Time-resolved measurement of optical emission line profiles from electron cyclotron resonance ion source plasma

Optical emission spectroscopy provides a non-invasive method to probe the properties of hot and highly charged magnetically confined plasmas. The optical emission line profiles enable, for example, to identify the different species and characterize the relative population densities and temperatures of the ions and neutrals forming the plasma. The feasibility of this approach has been demonstrated at the University of Jyväskylä accelerator laboratory by measuring the light emitted by Electron Cyclotron Resonance Ion Source (ECRIS) plasma with a high-resolution spectrometer setup POSSU (Plasma Optical SpectroScopy Unit). In these previous studies, the emission line profiles were measured by scanning the desired wavelength range by rotating the diffraction grating of the spectrometer. This process is slow compared to many interesting plasma phenomena, thus limiting the applicability of the setup. Recently, POSSU has been upgraded by changing the light sensor from a photomultiplier tube to a position-sensitive imaging sensor. As a result, it is possible to measure simultaneously a 1 - 2 nm wavelength range, with a spectral resolution in the order of picometers, without moving the grating. This enables a time-resolved study of the optical emission line profiles. By turning the grating, the measured wavelength region can be chosen between 370 nm and 870 nm, which covers the visible light spectrum. The time-evolution of optical emission line profiles emitted from the JYFL 14 GHz ECRIS plasma, during shifting plasma conditions induced by changing the gas balance, has been measured to demonstrate this new capability. The time-evolution of temperatures and emission intensities of selected ion species, correlated with extracted ion beam currents, are presented.


Introduction
The conditions within highly-charged Electron Cyclotron Resonance (ECR)-heated plasma are sensitive to external perturbations, which limits the methods available for plasma studies.Therefore, to acquire a deeper understanding of the plasma properties, non-invasive diagnostic methods are needed for reliable analysis of the plasma.There are multiple different ways of studying plasma without influencing it [1,2], for example using the escaping ion and electron populations or through the different types of radiation emissions, such as bremsstrahlung and microwaves [3].Plasma also emits radiation in the visible part of the light spectrum, which enables the use of optical emission spectroscopy (OES) as a non-invasive diagnostic method.The optical emission spectrum contains a lot of information about the plasma, such as its composition, temperature, and density [4,5].OES has a long history as a plasma diagnostic method.However, it has only recently started to be utilized more in ECR ion source plasma research, see e.g.[6,4,7,8,5].In the University of Jyväskylä accelerator laboratory a high-resolution spectrometer POSSU (Plasma Optical SpectroScopy Unit) has been used to (i) probe the temperature of the cold electron population [4], (ii) demonstrate the disassociation between the confined and extracted ion populations [7] and (iii) observe the thermal imbalance between the temperatures of the different ion species [5].In these studies, to measure an emission line profile, the spectrum had to be scanned by changing the angle of the spectrometer's diffraction grating.To get reliable results, scanning requires that the emission line profile doesn't change during the measurement process.Therefore, the setup couldn't be used to study temporally changing profile shapes, which led to the need to upgrade POSSU.By changing the light sensor of the spectrometer to a position-sensitive image sensor, a range of wavelengths can be measured simultaneously without the need to rotate the grating.This opens the opportunity to study the temporal effects of plasma phenomena that affect the whole emission line profile, such as changes in the ion temperatures.
This paper presents the upgraded POSSU and its capabilities in studying temporally changing plasmas.To demonstrate the new capabilities, the time-evolution of selected emission line profiles, along with extracted beam currents, from argon-oxygen ECR ion source plasma were studied during shifting plasma conditions induced by changing the balance of the two gases.

Upgraded Plasma Optical SpectroScopy Unit
As the main part of the upgrade, the light sensor of the high-resolution Fastie-Ebert type spectrometer POSSU [4] was changed from a photomultiplier tube (PMT) to a position-sensitive image sensor.The image sensor, Thorlabs CS2100M-USB, is attached to the output of the spectrometer, at the focal plane, with two linear stages enabling fine adjustment of the focus of the image measured by the sensor.As a result, the spectral resolution of the spectrometer is from single to tens of picometers depending on the wavelength of the light and the input aperture used.Moreover, the image sensor can distinguish sub-picometer changes in the width of the profile.The input fiber bundle of the spectrometer is optically coupled to the plasma through a radial pumping port of the JYFL 14 GHz ECR ion source [9] giving it a cross-sectional line of sight through the center of the confined plasma.The amount of light available from the plasma is the main factor limiting the measurement speed, and thus the temporal resolution that the system can achieve.With enough light the system can measure profile changes occurring in millisecond timescales.A custom software was also developed for spectrometer control and data analysis.
The analysis of the data acquired by the new image sensor starts by extracting the emission line intensity profile from the image data consisting of a matrix of intensities measured at different spatial positions of the sensor.The spatial dimension of the emission line intensity profile is then transformed to wavelength using a function derived from the geometry of the spectrometer and the grating equation [10].This overall line profile P, measured by the sensor, is the convolution of the emission line profile from the plasma and the instrumental line profile of the spectrometer [10].The shape of the emission line profile can be affected by multiple different mechanisms, such as Doppler and collision broadening, as discussed in Section 2 of Ref. [5].These mechanisms give typically either Gaussian G (λ; σ) or Lorentzian L (λ; γ) profile shapes, where σ is the Gaussian standard deviation, γ the Lorentzian scale parameter and λ the wavelength of light.When mechanisms contributing to both shapes are present, the resulting profile is called the Voigt profile V (λ; σ, γ) = G (λ; σ) * L (λ; γ).The instrumental line profile is a convolution of the different profile functions related to the optics of the spectrometer.In POSSU, the contribution from the 100 µm diameter input fibers dominate the instrumental line profile, and can be approximated with a profile function S ent (λ, W ) = (W/2) 2 − λ 2 .Here, W is the width of the image cast by the input fiber on the sensor in the spectral direction.The scale parameters σ and γ of the emission line profile can be obtained by fitting P = V (λ; σ, γ) * S ent (λ; W ) to the measured emission line intensity profile.In this study the contribution from the other optical components to the instrumental line profile is corrected using a calibration measured from well-known optical emissions of an Ar-ion laser in thermal equilibrium.This also provides a wavelength-dependent error estimate for the measured Gaussian standard deviation of the emission line profile.
The most prevalent mechanism in minimum-B ECR ion source plasma affecting the shape of the optical emission profile is Doppler broadening.As it's caused by the Maxwellian velocity distribution of the plasma particles, it has a Gaussian profile shape which standard deviation σ D is related to the temperature of the emitting particle species.This relation is of the form where k B is the Boltzmann constant, T i and m i the temperature and mass of the ion species, c the speed of light and λ 0 the rest wavelength of the emission.Also, the maximum optical emission intensity of a profile is related to the population density of the emitting ion species, the cold electron density, and the electron energy distribution function.

Demonstration measurement
To demonstrate the new capabilities of the upgraded POSSU, oxygen-buffered argon plasma was studied during shifting plasma conditions induced by changing the balance of the two gases.The JYFL 14 GHz ECR ion source was first optimized for Ar 9+ ion beam current such that the partial pressures of argon or oxygen gases could be varied in a wide range without triggering plasma instabilities.The gas-calibrated partial pressure of argon was measured to be 1.65 • 10 −7 mbar and of oxygen 2.90 • 10 −7 mbar at the optimum, and the ion beam was extracted with 10 kV acceleration voltage.
To study the effects of the changing plasma conditions, both high and low-charge state ion species were selected to be measured from the confined (optical emission) and escaping (beam current) ion populations.For optical emissions, the species selection is limited by the measurable wavelength range of the spectrometer and the emission intensity of the available transitions.As such, optical transitions of O + (441.490nm), Ar + (487.986nm), Ar 9+ (553.326nm), Ar 10+ (691.688nm) and Ar 13+ (441.256nm) were selected [11].Ideally, the ion beam currents would be measured for the same species as the optical transitions.However, the mass spectrometer can't deflect the lowest charge states enough, for them to be measured, due to the 10 kV acceleration voltage.As a compromise, O 3+ was measured instead of O + , and Ar 6+ instead of Ar + .Also, the argon ions whose mass-to-charge ratio overlaps with oxygen were avoided (Ar 5+ and Ar 10+ ).
Two sets of measurements were performed.In the first set, the partial pressure of the oxygen buffer was increased by about a factor of three from Ar 9+ optimum while keeping the other ion source settings unaltered.After the plasma had stabilized, the oxygen gas valve was slowly closed over a 10 min period, during which light emission, and mass-over-charge separated beam current, from a chosen ion species were measured simultaneously.A 1 s exposure time was used in the image sensor and the ion beam current was measured using a Faraday cup.This measurement was repeated for all the ion species mentioned above.The ion source drain current I D (related to the total extracted beam current), the bias disk current I BD and the plasma chamber pressure p PC (with plasma) were also recorded over each measurement.The microwave and bremsstrahlung radiation emitted by the ion source were monitored to ensure the ion source plasma remained stable during the measurements [3].In the second measurement set the process described above was repeated by varying the argon partial pressure while keeping the oxygen at Ar 9+ optimum.
The ion temperatures and the optical emission intensities were analyzed from the optical emission data using the method outlined in Section 2. The temperature and intensity data, along with the measured ion beam current, was then cleaned using 5 s rolling average, for which a 95 % confidence interval was calculated.The ion temperatures were corrected using the calibration measured from Ar-ion laser, and its error estimate was obtained by pooling the errors from the measurement and the calibration.The obtained optical and ion beam intensities were normalized with the maximum value of each measured transient.Ion temperature data is not presented for regions where low signal levels prevented reliable analysis.

Results and discussion
The results obtained from decreasing the oxygen buffer gas partial pressure are presented in Figure 1 on the left side, and from decreasing the argon gas partial pressure on the right side.
As plasma chamber pressure decreases, in both of the measurement sets the optical emission intensities I OE reach maxima in the order of increasing charge state.Similar behavior is also seen in the measured ion beam currents I FC , indicating that the plasma conditions shift to prefer the production of higher charge states when the total pressure decreases.The measured optical emission intensities are affected not only by the population density of ions, but also by the electron population density and the electron energy distribution.Currents I D and I BD also decrease over the measurement duration, implying that the plasma density decreases.The maxima of the optical emission intensities occur at different times compared to the maxima of the measured beam currents of the same ion species.This is particularly clear with Ar 9+ .This difference demonstrates that the confined and escaping ion populations are not fully correlated.When the oxygen buffer partial pressure is decreased, an instability region affecting all measurement signals is observed around the t = 50 s − 70 s region.During this time interval the plasma alternates between two different conditions.The region starts with condition preferring the lower charge states (Ar + , Ar 6+ , O + and O 3+ ), and stabilizes into preferring the higher charge states (Ar 9+ , Ar 10+ and Ar 13+ ).Based on the bremsstrahlung and microwave diagnostics this behavior is not due to kinetic plasma instabilities.Similarly, when the argon partial pressure is decreased the plasma undergoes multiple fast step-like changes both in optical emission intensities and ion beam currents.During these changes, the optical emission intensities of the lower charge states decrease, and higher increase.This suggests that the plasma needs to reach certain threshold conditions before high charge state ion population density can increase.These results demonstrate the capability of POSSU to measure the fast changes occurring in the plasma.
As the injection rate of either gas is decreased, the Ar 9+ and Ar 10+ ion temperatures also show a gradual decrease.This contradicts the generally understood effect of gas mixing, which is believed to cool down the heavier ions of plasma due to the mass difference between the gases in ion-ion collisions [12].The decrease in temperatures may be caused by changes in the ion confinement, which leads to the hot ion population escaping, thus effectively decreasing the temperature of the confined populations as the pressure decreases.However, these are just preliminary results and the behavior will be explored further in future studies.The temperatures of Ar 9+ and Ar 10+ ions were measured to be significantly higher than the rest, which may be a result of better electrostatic confinement of the high charge states due to the potential dip [12].However, it's unclear why this doesn't apply for the significantly colder Ar 13+ , which underlines the need for further studies of high charge state ion temperatures.
The results above demonstrate that the upgraded POSSU works well and is ready for research use, already providing encouraging results for further study.With the new position-sensitive image sensor, POSSU can concurrently measure the full emission line profile enabling the timeresolved study of plasma transitions not possible with PMT-based The upgraded capabilities of POSSU will enable the study of many fast plasma phenomena that have not been accessible before, with sufficient light even on a millisecond timescale.In the future, POSSU will be used to study the plasma conditions, especially the time-evolution of the ion temperatures during plasma breakdown and decay, or during kinetic plasma instabilities, to understand better the different ionization, confinement, and ion heating processes.

Figure 1 :
Figure 1: Measured ion temperature k B T i , normalized optical emission intensity I OE , normalized ion beam current I FC , drain current I D , bias disk current I BD and plasma chamber pressure p PC .The partial pressure of the oxygen gas was decreased on the left side figures, and the argon gas on the right side figures.Time t = 0 corresponds to the moment when the gas valve was set to close.Note that the timescale changes at the Ar 9+ optimum.The colored area around the rolling average, denoted by the solid line, represents the 95 % confidence interval of the measurement.In the left figure the peak beam currents I FC O 3+ 157 µA, Ar 6+ 19 µA, Ar 9+ 128 µA and Ar 13+ 40 µA, and in the right figure O 3+ 105 µA, Ar 6+ 59 µA, Ar 9+ 151 µA and Ar 13+ 28 µA.