# Full text of "Photon Self-Induced Spin to Orbital Conversion in TGG crystal at high laser power"

## See other formats

Photon Self-Induced Spin to Orbital Conversion in TGG crystal at high laser power. O <N ©jo: < <J1 o • 1—1 •*-> Oh q o ^ . > m oo <N 00 o o S. Mosca, 1 ' 2 B. Canuel, 3 E. Karimi, 1 B. Piccirillo, 1 ' 4 L. Marrucci, 1 ' 5 R. De Rosa, 1 ' 2 E. Genin, 3 L. Milano, 1 ' 2 and E. Santamato 1 ' 4 '^ 1 Dipartimento di Scienze Fisiche, Universitd di Napoli "Federico II", Compl. Univ. di Monte S. Angelo, 80126 Napoli, Italy 2 INFN Sezione di Napoli, Compl. Univ. di Monte S. Angelo, 80126 Napoli, Italy 3 EGO, European Gravitational Observatory, Via E. Amaldi, 56021 S. Stefano a Macerata - Cascina (PI) ITALY 4 CNISM-Consorzio Nazionale Interuniversitario per le Scienze Fisiche della Materia, Napoli 5 CNR-INFM Coherentia, Compl. Univ. di Monte S. Angelo, 80126 Napoli, Italy (Dated: July 2010) In this paper, we present experimental evidence of a newly discovered third-order nonlinear optical process Self-Induced Spin-to-Orbital Conversion (SISTOC) of the photon angular momentum. This effect is the physical mechanism at the origin of the depolarization of very intense laser beams prop- agating in isotropic materials. The SISTOC process, like self-focusing, is triggered by laser heating leading to a radial temperature gradient in the medium. In this work we tested the occurrence of SISTOC in a terbium gallium garnet (TGG) rod for an impinging laser power of about 100 W. To study the SISTOC process we used different techniques: polarization analysis, interferometry and tomography of the photon orbital angular momentum. Our results confirm, in particular, that the apparent depolarization of the beam is due to the occurrence of maximal entanglement between the spin and orbital angular momentum of the photons undergoing the SISTOC process. This expla- nation of the true nature of the depolarization mechanism could be of some help in finding novel methods to reduce or to compensate for this usually unwanted depolarization effect in all cases where very high laser power and good beam quality are required. PACS numbers: 42.65.-k, 42.50.Tx, 42.65.Jx There are several optics experiments in the world where both high laser power and excellent beam quality are simultaneously mandatory. For instance, this is the case for the next generation of optical interferometers used to detect gravitational waves, where high laser power, of the order of 200 W, in the fundamental mode, is required to increase the detector sensitivity [J-Q- Thermal effects appearing in high power lasers or in the bulk optical components exposed to high laser power due to non-negligible absorption can strongly affect the beam quality. In solid state lasers, thermal gradients within the laser medium cause refractive index changes leading to thermal lensing, aberrations and birefrin- gence. In particular, the thermally induced birefringence is known to introduce a depolarization of the light that becomes the limiting effect on power scaling [J-[6j. Ways to compensate for this effect have been recently proposed [6|-[8|, thus opening the possibility to realize high power and high quality continuous wave lasers that could be used in gravitational wave interferometers or in other applications. The Faraday Isolator (FI) is one of the components to be most strongly affected by thermal effects and is fundamental for the success of gravitational wave optical experiments. The magneto-optic crystal used in the FI is Terbium Gallium Garnet (TGG) which has a relatively high absorption (generally higher than 1000 ppm-cm -1 ) [9|. This absorption creates an overall K enrico.santamato@na.infn.it temperature increase of the magneto-optic crystal and generates a non-uniform temperature distribution over the transverse cross section of the optical element. Both effects can significantly impact Faraday isolation when going to high power. The first effect is associated with the Verdet constant change and is detailed in [10]. The second effect is associated to mechanical stresses induced by temperature gradient and gives the main contribution to the apparent depolarization of high power laser beams and to the consequent deterioration of the degree of isolation Q. In the present paper, we aim to demonstrate that the mechanism creating the depolarization and there- fore spoiling the Faraday isolator performance is a self- induced partial Spin- To- Orbital Conversion (STOC) of the input photon Spin Angular Momentum (SAM) into the Orbital Angular Momentum (OAM) flll4l3j. The Self-Induced STOC, or SISTOC, in fact, may put some photons of the input beam into particular states where the photon SAM and OAM are maximally entangled. The apparent depolarization of the beam is a direct man- ifestation of the decoherence of two quantum degrees of freedom (the photon SAM, here) when they are entan- gled. The depolarization is said to be "apparent" , here, because it could be removed, for example, by a quantum erasing apparatus [l4j . In order to gain a deeper insight into the SISTOC pro- cess we used three different techniques: polar imetry, in- terferometry and full OAM tomography in the photon spinorbit space. Our experiments were carried out with 2 the input laser beam polarized either linearly or circu- larly. As expected from the theory, SAM- OAM entan- glement and apparent depolarization of the photons con- verted by SISTOC was only found in the case of linearly polarized incident beam. In the case of the circular po- larization, there is no SAM-OAM entanglement and the SISTOC converted photons were found to be in OAM eigenstates with full circularly polarization with helicity depending on the helicity of the input beam. A. Photo-elastic effect by radial thermal gradient and induced birefringence In the case of a high intensity TEMoo Gaussian beam, the optical material experiences temperature gradients imprinted by the bell-shaped beam profile. These gradi- ents introduce a mechanical stress in the material creat- ing a birefringence with a radial symmetry. In isotropic materials [HI , the birefringence axis follows the radial di- rection, along the temperature gradient and the birefrin- gence optical retardation S(r) depends only on the radial coordinate r. The temperature-induced birefringence re- tardation S(r) can be found by sol ving the thermal and elastic problem in the material 0, fig . In the case of a Gaussian heat source, S(r) is given by S(r) aP LT -2r 2 /r 2r 2 /r (1) where A is the optical wavelength, tq is the beam radius at the rod position, a and k are respectively the ab- sorption coefficient and the thermal conductivity of the material, and T is an effective opto-elastic coefficient. Taking typical values for TGG (Tg T = 4, 7 • 10~ 6 K" 1 , a = 1500 ppm cm -1 , k = 7.4 W m _1 K 1 . L = 18 mm, and considering an incident power Pq = 125 W, we found | ~ 5.7°. We may then consider in Eq. ((4]) below as a small quantity. However, the key point in produc- ing the SISTOC effect is the radial direction of the local optical axis in the heated material, as shown in Fig. [U In fact, the indefinite birefringence direction located at the center of the heated medium creates a topological singularity of charge q = 1 which is tranferred into the phase of the optical beam leading to a vortex light beam carrying OAM. We may regard the heated material as a g-plate, an optical device recently developed for orbital angular momentum (OAM) manipulation exploiting the STOC process [ll|. The local orientation ip of the opti- cal axis in the plane of the (/-plate is generally written as i/j = q<j) + ao [11|, where (j) is the azimuthal angle, leading to the analogy between the (/-plate and the heated TGG for q = 1 and ao = 0. Q-plates are usually made with liq- uid materials and their optical retardation 5 is uniform, whereas the optical retardation of the heated TGG has the radial distribution (pQ). The effect of the heated TGG on a polarized beam can be calculated in the same way as for the (/-plate, using the Jones matrix formalism and \ \ \ V * 4 4 r 4 4 4 4 \ i i f r r t y / . \ i i t r r r * > * i i t r r * * < \ \ \ i f r > \ \ \ \ j / / > > v \ \ j / / v t y — s / f t \ \ x ^, d // / t\w>i 4 4 M \ \ v % M 4 4 i k i W V V V 4 4 I A ^ kVVVVV 4 4 I l i k V V V V V FIG. 1. Temperature distribution in the TGG transverse plane. The arrows show the temperature gradient as well as the local direction of the thermally induced birefringence taking into account the radial dependence of the optical delay. B. SISTOC process The Jones matrix F(r, (j)) representing the heated ma- terial is given by [ll| F(r,<f>) = R(c/>) ( e "7" ^)R(-4>), (2) \ e 2 J where S(r) is given by Eq. (pp) and R is the rotation matrix of angle (j) R cos <p sm ^ — sin 6 cos ( (3) Let us consider a high power TEMqo Gaussian beam impinging on the TGG rod. In the case of the cir- cular polarization, the input optical field is E in = Eo(r)e±, where e± = (e x d= ie y ) /y/2. The output field E ou t{r,(j)) = F(r,(j))Ei n transmitted beyond the heated rod is calculated from Eq. (j2j) as £ O ut(r,0) = Eo(r) ■ S(r) cos e± 2 isin^e^ 2 (4) The first term on the right of Eq. ((4]) proportional to cos^/2 has the same circular polarization and radial dependence of the input field. We will refer to this term as to the unconverted part of the input beam. The unconverted part of the beam carries no OAM. The second term, proportional to sin 5/2 has opposite circular helicity and presents the characteristic phase factor exp(±2i</>), corresponding to the definite OAM content of ±2Ti per photon. This term is the part of the input beam that was converted by the SISTOC process. 3 Each photon in the converted part of the beam has its SAM changed by ^2T% and its OAM changed by ±2ft, thus leaving the total (SAM+OAM) angular momentum conserved, which is a peculiar feature of the STOC process pjj. From Eq. (jlj we see that S(r) and, hence, the fraction of photons that are converted by STOC, depends on the power Pq carried by the beam itself, which is the characteristic of self-induced third-order nonlinear optical process. We may then regard the SISTOC as a thermally induced nonlinear process as self-focusing, but able to change the OAM content of the beam. In the case of the input beam being linearly polarized along the x axis the input optical field is Ei n = Eo(r)e x , and the field transmitted beyond the heated material is given by S(r) cos ——e x — 2 SoutM) =E (r) 5(r) i sin — ^— (cos 2<p e x + sin 2(j) e y ) (5) From Eq. (j5j) we see once again that the term propor- tional to cos 5/2 is the unconverted part of the input beam, while the other term, proportional to sin J/2, is the part converted by SISTOC. However, unlike the previous case, where the SISTOC converted photons were circularly polarized, in this case the converted part is a coherent, maximally entangled, superposition of the left- and right-circular polarizations and the ±27i eigenstates of the photon OAM. It is precisely this spinorbit entanglement the ultimate reason of the apparent complete depolarization [l^ of the converted field noticed in previous works Q. Our measurements were carried out in the far field be- yond the heated TGG material. In order to compare the experimental data with theory, we must Fourier- transform the field given in Eqs. (J2J) and (|5j). For a Gaus- sian input field, the result is TP SISTOC 37r<xPoZ/Yro e 2 (cos 2<j) e x + sin 20 e y ) (6) where p = (\^2ttwo j \fZ\z)r' , and r' is the radial coordi- nate in the far field transverse plane ad distance z. In calculating Eq. (|6j), we considered only the SISTOC part of the output beam in the limit of small 5. The upper row in Eq. © refers to the case of left circular polariza- tion of the input beam and the lower row the case of the linear polarization along x. In Fig. [3] we show the calcu- lated far-field transverse profiles of the SISTOC part of the beam. In the case of input left circular polarization (Fig. [3^i), the intensity takes the typical doughnut pro- file of the OAM eigenstates. In the case of input linear polarization the intensity profile of the x and y compo- nents of the far field have four-leaf clover shapes rotated by 45° each other (Fig. [3Jd,c, respectively). Finally, we notice that in view of Eq. © the power fraction con- verted by the SISTOC process scales as the square of the incident power, as shown in Fig. [2] and found in previous experiments [7]. FIG. 2. Fraction 7 of the incident power converted by the SISTOC process as a function of the incident power Po- The fit (solid curve) confirms the square dependence. The experimental study of SISTOC in a TGG crystal (a) (b) (c) • • • • • • • • FIG. 3. Expected far-field beam profile of SISTOC converted modes calculated from Eq. (|6]). Left-circular input polar- ization (a); x-linear input polarization x-component (b); y- component (c). In our experiments we used a <lll>-cut TGG rod (di- ameter =20 mm, length=18 mm) from Northrop Grum- man with an absorption of 1900 ppm cm -1 . The laser source was a high power diode-pumped Ytterbium fibre laser from IPG photonics delivering up to 200 W at a wavelength close to 1064 nm. We setup several experi- ments to analyze the SISTOC part of the beam emerging form the heated TGG material. When the input beam was linearly polarized along x, we isolated the converted part of the output beam using a half- wave plate at 45° and a linear polarizer aligned along x (see Fig. S]). Only the ^/-component of the emerging field was analyzed, be- cause the ^-component was overwhelmed by the uncon- verted part of the input beam. The far-field intensity profiles of the radiation converted by SISTOC was de- tected by a CCD camera. The experimental intensity profiles are shown in Fig. |4] for the case of linear in- put polarization and in Fig. [5] for the case of circular input polarization. As we see, the observed patterns in Figs. H] and [5] are in very good agreement with Figs.[3fc) and [31(a), respectively. The transition from the dough- nut to the four- leaf clover profile is evident. To prove HP-Laser FIG. 4. Measurement of the intensity mode profile of the SISTOC light in TGG with the linear input polarization at Po = 125 W. A half- wave plate placed in front of the second PBS is adjusted to obtain minimum of transmission beyond the second polarizer, in order to have only the SISTOC light transmitted. A four- leaf clover shape is observed beyond the second PBS. HP-Laser FIG. 5. Measurement of the intensity mode profile of the SISTOC light in TGG with the circular input polarization at Po = 125 W. The first quarter- wave plate makes the po- larization left circular on the TGG material and the second quarter- wave plate at 45° turns back the polarization into lin- ear along y to be reflected by the second PBS. A doughnut- shaped vortex beam is observed beyond the second PBS. that the SISTOC converted beam acquires the phase az- imuthal dependence exp(2z0) corresponding to the OAM eigenvalue 2% per photon, we arranged the interferome- ter shown in Fig. (|6j). To isolate the SISTOC converted field, we used the same setups shown in Figs. [5] and HI The only difference was the polarization quality of the second PBS, voluntarily chosen as worst so to transmit, together with the converted field generated by the heated material, a small fraction (about £=1/1000 intensity) of the a>polarized unconverted light. Therefore, at the en- try of the interferometer, we have the sum of two fields M FIG. 6. The SISTOC light was isolated as in the previous ex- periments. The second polarizer, however, transmits a small part of the unconverted light too. The converted and uncon- verted fields sent into the interferometer have orthogonal po- larizations. The unconverted light is used as reference beam and sent into the upper arm of the interferometer. The refer- ence beam passes twice through a quarter- wave plate to send back the whole field to the output of the interferometer. The mirror of this arm is placed on a translation stage to adapt the differential length of the two arms. A lens makes the ref- erence beam wavefront curved. The SISTOC beam is sent into the other arm of the interferometer. The reference and the SISTOC beam interfere in the output polarizer at 45°. The fringe pattern is observed by the CCD camera. into orthogonal polarizations, viz. ^interf = y/eE (r)e x + (^SISTOC 0, <j>) (7) where i2siSTOc(^ <t>) is given by Eq. (j6]). It is worth not- ing that the two terms on the right of Eq. ([7]) have a different OAM content and different polarization. The first term was unconverted by SISTOC and is left in the TEMoo mode with no OAM, while the second term is con- verted into a linear combination of the OAM eigenvalues ±27i, in general. The unconverted part of the field was used as reference and it was sent into the upper arm of the interferometer and the SISTOC converted field was sent into the other arm. The two fields were made to interfere in the polarizer oriented at 45° and the fringe pattern was detected by the CCD camera. The observed interference pattern for the linear input polarization is shown in Fig. ([7]) on the right. The interference pattern reveals a 7r-phase- shift between each lob of the four-leaf clover. This pattern is completely in agreement with that calculated from Eq. ([6]) and shown on the left of Fig. flTJ). The calculated and observed interference patterns for the circular input polarization respectively are shown on the left and right sections of Fig. ([5]). In this case, the two patterns exhibit the two-branch spiral shape characteris- tic of the OAM eigenvalue 2h. To complete the analysis 5 FIG. 7. (a) the simulated pattern; (b) interference pattern obtained on the CCD camera of the setup shown in Fig. [6] for P = 150 W and linear input polarization along x. The 7r-phase shift between each lob of the four-leaf clover may be noticed. FIG. 8. (a) the simulated pattern; (b) interference pattern obtained on the CCD camera of the setup shown in Fig. [6] for Po = 100 W and left circular input polarization. The two- branch spiral shape characteristic of the OAM eigenvalue 2ft may be noticed. of the OAM content of the SISTOC converted light, we made a full OAM tomography of the SISTOC light gen- erated by the heated TGG rod. The main advantage of the OAM tomography is its capability to measure both amplitude and relative phase of the OAM components of a light beam [18]. In our case, the polarization of the collected SISTOC light is fixed to be orthogonal to the input beam, so only the tomography of the OAM con- tent of the beam is required. Moreover, the cylindrical symmetry of the system, allows us to restrict the tomog- raphy to the two-dimensional Hilbert space spanned by the eigenvectors of the photon OAM with opposite eigen- values ±2ft. In our experiment, we used the holographic tomography technique [l8|, Qil already used in single pho- ton O AM-based experiments [20|, Hl| . This technique ex- ploits the six computer generated holograms shown in Fig. [9] We made these holograms by a photographic technique, starting from computed images. After chemi- cal bleaching, the first-order diffraction efficiency of our holograms was about 10% at 1064 nm wavelength. The measurements were made by carefully aligning each holo- gram with the SISTOC beam transmitted by the heated TGG rod and collecting the a TEM 00 -like" spot of the far n 1 II 1 1 i >J 1 II M H 'rl L I ii 'llllll tji mi III A FIG. 9. Patterns of the holograms used in the experiment. The holograms in the same column correspond to the two eigenstates of Pauli's operators a z , a x , a y in the 2D Hilbert space spanned by the OAM eigenstates of a z . Each hologram corresponds to the OAM state reported in the upper left cor- ner. The state notation is the same as used for the photon SAM, but the corresponding symbols are put in the lower case to indicate OAM. field in the first-order diffraction direction [l8j |. The far field spot was collected at the focal plane of a microscope objective and filtered through a small aperture iris. This technique, in fact, projects any unknown photon OAM state on the OAM state fixed by the hologram [l8j ]. The 2D OAM subspace considered here is isomorphic to the 2D space of the photon spin. We may think of the holo- grams in Fig. [9] as equivalent to polarizers acting in the spin space. The holograms in the first column corre- spond to polarizers selecting the left (1) and right (r) circular polarizations; the holograms in the second col- umn correspond to polarizers selecting the horizontal (h) and vertical (v) polarizations; the holograms in the third column correspond to polarizers selecting the antidiago- nal (a) and diagonal (d) polarizations, as indicated in the upper left corner of the images. In complete analogy to the polarization state analysis, we measured the "Stokes- like" parameters S3 in each one of the three above men- tioned bases so to reconstruct the density matrix of the OAM state [19[ . The real and imaginary parts of the den- sity matrix obtained from our measurements are shown in Fig. [10] for the circular input polarization (a) and the linear input polarization (b). The fidelity of SISTOC process for the case of circular and linear input polariza- tions are 0.98 and 0.86, respectively, which are promising results and show that our experimental results are in a very good agreement with our theoretical model [22| . As expected from Eq. [6l in the case of the circular input po- larization, the SISTOC converted photons are put into the OAM eigenstate 2ft, while in the case of the linear input polarization, the state of the ^/-component of the SISTOC photons is an equally- weighted antisymmetric superposition of the OAM eigenstates 2ft and —2ft. 6 (a) FIG. 10. Experimental density matrix for (a) left circular input polarization and (b) for linear input polarization along x. The left and the right charts show the real and imaginary parts, respectively. The OAM eigenvalues are in units of ft. D. Conclusions We have proved by a series of experiments, includ- ing full OAM state tomography that the apparent depo- larization observed when a very high power laser beam passes through a medium is due to a new thermally in- duced third-order process, namely Self-Induced Spin- To- Orbital Conversion (SISTOC), where a power-dependent fraction of the incident photons converts its angular mo- mentum from spin into orbital. Our experiments are in full agreement with a model where the SISTOC conver- sion is limited to the 2D OAM subspace spanned by the OAM eigenstates ±2% per photon. The SISTOC process is triggered by the birefringence induced in the material by radial temperature gradient due to light absorption. The fraction of light suffering the SISTOC process re- mains proportional to the square of the input laser power up to about 100 W. The light depolarization is appar- ent because it is not due to random dephasing of the polarization components, but to the entanglement be- tween the photon SAM and the OAM degrees of freedom. Suitable quantum erasing scheme could remove such en- tanglement so that the SISTOC component could be re- moved from the beam. We studied the SISTOC process in a TGG Faraday isolator, but the process is very general and it may occur in isotropic materials, as, for example, Nd: Glass rods under strong pumping conditions. [1] The Virgo collaboration, Advanced Virgo baseline de- sign, VIR-0027A-09, https://tds.ego-gw.it/ql/?c=6589, (2009). [2] M. A. Arain, and G. Mueller, Opt. Express 16, 10018- 10032 (2008). [3] H. Grote and the LIGO Scientific Collaboration, Class. Quantum Grav. 27, 084003 (2010). [4] J. D. Forster, and L. M. Osterink, J. Appl. Phys. 41, 3656-3663 (1970). [5] M. P. Murdough, and C. A. Denmann, Appl. Opt. 35, 5925-5236 (1996). [6] M. Frede, R. Wilhelm, R. Gau, M. Brendel, I. Zawischa, C. Fallnich, F. Seifert, and B. Willke, Class. Quantum Grav. 21, S895 (2004). [7] E. Khazanov, O. V. Kulagin, S. Yoshida, D. B. Tanner, and D. H. Reitze, IEEE J. Quantum Electron. 35, 1116 (1999) [8] M. Frede, R. Wilheim, M. Brendel, C. Fallnich, F. Seifert, B. Willke, and K. Danzmann, Opt. Express 12, 3581 (2004). [9] T. V. Zarubina, and G. T. Petrovskii, Opt. Zh. 59, 48 (1992) [Sov. J. Opt. Technol. 59, 700 (1992)]. [10] The Virgo collaboration, In vacuum Faraday isolation remote tuning, accepted for publication, Appl. Opt., 49, no. 24 (2010). [11] L. Marrucci, C. Manzo, and D. Paparo, Phys. Rev. Lett. 96, 163905 (2006) [12] E. Karimi, B. Piccirillo, E. Nagali, L. Marrucci, and E. Santamato, Appl. Phys. Lett. 94, 231124 (2009) [13] E. Karimi, B. Piccirillo, L. Marrucci, and E. Santamato, Opt. Lett. 34, 1225 (2009) [14] A quantum eraser based on C-NOT logic in the photon spinorbit space is under study and will be the object of a forthcoming work. [15] TGG is a "weakly anisotropic" cubic material, where the difference pn — p±2 — 2^44 among the photoelastic coef- ficients is very small (see [7|] for notation). [16] E. Khazanov et al, IEEE J. Quantum Electron. 40, 1500 (2004) [17] In a maximally spinorbit entangled state, the reduced quantum density matrix (known in optics as the co- herency matrix) of the SAM degree of freedom is pro- portional to the 2D unit matrix. [18] M. J. Padgett, and J. Courtial, Opt. Lett. 24, 430 (1999) [19] D. F. V. James, P. G. Kwiat, W. J. Munro, and A. G. White, Phys. Rev. A 64, 052312 (2001) [20] E. Nagali, F. Sciarrino, F. De Martini, L. Marrucci, B. Piccirillo, E. Karimi, and E. Santamato Phys. Rev. Lett. 103, 013601 (2009) [21] E. Nagali, F. Sciarrino, F. De Martini, B. Piccirillo, E. Karimi, L. Marrucci, and E. Santamato, Opt. Express 17, 18745 (2009) [22] The quantum fidelity is given by (ip\p\ip), where p and are the experimental density matrix and the expected theoretical state.