2. DIAGNOSTIC METHODS 2.1 Interferometry 2.1.1 Principle Measurements of the refractive index are often made by interferometry. The ordinary wave that is not necessary to consider the effect of external magnetic field and comparable to the isotropic case is used for electron density measurement by interferometry. The refractive index of the O-mode is given by Eq. (1.17) as 1/ 2 2pe (r ) No 1 2 1/ 2 n (r ) 1 e nc , (2.1) where nc 0me 2 / e is the “cutoff” density corresponding to the incident wave. The interferometry measures the phase difference (x ) between the wave propagating in the plasma and propagating in the outside of the plasma, which is given by. y ( x) 2 (k0 k p )dy y1 2 y2 y1 (1 NO )dy , (2.2) where k0 2 / and k p are the wave numbers in the vacuum and plasma, respectively. Assuming ne ≪ nc , (x ) is shown as the following formula. ( x) y2 2 a ne (r )dy n (r )(r 2 x 2 ) 1 / 2 rdr , nc y1 nc x e rx (2.3) When radial profile of the density is axisymmetric, we can obtain the profiles by Abel inversion of Eq. (2.26) as ne (r ) nc a d 2 2 1 / 2 ( x r ) dx , x r 2 r dx (2.4) Now we assume the plasma has parabolic distribution given by the following formula, as it is known empirically, -1- Fig. 2-1 Geometry of interferometry. r 2 ne r ne 01 a (2.5) Phase difference when the incident wave passes through the line x 0 is found by Eqs. (2.3) and (2.5), which is proportional to ne 0 2 a ne 0 . 3 nc (2.6) 2.1.2 Choice of incident wavelength The density gradient along the diameter causes a refractive effect, when the frequency of the incident wave becomes close to the electron plasma frequency. The value of the refraction angleδis maximum when the incident beam propagate at the chord of x / a 0.7 as sin 1 ne 0 / nc ne 0 / nc (2.7) Meanwhile, taking Gaussian beam theory into consideration, the beam expands along the -2- distance y shown in Fig. 2-1. 1/ 2 42 y 2 d d 02 2 2 d 0 (2.8) Let us take the distance to the first collecting optics as L, and assuming d0 2L / 1 / 2 , we obtain d 2d0 . Then, the conditions to allow measurement are led by Eq. (2.7) as L m L ne 0 / nc d 2L / 1/ 2 , (2.9) where nc 1.1 1013 2 ( is in the unit of cm) . On the other hand, the lower wave length limit is determined that parasitic fringe shift / has no effect on measurement accuracy. If it is 1% and below, F is fringe number due to the plasma density. Equation (2.9) leads range of incident wavelength as / 102 F Therefore, we obtain. 4.1 108 a ne 0 1.2 1010 Lne2 0 1 / 3 (2.10) In large devices (TFTR, JET, JT-60, DIII-D, LHD, KSTAR and so on) currently in operation or under construction, it is shown that 100 m (far-infrared region) is the best suited except low density region in periphery. -3- 2.1.3 Phase Detection An example of interferometer system and phase detector is shown in Fig. 1-4, and 1-5, respectively Fig. 2-2. Heterodyne interferometer using upconverter Fig. 2-3. Quadrature-type phase detector. -4- 2.2 Reflectometry 2.2.1 Density profile measurements A reflectometer consists of a probing beam propagating through a plasma and a reference beam. The microwave beam in the plasma undergoes a phase shift with respect to the reference beam. This phase difference as a function of the probing frequency is given by ( ) 2k rc ( ) N (r , )dr a (2.11) 2 within the WKB approximation, where k and are the wave number and frequency of the probing beam, N is the plasma refractive index, and a and rc() are the plasma and cutoff radii, respectively. The refractive indexes of the O-mode and the X-mode propagations are given by 12 2pe No 1 2 o o cko 12 , (2.12) 2pe x2 2pe ck N x x 1 2 2 2 x x x 2pe ce in the cold plasma approximation, where c is the speed of light, pe and ce are the electron plasma frequency and the electron cyclotron frequency, and subscript O and X correspond to the ordinary mode and the extraordinary mode incident waves, respectively. For the O-mode propagation, this integral can be analytically solved for the density profile using an Abel inversion. For the X-mode propagation, a numerical algorithm has been developed to invert the data. In an ultrashort-pulse reflectometer, a very short pulse (1-100 ps) is used as a probe beam. The pulse has a frequency component governed by the shape and duration. Each Fourier component reflects from a different spatial location in the plasma. The time-of-flight for a wave with frequency from the vacuum window position rw to the reflection point at rp is given by 2 ( ) c rp 2pe 1 2 rw 1 2 dr . (2.13) -5- In order to obtain the density profile from the time-of-flight data, the Eq. (2.13) can be Abel inverted to obtain the position of the cutoff layer, r ( pe ) c pe 0 ( ) d ( 2pe 2 )1 2 (2.14) . By separating different frequency components of the reflected wave and obtaining time-of-flight measurement for each component, the density profile can be determined. The movement of the plasma postion due to the fluctuations can be neglected during the reflection of ultrashort pulse (ps). 2.2.2 Fluctuation measurements Reflectometry has also been used in order to study plasma fluctuations. The instataneous phase shift between the local beam and the reflected beam is expressed as 0 , where 0 and are the phase shift depending on the cutoff layer due to density and density fluctuations of the plasma profile. In a simple homodyne reflectometer, the mixer output is given by V El Er cos(0 ) El Er (cos 0 cos sin 0 sin ) El Er (cos 0 sin 0 ) (2.15) assuming 1, where El and Er are the electric field amplitude of the local beam and the reflected beam, respectively. The time varying component of the mixer output depends on both amplitude and phase modulations, since Er can be time dependent due to changes in the cutoff layer. In general, the radial fluctuations of the cutoff layer produce the phase modulations and the poloidal (azimuthal) fluctuations cause amplitude modulations. Therefore it is important to identify both phase and amplitude fluctuations using, such as, heterodyne detection or quadrature type mixer. In a simple one-dimensional model, the phase changes in the O-mode and the X-mode propagations due to the small perturbations of the density and the magnetic field, ne and B at the critical density layer are given by o 2ko Ln (ne / ne ) (2.16) , and -6- x 2k x ne ne (ce x 2pe ) B B (2.17) 1 Ln (ce x 2pe ) LB When the wavelength of the fluctuation is much longer than the spot size of the incident wave, there is no attenuation or modulation of the reflected wave. The depth of the phase modulation approaches the 1D geometric optics limit. 2.2.1 Reflectometer systems Various types of reflectometer are shown in Fig. 2-4. Fast-Sweep FM Reflectometer AM Reflectometer ○high resolution with simple hardware ○minimal effect of density fluctuations ●phase runaway ●parastic reflections from wall and window f Source f+ (t) Short-Pulse Reflectometer Ultrashort-Pulse Reflectometer ○measurement of real-frozen plasma ○an impulse generator ●many sorces or sweep source with ●ultrashort pulse (<10 ps) for high density wideband switches Source f Pin Switch plasmas f±f/2 Pulse Generator Trigger BP Filter Time Delay Digitizer and/or TAC Fig. 2-4. Reflectometer systems -7- d ( ) d dt dt c 2 () 2c 1 pe2 r r 1 2 Lr dr Ls c p ( ) w( ) ant pe p () c rc rant d 1 2 2 2 0 pe -8- 2.3 Thomson Scattering 2.3.1. Collective scattering By using microwave as an incident wave, scattering parameter (k D ) 1 is usually larger than unity. Then the scattering feature corresponds to so called collective scattering. Most laboratory plasmas and magnetically confined plasmas have density fluctuations caused by various types of instability within the plasma. These fluctuations generally have wavelengths exceeding the Debye length and the fluctuation levels encountered far exceed the thermal levels. The scattered power per steradian and per radian frequency at the scattering angle s is written by Ps (k s , s ) pi neVs T S ( k , ) , (2.18) where k=ks-ki, =s-i, pi is the power density of the incident wave, ne is the mean electron density, Vs is the scattering volumn, T is the cross section of the Thomson scattering, and S(k,) is the power spectral density of the density fluctuation given by 1 S (k ) 2 2 n~k S (k , )d neVs ne , (2.19) where n~k is the amplitude of density fluctuations with wave number k. The wave number spectrum can be obtained by changing the scattering angle s. The density fluctuation level is then determined from the integration of k as. 4 1 n~e (r , t ) 2 ne S (k , )dkd 2 (2.20) For the thermal fluctuations, S(k)~1, however, S(k)>>1 for the non-thermal fluctuations. Assuming n~e / ne 102 103 , ne=1019 m-3, and Vs=10-5 m3, S(k)=108-1010. -9- 2.3.2. Microwave scattering - 10 - 2.3.3. Far-Infrared scattering - 11 - 2.3.4. Far-forward scattering - 12 - 2.4 Electron Cyclotron Emission 2.4.1 Determination of electron temperature In 1.3.1 it is shown that the intensity emitted by the plasma is given by Eq. (1.43). In actual experiment, the plasma is produced in a metal chamber. If we consider the effect of the reflection from the metal wall, it is known that the radiation intensity is modified as I n ( ) I B0 1 e n 1 ree n , (2.21) where re is the wall reflectivity (1> re > 0.9). Therefore, when 1 re ≪ n , In becomes nearly equal to the black body radiation, then we call as “plasma is optically thick”. On the other hand, when n ≪1 re , Eq. (2.21) becomes I B0 n 1 re I n ( ) (2.22) and “plasma is optically thin”. Let us consider a tokamak plasma case and the coordinate system as shown in Fig2-5, where B0 is the magnetic field intensity at the plasma center, R is the major radius. It is know that the troidal magnetic field, BT is a function of x as BT B0 R Rx (2.23) Therefore nce also varies accordingly. Elecron cyclotron emission (ECE) appears resonantly in width xn n d nce / dx 1 (2.24) with centering on x x which corresponds to nce . When the plasma is optically thick, the radiation power becomes proportional to its local electron temperature. On the other hand, when plasma is optically thin ≪1 re , I n ( ) I B0 1 n e (Te ) n 1 re 1 re (2.25) The radiation power is proportional to both ne and Te profiles. Therefore, when Te profile is obtained by different methods, we can determine ne profile, and vise visa. - 13 - Fig. 2-5. Electron cyclotron emission from tokamak plasma. Furthermore observing the ECE at the optically thin n and n+1 th harmonics, we can determine the electron temperature using the following formulas, k Te M n I n I ( ) (2.26) I n ( )m0 c 2 2en 2n 1 1 Mn 2n 3 2 n 1 2n n 1 n 1 (2.27) Similarly, observing the ECE at the optically thin O-mode and X-mode waves, we obtain the electron temperature as kTe mec 2 I n(0) ( ) / I n( x ) (2.28) Let us consider the case that non-thermal electrons (relativistic electrons or runaway electrons) and thermal electrons (bulk electron) coexist in a plasma. When we detect the ECE from outside of torus (low-field side), the radiation temperature is given by Tr T1(1 e 1 )e Te (1 e ) (2.29) - 14 - where 1 is the optical thickness for non-thermal electrons. In case we detect it from inside of torus (high-field side), Tr T1(1 e1 ) Te (1 e )e1 (2.30) is given. 2.4.2 ECE radiometry There are several types of diagnostic systems for ECE measurements, such as , i) Heterodyne radiometer, ii) Fourier-transform spectrometer, iii) Grating polychromator, iv) Fabry-Perot interferometer, and v) Multuchannel mesh filter Two representative methods will be described i) Heterodyne radiometer Conventional heterodyne technique is often used for the second harmonic of the ECE. This technique has very good frequency resolution. In the initial stage this had not been used to monitor the entire 2ce spectrum, however, wideband mixers having almost full band responsibility hav been developed, and most of the spectrum can be covered by a few mixers. Figure 2-6 shows an example of frequency down conversion section of an ECE heterodyne radiometer. As an IF section we have prepared 8 channel filter bankand amplifiers fabricated by microwave integrated circuit technology. This radiometer has 96 channels in 110-196 GHz range. . Fig. 2-6. Down conversion section of ECE heterodayne radiometer. - 15 - iv) Fourier-transform spectroscopy When electric filed of incident wave on interferometer is given by E (t ) i Ei cos it , the electric field En entering a detector is written by E En (t ) i cos it cosit i 2 i (2.31) Therefore, if we take mean square of En (t ) in time period T when i can be regarded constant, we obtain the following Ei2 1 T 2 1 2 E ( t ) dt E 4 i 4 cos i , n T 0 i i (2.32) where i i , and X / c . The second term of right-hand side of Eq. (2.32) is proportional to auto-correlation function RE ( ) , i.e. RE ( ) 1 T 1 E (t )E (t )dt Ei2 cos i 0 T 2 i (2.33) According to Wiener-Khinchine theorem, spectral density I ( ) is eventually obtained by Fourier transform of RE ( ) , such as. I ( ) 4 RE ( ) cos d (2.34) 0 In this method, the frequency resolution f is determined from maximum m X m / c , as f 1/ m . - 16 - Table 1. Microwave Sources Type of Oscillator Gunn Frequency 280GHz 150-100GHz 90-60GHz 300GHz IMPATT 70GHz HTO & Multiplier Application to Plasma Output Power Turning Bandwidth 10mW 50-100mW 1GHz(E..T.) 100-200mW 30GHz(M.T.) 2mW 500MHz(E..T.) 5GHz(M.T.) 1W 150-60GHz 5-10mW 375GHz 1W 70-50GHz 400-40GHz 10-100mW 110W, 0.6ms 150-60GHz 200-400kW Narrowband reflectormetry, ECE Multichannel interferometry Narrowband reflectometry 10-30GHz Fast-sweep reflectometry Collective scattering from waves BWO Gyrotron 50-75GHz 75-400GHz Broadband reflectometry Collective scattering from waves Collective scattering from ion thermal fluctuatuions E..T.: Electronic tuning M.T.:Mechanical turning Table 2. Comparison of Microwave Detectors Type Wavelength [mm]Response time Video NEP [W/Hz1/2] Remarks InSb 0.1-3 1-5ms ~10-13 Pyroelectric <1 1 ms 10 ms ~10 -9 ~10 Room temperature Ge:Ga 0.01-0.1 1ms ~10-13 Liquid helium cooled Schottky Barrier Dioder >0.07 <1ns ~10-10 Room temperature Response time depends on IF Liquid helium cooled -7 - 17 -