---
title: Globus-M2 Tokamak Diagnostics
url: https://www.emergentmind.com/topics/globus-m2-tokamak
type: topic
---

# Globus-M2 Tokamak Diagnostics

Searching arXiv for recent Globus-M2 tokamak papers to ground the article.
Tool call: arXiv search for "Globus-M2 tokamak"
Globus-M2 is a compact spherical tokamak with major radius $R_{0}=0.36\ \mathrm{m}$ and minor radius $a=0.22\ \mathrm{m}$, operated with toroidal magnetic field up to $1\ \mathrm{T}$ and plasma current up to $0.5\ \mathrm{MA}$, and used both for confinement studies in high-field spherical-tokamak regimes and for the development of Thomson-scattering diagnostics with ITER-grade components [2311.18723] [2509.02214]. In the reported operating domain, it spans core and scrape-off-layer electron conditions from a few eV to $\sim 5\ \mathrm{keV}$ and from $5\times 10^{17}\ \mathrm{m}^{-3}$ to $3.25\times 10^{20}\ \mathrm{m}^{-3}$, while hot-ion neutral-beam-heated plasmas at $B_{t}\approx 0.9\ \mathrm{T}$ reach central ion temperatures up to $4\ \mathrm{keV}$ and peak $T_{i}/T_{e}$ ratios exceeding $2.5$ [2311.18723] [2509.02214].

## 1. Machine configuration and operating domain

Globus-M2 is described as a compact spherical tokamak with $R_{0}=0.36\ \mathrm{m}$, $a=0.22\ \mathrm{m}$, and aspect ratio $A=R_{0}/a\approx 1.6$ [2509.02214]. In the diagnostic description, the machine is specified with toroidal field $B_{t}$ up to $1\ \mathrm{T}$ and plasma current $I_{p}\leq 0.5\ \mathrm{MA}$, while the hot-ion-mode study reports $B_{t}$ varied from $0.5$ to $0.95\ \mathrm{T}$ and $I_{p}$ up to $0.4\ \mathrm{MA}$, with typical hot-ion cases at $0.3$–$0.4\ \mathrm{MA}$ [2311.18723] [2509.02214]. Line-averaged density is reported as $n_{e}\approx (1.5$–$9)\times 10^{19}\ \mathrm{m}^{-3}$, with peaked core values $n_{0}\sim 1\times 10^{20}\ \mathrm{m}^{-3}$ in H-mode, and the Thomson-scattering system covers densities from the scrape-off layer to the core [2509.02214] [2311.18723].

Neutral-beam heating is a central actuator in the reported high-performance regimes. The machine operates with a single deuterium beam at $E_{nb}=45\ \mathrm{keV}$ and $P_{nb}$ up to $0.9\ \mathrm{MW}$, with $0.3$–$0.9\ \mathrm{MW}$ explored and two-beam scenarios with D or H injection also reported [2509.02214]. Ohmic heating is given as $P_{oh}\sim 0.4\ \mathrm{MW}$, and in hot-ion discharges $P_{oh}\approx P_{nb}$ [2509.02214].

| Parameter | Value | Context |
|---|---:|---|
| Major radius $R_{0}$ | $0.36\ \mathrm{m}$ | Machine geometry |
| Minor radius $a$ | $0.22\ \mathrm{m}$ | Machine geometry |
| Aspect ratio $A$ | $\approx 1.6$ | Spherical-tokamak regime |
| Toroidal field $B_{t}$ | $0.5$–$0.95\ \mathrm{T}$; up to $1\ \mathrm{T}$ | Transport and diagnostic studies |
| Plasma current $I_{p}$ | up to $0.4\ \mathrm{MA}$; $\leq 0.5\ \mathrm{MA}$ | Hot-ion and general operation |
| Neutral-beam heating | $45\ \mathrm{keV}$, up to $0.9\ \mathrm{MW}$ | Hot-ion regimes |
| Line-averaged density | $(1.5$–$9)\times 10^{19}\ \mathrm{m}^{-3}$ | Operational space |

These parameters place Globus-M2 in the small, low-aspect-ratio, high-beta-relevant spherical-tokamak class. Within the reported dataset, the machine is used to study both profile-resolved transport and the response of confinement to increasing toroidal magnetic field.

## 2. Thomson-scattering system architecture

The Thomson-scattering system on Globus-M2 is launched in the equatorial midplane, and the probing chord spans $R\approx 23\ldots 56\ \mathrm{cm}$, corresponding to $r/a\approx -0.5\ldots 0.8$, thereby covering the low-field-side scrape-off layer, the last closed flux surface, and the core [2311.18723]. The system is explicitly presented as recently developed for the Globus-M2 spherical tokamak and as a prototype relevant to ITER divertor Thomson scattering.

The beam injection system uses two ITER-grade pulsed lasers located in a dedicated laser room. The main laser is an Nd:YAG at $1064.5\ \mathrm{nm}$ with $330\ \mathrm{Hz}$ repetition rate, $0.1$–$3\ \mathrm{J}$ pulse energy, $10\ \mathrm{ns}$ FWHM, MOPA configuration, and beam divergence $0.1\ \mathrm{mrad}$. The auxiliary laser is an Nd:YLF at $1047.3\ \mathrm{nm}$, $50\ \mathrm{Hz}$, $2\ \mathrm{J}$, and $3\ \mathrm{ns}$ FWHM, used for multi-colour self-calibration tests. The beams are transported approximately $11\ \mathrm{m}$ via six relay mirrors, and a fraction of each pulse is picked off to an optical fibre connected to a synchronization and energy-monitoring unit. Probing is single-pass in the equatorial plane, and the beam is dumped outside the vessel [2311.18723].

Collection optics consist of an input lens with $f=2.5\ \mathrm{m}$ and $D=65\ \mathrm{mm}$, AR-coated and mounted on the vacuum vessel with in-vessel baffles to suppress stray light. A telecentric 5-lens objective collects scattered light from 18 possible chord locations; 9 are currently equipped, including 8 standard locations and 1 split bundle for enhanced edge resolution. Lens magnification varies from $4.1$ on the low-field side to $2.8$ on the high-field side, giving spatial chords of length $11$–$22\ \mathrm{mm}$ in routine mode and $2.5\ \mathrm{mm}$ with the split bundle. The fibre bundles are CeramOptec bundles with 188 fused-silica fibres each, of diameter $196/220\ \mu\mathrm{m}$, numerical aperture $0.22$, and length $18\ \mathrm{m}$; overall collection throughput is approximately $30\%$ [2311.18723].

Spectral analysis is performed by 10 identical polychromators, each with six interference-filter channels, of which five channels are used in multi-wavelength mode, centered at approximately $\lambda\approx 1056, 1039, 1007, 944, 818\ \mathrm{nm}$ with FWHM approximately $12\ldots 168\ \mathrm{nm}$ optimized for $T_{e}=5\ldots 5000\ \mathrm{eV}$. Detection uses Hamamatsu S11519-15 APDs with multiplication factor $M=100$, a custom preamplifier, bandwidth greater than $250\ \mathrm{MHz}$, and thermal drift compensation of $\pm 0.5\%$ over $17$–$35\,^{\circ}\mathrm{C}$. To cover a density span of $600\times$, each APD output is split into two high-pass channels with gains $G=2$ and $G=10$; low-pass outputs feed a $500\ \mathrm{kHz}$ ADC for background recording and auxiliary diagnostics such as $Z_{\mathrm{eff}}$ [2311.18723].

Data acquisition is built around eight CAEN V1743 digitizers using SAMLONG sampling at $3.2\ \mathrm{GS/s}$ and $12$-bit resolution, with a $320\ \mathrm{ns}$ record window that can capture both $1064\ \mathrm{nm}$ and $1047\ \mathrm{nm}$ pulses separated by $O(10\ \mathrm{ns})$. The system is compatible with burst mode up to $8\ \mathrm{kHz}$ and continuous acquisition up to $1.4\ \mathrm{kHz}$. Electronics are placed in an EM-shielded room that includes the fibre feedthrough, 10 polychromators, a VME crate with digitizers and synchronization hardware, a central computer, and a UPS. Data are streamed in real time for post-processing or feedback control [2311.18723].

## 3. Measurement model and calibration methodology

The theoretical basis used for the diagnostic is the spectral Thomson-scattering signal model
\[
\frac{d^{2}N}{d\Omega\,d\lambda}
=
E_{L}\,n_{e}\,\frac{d\sigma_{T}}{d\Omega}\;f(T_{e},\lambda-\lambda_{0},\theta)\;,
\]
where $E_{L}$ is the probe-laser pulse energy in photons, $d\sigma_{T}/d\Omega=(r_{e}^{2}/2)(1+\cos^{2}\theta)$ is the Thomson cross-section, and $f(T_{e},\Delta\lambda,\theta)$ is the normalized spectral form factor for a Maxwellian electron population [2311.18723]. For non-relativistic $T_{e}$, the summary gives the approximation
\[
f(T_{e},\Delta\lambda)\propto\exp\Bigl[-\frac{(\Delta\lambda)^{2}}{2\,\sigma_{\lambda}^{2}}\Bigr],\qquad
\sigma_{\lambda}=\frac{\lambda_{0}}{c}\sqrt{\frac{kT_{e}}{m_{e}}}\;.
\]
In practice, this spectral form factor is folded with the filter transmission $T_{i}(\lambda)$ of each channel, yielding
\[
S_{i}
=
C\,n_{e}\,\frac{d\sigma_{T}}{d\Omega}\;\int T_{i}(\lambda)\,f(T_{e},\lambda-\lambda_{0})\,d\lambda\;,
\]
where the global system constant $C$ absorbs lens solid angle, fibre throughput, detector quantum efficiency, digitizer gain, and related factors [2311.18723].

Absolute calibration proceeds by measuring the spectral transmission of each channel ex situ with calibrated light sources, thereby determining the system constant $C$. Laser energy monitors and pick-off fibres track the pulse-to-pulse laser energy $E_{L}$, and APD gains are regularly checked against dark and light standards [2311.18723]. This calibration chain supports conventional multi-channel least-squares fitting, in which $T_{e}$ is inferred from spectral shape and $n_{e}$ from absolute amplitude once $T_{e}$ is known.

A distinctive feature of the Globus-M2 implementation is its multi-colour self-calibration mode. In a reactor-like environment, the relative spectral sensitivity of inaccessible optics may drift; the multi-colour approach is intended to remove the need for absolute channel calibration. If two probe wavelengths $\lambda_{1}$ and $\lambda_{2}$ fire within $\lesssim 100\ \mathrm{ns}$ so that local plasma parameters remain constant, the channel ratio
\[
S_{i}\equiv \frac{S_{i}(\lambda_{1})}{S_{i}(\lambda_{2})}
\]
is compared to
\[
F_{i}(T_{e})=
\frac{\int_{\lambda_{\min,i}}^{\lambda_{\max,i}} P_{TS}(T_{e},\lambda,\lambda_{1})\,d\lambda}
{\int_{\lambda_{\min,i}}^{\lambda_{\max,i}} P_{TS}(T_{e},\lambda,\lambda_{2})\,d\lambda}\;,
\]
which depends only on $T_{e}$ and the known filter bandpasses, not on $C$ or channel-to-channel throughput [2311.18723]. The temperature is then obtained by minimizing
\[
\chi^{2}(T_{e})=\sum_{i=1}^{N}\frac{[S_{i}-F_{i}(T_{e})]^{2}}{\sigma_{i}^{2}}\rightarrow \min\;.
\]
Once $T_{e}$ is known, absolute calibration enters only the amplitude estimate for $n_{e}$, which can be referred to one of the wavelengths with known $T_{e}$ [2311.18723].

This calibration strategy is significant because it links Globus-M2 directly to ITER-relevant diagnostics constraints: inaccessible optics, long-term spectral drift, and the need for time-coincident multi-wavelength inversion without full re-access to the optical chain.

## 4. Measurement capability and first profile-resolved results

The reported Thomson-scattering operating range is $T_{e}=5\ \mathrm{eV}\ldots 5000\ \mathrm{eV}$ and $n_{e}=5\times 10^{17}\ldots 3.25\times 10^{20}\ \mathrm{m}^{-3}$ [2311.18723]. Spatial resolution is $10$–$20\ \mathrm{mm}$ in standard 188-fibre bundles and $\lesssim 2.5\ \mathrm{mm}$ at the edge using the split bundle. In routine calibration, single-point precision is reported as $\Delta T_{e}/T_{e}<10\%$ over most of the range and $\Delta n_{e}/n_{e}<5\%$. In the scrape-off layer, with reduced laser energy of $0.7\ \mathrm{J}$ and split bundles, the system remains reliable down to $n_{e}\gtrsim 5\times 10^{17}\ \mathrm{m}^{-3}$, although $T_{e}$ errors grow as $n_{e}$ falls [2311.18723].

The first core measurements are illustrated by sawtooth-oscillating discharge \#39627 at $I_{p}=210\ \mathrm{kA}$ and $B_{t}=0.7\ \mathrm{T}$. Typical raw waveforms showed $\mathrm{S/N}\gg 1$ even at $0.7\ \mathrm{J}$ probe energy and $n_{e}\approx 2.8\times 10^{19}\ \mathrm{m}^{-3}$. A multi-channel fit at $R=39\ \mathrm{cm}$ gave $T_{e}\approx 920\pm 30\ \mathrm{eV}$ and $n_{e}\approx 3.6\pm 0.5\times 10^{19}\ \mathrm{m}^{-3}$. Radial profiles $T_{e}(r)$ and $n_{e}(r)$ were mapped through the sawtooth cycle, with mixing radius $R_{\mathrm{mix}}\approx 49\ \mathrm{cm}$, corresponding to $r/a\approx 0.44$, in excellent agreement with the ASTRA equilibrium $q=1$ surface. The stored electron energy $W_{e}$ inferred from Thomson scattering matched ASTRA and interferometer data [2311.18723].

The same system resolved scrape-off-layer conditions in a multi-shot scan over discharges \#39586 to \#39590. By shifting the separatrix radially in flattop, data from $r/a>1$ were obtained. At the separatrix, the measured values were $n_{e}\approx 1\times 10^{19}\ \mathrm{m}^{-3}$ and $T_{e}\approx 10\ \mathrm{eV}$, and behind the separatrix $T_{e}$ decayed exponentially into the ELMy SOL. The lower sensitivity bound was reported as $n_{e}\gtrsim 5\times 10^{17}\ \mathrm{m}^{-3}$ for full probe energy and $\gtrsim 3\times 10^{18}\ \mathrm{m}^{-3}$ in reduced-energy trials [2311.18723].

A dedicated multi-colour test in discharge \#40204 used synchronized $1064.5\ \mathrm{nm}$ and $1047.3\ \mathrm{nm}$ pulses with $\Delta t\approx 60\ \mathrm{ns}$, both recorded within one $320\ \mathrm{ns}$ digitizer window. Independent classical fits at each wavelength agreed on $T_{e}(r)$ to within $\lesssim 10\%$. The multi-colour inversion without spectral calibration yielded $T_{e}$ with approximately $22\%$ RMS error and approximately $8\%$ mean offset relative to the calibrated $1064\ \mathrm{nm}$ result; Monte Carlo simulations of synthetic Thomson-scattering signals reproduced these error levels. The ratio of laser energies $y=E_{1}/E_{2}$ was found with $\lesssim 3$–$4\%$ uncertainty [2311.18723].

## 5. Hot-ion operational regimes and confinement scalings

The hot-ion-mode study describes an operational sequence in which early neutral-beam injection during the plasma-current ramp, while $q_{\min}>1$, deposits large fast-ion torque, builds strong co-current toroidal rotation with $V_{\phi}\sim 50$–$100\ \mathrm{km/s}$, and establishes peaked density and temperature profiles [2509.02214]. Ion-scale turbulence is interpreted as stabilized by $E\times B$ shear from beam-driven rotation and by increased neoclassical ion damping at higher $B_{t}$. A transition to H-mode occurs before sawtooth onset, signaled by ELMs and a $D_{\alpha}$ drop. Peak $T_{i}/T_{e}\gtrsim 2.6$ is reached before MHD activity, including sawteeth and TAEs, degrades core confinement [2509.02214].

Within the reported high-field regime, central ion temperatures $T_{i}(0)$ up to $4\ \mathrm{keV}$, approximately $50\ \mathrm{MK}$, are routinely reached at $B_{t}=0.9\ \mathrm{T}$ and $I_{p}=0.4\ \mathrm{MA}$, with peak $T_{i}/T_{e}$ ratios $\gtrsim 2.5$. Core electron temperatures $T_{e}(0)$ rise to approximately $1.5\ \mathrm{keV}$ in H-mode, and the Thomson-scattering profiles measured at $330\ \mathrm{Hz}$ are reported to agree with CXRS for ions [2509.02214].

Multi-shot regression over $I_{p}=0.15$–$0.4\ \mathrm{MA}$, $B_{t}=0.5$–$0.95\ \mathrm{T}$, $n_{e}=1.5$–$9\times 10^{19}\ \mathrm{m}^{-3}$, and $P_{nb}=0.3$–$0.9\ \mathrm{MW}$ yields the following confinement scalings:
\[
W_{\mathrm{MHD}} \propto I_{p}^{1.1} B_{t}^{0.6} n_{e}^{-0.63} P_{nb}^{0.08}
\]
with mean absolute percentage error approximately $12\%$, and
\[
W_{e} \propto I_{p}^{1.16} B_{t}^{0.5} n_{e}^{-0.20} P_{abs}^{1.0}
\]
with MAPE approximately $10\%$ [2509.02214]. Assuming $P_{abs}\approx P_{nb}+P_{oh}$, the implied energy-confinement time in the hot-ion regime follows approximately
\[
T_{E}\propto I_{p}^{0.53} B_{t}^{1.05} P_{abs}^{-0.92} n_{e}^{-0.63}\;,
\]
which is stated to fit the $30\%$-increase test from $0.7\rightarrow 0.9\ \mathrm{T}$ and $0.3\rightarrow 0.4\ \mathrm{MA}$, where $T_{E}$ rose from $7.5\ \mathrm{ms}$ to $11.5\ \mathrm{ms}$ [2509.02214].

The study also reports a previous Globus-M2 fit, denoted GLB_2020,
\[
T_{E}\propto I_{p}^{0.43} B_{t}^{1.19}\;,
\]
and contrasts it with the IPB98(y,2) scaling,
\[
T_{E,\mathrm{IPB98}}\propto I_{p}^{0.93} B_{t}^{0.15} P_{nb}^{-0.69} n_{e}^{-0.41}\;.
\]
In dimensionless form, the reported confinement correlation is
\[
B_{t}\,T_{E}\propto \nu_{\mathrm{eff}}^{*-0.45}\;,
\]
with $\nu_{\mathrm{eff}}^{*}\propto n_{e}/Z_{\mathrm{eff}}\,T_{e}^{-2}$, stated to be close to both gyro-Bohm and neoclassical trends in the collisionality range $0.01$–$0.1$ [2509.02214].

## 6. Transport interpretation and comparison with higher-field spherical tokamaks

The reported transport picture separates ion and electron channels. Ion heat diffusivity is given as $\chi_{i}\approx 1\ \mathrm{m}^{2}/\mathrm{s}$, within $10$–$20\%$ of neoclassical predictions at low $\nu^{*}$ and in D-beam cases, so that $\chi_{i}/\chi_{i,\mathrm{NC}}\rightarrow 1$ [2509.02214]. Electron heat diffusivity $\chi_{e}$ remains anomalous but decreases with increasing $B_{t}$, and an effective $\chi_{e}$ is inferred from power balance and the gradient scale $R/L_{T_{e}}$ [2509.02214].

The mechanisms proposed for electron-heat-transport suppression are specific. At low $B_{t}$, microtearing modes and ETG are identified as drivers of $\chi_{e}$; higher $B_{t}$ lowers collisionality and drives $\nabla T_{e}$ below ETG and TEM thresholds. At $r/a=0.7$, the normalized gradient $R/L_{T_{e}}$ is reported as $20$–$40\%$ below TEM thresholds and only moderately above linear ETG thresholds. GENE gyrokinetic runs predict strong ETG growth at $k_{\perp}\rho_{e}\sim 0.1$–$0.2$, but $E\times B$ shear and increased $B_{t}$ reduce fluctuation amplitude [2509.02214]. Diagnosis of this regime combines Thomson-scattering profiles, analytic ETG and TEM thresholds, and the dimensionless fit $B_{t}T_{E}\propto \nu^{*-0.45}$ [2509.02214].

The comparison with ST40 is used to frame the limits of simple magnetic-field extrapolation. ST40 is reported to achieve $T_{i}>9\ \mathrm{keV}$ and confinement enhancement factor $H\equiv T_{E}/T_{E,\mathrm{IPB98}}\approx 1.4$ at $B_{t}=1.7$–$2\ \mathrm{T}$, while Globus-M2 rises from approximately $H\sim 0.7$ at $0.6\ \mathrm{T}$ to approximately $H\sim 1.4$ at $0.9\ \mathrm{T}$ [2509.02214]. However, above $1\ \mathrm{T}$ no further gain in $H$ is observed when compared to ST40, and both machines exhibit similar $\chi_{i}$ and $\chi_{e}$ versus $\nu^{*}$ despite a twofold difference in $B_{t}$ [2509.02214].

This comparison is relevant to a common expectation that confinement in spherical tokamaks should continue to improve strongly with toroidal magnetic field. The reported Globus-M2–ST40 comparison does not support that extrapolation: ion transport is already at neoclassical level, while electron transport remains limited by ETG and microtearing and is only weakly reduced above $B_{t}\sim 1\ \mathrm{T}$ [2509.02214]. The saturation arguments given in the study are that rigid ETG and turbulence thresholds become insensitive to further $B_{t}$ increase once $R/L_{T_{e}}$ is pinned, and that $\rho_{i}^{*}\approx 0.06$–$0.1$ and $\nu^{*}\approx 0.01$–$0.1$ in Globus-M2 already match the dimensionless regime of ST40 [2509.02214].

## 7. ITER-oriented diagnostic role and projected development

The Thomson-scattering system on Globus-M2 is explicitly framed as a prototype for ITER divertor Thomson scattering. Its distinctive features are the use of spectrometers, acquisition system, and lasers that meet the base requirements for ITER Thomson-scattering diagnostics, and the system is stated to meet the ITER divertor and core Thomson-scattering specifications in several key respects [2311.18723]. These include pulse width $\leq 10\ \mathrm{ns}$, repetition rate up to $330\ \mathrm{Hz}$, with an indicated path to $100\ \mathrm{Hz}$ for ITER, energy up to $3\ \mathrm{J}$, detector bandwidth of at least $250\ \mathrm{MHz}$ with $3.2\ \mathrm{GS/s}$ digitizers for full time-trace recording, and demonstrated multi-colour self-calibration with $O(10\ \mathrm{ns})$ synchronization tolerance [2311.18723].

The significance assigned to the Globus-M2 implementation is twofold. First, it validates hardware choices based on LD-pumped Nd:YAG and Nd:YLF lasers, GHz digitizers, polychromators, and APDs of ITER-prototype or grade quality. Second, it validates the spectral-ratio algorithms proposed for ITER to maintain calibration over long-term operation with inaccessible optics [2311.18723]. In that sense, Globus-M2 functions not only as a confinement experiment but also as a reactor-relevant testbed for profile diagnostics under conditions where direct spectral recalibration may not remain feasible.

The next steps stated for the diagnostic program are to expand spatial channels to cover the entire poloidal cross-section, increase repetition rate to $100\ \mathrm{Hz}$, and integrate real-time inversion for feedback and control [2311.18723]. A plausible implication is that the same machine geometry and plasma conditions that make Globus-M2 useful for high-gradient, low-collisionality transport studies also make it a stringent environment for validating fast, calibration-robust profile diagnostics.

Source: https://www.emergentmind.com/topics/globus-m2-tokamak