Antiresonant ring interferometry as a sensitive technique for measuring nonlinear optical properties of thin films Parinda Vasa a, B.P. Singh b, Praveen Taneja a, Pushan Ayyub a a,* Department of Condensed Matter Physics and Material Science, Tata Institute of Fundamental Research, Homi Bhabha Road, Mumbai 400005, India b Physics Department, Indian Institute of Technology Bombay, Powai, Mumbai 400076, India Abstract There is a critical need for a simple technique for the accurate measurement of weak optical nonlinearities such as the nonlinear coefficients of thin films. We discuss the experimental set-up and provide a realistic analysis for a sensitive and single beam technique based on an antiresonant ring interferometer for measuring nonlinear optical coefficients in thin films. The technique was benchmarked using toluene and its superiority was demonstrated by measuring the effective nonlinear absorption coefficient of a 1.3 lm thick CdS film, which could not be detected using standard techniques such as z-scan. We show that this technique can, in principle, be used for films with thickness down to the nanometer regime. PACS: 42.65.An; 42.70.Nq; 78.66.Hf 1. Introduction Considerable scientific research has been focussed on the development of novel materials for nonlinear optical and optoelectronic device applications. These include band gap engineered low-dimensional quantum confined semiconductors and polymers. It is particularly advantageous to grow these materials in the form of thin films, a geometry that inherently favors the integration of devices in the system architecture. Various techniques have been developed to study the nonlinear optical response of materials. Though z-scan is one of the simplest and most popular single beam techniques [1], it does not provide the sensitivity usually necessary for studying thin films. In a 1 lm thick film, for example, z-scan either fails to detect the nonlinearity or requires intensities high enough to cause undesirable processes such as excited state absorption, thermal effects [2] and optical damage [3]. Other methods such as degenerate four wave mixing (DFWM) [2] or interferometric 298 measurements [4,5] are more sensitive but difficult to implement, requiring careful spatial and temporal alignment of two or more beams. Lee and Hughes (LH) [6] proposed a simple, sensitive, single beam technique based on an antiresonant ring (Sagnac) interferometer for simultaneously measuring the real and imaginary contributions to optical nonlinearity, and called it Antiresonant Ring Nonlinear Spectroscopy (ARINS). Used for measuring very small changes in a variety of physical properties, the antiresonant ring (ARR) is sensitive enough to detect changes of less than 0.005% in the laser amplitude. Due to its high sensitivity, ARR allows measurements to be performed at low intensities, eliminating unwanted processes that may complicate measurements. It also offers selective probing of the nonlinear optical processes by discriminating against their response time. As both the beams in the ARR traverse the same path, it is simpler to align than other interferometric techniques and is intrinsically immune to thermal fluctuations and mechanical vibrations [7]. Such a combination of properties makes ARR an ideal technique for studying the nonlinear response in thin films. This technique utilizes the dressing of two unequal-intensity counter-propagating beams with differential nonlinear phase, which occurs upon traversing the sample. This difference in phase manifests itself in the intensity dependent transmission. Thus it is important to note that the measurement of cubic nonlinearity requires the use of pulsed laser beams only. Most nonlinear optical measurements are performed using ultrashort laser pulses with Gaussian temporal and spatial profile. The transmission of the ARR resulting from the interference of the two counter-propagating pulses varies nonlinearly both in space and time in a complex manner. Photodetection of the transmission of the ARR yields spatially and temporally integrated response. Consequently, it is the pulse energy and not the instantaneous power that is detected. The analysis of ARINS by LH [6] does not cover these crucial aspects and treats the transmission of ARR for CW beams. As pointed out above, it is not actually possible to make the measurement with a CW beam or even with pulsed beams having a long pulse duration. The treatment of LH is thus applicable to the somewhat academic problem of square pulses with a top hat spatial profile. Since LH had applied their analysis to data obtained using Gaussian pulses from an ultrafast Ti:sapphire laser, the results obtained were probably not accurate. This aspect presumably went unnoticed because LH had not benchmarked the technique using a standard nonlinear sample. We have, in fact, confirmed that in the case of a common system such as toluene, their analysis yields a value of the thirdorder nonlinearity that is lower than the literature value by a factor of 25. It is, therefore, imperative to modify the analysis to consider pulsed excitation with realistic shapes. Here we present a generalized analysis of ARINS valid for the measurement of real and imaginary components of nonlinear susceptibility and discuss the interpretation of our results. We incorporate linear as well as nonlinear absorption (two-photon absorption or saturation of absorption). We have also taken account of the actual reflectivities of the two mirrors (viz., 80% and 4%) used in the ARR, while LH had assumed both mirrors to have 100% reflectivity, a situation under which ARINS cannot function. To test the efficacy of our formulation and benchmark the technique, we have measured the nonlinear susceptibility of toluene. Our results are in excellent agreement with earlier values of the electronic nonlinearity. We also report the measurement of the effective nonlinear absorption coefficient (bÞ of a 1.3 lm thick CdS film, which demonstrates the superior sensitivity of ARINS compared to z-scan. Even though the ARINS technique was proposed more than a decade ago and has found applications in mode locking [8] and fiber optics [9–11], it has subsequently never been applied to the measurement of nonlinearity in spite of its simplicity, elegance, versatility and extremely high sensitivity. We believe that the physically realistic formulation proposed here will allow this powerful technique to be used widely for the determination of small nonlinear optical responses (such as in thin films) and help to elucidate their corresponding mechanisms. 299 2. Experimental setup and formulation The ARR can be setup in different configurations based on the Sagnac interferometer [12]. Fig. 1 shows the ARINS setup used in our experiments. A 50–50 beam splitter divides the input pulse into two counter-propagating pulses having p phase difference. Reflections from a high reflectivity mirror and an uncoated flat with 6° wedged rear surface close the ring and the two pulses again recombine at the beam splitter to yield ARR 2 2 transmission, jEout j / jEcw þ Eccw j . Here Ecw and Eccw are the optical fields traveling in clockwise and counterclockwise directions respectively. Inside the ring there is a unit magnification telescope comprised of a pair of identical lenses in 2f configuration. One of the lenses focuses the pulses into the sample placed at its focus while the other one re-collimates them. Two counter propagating fields traversing the ring acquire linear as well as intensity dependent nonlinear phase shifts. Since both the fields traverse the same optical path through the ring and encounter the same interactions with the optical elements, linear interactions would affect their amplitude and phase in identical manner. For an exactly 50% beam splitter, in the absence of any nonlinear interactions, the two returning fields with the same amplitude and phase difference will interfere destructively at the beam splitter to yield zero transmission. In this ÔbalancedÕ condition all the input power is reflected back to the incident direction. Measurement Fig. 1. Schematic diagram of the experimental set-up for antiresonant ring nonlinear spectroscopy (ARINS). against this dark background provides the basis for the improved sensitivity essential for measuring relatively weak signals. Any small deviation from the ideal splitting ratio (d) results in a leakage from the ARR and is responsible for the background that limits the sensitivity of the measurement. In the case of nonlinear interaction, the two unequal intensity counter propagating pulses in the sample will undergo different phase changes. Their superposition on the beam splitter will result in the intensity dependent transmission of the ARR, which is related to the nonlinear response of the sample. The difference of nonlinear phases picked up by two counter propagating pulses and hence the transmission of ARR will be maximum when only one of these is intense enough to cause an appreciable nonlinear polarization in the medium. This can be achieved if one of the pulses emerging from the 50–50 beam splitter is sufficiently attenuated by a neutral density filter before passing through the sample. In our setup this is implemented for the clockwise beam using an uncoated flat as a 4% reflector. We point out that the differential dressing of counter propagating pulses with nonlinear phases is possible only when the two do not interact simultaneously in the sample. Else, the same nonlinear phase will be impressed on both by the cross action phenomenon. The temporal overlap of the two pulses in the sample can be prevented by spatially offsetting the sample with respect to the center of the ARR. The time difference between the arrivals of the two pulses (Dsarr ) determines the nature of the nonlinear optical process that can be studied depending on its response time. Nonlinear processes with decay time longer than Dsarr do not contribute to the intensity dependent transmission of the ARR as both the pulses are affected identically. The delay window thus acts as an ultrafast gate. This gives ARR the unique ability to filter the nonresonant electronic contributions from integrating (or slow, e.g. thermal) nonlinearities or those arising from long lived (resonant) states and makes it ideal for time resolved studies and ultrafast gating. If the arrival times of the two pulses in the sample are reversed, one would measure the unfiltered response. We also point out that the weak nonlinear lens induced by the tightly focused Gaussian beam does not significantly affect the beam parameters 300 upon propagation, when the sample is situated at the beam waist (f the Rayleigh range) [13]. This is amply illustrated by the fact that no variation in transmittance occurs in closed aperture z-scan when the sample is at the focus [1]. Thus the spot sizes at the beam splitter can be taken to be the same as in the linear interaction case. It then transpires that for the evaluation of the transmitted pulse energy one only needs to consider the nonlinear phase change due to the sample. Any linear phase effects due to propagation inside the ring can be completely omitted or at best an arbitrary phase function can be plugged into the field expression for the sake of completeness. To calculate Iout we consider a collimated, spatially and temporally Gaussian pulse with electric field amplitude E0 , incident on the beam splitter, where it is split into two counter-propagating beams with electric fields Ecw and Eccw . The general form for the electric field of a Gaussian beam is: w0 r2 Eðz; r; tÞ ¼ E0 exp exp ½i/ðzÞF ðtÞ; wðzÞ w2 ðzÞ ð1Þ where z is the distance of propagation, r is the transverse coordinate, w0 is the beam waist (z ¼ 0), /ðzÞ ¼exp½iðkz tan1 ðz=z0 ÞÞ is the phase of the Gaussian beam (with pulse duration sp and wave vector k and the function F ðtÞ ¼ exp½ð2 ln 2Þt2 =s2p gives the temporal variation. The spot size at a distance z is w2 ðzÞ ¼ w20 ½1 þ ðz2 =z20 Þ, where z0 is the Rayleigh range. For simplicity, we assume that the Rayleigh range is larger than the sample thickness (thin sample approximation). We ignore the slight difference in the spot sizes for the CW and CCW beams at the lenses (which arises because the sample is offset from the center of the ARR). If the intensity splitting ratio in CW and CCW directions are (1=2 d) and (1=2 þ d) respectively and the spot size at the focal point is w0 , then the electric field at the incident face for the two counter propagating beams can be expressed as: pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffi Ecw ¼ 1=2 dE0 expðr2 =w20 ÞF ðtÞ R and pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Eccw ¼ 1=2 þ dE0 expðr2 =w20 ÞF ðtÞ; ð2Þ where R is the reflectivity of the uncoated flat. The field propagation through the sample is governed by dI=dz ¼ aðIÞI and d/=dz ¼ knðIÞ. For the weaker CW beam, aðIÞ ¼ a, the coefficient of linear absorption and nðIÞ ¼ n0 , the linear refractive index. For the CCW beam aðIÞ ¼ a þ bI, and nðIÞ ¼ n0 þ n2 I, where b is the effective nonlinear absorption coefficient and n2 is the nonlinear index of refraction. The electric field of a pulsed Gaussian beam at the exit face of the sample (thickness ¼ L) with a nonlinear absorption and nonlinear refractive index is E0 ðr; tÞ Eexit ðr; tÞ ¼ pffiffiffiffiffiffiffiffiffiffiffi expðaL=2Þ expðikn0 LÞ 1þq expðikn2 lnð1 þ qÞ=bÞ; ð3Þ where E0 ðr; tÞ is the incident electric field, q ¼ bIin Leff , Leff ¼ ½1 expðaLÞ=a is the effective length of the sample and Iin ¼ ð1=2 þ dÞKI0 is the intensity incident on the sample, I0 is the intensity incident on the beam splitter and K is a constant (<1) accounting for the reflection losses at the sample and lens surfaces. In our experimental setup Iin 0:4I0 and jdj 0:02. Using Eqs. (2) and (3), the electric field of the two counter propagating beams at the corresponding exit faces of the sample can be expressed as pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi exit Ecw ¼ ð1=2 dÞE0 2 r exp expðaL=2Þ w20 pffiffiffi ð4Þ expðikn0 LÞF ðtÞ R; 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi E0 r exit Eccw ¼ ð1=2 þ dÞ pffiffiffiffiffiffiffiffiffiffi exp expðaL=2Þ w20 1þq expðikn0 LÞ expðikn2 Iin Leff ÞF ðtÞ; ð5Þ where q 1 and lnð1 þ qÞ q. When the two beams arrive again at the beam splitter after one trip round the ring, the electric fields in the transmission branch are given by multiplying Eqs. (4) and (5) by the respective splitting ratios t Ecw ¼ ð1=2 dÞE0 w20 r2 exp w2 ðzÞ w2 ðzÞ pffiffiffi expðaL=2Þexpðikn0 LÞexpði/ðzÞÞF ðtÞ R; ð6Þ 301 " 2 E0 w0 t Eccw ¼ ð1=2 þ dÞ pffiffiffiffiffiffiffiffiffiffiffi 1 þ q w2 ðzÞ r2 exp expðaL=2Þ w2 ðzÞ expðikn0 LÞ expði/ðzÞÞ pffiffiffi expðikn2 Iin Leff ÞF ðtÞ R; bLdIin pffiffiffi 2 2 " 2 # 2 2 # 2 b kn2 L Iin pffiffiffi : þ þ 4 2 3 3 Iout ¼ ðexpðaLÞRIÞ 4d2 þ ð7Þ where wðzÞ is the spot size at the beam splitter after one round trip. Both wðzÞ and w0 can be determined experimentally. As mentioned earlier, the ARR 2 2 t t leakage is jEout ðr; tÞj ¼ jEcw ðr; tÞ þ Eccw ðr; tÞj . Sub2 stituting Eqs. (6) and (7), jEout j can be expressed as 2 w0 2r2 2 2 jEout j ¼ expð aLÞ exp F ðtÞ w2 ðzÞ w2 ðzÞ " 2 # 2 1 ð1=2 þ dÞ 1 2 þ 2d d þ 2 2 ð1 þ qÞ 2 jE0 j R cosðkn2 Iin Leff Þ ð8Þ for q 1, Eq. (8) becomes 2 w0 2r2 2 2 jEout j ¼ expð aLÞ exp ðtÞ F w2 ðzÞ w2 ðzÞ " " 2 bIin 4d2 þ bIin Ld þ 4 2 # # kn2 Iin 2 þ ð9Þ L2 jE0 j R: 2 As has been stated earlier, the relevant measured quantity in this experiment is the transmitted pulse energy: Z 1Z 1 2 W ¼ Eout 2pr dr dt: ð11Þ For low intensity levels (q 1), the ARR transmission Iout as a function of Iin is a cubic polynomial. The coefficients are directly related to the coefficient of nonlinear absorption (two-photon or saturation or excited state absorption) and to n2 . As mentioned earlier, d is related to the linear leakage and limits the sensitivity. For d ¼ 0, Iout / Iin3 , in which case the simultaneous evaluation of b (or Is ) and n2 becomes difficult. However, a small nonzero d makes the analysis much simpler as each term of the polynomial can be evaluated directly. In our geometry, Eccw Ecw , and if the sample exhibits nonlinear absorption (not saturation), Eccw will be attenuated more than the Ecw . For d > 0 and Eccw Ecw , all the coefficients of the polynomial will be positive and Iout will be a continuously increasing function of Iin . If the sample exhibits saturation of absorption then Eccw will be attenuated less than Ecw and the sign of d opposes the increase in nonlinear leakage. The quadratic term in the polynomial becomes negative, and Iout shows saturation at a relatively low intensity, where the effect of d is cancelled by the saturable absorption. At higher intensities Iout increases continuously. For d < 0, the curvature of ARR leakage as a function of Iin is reversed for the two processes. Thus, by carefully choosing d, both the origin of nonlinearity and the values of nonlinear coefficients can be conveniently obtained. 1 0 Using Eq. (9) it can be shown easily that pffiffiffi 2 p pw ðzÞRsIðexpðaLÞÞ pffiffiffiffiffiffiffi W¼ 2 ln 2 " " 2 # 2 2 # 2 bLdIin b kn2 L Iin 2 pffiffiffi : 4d þ pffiffiffi þ þ 4 2 3 3 2 2 ð10Þ 1=2 Defining Iout ¼ 2ðln 2Þ W =p3=2 sw2 ðzÞ which has the dimensions of intensity, Eq. (10) can be expressed as 3. Results and discussion To verify our formulation and demonstrate the application of ARINS, we measured the real and imaginary contribution to the third order nonlinearity of toluene at 774 nm. A tunable, femtosecond Ti:Sapphire laser was used to generate a 100 MHz pulse train with a pulse duration of 68 fs. A Pockels cell reduced the repetition rate to 3 Hz. A lower repetition rate prevents the contribution of thermal effects to the nonlinearity. The sample cell 302 2 ARR Output Intensity (KW/cm ) 600 500 400 Transmision (Normalized) was a 1 mm path length quartz cuvette. Fig. 2 shows the ARINS leakage as a function of incident intensity, Iin , which was varied by a variable attenuator. Table 1 gives the nonlinear coefficients (b and vð3Þ ) of toluene measured by us, and the values calculated using our data but the old ARINS formulation. For comparison, we also provide the corresponding values reported earlier by Meredith ð3Þ et al. [14], who measured vxxxx ð3x; x; x; xÞ by using third harmonic generation of 1.91 lm fundamental wavelength, a technique that would measure purely the electronic part of the nonlinearity. As all the interacting wavelengths are far away from the resonance in toluene (nonresonant ð3Þ condition), vð3Þ xxxx ð3x; x; x; xÞ ffi vxxxx ðx; x; x; 1.08 1.04 1.00 0.96 Closed aperture Open aperture 0.92 -100 -50 0 50 100 Z (Arb. Units) 300 200 Quadratic fit Linear transmission ( β = 0) Experimental 100 0 0.0 0.5 1.0 1.5 2.0 2 .5 2 Incident Intensity (GW/cm ) Fig. 2. Antiresonant ring leakage signal for toluene. Inset shows the closed and open aperture z-scan data (Imax ¼ 3 GW/ cm2 ) for the same sample. xÞ. The fact that this indeed is the case is established by the excellent agreement of their ð3Þ value detervð3Þ xxxx ð3x; x; x; xÞ value with the v mined by Saikan and Marowky [15] using the FWM (CARS Maker fringe) technique. Clearly, the value of vð3Þ for toluene calculated using our formulation is in very good agreement with these reported values. However, the value calculated from our experimental data but using the treatment of LH [6] is lower by a factor of 25 and is obviously incorrect. We point out that in our experiment, the maximum intensity at the sample (Iin ) was only 0.8 GW/cm2 , whereas a z-scan measurement with 3 Hz repetition rate pulses fails to detect the nonlinear refractive index and two-photon absorption in toluene even at the much higher intensity of 3 GW/ cm2 (Fig. 2, inset). The scatter in the z-scan data is about 4%, which obviously limits the sensitivity of the technique. Note that the ARR leakage exhibits a point of inflexion, which – as explained earlier – is due to the value of d being negative. The values obtained in our experiments are estimated to have an error of about 7%. Various external sources not related to ARINS, such as variation in the sample thickness and reflectivity also contribute to the error, often in ways difficult to quantify. On the other hand, z-scan experiments performed with 100 MHz pulse train yield vð3Þ 2 1012 esu. This would imply that the measured nonlinearity in this experiment is thermal in nature as a 100 MHz pulse train can cause significant sample heating. However, the ARINS experiment with 100 MHz pulse train still yields the same vð3Þ value as obtained with 3 Hz pulses. Table 1 Comparison of the values of the nonlinear coefficients for toluene and CdS from our present data (column 3) and literature (column 5) Sample Parameter ARINS (our analysis) ARINS (analysis of LH [6]) Values reported earlier Toluene viiii (x; x, x; xÞ esu 1014 b cm/GW ð3Þ 10.1 0.378 0.046 0.001 9.81 [12], 9.4 [13] – CdS thin film viiii (x; x, x; xÞ esu 1014 b cm/GW ð3Þ – – 62 0.2 95 ([2], single crystal) 6.4 ([2], single crystal) Values obtained from our data using the older analysis of Lee and Hughes (LH) [6] are given in column 4. 30 20 Transmission (Normalized) 2 This establishes that ARINS can discriminate against slow nonlinear processes and can yield purely the electronic nonlinearity. Having benchmarked the ARINS technique as well as our modified analysis procedure, we now demonstrate the usefulness of ARINS for measuring the relatively weak nonlinear response in thin films. We selected an optically flat, 1.3 lm thick polycrystalline CdS film, rf-sputtered on a 1 mm thick quartz substrate [16]. The CdS film showed a bandgap of 2.5 eV, which is close to its bulk bandgap. The measurement was made at 774 nm using a pulse duration of 80 fs, and a low pulse repetition rate of 3 Hz. The quartz substrate and the sample cell did not show any nonlinearity at the intensity levels used in our experiment. Fig. 3 shows the ARR leakage for the CdS film. The monotonically increasing ARR output is related to the positive d in this case. In the CdS thin film we could not detect any intensity dependent refractive index. In our experiment Dsarr was 2.5 ns, which indicates that the relaxation time in CdS is longer. The response in toluene could be detected since it is purely nonresonant electronic nonlinearity. Fig. 3 (inset) shows that closed and open aperture z-scans – performed at an intensity of 3 GW/cm2 – cannot detect the nonlinearity in CdS. This indicates the superiority of ARINS over z-scan for thin films. Table 1 shows the values of the nonlinear absorption cross-section for CdS measured by us, as well as those reported earlier [2]. Values obtained using LH formulation are much lower than the literature values because of problems in their formulation and because the technique was not benchmarked using a standard material. The value of the effective b obtained by us for the nanocrystalline CdS film (average grain size ¼ 5 nm) is much larger than that reported for the nonlinear absorption coefficient in single crystal CdS. Further experiments (infrared photoconductivity, excited state absorption and dispersion of bÞ aimed to understand the origin of the large observed nonlinear coefficients in CdS thin films suggest the existence of mechanisms different from that in bulk CdS [17]. The enhancement can be ascribed to the presence of mid-band gap defect states in the sputter deposited nanocrystalline films [18]. Such defect states can act as intermediate states for 774 ARR Output Intensity (KW/cm ) 303 1.10 1.05 1.00 0.95 0.90 -60 -40 -20 Open aperture Closed aperture 0 20 40 60 z (Arb. Units) 10 Quadratic fit Linear transmission ( β = 0) Experime ntal 0 0.0 0.5 1.0 1.5 2.0 2 Incident Intensity (GW/cm ) Fig. 3. Antiresonant ring leakage signal for CdS thin film. Inset shows the closed and open aperture z-scan data (Imax ¼ 3 GW/ cm2 ) for the same sample. nm incident photon energy and can increase the two-photon absorption cross-section significantly. Excited state absorption due to the defect states in semiconductors also contributes to the nonlinear coefficient. The presence of such states was confirmed by a measurement of the single-photon photocurrent at IR wavelengths. In fact, nanocrystalline CdS thin films also show a stable photocurrent response at visible wavelengths [16]. Though we have discussed the use of ARINS for ð3Þ measuring the contribution to vjklm ðx; x; x; xÞ, the method can be suitably modified to incorporate different wavelengths and/or polarizations to meað3Þ sure contributions to vjklm ðx4 ; x1 ; x2 ; x3 Þ. The lowest value of b that could be successfully measured is 0.38 cm/GW for a 1 mm sample length of toluene. For a polycrystalline CdS thin film, we obtained b of 62 cm/GW. This implies that the ARINS techniques can be used to measure the value of b in CdS film with a thickness as low as 60 nm. This is therefore an extremely powerful technique for measuring nonlinearity in films having thickness in the nanometer range. Another technique that can measure such low nonlinearities is DFWM, but the difficulty with this is that it in- 304 volves critical alignment of more than one beam. It also does not give both real and imaginary parts of the nonlinearity, which can be obtained simultaneously using ARINS. In summary, we have provided a general formalism for using a simple, single beam interferometric technique to measure very weak nonlinear signals. The ARINS technique is highly sensitive, stable against thermal and mechanical perturbations, and can simultaneously measure real and imaginary susceptibilities. We have also experimentally demonstrated the superiority of the ARINS technique over the conventional z-scan technique in detecting nonlinear optical coefficients in thin films. References [1] M. Sheik-bahae, A.A. Said, T.H. Wei, D.J. Hagan, E.W. Van Stryland, IEEE J. Quant. Electron. 26 (1989) 760. [2] R.L. Sutherland, Handbook of Nonlinear Optics, Marcel Dekker, New York, 1996, Chapters 6, 7, 8, 11. [3] A. Ashkin, G.D. Boyd, J.M. Dziedzic, R.G. Smith, A.A. Ballman, J.J. Levinstein, K. Nassau, Appl. Phys. Lett. 9 (1966) 72. [4] S.M. Saltiel, B. Van Wonterghem, P.M. Rentzepis, Opt. Lett. 14 (1989) 183. [5] T. Hattori, H. Okawa, T. Wada, H. Sasabe, Opt. Lett. 17 (1992) 1560. [6] H.W.H. Lee, R.S. Hughes, Opt. Lett. 19 (1994) 1708. [7] R. Trebino, C.C. Hayden, Opt. Lett. 16 (1991) 493. [8] A.E. Siegman, Opt. Lett. 6 (1981) 334. [9] E.A. Kuzin, B.I. Escamilla, D.E. Garcia-Gomez, Opt. Lett. 26 (2001) 1559. [10] P. Senthilkumaran, B. Culshaw, G. Thursby, J. Opt. Soc. Am. B 17 (1997) 1914. [11] A.N. Starodumov, L.A. Zenteno, D. Monzon, E. De La Rosa, Appl. Phys. Lett. 70 (1997) 19. [12] M.C. Gabriel, N.A. Whitaker Jr., C.W. Dirk, M.G. Kuzyk, M. Thakur, Opt. Lett. 16 (1991) 1334. [13] P.W. Milonni, J.H. Eberly, Lasers, John Wiley and Sons, New York, 1988, p. 494. [14] G.R. Meredith, B. Buchalter, C. Hanzlik, J. Chem. Phys. 78 (1983) 1543. [15] S. Saikan, G. Marowky, Opt. Commun. 26 (1979) 466. [16] P. Vasa, P. Taneja, P. Ayyub, B.P. Singh, R. Banerjee, J. Phys.: Cond. Matt. 14 (2002) 281. [17] P. Vasa, B.P. Singh, P. Ayyub, in preparation. [18] U. Woggon, Optical Properties of Semiconductor Quantum Dots, Springer, Berlin, 1997, p. 176.
© Copyright 2025 Paperzz