Abstract
A generalized lock-in detection method is proposed to extract amplitude and phase from optical interferometers when an arbitrary periodic phase or frequency modulation is used. The actual modulation function is used to create the reference signals providing an optimal extraction of the useful information, notably for sinusoidal phase modulation. This simple and efficient approach has been tested and applied to phase sensitive spectroscopy and near-field optical measurements. We analyze the case where the signal amplitude is modulated and we show how to suppress the contribution of unmodulated background field.
© 2014 Optical Society of America
1. Introduction
Precise, fast and cost effective phase measurement is an important issue for many optical instruments and sensors. Among the large number of interferometric modulation/demodulation techniques [1], phase modulations based on optical mirror vibration or piezo-strecher are particularly interesting in the exciting context of phase-sensitive microscopy and near-field imaging [2–4]. The advantages of mirror based modulation compared to acousto-optics frequency modulators, also used in optical microscopy [5–7], include excellent achromaticity which is needed for spectroscopic application, simplicity, low cost while keeping a low phase noise.
The general principle consists in modulating the phase in the reference arm of an interferometer by adding a variable delay length and measuring the resulting interferences. When the phase modulation is linear an harmonic beating with the signal arm occurs, and any phase change in the signal arm produces a phase lead or delay which can be recorded with an electronic phase-meter. Because a phase modulation can be described as a Doppler frequency modulation, such methods are often referred to as pseudo-heterodyne although this term was initially used to describe a detection scheme where the optical source itself is swept in frequency [8, 9] or in phase [10] in unbalanced interferometers. Since it is much easier to obtain a sinusoidal phase modulation by vibrating a mirror or stretching a fiber, than a linear one, pseudo-heterodyne methods have been generalized to such non linear periodic phase modulation. However, in these cases the detected intensity presents multiple harmonics and the phase is not retrieved with a conventional phase meter.
Different approaches have been proposed to tackle this issue, most notably in the context of fiber optic sensors [11,12] and near-field optics [3,4,13] where the signal is particularly weak. For a sinusoidal modulation function, lock-in amplifier (LIA) detections on two consecutive harmonics [3] present in the detected signal have been used successfully. However, since the useful signal is spread over a larger number of frequencies, more efficient or simpler methods are considered [14,15]. In this paper, we propose a generalized LIA approach to recover amplitude and phase with an optimum signal-to-noise ratio for any modulation function. Instead of using multiple LIA at selected harmonic references (M-LIA), our lock-in detector is fed by references having all the frequencies of interest. The basic operation principle of this Generalized Lock-In Amplifier (G-LIA) is first presented and applied to phase sensitive spectroscopy. The approach is then extended to the case where the signal field presents an additional amplitude modulation. This extension is notably important in the context of near-field nanoscopy where a nano-probe provides a weak modulated signal in the signal arm. The practical implementation of the extended G-LIA in near-field nanoscopy is then detailed, and the method advantages are presented. Some calculation details and a summary table for different modulation functions are given in appendices.
2. Interferometry using G-LIA
2.1. Principle
We consider a monochromatic two-arm interferometer with a phase modulated reference field. We denote by ω the angular frequency of the light, is the signal field with unknown amplitude term ES and phase ϕS, and is the reference field with a constant amplitude and a time-modulated phase ϕR(t). In practice ϕR(t) can be produced by either a phase, or an equivalent frequency modulation [8, 9]. With these notations the detected intensity proportional to 〈(ES(t) + ER(t))2〉 is:
In order to determine ES and ϕS, the phase ϕR(t) is typically varied at a speed much higher than the expected variation of these two quantities. The intensity measured on a sufficiently short time can then be written as the sum of a constant or quasi-constant term and a modulated one : with I0 ≃ const. during a measurement, and From Eqs. (3), we see that the interesting term Imod can be decomposed on the basis (cos(ϕR), sin (ϕR)). An ideal way to determine amplitude and phase is then to achieve a linear phase variation ϕR = Ωt and perform a synchronous detection by multiplying the detected signal by the two orthogonal harmonic references sin(Ωt) and cos(Ωt) and integrating over a sufficient time period. Such operation is typically performed by a dual output LIA locked at the angular frequency Ω. The quantities (ES, ϕS) are directly inferred from the two signal outputs (X, Y): where tint is the integration time. Unfortunately, achieving a purely linear phase variation using phase modulators is not possible as they necessarily present a finite range of phase modulation. A well-known approach to overcome this difficulty is to use serrodyne modulation, or in other words, to use a sawtooth modulation of the optical path, e.g. in the reference arm. A quasi-harmonic modulation of Idet is then obtained if the full amplitude of the sawtooth corresponds to an integer number N times 2π (i.e. an optical path modulation of Nλ). Although this approach works, some errors are induced due to the non negligible “flyback” time at the end of the sawtooth modulation. Therefore, serrodyne modulation is not highly recommended for fast or precise experiments, and in many cases a sinusoidal function ϕR = a sin Ωt is preferred. In the latter case, the Fourier spectrum of Idet presents harmonic sidebands at radial frequencies mΩ. The amplitudes of these sidebands are obtained using Jacobi-Anger expansion of cos(ϕR) (where m is only even) and sin(ϕR) (odd m only) [16]. Again, a synchronous detection at the harmonics mΩ provides: where Jm is the m-th Bessel function. As the signal is spread over several harmonics, we need to perform a lock-in detection on at least 2 harmonics (odd and even) to recover amplitude and phase information. We can note also that some care should be taken regarding the best value of a, as a given amplitude of modulation can lead to a large signal value of Jm for m odd, but a small value for m even, and vice versa.Here, we propose another approach where the detected intensity is not multiplied by consecutive single harmonic references such as in the M-LIA method [3]. Instead the signal is multiplied by two orthogonal references having the same set of harmonic frequencies than Imod, namely cos(ϕR) and sin (ϕR). As for the serrodyne modulation, this operation can be performed with an electronic multiplier and integrator:
As long as the two references do not contain a DC term the contribution from Io is filtered, and according to Eqs. (3) XϕR and YϕR are also proportional to ESER cos (ϕS) and ESER sin (ϕS) respectively. To determine amplitude and phase from these two outputs, the two proportionality constants (kX, kY) must be analytically or numerically evaluated for the considered modulation function ϕR.In order to work with any modulation function, the DC terms in Idet can be simply filtered using a high pass filter. Amplitude and phase information are then obtained via the magnitude and angle of (XϕR (Ĩdet)/k̃X + i * YϕR (Ĩdet)/k̃Y) where Ĩdet denotes the filtered detected intensity and (k̃X, k̃Y) the associated proportionality constants. In this way, all the available modulated power is used to recover the signal field information. Moreover, many phase modulation functions can be considered: sinusoidal, symmetric or asymmetric triangular functions, etc., regardless of the phase modulation amplitude a. For such simple modulation functions the constants have simple analytical expressions and a proper choice of a actually leads to kX = kY = 1 as for the perfect serrodyne case. The expressions of these constants are given in appendices A and B for sine and triangular modulation functions.
More generally, the actual phase modulation ϕR = ϕreal can deviate from the required drive excitation. This real phase modulation induced by the modulator can usually be measured during the experiment, for example, using piezoelectric modulator equipped with strain gauges or other displacement sensors. Empirically, the same approach can also be used in this case: the integrals (8–9) are numerically evaluated with the measured ϕR in order to determine the proportionality constants (k̃X, k̃y).
2.2. Test and application to phase resolved spectroscopy
The Michelson interferometer shown in Fig. 1(a) was built to assess the validity of the technique for precise phase measurements in presence of arbitrary phase modulation. To create the phase modulation, the reference mirror, mounted on a piezoelectric crystal was excited sinusoidally at an angular frequency of 200π rad/s and an arbitrary amplitude a = 2.4 rad. For demonstration purposes, the second mirror mounted on a piezoelectric crystal was used in the signal arm to obtain controllable changes of the phase ϕS, monitored via a capacitive displacement sensor. The detected signal Idet containing many harmonics was filtered from its DC component by a high pass filter and the resulting signal Ĩdet was sampled with a 16-bit acquisition card. The G-LIA operation expressed by (8–9) was then performed numerically using LabVIEW [17], and the phase and (constant) amplitude were retrieved using the expression of (k̃X, k̃Y) given in the appendix A.
Figure 1(b) shows the phase measurements obtained for a symmetric triangular displacement of the signal mirror. A remarkable agreement is observed between the G-LIA based interferometric measurement and the the actual phase change obtained through the displacement sensor. With this approach similar results were obtained with sine phase modulation frequencies of few kHz, only limited by the bandwidth of the loaded piezoelectric actuator.
Figure 1(c) shows a similar setup where a single-mode laser diode operating near λ = 650.0 nm was used to perform phase-resolved spectroscopy. A gas cell of iodine was added in the signal arm and the wavelength was tuned over few picometers across an absorption line of the gas with ramps of current. As for the previous experiment an arbitrary sine phase modulation was created in the reference arm (2000π rad/s, a = 1.2 rad) and amplitude and phase were retrieved through the G-LIA operation. The corresponding results are shown in Fig. 1(d). The observed magnitude and phase spectra actually reflect the expected variation of the real and imaginary parts of the gas refractive index across a Lorentzian absorption peak slightly convoluted with the source linewidth.
3. Extension of the G-LIA to amplitude modulated signals
We consider now the situation where both the signal field amplitude and the phase ϕR of the reference field are modulated. An amplitude modulation is particularly interesting to recover a weak modulated signal field, and to get rid of unwanted (and not modulated) coherent light impinging on the detector in addition to ES and ER. This approach is notably used successfully in near-field nanoscopy, where heterodyne and pseudo-heterodyne interferometric detection schemes are used to determine the amplitude and phase of the field scattered by sharp oscillating nano-probes. In such experiments the angular frequency ΩA of the amplitude modulation is typically one or few order of magnitudes higher than the characteristic angular frequency Ω of ϕR.
To simplify the mathematical description, we note ΩA=nΩ, where n is typically much greater than one. The part of the signal field modulated at ΩA is then noted to be . With this expression, the useful modulated term Imod of relation (2) is now
that can be expressed as to emphasize the presence of sidebands at frequencies above and below nΩ, and to facilitate the analytical integral calculations. On the other hand, the term in Eqs. (2) carries no information on the phase, and exhibits a modulation at the frequency 2nΩ only. Consequently, we can safely ignore I0 as the reference signals that we will now consider don’t have such frequencies in their spectrum. As before, we multiply the detected intensity by the two orthogonal reference signals C(t) = 2cos(nΩt)cosϕR and S(t) = 2cos(nΩt)sinϕR having the same set of frequencies as Idet, and we integrate to obtain: To determine an expression for the proportionality constants (kX and kY), C(t) and S(t) can also be expanded in cos(nΩt + ϕR) + cos (Ωt − ϕR) and sin (Ωt + ϕR) − sin(Ωt − ϕR), respectively. For the sinusoidal phase variation case, the two integrals then have simple analytical solutions obtained from the integral representation of n-th Bessel functions: Amplitude and phase are then retrieved directly from these two outputs for any phase modulation amplitude a. We note that J2n(2a) is actually negligible for large values of n and reasonable phase modulation amplitude a (for example for n = 10 and a < π, or n = 20 and a < 6π). It is therefore convenient in this case to choose a modest phase modulation a such that J0(2a)= 0, so that the outputs are again identical to the perfect serrodyne case given by Eqs. (4–5). In appendix C, we show that in presence of an additional (unwanted) coherent background field, the value of a can be adjusted to cancel its contribution. This is typically obtained for J0(a) = 0, e.g. for a ≃ 2.404 rad.4. Test and application to infrared nanoscopy
In the second configuration depicted in Fig. 2(a), the signal field is modulated in amplitude which is particularly useful to extract a weak signal from noise. To test and assess the quality of the phase measurement in this second mode, an amplitude modulation was achieved by placing in the signal arm a tuning fork vibrating at ΩTF. In this way the signal intensity < > is sinusoidally modulated and therefore the signal field also presents modulation at the frequency ΩTF (and harmonics). The signal mirror was then moved back and forth in a controlled triangular motion (±0.80μm corresponding to ±1.0 rad). To recover the corresponding signal phase, the sine phase modulation ϕR = acos(Ωt) and the tuning fork excitation signal proportional to cos(ΩAt) were used to build the references C(t) and S(t) with ΩA = ΩTF. The corresponding phase measurement obtained with such extended G-LIA method is shown as an insert in Fig. 2(b), proving the validity of the approach.
The same setup (including the source) was then used in a near-field optics experiment where the bare tuning fork and signal mirror were replaced by a scanning probe microscope. The system is sketched in Fig. 2(c). In this case, the light periodically scattered by a nano-probe produces the modulated signal field interfering with the phase-modulated reference field. As for the previous experiment, shown in Fig. 2(a), an acquisition card featuring high resolution (22 bits) was used for the G-LIA operation to offer the performance level of a high quality digital LIA. A tunable infrared quantum cascade laser [18] emitting near 10.1μm with a rather low but highly stable power of few mW was used. The laser polarization was set along the probe axis with an incident angle of about 60°. More details on the optical system itself can be found in [19].
An example of highly resolved phase sensitive nanoscopy measurement is shown in Fig. 2(d). The images were obtained on a highly sub-wavelength surface grating made of slightly oxidized copper lines (native oxidation) embedded in silicon as depicted in the insert of Fig. 2(c). Two copper lines imaged in the mid-infrared are clearly visible, both in the amplitude (dark lines) and in the phase (bright lines). While a higher signal is typically obtained above metal structures in infrared near-field nanoscopy due to an increased effective polarizability of the probe [20], very sharp probes are actually mostly sensitive to the first nm of the sample material (CuO2). Additional experiments were carried out on the same sample with a similar setup using a more powerful CO2 laser and different probe sizes [21]. A stronger amplitude signal was indeed obtained for bigger probes as shown in Fig. 2(e). We point out also that given the size of the probe, phonon confinement effects could also play an important role on the permittivity of the sharpest probe and impact the observed signal [21, 22]. The description of possible specific probe-sample interaction leading to the observed contrasts should be the subject of a separate paper.
The extended G-LIA method was also tested on a second subwavelength grating made of gold and polymer (S1813 photopolymer). This structure is referred as sample2 in Fig. 2(c), and the signal profiles are given in Fig. 2(f). The G-LIA amplitude signal exhibits an expected higher signal on the gold material [23] compared the the polymer signal and a sharp drop as the probe moves away from the metal surface.
Several precautions were taken to minimize any background contribution. Experimentally, one of the remarkable advantages of the mounted setup is the implementation of an aberration-corrected and diffraction limited Schwarzschild objective in the signal arm. This afforded the ability to highly focus the beam to ≈ λ2 (as measured by an IR-camera) on the tip apex. The probe was electrochemically etched from a straight tungsten wire [24], and was attached to a tuning fork operating in tapping mode. The tip end diameter was smaller than 30 nm with an angle of about 10°. The small size of the probe body and the absence of cantilever was found to be useful to strongly reduce eventual background light modulation. Another crucial issue, was the usage of an ultra-small oscillation amplitude for the probe (≤ 5 nm, as measured by interferometry). This unconstrained the necessity to use detection at the probe’s higher harmonics frequencies [25, 26] often required to suppress modulated background light, the two approaches being actually nearly equivalent [27]. The setup was built to operate in the mid-IR range where the scattering cross-section is essentially low. The G-LIA method, however, can be easily adapted if required to detect the signal at the kth harmonic of the probe oscillation frequency by setting ΩA = k × Ωtip, i.e. simply by multiplying the factor n by k.
In these experiments, the amplitude modulation frequency was imposed by the experimental resonance frequency of the tuning fork (≃ 30kHz) on which the scattering nano-probe is glued to operate in tapping mode (cf. Fig. 2(b)). The frequency of the vibrating reference mirror was arbitrarily chosen equal to 1 kHz leading to a frequency ratio of n = ΩA/Ω ≃ 30. Although the frequency ratio could be set to an integer value with high precision, we point out that the benefit is actually negligible for such high values of n, as exemplified in Fig. 3(a).
The two references signals C(t) and S(t) were built from the normalized modulation signals ei(t):
with ēi (t) = ei (t) /max (ei (t)) and a the theoretical phase modulation taken equal to 2.404 rad to remove the background contribution cf. Fig. 3(b). The actual amplitude of eΩ(t) was easily adjusted to obtain this phase modulation with a precision of about 0.05 rad. The experimental phase modulation amplitude could have been monitored with a position sensor but was here determined from the form of the detected intensity signal. Fig. 3(c) shows a simulated comparison of the signal-to-noise ratio (SNR) between G-LIA and M-LIA as a function of the modulation amplitude a. As can be seen for the used amplitude modulation of 2.4 rad we are gaining more than twice in terms of SNR.Finally, it is important to note that the phase of the mechanical probe oscillation can be strongly phase shifted with respect to the excitation signal, notably for cantilever probes. While such shift does not affect the measured signal phase it attenuates the retrieved amplitude. Therefore it is better to use the measured mechanical modulation signal rather the excitation one. Alternately, it is clear that the G-LIA can be applied twice using cos(ΩAt) and sin(ΩAt) as modulation functions to recover the full amplitude signal whatever the mechanical phase shift. This approach also provides the mechanical phase shift and is described in appendix D. A general summary table is given in appendix E.
5. Conclusion
In conclusion we have described a modified lock-in detection method to recover amplitude and phase in optical interferometers, where the reference beam can be non linearly modulated in phase or frequency. This method is optimum in the sense that it fully exploits the detailed spectrum of the detected beating between signal and reference, thus providing the highest signal-to-noise ratio possible. The principle was detailed for sinusoidal and triangular phase modulations and was experimentally tested, showing high stability and reliability. The method was extended to amplitude-modulated signals in the common case of a sinusoidal phase modulation. This second approach, useful in the the case of very low signal experiments, was successfully tested in mid-IR near-field nanoscopy, leading to a high signal to noise ratio under low illumination power conditions. In addition, the conditions to obtain efficient background light suppression in this second approach were derived. Although the tests were made using vibrating mirrors, much faster phase modulation can be achieved, for example using a slight wavelength modulation of the laser source in an unbalanced interferometer.
Appendices
A. Sine wave phase modulation
We start from Eqs. (8–9) where the references cos(ϕR) and sin (ϕR) contain all the frequencies of interest contained in the intensity. In the case of a sinusoidal modulation ϕR = a sin(Ωt), the two integrals have well known analytical solutions based on the integral representation of Bessel functions [16]. Neglecting for now the contribution of the constant term I0, i.e. considering that it can be subtracted, we have:
where J0 is the 0 – th Bessel function. Although the phase can be retrieved from these two Eqs., it is convenient to choose a phase modulation a such that J0(2a)= 0 (e.g. for a ≃1.20 rad, 2.76 rad, etc.). The outputs are then identical to the perfect serrodyne case cf. relations (4–5), i.e. kX = kY = 1.While it is possible to obtain Imod by subtracting I0, such operation complicates the experimental setup, unless the signal is small enough to consider . The constant term I0 is easily filtered by a simple DC filter, but such filter can affect Imod since the reference cos (ϕR) may exhibit a non negligible DC component as can be seen in Fig. 4(c).
This is also the case for a sinusoidal modulation if we consider an arbitrary amplitude a (cf. Eqs. (6) with m = 0). For this reason we evaluate the signal outputs for a filtered detected signal Ĩdet:
from where (Es, ϕs) can again be retrieved directly. We note that for 2a corresponding to a k-th zero of J0, the term is actually small for k even, and especially for large value of k, so that we can have k̃X ≃ k̃Y ≃ 1. The corresponding expressions for a triangular modulation are given in appendix B hereafter. Examples of orthonormal references are given in Fig. 4 for serrodyne 4(a), sinusoidal 4(b) and triangular 4(c) modulation in direct and frequency space.B. Triangle wave phase modulation
For a triangle wave modulation ϕR = aTr(Ωt), the modified lock-in operation performed on the modulated intensity term Imod = 2ERES cos (ϕS − ϕR) is
where Tr(Ωt) is the triangle wave ranging from −1 to 1 in phase with sin(Ωt). It is then convenient to choose a phase modulation a equals to a positive integer number times π/2 in order to obtain the same output as the ideal serrodyne given by (4–5), and simple reference signals cf. Fig. 4(c). Since generally Imod can not be obtained by a DC filtering of the detected intensity I, it is more convenient to consider a modified lock-in operation on the filtered detected intensity Ĩdet : The additional term is small for large values of the phase modulation amplitude a, or values close to an integer number times π, so that we can also have k̃X ≃ k̃Y ≃ 1.C. Background elimination for amplitude-modulated fields
In addition to ER(t) and EP(t) we consider now that we may have a third additional (unwanted) coherent background field , comparable or stronger than the modulated signal field . This constant background field can interfere with EP(t) and gives an additional intensity term modulated at radial frequencies nΩ:
Since C(t) and S(t) also have frequency components at nΩ, we have to evaluate the contribution of this term in the extended G-LIA operation: The contribution of these two integrals are found to be proportional to J2n(a)+ J0(a). From this result we see that the contribution of the unwanted terms modulated at nΩ can be canceled out by setting a modulation amplitude a corresponding to a zero of the 0th Bessel function, J2n(a) being extremely small for a sufficiently large n.It is worth noting that the additional unwanted interference term between EBg(t) and ER(t), and the self interference term proportional to are not modulated at nΩ, and are therefore efficiently filtered without special precaution.
D. Double generalized lock-in detection
We consider an amplitude modulated signal field Es(t) at an angular frequency nΩ, where Ω is still the phase modulation frequency. In general, the amplitude modulation of this field can be phase shifted with respect to the excitation signal. We denote by the full signal field. A quite general expression for the modulated amplitude is + harmonics. The phase ψ accounts for the eventual phase shift existing between the fundamental term and the driving signal in phase with cos(nΩt). To obtain an optimal determination of the amplitude and phase (ES, ϕS) without knowing ψ, the G-LIA can be applied twice as part of the fundamental term is modulated proportional to cos(nΩt) and the other part proportional to sin(nΩt):
Evaluation of the terms gives: where kX, kY are the already known proportionality constants corresponding to a sine phase modulation in Eqs (15) and (16). The amplitude, the phase ϕs but also the unknown phase shift ψ are then calculated from these four outputs.E. Summary table
The values of all the proportionality constants (kX, kY) and (k̃X, k̃Y) corresponding to the different modulation functions mentioned in this paper are summarized in Fig. 5.
Acknowledgments
Authors gratefully acknowledge the DRRT and the Champagne-Ardenne (G-mini) for financial funding, the Nano’mat platform for the infrared laser source and the National Research Agency (ANR) for financing complementary near-field and interferometric equipments ( ANR-09-BLAN-0168-01 and ANR-09-BLAN-0124-01).
References and links
1. Y. Shizhuo, F. T. Yu, and P. Ruffin, Fiber Optic Sensors (CRC Press, 1998).
2. P. S. Carney, B. Deutsch, A. A. Govyadinov, and R. Hillenbrand, “Phase in nanooptics,” ACS Nano , 6(1), 8–12 (2012). [CrossRef] [PubMed]
3. N. Ocelic, A. Huber, and R. Hillenbrand, “Pseudoheterodyne detection for background-free near-field spectroscopy,” Appl. Phys. Lett. 89(10), 101124 (2006). [CrossRef]
4. M. Vaez-Iravani and R. Toledo-Crow, “Phase contrast and amplitude pseudoheterodyne interference near field scanning optical microscopy,” Appl. Phys. Lett. 62(10), 1044–1046 (1993). [CrossRef]
5. R. Hillenbrand and F. Keilmann, “Complex optical constants on a subwavelength scale,” Phys. Rev. Lett. 85(14), 3029–3032 (2000). [CrossRef] [PubMed]
6. R. Debdulal, S. Leong, and M. Welland, “Dielectric contrast imaging using apertureless scanning near-field optical microscopy in the reflection mode,” J. Korean Phys. Soc. 47, 140–146 (2005).
7. Y. Sasaki and H. Sasaki, “Heterodyne detection for the extraction of the probe-scattering signal in scattering-type scanning near-field optical microscope,” Jpn. J. Appl. Phys. 39(4A), L321 (2000). [CrossRef]
8. D. Jackson, A. Kersey, M. Corke, and J. Jones, “Pseudoheterodyne detection scheme for optical interferometers,” Electron. Lett. 18(2), 1081–1083 (1982). [CrossRef]
9. G. Conforti, M. Brenci, A. Mencaglia, and A. G. Mignani, “Fiber optic vibration sensor for remote monitoring in high power electric machines,” Appl. Opt. 28(23), 5158–5161 (1989). [CrossRef] [PubMed]
10. A. Kersey and D. Jackson, “Pseudo-heterodyne detection scheme for the fibre gyroscope,” Electron. Lett. vol. 20(2), 368–370 (1984). [CrossRef]
11. A. Dandridge, A. Tveten, and T. Giallorenzi, “Homodyne demodulation scheme for fiber optic sensors using phase generated carrier,” IEEE Trans. Microwave Theory Tech. 30(10), 1635–1641 (1982). [CrossRef]
12. B. Lee and Y. Jeong, Fiber Optics Sensors (CRC Press, 2008), Chap. 7.
13. S. Pilevar, W. Atia, and C. Davis, “Reflection near-field scanning optical microscopy: an interferometric approach,” Ultramicroscopy. 61(1), 233–236 (1995). [CrossRef]
14. L. Stern, B. Desiatov, I. Goykhman, G. M. Lerman, and U. Levy, “Near field phase mapping exploiting intrinsic oscillations of aperture NSOM probe,” Opt. Express 19(13), 12014–12020 (2011). [CrossRef] [PubMed]
15. B. Deutsch, R. Hillenbrand, and L. Novotny, “Near-field amplitude and phase recovery using phase-shifting interferometry,” Opt. Express 16(2), 494–501 (2008). [CrossRef] [PubMed]
16. H. J. Weber, L. Ruby, and G. B. Arfken, Mathematical methods for physicists (Harcourt/Academic, 2000).
17. C. Elliott, V. Vijayakumar, W. Zink, and R. Hansen, “National instruments LabVIEW: A programming environment for laboratory automation and measurement,” Journal of the Association for Laboratory Automation 12(1), 17–24 (2007). [CrossRef]
18. Y. Yao, A. Hoffman, and C. Gmachl, “Mid-infrared quantum cascade lasers,” Nat. Photonics 6, 432–439 (2012). [CrossRef]
19. Z. Sedaghat, A. Bruyant, M. Kazan, J. Vaillant, S. Blaize, N. Rochat, N. Chevalier, E. Garcia-Caurel, P. Morin, and P. Royer, “Development of a polarization resolved mid-IR near-field microscope,” Proc. SPIE 7946, 79461N (2011). [CrossRef]
20. S. Amarie and F. Keilmann, “Broadband-infrared assessment of phonon resonance in scattering-type near-field microscopy,” Phys. Rev. B 83(4), 045404 (2011). [CrossRef]
21. Z. Sedaghat, “A near-field study of the probe-sample interaction in near and mid-infrared nanoscopy,” Ph.D. thesis, University of Technology of Troyes (2012).
22. M. Kazan, A. Bruyant, Z. Sedaghat, L. Arnaud, S. Blaize, and P. Royer, “Temperature and directional dependences of the infrared dielectric function of free standing silicon nanowire,” Phys. Status Solidi C 8(3),1006–1011 (2011). [CrossRef]
23. B. Knoll and F. Keilmann, “Enhanced dielectric contrast in scattering-type scanning near-field optical microscopy,” Opt. Commun. 182(4), 321–328 (2000). [CrossRef]
24. M. Kulawik, M. Nowicki, G. Thielsch, L. Cramer, H.-P. Rust, H.-J. Freund, T. P. Pearl, and P. S. Weiss, “A double lamellae dropoff etching procedure for tungsten tips attached to tuning fork atomic force microscopy/scanning tunneling microscopy sensors,” Rev. Sci. Instrum. 74(2), 1027–1030 (2003). [CrossRef]
25. A. Apuzzo, M. Fvrier, R. Salas-Montiel, A. Bruyant, A. Chelnokov, G. Lerondel, B. Dagens, and S. Blaize, “Observation of near-field dipolar interactions involved in a metal nanoparticle chain waveguide,” Nano Lett. 13(3), 1000–1006 (2013). [CrossRef] [PubMed]
26. R. Salas-Montiel, A. Apuzzo, C. Delacour, Z. Sedaghat, A. Bruyant, P. Grosse, A. Chelnokov, G. Lerondel, and S. Blaize, “Quantitative analysis and near-field observation of strong coupling between plasmonic nanogap and silicon waveguides,” Appl. Phys. Lett.100(23), (2012). [CrossRef]
27. A. Bruyant, “Etude de structures photoniques en champ proche par microscopie optique à sonde diffusante,” Ph.D. thesis, University of Technology of Troyes, (2004).