Development of the upgraded single crystal dispersion interferometer (SCDI-U) and its first measurements of the line integrated electron densities in KSTAR during shattered pellet injections

Dispersion interferometers (DI) are widely used to measure line integrated electron densities in many fusion devices. A recent development of a heterodyne single crystal DI (SCDI) with a laser wavelength of 1064   nm (Lee et al 2021 Rev. Sci. Instrum. 92 033536) allows an easier and simpler optical setup by using only one, instead of two, nonlinear crystal. It is found that the reported heterodyne SCDI with an acoustic-optical modulator (AOM) has different beam paths between the frequency-shifted, via the AOM, fundamental and second harmonics which act as the reference beams. Such a separation of the reference beams inevitably produces non-removable phase shifts associated with mechanical vibrations, resulting in a reduction of the removing efficiency of the mechanical vibrations that DI systems can provide. By utilizing the fact that the diffraction angle due to the AOM is inversely proportional to the frequency of the laser beam and linearly proportional to an order of the frequency-shift, the SCDI-Upgrade (SCDI-U), which has complete overlap of the optical paths for both probing and reference beams from the laser source to the detectors, is proposed in this work. Its first measurements in KSTAR during shattered pellet injections are reported, and results obtained by the SCDI-U are compared with those from the existing two-color interferometer (TCI) in KSTAR. It is found that the SCDI-U measures the electron density more reliably during such an abrupt and large density change than the TCI does. Qualitative analyses on the effects of different injection schemes of the shattered pellets and possible application of the SCDI-U for ITER are also discussed.


Introduction
It is expected that few systems can survive and be utilized as diagnostic systems in reactor-grade fusion devices due to high particle, heat and neutron fluxes [1,2], in which optics-based diagnostics systems can be accessible and applicable.Being of such a kind, dispersion interferometers (DIs) [3,4] can play a major role in high performance fusion devices so that fusiongrade hot plasmas can be operated with a certain degree of controllability.
A DI system in a magnetic confinement device, similar to a conventional interferometer, measures phase shifts due to plasmas from which the line integrated electron density can be obtained.As the phase shifts in an interferometer can be caused by mechanical vibrations in addition to the plasmas, the mechanical vibration is a source of noises.Unlike a conventional two-color interferometer (TCI) where the phase shifts due to mechanical vibrations are suppressed through data analyses on the measured phases [5][6][7], the DI suppresses mechanical vibrations optically [3].Nonlinear crystals that generate the second harmonic beam within the optical setup are key components of a DI system which make it capable of optically suppressing the noise.This inherent suppression of the mechanical vibrations is advantageous, and many fusion devices such as DIII-D [8], LHD [9], TEXTOR [10], W7-X [11] and KSTAR [12] operate DI systems.
A homodyne DI system was first developed [3] without any modulation schemes that caused uncertainty in the detected phase shifts due to fluctuations of the laser intensity and ambiguity associated with the direction of the phase shifts, i.e. increase or decrease of the line integrated electron density.Later, a phase modulated DI system was developed with a photo-elastic modulator (PEM) [9][10][11]13], resolving the uncertainty and ambiguity that the homodyne DI system has.It was, then, reported [8] that due to the limited bandwidth of the system with a PEM up to a few hundreds kHz, a PEMbased DI system is insufficient to measure fast density fluctuations.To increase the bandwidth of the system up to several MHz a heterodyne DI system with an acousto-optic modulator (AOM) has been developed [8,12], and the density fluctuations induced by magnetohydrodynamics instabilities [8] and by shattered pellet injections [12] were successfully measured.These different types of interferometer systems used in fusion research are summarized in table 1.
The first kinds of DI systems use two nonlinear crystals, where one is located right after the laser source and the other after the plasmas.Aligning and focusing both the fundamental and second harmonic beams to the second nonlinear crystal located after the plasmas requires a high level of precision because the beams are typically traveling more than a few tens of meters before being focused onto the nonlinear crystal whose aperture is less than a couple of mm [4].To alleviate the required precision, a single crystal DI (SCDI) has been proposed [12] as well as its forward model for Bayesian analysis of the measured data [14].The SCDI replaces the second nonlinear crystal with a frequency doubler, which is a simple electronic component, thereby removing the task of aligning and focusing the beams after the plasmas.
In the SCDI system, being a heterodyne DI using an AOM, the fundamental and second harmonic probing beams that go through the plasmas are overlapped onto each other, while those of the reference beams, frequency-shifted by the AOM, are not overlapped [12].This is because the diffraction angles by the AOM are different for the fundamental and second harmonic beams, i.e. the diffraction angle of the second harmonic beam is half of the diffracted angle of the fundamental beam given the same order of diffraction.One may argue that such non-overlapped reference beams do not deteriorate overall performance of the SCDI since travel distance of the reference beams can be short.Nevertheless, such an imperfection can cause a non-negligible noise level due to mechanical vibrations as the reported SCDI uses a 1064 nm laser source, which is a shorter wavelength compared to other DI systems, e.g.10.6 µm.
We have specifically chosen such a shorter wavelength laser such that an abrupt and large density change induced by a shattered pellet injection [15] can be followed by the SCDI system with less complication associated with the fringe jumps [12].Note that when a shattered pellet is injected into a tokamak, it creates a sudden increase in the electron density which can, in turn, mitigate the plasma disruption [16][17][18][19].Shattered pellet injection is considered as a key technique to mitigate plasma disruption and selected as the basis of the disruption mitigation system on ITER [20].Therefore, it is important to be able to measure changes in the electron density due to a shattered pellet injection, and the DI reported in this work is designed for such a measurement.
In this work, we report a newly developed upgraded SCDI (SCDI-U) where both the fundamental and second harmonic reference beams are completely overlapped along the whole beam path.This is achieved by using the first order diffraction for the fundamental beam and the second order diffraction for the second harmonic beam such that the diffracted angles of the frequency-shifted reference beams are the same.Using the different orders of diffraction is also done by the DLR (German Aerospace Center) for sensing relative atmospheric pressures [21].The SCDI-U has complete overlap of the fundamental and second harmonic beams for both probing and reference beams from the laser source to the detectors, which, we believe, is one of the finest optical configurations of the DI systems so far developed for magnetic confinement devices.We first describe the scheme of the developed SCDI-U in section 2, and the actual optical setup of the KSTAR SCDI-U is presented in section 3. First measurements of the SCDI-U during the shattered pellet injections and its comparisons with the existing TCI system are discussed as well as its relevance to ITER in section 4. Here, we also discuss how the SCDI-U can provide critical information to develop disruption mitigation scenarios using shattered pellet injections [15] in ITER.The summary is given in section 5.

Scheme of the SCDI-U
The SCDI-U can be configured with either two detectors or a single detector as shown in figures 1(a) and (b), respectively.The only difference between the two is where the interfered signals are filtered, i.e. before or after the detector(s).

Scheme of the SCDI-U with two detectors
As shown in figures 1(a) and (b), a laser beam with the fundamental frequency of ω (red line) enters into the nonlinear crystal located right in front of the laser source, and the second harmonic beam with a frequency of 2ω (green line) is generated.These two beams are overlapped to each other which is depicted as the purple line in the figure.
The overlapped beams go through the AOM where the frequency-shifted beams are additionally produced.The frequency-shifted beams are diffracted, and the diffraction angle of the mth order, denoted as θ m , is [21] where m takes an integer, λ is the wavelength of an incident beam and N the refractive index of the AOM crystal.Λ is the acoustic wavelength within the AOM, which is inversely proportional to the frequency shift ∆ω.It is clear that using the first order (∆ω) and second order (2∆ω) diffracted beams for the fundamental (ω) and second harmonic (2ω) beams, respectively, the frequency-shifted beams, i.e. ω + (∆ω) and 2ω + (2∆ω), can be overlapped to each other in the SCDI-U, and second harmonic beams.The blue dashed line is the first order diffraction of the second harmonic beam by the AOM which was used in the SCDI [12] scheme.The SCDI-U uses the second order diffraction, resulting in a complete overlap of the frequency-shifted reference beams, i.e. the purple dashed lines.
shown in as the purple dashed line in figure 1. ∆ω typically ranges from a few tens to hundreds of MHz.These frequencyshifted beams act as the reference beams.Note that the SCDI used the first order diffraction of the second harmonic beam [12], i.e. 2ω + (∆ω) (blue dashed line in figure 1), causing a separation of the two reference beams.
The zeroth order components, that is m = 0, are neither frequency-shifted nor diffracted; thus, the fundamental and second harmonic beams are overlapped to each other as well.We use these beams as the probing beams that go through the plasmas (purple solid line in figure 1).The probing and the reference beams are superimposed at the beam splitter and passed onto two separate detectors, where one is receiving the fundamental components, i.e. ω and ω + (∆ω), and the other for the second harmonic components, i.e. 2ω and 2ω + (2∆ω).
The detected phases of the interferometer signals for the fundamental components ϕ 1 and for the second harmonic components ϕ 2 are, then, where the superscripts r and p denote the reference and probing beams, respectively, and the subscript f means the phase shifts due to fluctuations of the optical path lengths caused by, for instance, mechanical vibrations.The phase shift due to such fluctuations is where N is the refractive index of the medium in which the beam travels.∆ϕ pl (ω) is the phase shift due to the plasmas, that is, with the electron charge e, the speed of light in vacuum c, the permittivity of free space ε 0 , the electron mass m e , and the electron density n e .The integrations in equations ( 3) and ( 4) are performed along the beam path.
From equation (3) we see that 2∆ϕ p f (ω) = ∆ϕ p f (2ω) because the fundamental and harmonic components of the probing beams are completely overlapped along the whole beam path, and similarly for the reference beams, that is 2∆ϕ r f (ω + ∆ω) = ∆ϕ r f (2ω + 2∆ω), which is the condition that can be achieved only with the SCDI-U, allowing us to remove the phase shifts due to the fluctuations.
The detected phases ϕ 1 and ϕ 2 are RF signals in the order of MHz, which means that we can utilize various RF electronic components.We use a frequency doubler to get 2ϕ 1 , which replaces the second nonlinear crystal used in other conventional DI systems [3,4,8,10].An IQ demodulator is used to extract phase information from the SCDI-U, denoted as ϕ SCDI-U , from which we can obtain the line integrated electron density.We see that the SCDI-U takes an advantage of using an AOM, that has a wider bandwidth compared to PEM-based DI systems, yet accomplishes complete overlap of the frequencyshifted reference beams that allows us to remove effects of mechanical vibrations and any fluctuations in the optical path lengths.

Scheme of the SCDI-U with a single detector
The SCDI-U can operate with only a single detector as well, as shown in figure 1(b).It is almost same as the two detector system mentioned above.Difference is that we detect both the fundamental and second harmonic components with a single detector and separate them using a low-and a high-pass filters.The low-pass filtered signal goes through the frequency doubler.We, then, have 2ϕ 1 and ϕ 2 which are fed into the IQ demodulator, providing us ϕ SCDI-U .

With two detectors
Figures 2 and 3 show the optical setup and a picture of the KSTAR SCDI-U with two detectors.We use 1064 nm beam with 3 W power from a cw diode pumped laser (Cobolt Rumda).The second harmonic component, i.e. 532 nm beam, is generated with the nonlinear crystal (MSHG1064-1.0-10).Note that we use one laser source to operate two channels of SCDI-U; hence, we have a beam splitter after the nonlinear crystal.Both the 1064 nm and 532 nm beams pass through the AOM (MT110-A1.5-vis)operated at 110 MHz, generating the frequency-shifted components.As we use the first order diffraction for the fundamental component and the second order diffraction for the second harmonic component, their frequency-shifts are 110 MHz and 220 MHz, respectively.Table 2 lists the measured beam powers for both the fundamental and second harmonic components at the locations indicated by the circled numbers in figure 2.
The probing beams are carried to inside KSTAR vertically (see figure 5 in [12]) and reflects back from the top of KSTAR to the same optics table.Travel distance of the probing beams are approximately 25 m, and to make sure that diameters of the beams are maintained less than 1 inch in the whole beam path, we use a curved mirror whose focal length is 3 m.
A dichroic mirror (DMSP805) that reflects 1064 nm and passes 532 nm is placed in front of the 532 nm detector (APD430A2/M).It is not necessary to have a dichroic mirror in front of the 1064 nm detector (PDA10A2) for two reasons.First, the power of the 532 nm beam is negligible compared to that of the 1064 nm beam as shown in table     After the detectors, all the signals are processed with RF electronic components.The frequency doubler (FD-2+) doubles the phase of 110 MHz beat signals.Then, the two beat signals are converted into 40 MHz before the IQ demodulator (QPD45-s).We convert the signals down to 40 MHz so that we can examine possibility of using a field programmable gate array (FPGA) to extract phase information, where the FPGA in the SCDI-U operates with 40 MHz signal.Evaluating performance of the FPGA is not within a scope of this work.

With a single detector
Figure 4 show a picture of the KSTAR SCDI-U with a single detector.Same optics and electronic components used in the two-detector system are used in the single detector system as well.The detector we use is the 532 nm detector (APD430A2/M).The same dichroic mirror (DMSP805) is also used.According to the specification of DMSP805, 1.038 % of 1064 nm beam are transmitted.Since the fundamental component has much stronger intensity compared to the second harmonic component as listed in table 2, this 1 % transmission

Resolution of the line integrated electron density
The KSTAR SCDI-U uses Nd:YAG laser whose wavelength is 1064 nm, which is much shorter than using CO 2 laser with the wavelength of 10.6 µm or 9.57 µm used in many other TCI or DI systems [7][8][9][10][11]22].We have intentionally chosen to use a shorter wavelength so that it can measure abrupt and large changes in electron density induced by shattered pellet injections [12,23] to investigate effects of the shattered pellets on mitigation of plasma disruptions [15].From equation (4), it is clear that a shorter wavelength is favorable for measuring abrupt and large density changes, otherwise fringe jumps would complicate the data interpretation which is indeed the case we find from the KSTAR TCI system that uses 10.6 µm laser (see figure 6

(b)).
A drawback of using the shorter wavelength is that it is more prone to mechanical vibrations than using a longer wavelength.In order to measure a line integrated electron density with an interferometer, ∆ϕ pl (ω) must be greater than ∆ϕ f (ω).Thus, the minimum measurable line integrated electron density can be obtained with the condition of ∆ϕ pl (ω) = ∆ϕ f (ω), which is (ˆn ) ∆

(ˆN dℓ
) .(6) Therefore, provided that everything else is the same except the wavelength of an interferometer system, a 1064 nm based interferometer system has approximately a hundred times worse density resolution than a 10.6 µm based system.This is where the SCDI-U becomes compelling, which can optically suppress the fluctuations of the optical path lengths with complete overlap of the probing and the reference beams between the fundamental and second harmonic components.Actual resolution of the line integrated electron density can be estimated as the standard deviation of the measured data over approximately one second before or after plasma discharges [8].We have followed the same approach for the KSTAR SCDI-U and TCI. Figure 5 shows measured line integrated electron density (a) by the SCDI-U and (b) by five channels of TCI from KSTAR shot #29316.Note that a shattered pellet is injected at 4.985 s, inducing fast and large density changes.Using the data from −1 s to 0 s, we find that the line integrated density resolution for the SCDI-U is 1.6 × 10 19 m −2 .The resolutions of the TCI channels are estimated to range from 8.4 × 10 17 m −2 to 9.7 × 10 17 m −2 .If the SCDI-U were using the CO 2 laser instead of the Nd:YAG laser, the density resolution would have been 1.6 × 10 17 m −2 according to equation (6), which is five to six times better than the TCI system.
We mention that the achieved line integrated density resolution with the KSTAR SCDI-U using the 1064 nm Nd:YAG laser, i.e. 1.6 × 10 19 m −2 , is applicable for ITER.It is expected that the total path length of a tangential line of sight in ITER plasmas is ∼24 m [8].This means that the KSTAR SCDI-U will have the line averaged density resolution of ∼6.7 × 10 17 m −3 in ITER, which is 0.17 %-1.8 % of the electron density (3.8 × 10 19 m −3 and 4 × 10 20 m −3 [8]) during a flat-top phase of a standard H-mode scenario in ITER.
As a proposed DI system in ITER is foreseen to have a total beam path of ∼100 m [8,24], we believe that using the scheme of the SCDI-U is beneficial since it removes an arduous task of aligning and focusing the beams onto the second nonlinear crystal with an aperture of less than a couple of mm after traveling ∼100 m.Furthermore, the SCDI-U is a heterodyne type with a wide bandwidth achieved by using an AOM while spatial overlaps of the beams are not sacrificed at all.

SCDI-U vs. TCI in KSTAR during a shattered pellet injection
As the beam paths of the SCDI-U and the TCI in KSTAR are vertical and horizontal, respectively, it is difficult to make direct quantitative comparisons between the two.Thus, we examine overall qualitative behavior of the density evolution measured by the two interferometer systems.For the sake of easy comparisons, we provide vertical dashed lines in figure 5 at 0 s, 1.6 s and 3 s, where noticeable changes are observed.Note that a slow and small change of the density is more reliably measured by the TCI system since it has a smaller density resolution as discussed above.
The TCI data in figure 5(b) show that there is a slow build-up of the density from 0 s to 1.6 s, from where a faster increase is observed until ∼2 section.Then, a constant density with a small dip before 3.0 s is observed.After 3.0 s, we have an almost constant density, perhaps with very small increase of density, until a shattered pellet is injected at 4.985 section.We see that, although not exactly the same, similar density evolution can be observed in the SCDI-U data in figure 5(a).Given that the SCDI-U system uses the laser whose wavelength is ten times smaller than that of the TCI system, it clearly demonstrates how well the SCDI-U suppresses mechanical vibrations and that the SCDI-U is capable of measuring slow and small density changes with a reasonable degree of accuracy.
The SCDI-U becomes indispensable for measuring an abrupt and large density change.At 4.985 s a shattered pellet is inject, and we see that the plasmas respond after ∼0.04 s as shown in figure 6(a) measured by the SCDI-U system.It also shows how long it takes for the density to start decaying down with a finite duration of the density fluctuations.With many more experiments we can create a database on the plasma response time and the density decay start time with its rate due to shattered pellet injections, which is crucial information to develop disruption mitigation scenarios for ITER.The SCDI-U can measure such density evolution because it is based on the 1064 nm laser.This is evident since the TCI system with the 10.6 µm laser fails to measure such density changes caused by irremediable fringe jumps as shown in figure 6(b).Some of the TCI channels even fall in the region of negative density, which is non-physical.induced by various schemes of shattered pellet injections measured by the KSTAR SCDI-U.A pure deuterium pellet is injected from the G-port (black, KSTAR shot #28922) and from the O-port (blue, KSTAR shot #28923).Two pure deuterium pellets are simultaneously injected from both the G-port and O-port (red, KSTAR shot #28912).All the pellets are injected at 4.985 section.Note that the vertical line of sight for the KSTAR SCDI-U is located next to the G-port (see figure 5 in [12]).

Density evolutions in various shattered pellet injection schemes obtained by the KSTAR SCDI-U
Dual symmetric shattered pellet injection systems are installed at the G-port and the O-port that are toroidally separated by 180 • to each other in KSTAR [23].How plasmas respond to various scenarios of the pellet injections is an on-going active research area, and the SCDI-U provides essential information on behavior of the line integrated density evolution.
Although the scope of this work does not include detailed analyses on the plasma responses, we present effects of different schemes of the pellet injections to substantiate capability of the SCDI-U system.Figure 7 shows three different KSTAR discharges with the same plasma conditions except how the shattered pellets are injected.In all cases, pure deuterium pellets are used and injected at 4.985 section The difference is that a pellet is injected from the G-port in shot #28922 (black line) or from the O-port in shot #28923 (blue line), while two pellets are simultaneously injected from the G-port and the Oport in shot #28912 (red line).Note that the vertical line of sight for the SCDI-U is located next to the G-port (see figure 5 in [12]); therefore, we expect to observe a stronger plasma response when a same pellet is injected from the G-port compared to an injection from the O-port.
First of all, the plasma response times are much faster for these three shots as well as the decaying rates compared to shot #29316 (see figure 6).To investigate what and how the time scales are determined, we are in the process of creating database from which we can perform various correlation analyses.
Focusing on the three KSTAR shots in figure 7, we find that qualitative behavior of the density evolution for a single pellet injection is similar (see black and blue lines).We see this result as a consequence of high precision on toroidal symmetry that KSTAR has [25].Note that we prefer not to make comprehensive quantitative comparisons between the two due to the biased position of the line of sight for the SCDI-U system towards the G-port.When two same pellets are simultaneously injected from the both ports (red line), we see that the result is not a mere linear combination of the two independent results (black and blue lines).The simultaneous injections create a faster density increasing rate (around 5.005 s) and fine structures before the decay starts after 5.010 section.
Developing a physics-based numerical model for the shattered pellets is probably insurmountable due to strong nonlinearities and ample dynamics; thus, we need to produce a rich database on how the plasmas respond to shattered pellet injections, so that disruption mitigation scenarios using such pellets ITER can be developed.The SCDI-U is a key diagnostic system for this purpose.

Summary
We have developed an SCDI-U system that measures a line integrated electron density along the probing beam path.Similar to a conventional DI, the SCDI-U system generates the second harmonic beam with a nonlinear crystal located right in front of the laser source, and optically suppresses noises caused by fluctuations of the optical path length such as mechanical vibrations.Because the SCDI-U, like an SCDI, replaces the second nonlinear crystal, which is located after the plasmas in a conventional DI system, with a simple electronic frequency doubler, optical setup is much easier even with a long beam path of ∼100 m expected for an ITER DI system, i.e. the task of aligning and focusing the beams onto a small aperture (typically less than a couple of mm) of the second nonlinear crystal after traveling such a distance is completely removed.
Additionally, the SCDI-U using an AOM achieves complete overlap of the fundamental and second harmonic beams by using different diffraction orders from the laser source to the detectors.This was not achieved in the previously reported SCDI system because of the different diffraction angles for the fundamental and second harmonic beams.This allows the SCDI-U to have a wide bandwidth as the AOM operates in a range of several MHz.Complete overlap of the beams also provides less sensitivity to the mechanical vibrations; thus, it allows us to use a laser beam with a shorter wavelength such as 1064 nm so that one can measure an abrupt and large density change induced by, for instance, a shattered pellet injection which is a critical technique to mitigate plasma disruptions in high performance tokamaks.
We have described optical setup of the SCDI-U and how it is installed in KSTAR.Resolution of the line integrated electron density is estimated to be 1.6 × 10 19 m −2 , which is five to six times better than the existing TCI system if the TCI were using a 1064 nm laser beam instead of 10.6 µm.As the resolution is found to be 0.17 %-1.8 % of the expected electron density during a flat-top phase of a standard H-mode scenario in ITER, we believe that the SCDI-U is applicable for ITER.
One of the biggest advantages that SCDI-U can provide is its capability of measuring density evolution right after a shattered pellet injection that causes abrupt and large density changes.We have shown that the KSTAR TCI system with the 10.6 µm CO 2 laser fails to follow the density evolution as the fringe jumps complicate the data analyses, while the SCDI-U captures behavior of the evolution clearly, for instance, how long the plasmas take to respond to the injected pellets, how fast the density increases and decreases, how the plasmas develop fine temporal structures, and so on.This information is of paramount importance for developing valid scenarios of disruption mitigations using shattered pellet injections for ITER.KSTAR is equipped with dual symmetric shattered pellet injection systems to support the ITER program, and the newly developed SCDI-U system is expected to provide necessary and valuable information.

Figure 1 .
Figure 1.Schematic diagrams of the SCDI-U using (a) two detectors and (b) a single detector.Solid lines correspond to the probing beams that go through the plasmas, whereas dashed lines correspond to the reference beams which are frequency-shifted by the AOM.Red and green lines are the fundamental and second harmonic beams, respectively, where the second harmonic beam is generated by the nonlinear crystal (NLC) located right in front of the laser source.Purple lines indicate the overlapped fundamental and second harmonic beams.The blue dashed line is the first order diffraction of the second harmonic beam by the AOM which was used in the SCDI[12] scheme.The SCDI-U uses the second order diffraction, resulting in a complete overlap of the frequency-shifted reference beams, i.e. the purple dashed lines.

2
Figures2 and 3show the optical setup and a picture of the KSTAR SCDI-U with two detectors.We use 1064 nm beam with 3 W power from a cw diode pumped laser (Cobolt Rumda).The second harmonic component, i.e. 532 nm beam, is generated with the nonlinear crystal (MSHG1064-1.0-10).Note that we use one laser source to operate two channels of SCDI-U; hence, we have a beam splitter after the nonlinear crystal.Both the 1064 nm and 532 nm beams pass through the AOM (MT110-A1.5-vis)operated at 110 MHz, generating the frequency-shifted components.As we use the first order diffraction for the fundamental component and the second order diffraction for the second harmonic component, their frequency-shifts are 110 MHz and 220 MHz, respectively.Table2lists the measured beam powers for both the fundamental and second harmonic components at the locations indicated by the circled numbers in figure2.The probing beams are carried to inside KSTAR vertically (see figure5in[12]) and reflects back from the top of KSTAR to the same optics table.Travel distance of the probing beams are approximately 25 m, and to make sure that diameters of the beams are maintained less than 1 inch in the whole beam path, we use a curved mirror whose focal length is 3 m.A dichroic mirror (DMSP805) that reflects 1064 nm and passes 532 nm is placed in front of the 532 nm detector (APD430A2/M).It is not necessary to have a dichroic mirror in front of the 1064 nm detector (PDA10A2) for two reasons.First, the power of the 532 nm beam is negligible compared to that of the 1064 nm beam as shown in table2.The other reason is that the detected interference signals from the 1064 nm and 532 nm detectors are 110 MHz and 220 MHz beat signals, respectively.The phases of detected signals by the 1064 nm detector is doubled by the frequency doubler which generates 220 MHz and 440 MHz signals.When these signals are demodulated with the signal from the 532 nm detector, the DC output of the IQ demodulator is generated by the 220 MHz signal, while the 440 MHz signal does not contribute to the DC output.
Figures2 and 3show the optical setup and a picture of the KSTAR SCDI-U with two detectors.We use 1064 nm beam with 3 W power from a cw diode pumped laser (Cobolt Rumda).The second harmonic component, i.e. 532 nm beam, is generated with the nonlinear crystal (MSHG1064-1.0-10).Note that we use one laser source to operate two channels of SCDI-U; hence, we have a beam splitter after the nonlinear crystal.Both the 1064 nm and 532 nm beams pass through the AOM (MT110-A1.5-vis)operated at 110 MHz, generating the frequency-shifted components.As we use the first order diffraction for the fundamental component and the second order diffraction for the second harmonic component, their frequency-shifts are 110 MHz and 220 MHz, respectively.Table2lists the measured beam powers for both the fundamental and second harmonic components at the locations indicated by the circled numbers in figure2.The probing beams are carried to inside KSTAR vertically (see figure5in[12]) and reflects back from the top of KSTAR to the same optics table.Travel distance of the probing beams are approximately 25 m, and to make sure that diameters of the beams are maintained less than 1 inch in the whole beam path, we use a curved mirror whose focal length is 3 m.A dichroic mirror (DMSP805) that reflects 1064 nm and passes 532 nm is placed in front of the 532 nm detector (APD430A2/M).It is not necessary to have a dichroic mirror in front of the 1064 nm detector (PDA10A2) for two reasons.First, the power of the 532 nm beam is negligible compared to that of the 1064 nm beam as shown in table2.The other reason is that the detected interference signals from the 1064 nm and 532 nm detectors are 110 MHz and 220 MHz beat signals, respectively.The phases of detected signals by the 1064 nm detector is doubled by the frequency doubler which generates 220 MHz and 440 MHz signals.When these signals are demodulated with the signal from the 532 nm detector, the DC output of the IQ demodulator is generated by the 220 MHz signal, while the 440 MHz signal does not contribute to the DC output.

Figure 2 .
Figure 2. Optical setup of the KSTAR SCDI-U with two detectors.The circled numbers indicate locations where we measure beam power, which are listed in table 2.

Figure 4 .
Figure 4. Picture of KSTAR SCDI-U with a single detector.

Figure 5 .
Figure 5. Measured line integrated electron densities (m −2 ) (a) by the KSTAR SCDI-U (using the 1064 nm Nd:YAG laser) and (b) by five channels of the KSTAR TCI (using the 10.6 µm CO 2 laser) from KSTAR shot #29316.A shattered pellet is injected at 4.985 section To aid comparisons between the SCDI-U and TCI measurements, vertical dashed lines at 0 s, 1.6 s and 3 s are drawn to indicate noticeable changes in the overall behavior of the density evolution.

Figure 6 .
Figure 6.Enlarged version of figure 5 from 5.01 s to 5.04 s where detailed responses of the line integrated electron density can be seen right after a shattered pellet injection at 4.985 section Grey highlighted regions indicate negative density, which is non-physical.

Figure 7 .
Figure 7. Evolution of the line integrated electron density (m −2 )induced by various schemes of shattered pellet injections measured by the KSTAR SCDI-U.A pure deuterium pellet is injected from the G-port (black, KSTAR shot #28922) and from the O-port (blue, KSTAR shot #28923).Two pure deuterium pellets are simultaneously injected from both the G-port and O-port (red, KSTAR shot #28912).All the pellets are injected at 4.985 section.Note that the vertical line of sight for the KSTAR SCDI-U is located next to the G-port (see figure5in[12]).

Table 2 .
Beam powers at the positions indicated by the circled numbers in figure 2.