Upgrade of vacuum ultraviolet spectroscopy system on J-TEXT tokamak

Spectroscopy diagnostics, as one of the most fundamental diagnostic systems on tokamaks, has long been used in the study of impurity behaviors. Because many strong spectral lines fall in the vacuum ultraviolet (VUV) region, VUV spectroscopy is commonly used to monitor impurity species and obtain the impurity distribution.^1–4 

The previous VUV spectroscopy system on the Joint Texas Experimental Tokamak (J-TEXT) consisted of a McPherson normal-incidence VUV monochromator with a concave grating of 1 m focal length and an Andor high-energy 1024 × 1024-pixel² charge-coupled device (CCD). For the detailed layout and configuration of this system, see Refs. 5 and 6. The response time of the CCD is ∼40 ms with a 20-vertical-track setup due to its relatively low readout rate (3 MHz) and slow vertical shift speed. As a typical J-TEXT discharge can last for 800 ms with the particle confinement time ranging from 10 to 20 ms, the previous system cannot resolve the impurity evolution in the transient modulated experiments on the J-TEXT, such as those studies using external resonant magnetic perturbation (RMP) and electrode biasing. It raises the requirements of operating the VUV system at a high-time response of about 5 ms. In addition, the significance of the role of three-dimensional (3D) MP fields induced by RMP is highlighted on the J-TEXT with its impacts mainly located in the edge region (r > 18 cm).^7–9 It makes a demand on the spatial resolution for about 1 cm.

There have been various attempts to achieve high temporal resolution in tokamak VUV systems.^10–12 For example, the detector of the VUV system on the TEXTOR tokamak consists of an open multichannel-plate detector with a subsequent first-generation light amplifier and a linear photodiode array camera head, which allows continuous recording of spectra at a full-spectrum rate of 1000 per second.¹¹ The detectors applied on the JET tokamak consist of two chevron microchannel plates (MCPs) and several multiwire anodes. Of these, the central six are clocked to give fast measurements with 1 ms time resolution for the vertical system and 10 ms for the horizontal system.¹² 

On the J-TEXT, the VUV spectroscopy system has been upgraded using a one-stage MCP to convert the invisible VUV light to visible emission and the first application of an electron-multiplying charge-coupled device (EMCCD) to achieve fast acquisition while retaining an appropriate spatial resolution.

The remainder of this article is structured as follows: The details of the VUV spectroscopy system’s upgrade, including the equipment configuration and an efficiency assessment, are given in Sec. II. The system test and preliminary results on J-TEXT are presented in Sec. III. Finally, a closing summary is given in Sec. IV.

In a fast-acquisition condition, the short exposure time may lead to an unworkably weak signal, and this commonly limits the temporal resolution of many spectroscopic systems. To increase the response speed, the VUV spectroscopy system was upgraded with the aim of enhancing the signal intensity. The CCD detector was replaced with a new detector system consisting of a one-stage MCP, a lens relay, and an EMCCD camera, as shown in Fig. 1. Some key components are highlighted in yellow in the equipment photograph.

The MCP (McPherson Co.) has an active area with a plate diameter of 40 mm and a pore diameter of 10 µm. It is configured with a CsI-coated photocathode to obtain a high sensitivity in the wavelength range 1–200 nm, and it includes P43 phosphor mounted with a fiber-optic image-transfer system. The VUV light can thus be converted to visible light with a peak phosphor emission at 545 nm. Additionally, the MCP is opened to the VUV monochromator and works with high-voltage power supplies (1 kV for the MCP and 5 kV for the phosphor). This brings a significant constraint for the operation of the system: the MCP front-end must operate at a vacuum pressure lower than 1 × 10⁻⁴ Pa to avoid damage caused by high-voltage electric arcing. Therefore, we added a molecular turbopump assembly mounted on the VUV monochromator main chamber, as presented in Fig. 1. This can maintain the vacuum pressure at a level of 10⁻⁵ Pa.

The EMCCD camera (Andor iXon Ultra 888) has the same sensor size as the previous high-energy CCD: 1024 × 1024 13-µm pixels. It is optimized to achieve a high frame rate through a fast pixel-readout rate of 30 MHz, a short vertical shift time of 0.6 µs, and the use of the frame-transfer technique. The frame rate can reach 1400 f/s in a full vertical binning mode and ∼500 f/s with a ten-vertical-track setup. Its quantum efficiency peaks at ∼93% at a wavelength of ∼550 nm, in line with the phosphor’s peak emission wavelength. The application of an EMCCD makes the system capable of achieving a continuous rapid response. Such EMCCDs have been widely used in visible spectroscopy diagnostics on tokamaks. However, this is the first attempt to introduce it to the VUV spectroscopic system.

A lens relay with a focal length of 50 mm and an aperture ratio of f/2.8 is used to image the phosphor screen (fiber optics) on the EMCCD sensor. Since the MCP’s active area is larger than the EMCCD’s sensor area, the lens relay is configured to achieve a magnification of 0.5, which extends the viewing field of the VUV system but will somewhat reduce the spectral resolution.

Since the high-energy CCD of the previous VUV spectroscopy system could offer a valuable signal on the J-TEXT, here we assess the VUV system performance by comparing the sensitivity of the upgraded detector assembly with that of the previous detector. For photons with a wavelength of ∼100 nm, the Andor high-energy CCD’s quantum efficiency is ∼15%, which determines the detector sensitivity for the previous VUV system. In the upgraded detector assembly, the efficiency of the CsI-coated cathode is ∼20%, and there are several signal-enhancement measures as follows:

1. Electron gain on the MCP: ∼1000×.

2. Electron-to-photon rate of P43 phosphor: ∼150 ph/e⁻@5 eV.

3. Electron multiplication on EMCCD: 1000 (maximum).

Accordingly, the signal from plasma impurity emission will be partly weakened from the coupling among the fiber optic array, the lens relay, and the EMCCD. The overall sensitivity of the detector assembly can be described as

where k_c, g_m, g_p, k_f, k_l, g_d, and k_d denote the cathode efficiency, MCP gain, phosphor gain, fiber-optic transmission, lens-relay transmission (including the effect of the aperture), EMCCD gain, and efficiency, respectively. By ignoring the EMCCD gain, it is estimated that the S value in Eq. (1) is around 50, more than 300 times the high-energy CCD’s efficiency. Thus, the upgraded detector assembly is believed to be promising for short-exposure-time application on the J-TEXT.

For the detector assembly, the MCP’s response time is ∼1 ns, but the decay time of the P43 phosphor is ∼2 ms (1.5 decay to 10% and 3 ms decay to 1%), which is the shortest response-time limit. As introduced in Sec. II A, the EMCCD’s frame-interval time with a ten-vertical-track setup is in line with the phosphor response time. The EMCCD frame rate is constrained by the required exposure time. Practically, the exposure time and the multi-track setup strategies are determined by the spectral line intensity, taking into account the specific experimental purpose.

There are two imaging processes in the upgraded system. The first is imaging the MCP’s screen on the EMCCD sensor via the lens relay; the second is the imaging on the MCP’s cathode plane via a concave spherical grating. This two-stage imaging, along with the high-vacuum environment requirement, makes it much more challenging to achieve system focusing.

We customized a reference plate with a slit, as shown in Fig. 2(a), which can replace the MCP with a slit at exactly the same position as the phosphor screen surface to achieve the first-step focusing. With a 532 nm green light source illuminating the slit, the imaging on the EMCCD is focused separately. Figure 2(b) illustrates the slit images acquired by the EMCCD. Pattern no. 5 has the clearest edge and smallest size for 39 × 390 pixels² [approximately equal to (0.5 × 5.0) mm²]. Considering the size of the slit is (1 × 10) mm², the magnification of the lens relay can be deduced as 0.5.

After the first focus, the EMCCD is mounted back onto the MCP assembly. Under the high-vacuum condition, the VUV system is operated during the Taylor cleaning phase, when the device is accessible and the plasma can provide strong hydrogen lines in the VUV band. The second focus is achieved on-site based on the hydrogen Lyman-alpha line (121.57 nm), as shown in Fig. 3. The linear dispersion was tested to be 1.65 nm/mm, and the upgraded VUV system obtains a spectral resolution of 0.13 nm.

The VUV spectrometer system was tested in the J-TEXT normal-operation phase. Figure 4 presents shot-by-shot scanned spectra in the range of 30–210 nm with a 1200 g/mm grating. The concave grating is rotated and translated automatically to achieve focus at the desired wavelength in each shot. The bandpass in one frame is ∼20 nm, which is twice that before the upgrade due to the magnification of 0.5 with the same CCD sensor size. It can be concluded that the upgraded detector shows a higher sensitivity in this range, especially for wavelengths under 60 nm, where the previous detector could rarely obtain an observable line.

The calibration in the spatial direction is carried out by means of optical path alignment with the application of an alignment plate.⁶ The 15 rectangular apertures opened on the plate are compiled for easy recognition, as shown in Fig. 5(a). Each aperture and the spatial resolving slit determine a line of sight, which projects to a fixed vertical position at the middle vertical line of the plasma cross section. The projected vertical position is −25.5 cm for center of spot No. 10 on the detector, and center of spot No. 6 corresponds to a vertical position of approximately −8.5 cm (minus indicates the location below the middle plane).

The spatial resolution, determined by the system aberration and the space-resolving slit width, can be evaluated based on the spot-blurring rate. As shown in Fig. 5(b) with the space-resolving slit’s width of 1 mm, the height of spot 7 (corresponding to the image of aperture 7 on the alignment plate) is ∼0.41 mm marked in red and the gap between spots 7 and 8 is ∼0.51 mm, as marked in blue. Taking into account the objectives’ relative height and the projection’s magnification in the plasma, the spatial resolution is deduced as ∼4 mm according to the spot-blurring rate via the geometrical analysis. Nevertheless, the final practical spatial resolution is also limited by the vertical grouping configuration on the EMCCD.

The viewing field of the upgraded system covers a wider region of the half section from r ∼ 6 to 28 cm, as shown in Fig. 6. Taking C III (97.7 nm) and O VI (103.19 nm) as examples, the cord-integrated radial profiles were obtained under a 20-vertical-track setup. This provides a spatial interval of 1.2 cm. The point at r ∼ 5.5 has been marked in gray for its location out of the edge of the viewing field. It is told that the C III radiation band is located at r ∼ 26 cm; the O VI radiation band is located at r ∼ 21 cm.

The acquisition interval is tested as short as possible with a good signal-to-noise rate, as shown in Fig. 7(a). The intensity of the O VI (103.19 nm) spectral lines detected in the 2 ms step with the ten-vertical-track setup is strong enough to be analyzed. The significant increase in impurity intensity comes with the application of the electron cyclotron heating system (ECRH) at 0.3 s. In this situation, the evolution of the O VI intensity is shown in Fig. 7(b). It can be concluded that the upgraded system can effectively depict the evolution of the impurity emission intensity with a 2 ms interval.

The VUV spectroscopy system on the J-TEXT was upgraded by replacing the detector with an MCP assembly to fulfill the need for experiments with fast evolution. The focusing of the two-stage imaging process in the updated system was achieved through a two-step focusing procedure, resulting in a spectral resolution of 0.13 nm. The cord-integrated radial profiles of C III (97.7 nm) and O VI (103.19 nm) were obtained under a 20-vertical-track setup with coverage of the viewing field. Based on the spot-blurring rate in the image of an alignment plate, the upgraded system’s spatial resolution was evaluated as ∼4 mm. The 2 ms acquisition interval with a ten-vertical-track setup was achieved after spatial calibration for some relatively strong lines, such as O VI (103.19 nm), effectively showing the evolution of the intensities of impurity lines with the application of ECRH.

The authors are very grateful for the help of the J-TEXT team. This work was supported by the National Key R&D Program of China under Grant Nos. 2017YFE0302000 and 2017YFE0301802 and the National Natural Science Foundation of China under Grant No. 51821005.

The data that support the findings of this study are available from the corresponding author upon reasonable request.