Assessment of the method

The method to measure the nonlinear thickness in a poled glass presented here is based on the measurement of two Maker fringe patterns. One pattern is recorded using a poled glass, and the second one is measured using a stack made of the measured poled glass and an identical one that could be fabricated either by poling it in the same conditions as the first glass or by cutting the first glass in two halves. The method also requires that the two nonlinear regions are facing each other in the stack. According to eq.(4.21) the ratio of the SH power measured at each angle gives the nonlinear thickness L. Finally, havingL, the χ(2) is estimated with reference with to a α-quartz crystal as described in section 4.1.7.

0 10 20 30 40 50 60 70 80 90 0 2 4 6 8 10 12 14 16 18

Angle of incidence ϑ (deg)

Nonlinear thickness L

o

(

µ

m)

Figure 4.21: Thickness of the nonlinear layer in a set of silica plates poled for different times: (diamonds) 7 min, (square) 15 min, (up -triangle) 30 min, (down triangle) 45

min, (circles) 90 min.

0 20 40 60 80 100 120 140 160 180 0 2 4 6 8 10 12 14 16 18

poling time (min)

Nonlinear thickness L

o

(

µ

m)

Figure 4.22: Measured values for the nonlinear thickness against poling time. The solid curve is the best stretched-exponential fit for the experimental data

silica plates (Herasil 1), poled in air for 7, 15, 30, 45 and 90 minutes at 280◦_{C, with}

4 kV applied voltage. The samples were tested for SHG using a mode-locked and Q- switched Nd:YAG laser (λ= 1064 nm) with a focused spot radius of w_o = 26µm. The corresponding Rayleigh range (2z0 ∼4 mm) is long enough to encompass the whole stack,

so that the effect of spherical aberrations and astigmatism is reduced. No index matching fluid is used between the samples, since its larger dispersion (compared to air) would introduce a phase shift between the SH beams generated by the two layers, randomly modifying their interference condition and making it impossible to retrieve the value of L. According to eq. (4.40) the thickness of the nonlinear optical layer is calculated, for each sample, at different angles (Fig.4.21). As in the standard Maker’s fringe technique, once determined, the value of L is used to fit the MFT pattern (Fig. 4.18) and χ(2) is evaluated by comparison with α-quartz crystal (d11 = 0.3 pm/V). A χ(2) of 0.14

pm/V is determined in the sample poled for 90 minutes. The evolution ofLwith poling time agrees well with the result of previous studies [75] (Fig. 4.22). Each datum for L

presented in Fig. 4.22 is the average of the values obtained at different angles (Fig. 4.21). The experimental errors, evaluated from the standard error of the mean, are in general below 10%.

Evaluation of poling methods and

materials for devices

The value of the induced second-order nonlinearity in thermally poled silica glass is more than one order of magnitude smaller than theχ(2)in nonlinear crystals such as LiNbO3.

Two main possibilities to reduce this gap can be envisaged: the first possibility is to use a different technique from thermal poling (see section 3.6). Corona poling [23, 157, 160], UV-poling [27, 156],electron-beam poling [24, 161], all-optical poling [162] have been reported so far. The second possibility is to either change or modify the material to make it more prone to the poling process. Several groups are now working on poling of glasses other than silica (see Section 3.5). Glasses with high χ(3) _{[8, 100, 159] are}

the most promising in order to obtain a larger χ(2) _{(see χ}(2) _{= 3χ}(3)_E

dc). Moreover,

pre-treatment of silica glass, with the aim of preserving its excellent optical properties whilst including new features, has already proven to be effective. For example it was shown that the creation of point defects in synthetic silica by UV exposure led to a ten fold increase in the value of the nonlinearity induced by thermal poling [80].

This thesis led to the discovery of a new poling method [28]: UV-fs poling, where two- photon absorption of femtosecond UV pulses in the presence of an applied electric field creates a permanent space-charge in pure silica glass. This technique is described in section 5.1. Various nanostructured glass-ceramic families have also been character- ized, both using standard MFT technique and the new stack Maker’s fringe technique introduced in section 4.2. In particular, thermal poling as efficient as in silica, was demonstrated in photosensitive tin-doped silica glass (Section 5.2). The addition of pho- tosensitivity to the optical properties of poled silica makes this new material attractive for planar poled devices where waveguides can be written by UV exposure. The gener- ation of point defects in silica and their effect on the poling process have been studied by exposing the glass to eitherγ-ray or IR-femtosecond pulses. γ-ray pre-treatment has been experimentally investigated in both type-II and type-III silica glass, in a wide range

Figure 5.1: Schematic of the UV-assisted poling experiment with induced photocur- rent measurement.

of doses and dose-rates of exposures to γ-rays. This study is presented in Section 5.3. An original contribution of this thesis is also the discovery that IR femtosecond pre- treatment of silica enhances the capability of the glass to freeze an electric field, leading to higher nonlinearity (Section 5.4).

The issue of the stability of the nonlinearity in poled glass, that is fundamental for the development of practical devices, has been addressed in Section 5.5. Glass compositions, whose second-order nonlinearity is far more stable than for poled silica, were identified. In Section 5.6, the compatibility of each poling method, glass, and glass pre-treatment with the fabrication of a glass based frequency doubler, is critically discussed on the basis of both literature review and of the experiments presented in this chapter.