cmy3 are produced by indiffusion of iron, whereas the sample with the lowest iron concentration (cFe 0:13 1020 cmy3) is fabricated on a LiNbO3 : Fe substrate that has been homogeneously doped in the melt. For the different modes of a waveguide, i.e., for different propagation depths, we obtain almost the same dark conductivity. This is expected from the nearly constant iron concentration in the waveguiding layer, and points to a minor influence of titanium on dark conductivity, because the titanium concentration strongly decreases with increasing propagation depth [6]. The measured dependence ln s cFe can be fairly well approximated by a straight line, including the sample prepared on the melt-doped substrate, thus obeing the relation sd cFe sd exp cFe=cFe, Rapid Research Notes R3