The Laboratory Astrophysics Programme at Imperial College London

Advancements in ground- and space-based telescopes have resulted in an acute need for and improvement in the breadth and accuracy of the atomic data vital for the analysis of astronomical spectra. Many data, even for astrophysically important elements such as the iron group, have analyses dating back over 50 years, which are no longer suitable for the interpretation of modern, high-resolution astrophysical spectra. The Imperial College Spectroscopy group is addressing this need by measuring new atomic data and improving the accuracy of existing data using Fourier transform spectroscopy from the infrared to the vacuum ultraviolet. Analysis of these high-resolution spectra has led to new and improved atomic data for a wide range of astrophysically important elements. Reductions in uncertainties of transition wavelengths and energy levels are routinely over an order of magnitude and transition probabilities are measured to a few percent of uncertainty. Our work has increased both the quantity and quality of available atomic data. In this paper, we provide an update of our recent investigations and research plans. We also detail the improvements being made to our world-class laboratory to ensure we are able to meet the future data needs of the astrophysical community.


Introduction
In recent decades, astronomical telescopes and their spectrographs have undergone significant improvements in capability, capturing high-resolution spectra from the infrared (IR) to the vacuum ultraviolet (VUV, λ < 200 nm) regions. Current generation telescopes include the Near-Infrared Spectrograph (NIRSpec) on JWST [1], the High Resolution Echelle Spectrometer (HIRES) observing in the visible at the Keck Observatory [2], and the Space Telescope Imaging Spectrograph (STIS) on the HST [3] which records UV-VUV. With multi-object spectrographs on many large telescopes (e.g., WEAVE [4], 4MOST [5]), the spectra of multiple astronomical sources can be recorded within a single exposure, enabling current and future spectroscopic surveys (including APOGEE [6], Gaia-ESO [7] and GALAH [8]) to collect large numbers of highresolution spectra and measure high precision stellar and Galactic chemical abundances [9].
Florence Concepcion and Christian P. Clear have contributed equally to this work. a e-mail: florence.concepcion@imperial.ac.uk (corresponding author) b e-mail: christian.clear@imperial.ac.uk (corresponding author) However, the advancements of telescopes have resulted in astronomical spectra now more accurate than large amounts of the laboratory atomic data needed for their analysis and have created an everincreasing demand for more accurate atomic data [9][10][11][12]. As technology continues to advance, future telescopes such as the Extremely Large Telescope (ELT) [13] will only increase these demands.
To ensure that observations, made at great effort and expense by multi-million dollar instruments, can be analyzed to their full potential, the atomic data vital for the interpretation of their spectra are in urgent need of expansion and improvement in accuracy. Even for astronomically important element groups, such as the iron-group, many of the latest data were measured more than 30 years ago with lower-resolution techniques than are available now.
The Imperial College (IC) Spectroscopy group is measuring new atomic data and increasing the accuracy of available laboratory data using Fourier transform (FT) spectroscopy. FT spectrometers can observe from the IR to the VUV, with higher resolutions and accuracies than most current telescope spectrographs. This broad spectral range, allowing for measurements of the thousands of spectral lines in complex atomic spectra, combined with its high resolving power makes FT spectroscopy an excellent technique for the measurement of atomic data for astronomical applications. The successful analyses of astronomical spectra rely on accurate atomic data across the entire spectral region, from the VUV to the IR. While there is now a focus on IR measurements with telescopes such as JWST, laboratory-measured data are needed for spectral lines across the entire spectrum. For example, highly-redshifted objects such as quasars will have spectral lines that may be emitted in one spectral region and observed in another [14].
To expand the capabilities of the world-class facility at IC, we have received investments, including from the STFC, to fund a new FT spectrometer, optimized for emission spectroscopy in the IR. The new instrument will complement our existing VUV FT spectrometer, which holds the record for the shortest wavelength measured using FT spectroscopy [15]. The IC Spectroscopy group now has the capability to measure across the entire IR-VUV spectral range at the high resolution and accuracy required for modern astronomical analyses. The new instrument is discussed in more detail in Sect. 2.2.
In this paper, we will discuss the recent contributions of the IC group and our future plans for improving the atomic database.

Current experimental setup
The IC Spectroscopy group uses high-resolution spectroscopy of laboratory sources to determine fundamental atomic data such as transition wavelengths, energy level values, transition probabilities and line broadening effects. We generally use plasma discharge sources where radiation is emitted from a plasma generated inside of a lamp. We use two main discharge sources to produce the desired radiation: a hollow cathode lamp (HCL) or Penning discharge lamp (PDL) [16].
Within the lamps, a large current (typically 0.5-2 A) is used to ionize a carrier gas (usually a noble gas) and accelerate gas ions into the surface of a cathode made of the material under study (e.g., iron). The cathode material is sputtered into the lamp plasma where it is excited, emitting radiation. The HCL and PDL both use this method of radiation generation, but the PDL employs a static magnetic field to confine the plasma, enabling lower pressures to be reached and therefore higher ionization states. HCLs generally produce radiation from neutral and singly ionized species while PDLs are able to reach singly and doubly ionized states [16,17].
Lamp currents, carrier gas species, and gas pressure are all carefully optimized to obtain the greatest intensity of spectral lines of the target element and ionization stage, while avoiding self-absorption (where the plasma becomes optically thick and radiation from strong lines is reabsorbed by the plasma, reducing line intensities).
Commercial lamps usually run at low currents and have fixed gas pressures, so to allow for greater variation of lamp conditions, IC uses custom HCL and PDLs, both water-cooled to reduce the Doppler widths of the emitted radiation. When conditions have been optimized, these lamps run very stably for long periods of time.
To study the light generated by our lamp sources, IC uses FT spectroscopy. A detailed description of this technique can be found in [18,19]. In previous decades, atomic data were measured extensively using grating, or even prism, spectrometers. The accuracy of the majority of atomic data produced by these techniques are not sufficient for modern astrophysical analyses. FT spectroscopy has a large number of advantages which makes it an ideal technique for measuring atomic data at high accuracy, including: • High resolution-The resolving power of an FT spectrometer is determined by the maximum path difference of the instrument. This is typically more than an order of magnitude greater than the very largest grating spectrometers. A high resolving power means individual lines can be fully resolved and line wavelength uncertainties reduced. • Broad spectral range-FT spectrometers are able to record spectra across a wide wavelength region, an advantage over line-by-line techniques when dealing with atomic spectra comprising tens of thousands of spectral lines. • Linear wavenumber scale-Fitted spectral line wavelengths can be calibrated to an absolute scale with very few calibration standard lines [20]. This is another large advantage over grating spectra which, being dispersive, are not linear and require reference lines throughout each grating spectrum for calibration, increasing the uncertainty of grating spectral line wavelengths. • Slowly varying intensity response-This allows radiometrically-calibrated continuum standard lamps to be used to generate instrument response functions allowing intensity calibration of FT spectra. This calibration is key to the branching fraction technique of determining transition probabilities [17].
Our VUV FT spectrometer measures spectra from the visible to the VUV, down to a lower wavelength limit of 135 nm (74, 000 cm −1 ) [11]. With a maximum path difference of 20 cm, the IC VUV FTS has a maximum resolving power of 2, 000, 000 at 200 nm (50, 000 cm −1 ). This is sufficient to fully resolve spectral lines with widths of several hundredths of a wavenumber, typical for the Doppler widths of atomic emission lines. For comparison, the normal incidence vacuum spectrograph (NIVS) at the National Institute of Standards and Technology (NIST) [21], a world-class highresolution grating spectrometer, has a resolving power of ≈150,000, an order of magnitude less than our FT spectrometer. The increase in resolving power of FT spectroscopy and its effect on the resolution of spectral   [23] lines can clearly be seen in Fig. 1. The current experimental arrangement of the IC VUV FT spectrometer is shown in Fig. 2. Further details of the instrument and the FT spectroscopy technique are given in [22].
For VUV spectra beyond the wavelength limit of FT spectroscopy, we supplement our FT spectra with data from other sources, such the NIVS at NIST which has a spectral range of 14.6-438 nm (22,800-685,000 cm −1 ). We also use archival data where suitable, such as IRvisible spectra from the Kitt Peak National Observatory, and combine our measurements with theoretical calculations where appropriate, e.g., using theoretical level lifetimes for the calculation of transition probabilities where experimental data are not available.

New instrumentation
To extend the capabilities of the IC Spectroscopy group, we have recently purchased a Bruker IFS 125HR spectrometer, due to delivery mid-2023. The spectrometer is optimized for measurements from the mid-IR to the visible spectral regions, from 1850-25,000 cm −1 (400 nm to 5.4 μm). The new instrument will complement the existing UV and VUV capability of our IC-VUV-FTS spectrometer and will enable the group to measure intensity-calibrated spectra across the entire VUV to mid-IR spectral regions.
Several modifications to the standard Bruker IFS 125HR are planned to optimize the base instrument for atomic emission observations. The first modification will be the addition of photomultiplier tube (PMT) detectors mounted externally to the main FTS, which will be purged with N 2 for IR observations. PMTs combine ultra-fast responses and high sensitivity, making them ideal for emission measurements. The second modification will be a "fore-optics" chamber positioned externally to the FTS to allow motorized selection between the HCL/PDL and the intensity calibration standard lamps (W or D 2 ). A rotating selection mirror will direct radiation from these sources onto a concave focussing mirror which will ensure that the FTS input aperture is fully illuminated to reduce finite aperture error.

Atomic data needs
The accuracy of atomic data plays a crucial role in understanding astrophysical spectra. Most lines in stellar spectra are parts of blended features and require high-accuracy reference wavelengths to untangle them. Chemical abundance analyses rely on wavelengths, transition probabilities and line broadening data. Failing to consider all these atomic data, due to inaccurate or missing data, leads to incorrect conclusions or results with large uncertainties [24].
Theoretical atomic structure calculations use experimentally-measured atomic data to predict line wavelengths and transition probabilities. However, these predictions often lack accuracy, particularly for energy levels and transition wavelengths, or have large uncertainties and so calculations are not a substitute for experimentally-measured data [25]. However, calculations are a key technique for providing data where it would be impossible in the laboratory, for example in transition probabilities of parity-forbidden or very weak lines, which could not be produced in laboratory discharge sources. Calculations are also one of the tools used by experimentalists as a guide to the underlying atomic structure of their measurements, for example enabling the identification of potential new energy levels [26,27].

Wavelengths and energy levels
The wavelengths of spectral lines are measured to a high accuracy using spectra recorded by FT spectroscopy and then identified according to the upper and lower atomic energy levels of the corresponding transition. Identified lines are used to determine and optimize atomic energy level values in a process known as term analysis [28,29].
High-accuracy wavelengths are used directly by astronomers as standards for the calibration of astro-nomical spectra, and optimized energy level values are used to calculate Ritz wavelengths which generally have lower uncertainties than observed values. Ritz wavelengths are particularly important for lines which were not observed in the experimental spectra, such as weak and parity-forbidden lines, as they cannot directly be measured in a laboratory setting but are crucial to the study of low-density astrophysical plasmas [30][31][32].
The IC VUV FT spectrometer measures wavelengths and atomic energy levels to an accuracy of at least a part in 10 7 (30 ms −1 , 0.15 mÅ at 1500Å) [11], making it an ideal instrument for measuring atomic data for astrophysical applications.

Transition probabilities
Comparison between intensities of transition lines from a given upper energy level in intensity-calibrated FT spectra determine the branching fractions of those lines [9]. Transition probabilities are then calculated by combining these branching fractions with atomic level lifetimes, either measured experimentally or through atomic structure calculations. Transition probabilities for strong lines measured using FT spectroscopy are accurate up to 5 percent [33][34][35]. Further details about transition probabilities and their experimental determination are given in [9].
One major challenge with measuring experimental transition probabilities are that all reasonably strong lines from a given upper energy level must be measured to ensure that accurate branching fractions are determined. Especially for higher-lying energy levels, these transition lines can lie across a wide spectral range, from the IR to the VUV. Any experimental technique must therefore be able to produce intensity calibrated spectra across this range.

Hyperfine and isotope structure
Hyperfine and isotope structure splitting of energy levels occurs due to the interaction of electrons with the nucleus of the atom. Hyperfine splitting is observed in atoms with a non-zero nuclear magnetic moment, which interacts with electron spins. Isotope splitting arises due to differences in mass and charge distribution between the nuclei of different isotopes of the element. These small shifts in atomic energy levels cause spectral lines to split into multiple components. The study of these line broadening effects is important for high-resolution astrophysical analyses because in the blended features found in astrophysical spectra, these small-scale line splittings appear as a broadening of the spectral line. The broadening is often asymmetrical and must be accounted for in order to accurately calculate wavelengths and astrophysical chemical abundances.
FT spectroscopy is able to study these broadening effects thanks to its high resolution, which is capable of resolving individual hyperfine and isotope structure components, limited only by the Doppler widths of the lines themselves. An example of an intensity profile of

Atomic data accessibility
To communicate our latest atomic data to the astrophysicists, astronomers and theorists who urgently require it, our atomic data are included in atomic databases, such as NIST ASD [37], Kurucz [38] and VALD [39]. These databases may be directly used by software for the analysis of astronomical spectra. For example, the tardis [40] code utilizes data from CHI-ANTI, NIST ASD, and Kurucz databases to generate synthetic spectra for the analysis of supernovae and kilonovae [41,42].
Our spectral wavelengths are also used in the creation of linelists for many spectroscopic surveys. Most recently, the IC group provided atomic data to the Gaia-ESO Spectroscopic Survey for the study of Galactic Evolution [43,44], including the addition of our Co I HFS [45] and Fe I transition probabilities data [34] to the Gaia-ESO line list [25].

Recent work at Imperial College
The complex atomic structure of the iron-group (3d transition group) elements produce rich, dense spectra, which combined with their high relative abundance makes the iron group of critical importance in many astrophysical sources. As such, there is a pressing need to obtain precise atomic data for the iron-group elements. Thirty years ago, the only available atomic data for the iron-group was measured using grating spectroscopy, which produced data at lower resolutions and accuracies than FT spectroscopy. Since then, the IC Spectroscopy group and its collaborators have been working on producing experimentally-measured atomic data for many neutral, singly and doubly ionized irongroup species. The progress of investigations into irongroup elements until 2019 is reported in [11]. This paper focusses on our recent results and current investigations into wavelengths, energy level values, transition probabilities and line broadening data for the iron group.
Following the first detection of gravitational waves from neutron star mergers, accurate atomic data for the lanthanide and actinide groups have gained much greater importance and are now an area of increased focus for the IC Spectroscopy group. The rare-earth elements (as they are known) are essential in the study of chemically peculiar stars [46] and neutron star mergers [41,47,48]. However, the study of the first recorded kilonova due to a neutron star merger, AT2017gfo, has been limited by the availability of lanthanide atomic data [41,42,48]. As there are very little previously experimentally-measured atomic data for some of these elements, we have begun investigating the lanthanide group, currently focussing our efforts on doubly ionized neodymium (Nd III).
In addition to our focus on the emission spectroscopy of the iron group and rare-earth metals, the IC spectroscopy group is also involved with the measurement and analysis of other data. Some of these projects include studies on the possible time variation of the fundamental constants, high-resolution FT molecular spectroscopy and the industrial applications of discharge sources.
All these projects are detailed further in the following sections. It should be noted that many of these projects involve collaboration, particularly with the atomic spectroscopy groups at NIST (USA) and Lund University (Sweden).

Wavelengths and atomic energy levels
The following sections detail our recent work producing high accuracy transition wavelengths and term analyses, optimizing previously reported energy level values and discovering new energy levels.

Vanadium
During the term analyses of V I [49] and V II [50], highresolution spectra of a vanadium HCL were recorded using FT spectrometers at IC (visible-VUV) and at the Kitt Peak National Observatory (IR-visible). Although thousands of spectral lines were classified as vanadium transitions at the time, there remained an unpublished linelist of unidentified spectral lines.
Using optogalvanic spectroscopy and laser-induced fluorescence spectroscopy, the unidentified lines in the visible-IR vanadium spectrum were investigated and confirmed to be V I transitions. These identified lines then led to the discovery of four previously unknown even, fine structure energy levels of V I and the classification of nine previously unidentified lines. This research has been submitted for publication [51].

Manganese
The term analysis of neutral manganese (Mn I) is underway. Spectra of manganese-neon and manganese argon HCL and PDLs have been recorded by FT spectrometers at IC in the visible-VUV region and at NIST in the IR. At least, 1284 lines have been newly classified as Mn I transitions following this analysis and new energy levels are currently being found and optimized.
Using FT and grating spectroscopy, Mn II spectra were recorded over the range 82-5500 nm (1820-121,728 cm −1 ). Analysis of these spectra resulted in the identification of 6019 Mn II lines [28]. A comprehensive term analysis has reduced the uncertainties of wavelengths and energy levels by at least an order of magnitude compared to previously published values. 57 Mn II energy levels, previously identified solely through observation of stellar spectra, were verified and improved in accuracy and 52 entirely new Mn II energy level values were determined. The availability of this data will facilitate a more reliable analysis of Mn II spectral lines in astrophysical spectra.
Using the results of the term analysis, Ritz wavelengths and transition probabilities for 1130 parityforbidden Mn II lines were derived in the range between 237 nm and 170 μm (42,125-58 cm −1 ) [31]. These results are essential in the investigation of dilute astrophysical plasmas.

Iron
Extending our work on the doubly-ionized iron-group elements [52,53], the IC group has recorded the first UV-VUV spectra (33,826-65,508 cm −1 , 153-296 nm) of Fe III using FT spectroscopy. Line wavenumbers were calibrated to an accuracy of 5-8.5 parts in 10 8 , with wavenumber uncertainties for strong lines within the range of 0.0024-0.0074 cm −1 . Short wavelength grating spectra (55,090-111,121 cm −1 , 89.9-181.5 nm) measured using the NIST NIVS supplemented the FT data by extending observations to higher wavenumbers and weaker lines. These combined spectra enabled the term analysis of the lower-lying atomic energy levels of Fe III.
To date, this term analysis has resulted in the optimized values of 305 Fe III energy levels in the configurations 3d 6 , 3d 5 4 s and 3d 5 4p, resulting in 1599 Ritz wavenumbers, which are at least an order of magnitude more accurate than the best previously published values [54,55]. The results of this analysis have led to 456 Ritz wavelengths that can be used directly as wavelength standards for the calibration of VUV spectra of hot stars, the re-assignment of 12 atomic energy level labels, and the calculation of Ritz wavenumbers for 432 parity-forbidden lines. Papers detailing these results are in preparation for publication.
An expansion of the term analysis of Fe III is planned to include energy levels with configurations 3d 5 4d and 3d 5 5 s. As the wavelength range for the transitions from these levels to the previously-optimized 3d 5 4p levels is outside the IC VUV FT spectrometer's range, grating data will be used to determine and optimize these energy level values. A suitable grating linelist has been prepared using the data recorded on the NIST NIVS, with some Fe III lines already identified as transitions from the 3d 5 4d, 3d 5 5 s, and 3d 5 5d configurations. The results of this investigation will also be used by astronomers for the identification and analysis of Fe III spectral lines in astrophysical spectra.

Cobalt
Since our major analyses of Co I and Co II using FT spectroscopy [26,27,56,57], we had intended to continue the search for new high-lying levels in Co II. Grating spectra of a cobalt-neon HCL have been recorded using the NIST NIVS. Lines have been identified as Co I and Co II by comparison to Ritz wavelengths calculated using energy levels determined from FT measured data. Analysis of the remaining lines is ongoing to identify potential new cobalt energy levels.

Nickel
High-resolution FT spectra of singly ionized nickel (Ni II) in the range 143-5555 nm (1800-70,000 cm −1 ) were recorded at IC and NIST or obtained from archival spectra recorded at the Kitt Peak National Observatory. These spectra were used to measure and identify 1424 Ni II transitions. Following an extensive term analysis, 283 energy levels, previously reported by [58], have been confirmed and optimized, with their uncertainties reduced by at least one order of magnitude [29]. 25 new energy levels were also identified for the first time. The Ni II data reported will enable more accurate and reliable analyses of Ni II in astrophysical spectra.
Using the energy levels established with highresolution FT spectroscopy, the Ritz wavelengths and transition probabilities for 126 parity-forbidden Ni II transitions have been calculated with uncertainties significantly lower than previously published values, by up to two orders-of-magnitude [30].
Extension of the term analysis to the higher-lying energy levels of Ni II has recently been completed. 1016 lines were identified as Ni II transitions from energy levels the 3d 8 ( 3 F ) 5f to 3d 8 ( 3 F ) 9s configurations. Term analysis has confirmed and optimized 208 previously reported energy levels with reductions in uncertainty by at least an order of magnitude. Many of the newly-identified transitions lie in the IR region and enabled the establishment of 33 completely new energy levels. Using highly-excited levels of the 3d 8 ( 3 F ) 5g, 3d 8 ( 3 F ) 6g and 3d 8 ( 3 F ) 6h configurations, a new ionization energy for Ni II has also been derived. These results are currently in preparation for publication.
Grating spectra of a nickel-helium HCL have also been recorded using the NIST NIVS, to extend the spectrum of Ni II beyond the FTS lower wavelength limit. Lines in this region will connect already measured energy levels of the singly excited system of Ni II to those of the doubly excited system. Calibration of the grating spectra has been completed using Pt standard lines and Ni II Ritz wavelengths from the FT spectroscopy term analysis. The grating lines are currently being analyzed to identify transitions and to determine the energy levels of the doubly excited system.

Neodymium
The spectrum of Nd III has been recorded using the IC VUV FT spectrometer in the region 11,500-62,000 cm −1 (8695-1613 nm) using both a HCL and, for the first time, a PDL. As a heavy element, the spectrum of neodymium is extremely dense, with almost 21,000 lines across the VUV-visible spectral range observable in the FT spectra. To date, 568 lines have been identified as Nd III transitions and have enabled the determination of 180 atomic energy levels with uncertainties of a few parts in 10 8 . Previously, only 35 experimentallymeasured atomic energy levels had been published for the entirety of Nd III [46]. Term analysis to identify further transitions and energy levels is ongoing. The newly identified Nd III lines are also being seen in stellar spectra.

Neon
The NIST NIVS was used to record grating spectra of several transition metals and neon HCLs. Comparison of the measured transition lines in these spectra and with spectra measured using other carrier gases (helium, argon etc.) has produced a linelist of probable neon transitions. These lines are undergoing analysis to determine new energy levels of neon.

Transition probabilities
The following sections detail our recent and current work measuring transition probabilities using the branching fraction method.

Manganese
The measurement of transition probabilities for Mn II is planned for the summer of 2023. Spectra of manganeseneon HCLs from the IR to the UV have already been recorded at IC and at NIST using FT spectrometers.
The spectra were intensity calibrated using radiometric deuterium and tungsten standard lamps. Measurements of line intensities from the calibrated spectra will be used to determine Mn II branching fractions which will be combined with calculated level lifetimes to produce transition probabilities. These new data will provide astronomers with more important tools for the study of early-type stars, where low ionization stages of the iron group are prevalent. In particular, Mn II is an important species in the study of late B-type, chemically peculiar, HgMn stars, where lines are sharp due to the low rotation velocities and weak magnetic fields of the stars [59]. High-accuracy reference wavelengths for these lines are essential to study these sharp lines.

Nickel
New intensity-calibrated FT spectra of a nickel-helium HCL have recently been recorded in collaboration with NIST. The analysis of these spectra is underway to produce branching fractions for a large number of Ni II lines from levels of the 3d 8 4d configuration. These branching fractions will be combined with experimental upper level lifetimes measured by our collaborators at Lund University [60] where possible and with calculated values from the recent work of Clear et al. [29] elsewhere. As the iron-group element with the second greatest solar abundance, nickel is an important element in many astrophysical bodies and these new data will offer new avenues for detailed analysis of these objects of interest, including supernovae [61], Galactic planetary nebulae [62], and quasi-stellar objects [63].

Scandium
In collaboration with Lund University, spectra of our water-cooled HCL were measured in the region 23,500-63,100 cm −1 (158-426 nm) [64]. The lamp cathode was an iron cylinder into which a small piece of scandium was placed. This cathode configuration was chosen due to the difficulty of machining scandium into a cathode cylinder itself. The spectra were intensity calibrated using deuterium and tungsten standard lamps. Branching fractions for 57 transitions were determined from the intensity calibrated spectra and combined with level lifetimes measured using laser-induced fluorescence spectroscopy to produce transition probabilities for these transitions. The transition probabilities in this work were used to calculate a new solar scandium abundance, bringing it into agreement with the established meteoric value.

Cobalt
The hyperfine structure (HFS) constants of 292 energy levels for singly ionized cobalt (Co II) have recently been measured [36]. The spectra of cobalt were recorded using FT spectroscopy at IC for the visible-UV region and at National Solar Observatory (US) for the infrared region. Over 700 spectral lines were fitted and used to determine 292 magnetic dipole hyperfine interaction A constants. The electric quadrupole hyperfine interaction B constant was also estimated for one energy level.
This analysis represents a large increase in the knowledge of HFS in Co II, as there were previously only A values available for 28 Co II energy levels [65,66] which provided unexpected results in studies of cobalt stellar abundances. Using our newly determined HFS A constants, Roederer et al. [67] found that the derived Co I and Co II abundances behaved as expected and recommended the Co II data as the best cobalt abundance indicator in their study.

Studies of the variability of physical constants
The possible time variation of the fundamental constant, the fine structure constant, α, at high redshifts can be probed by comparing the laboratory-measured wavelengths of atomic lines to their observed wavelengths in quasar absorption spectra. Eleven Ni II lines were identified as key data to probe this possible time variation [68]. Seven of these lines were previously measured using grating spectroscopy [69] and the remaining using FT spectroscopy by Pickering et al. [70]. However, thanks to our recent Ni II term analysis [29], the uncertainties of these key wavelengths have been further reduced, by at least an order of magnitude for the previously measured grating lines and by half for the FT-measured lines (due to improvements in experimental conditions) [14].

Atomic excitation mechanisms
Common excitation mechanisms can be studied by comparing the high-resolution emission spectra of various elements from a Grimm-type Glow discharge (GD [71]) [72][73][74]. For these studies to be as complete as possible, observed spectral lines must be resolved in intensitycalibrated spectra over a broad range of wavelengths, ensuring that the majority of the radiative deexcitation paths for the species under study are observed. Currently, only a high-resolution visible-VUV FTS spectrometer is suitable for this type of study, meeting all the highly demanding requirements [73].
In a study of charge transfer excitation mechanisms, the high-resolution emission spectra from a GD was measured for the elements Mn, Fe, Ti, Cr and Cu in a neon discharge on the IC VUV FT spectrometer in the range 160-600 nm with radiometric standard tungsten and deuterium lamps used to calculate relative spectral line intensities. These results [73] were the first to highlight the involvement of the charge transfer mechanism in a GD. In another recent investigation to compare excitation mechanisms [74], the differences in plane cathode and hollow cathode GD with titanium and iron in argon were studied using spectra recorded on the IC VUV FT spectrometer at 160-600 nm. This is the first extensive comparative study to understand the differ-ences in plane cathode and hollow cathode excitation mechanisms in GD spectroscopy [74].
This research into GD excitation mechanisms is a broad discipline with numerous applications in diverse fields, including analytical chemistry [75], plasma deposition processes [76], hollow cathode metal ion lasers [77], and various other areas of plasma science and technology.

Sulfur dioxide
Although at present the IC spectroscopy group is focussing on atomic emission spectroscopy, our instrumentation is also very suitable for molecular absorption measurements. Most recently, the absorption spectrum of sulfur dioxide (SO 2 ) was recorded in the VUV region to investigate pressure-broadening effects which can be used to place constraints on processes in early Earth's (pre-oxygenated) atmosphere [78].

Summary
The Imperial College Spectroscopy group is continuing our work measuring high accuracy atomic data. We are a major contributor to many atomic databases and our atomic data are not only used directly by astronomers, but also in the creation of astrophysical atmospheric and plasma models and for advancing atomic structure physics and quantum mechanics. The latter of these uses our contributions to improve atomic structure calculations. To ensure that we are able to maintain a world-class facility and meet the increasing demand for IR atomic data, we are installing a new FT spectrometer specifically optimized for IR work. We are focussing our efforts on the elements most urgently requested by the astronomy community. At present, these are primarily from the iron group and rare earth group elements.
We are always open to atomic data requests and would encourage astronomers and astrophysicists to contact the group for further discussions. Bequest of Prof. Edward Steers, and the Physics Department of Imperial College London for financial support for our new IR-visible FT spectrometer this year. We acknowledge Grants in support of the research in this paper from the STFC (UK), the Royal Society, the EU Research Training Network GLADNET, the Leverhulme Trust, and the bequest of Prof. Edward Steers.

Author contributions
FC and CPC: writing (original draft preparation). JCP: funding acquisition and writing (critically reviewed and edited). All authors contributed to the literature research.
Data Availability Statement This manuscript has no associated data or the data will not be deposited. [Authors comment: This manuscript provides an overview of recent research and so has no associated data. Datasets created in the research mentioned in this paper can be found at the relevant references.] Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecomm ons.org/licenses/by/4.0/.