Broad band high resolution rotational spectroscopy for Laboratory Astrophysics

Abstract

We present a new experimental setup devoted to the study of gas phase molecules and processes using broad band high spectral resolution rotational spectroscopy. A reactor chamber has been equipped with radio receivers similar to those used by radio astronomers to search for molecular emission in space. The whole Q (31.5-50 GHz) and W bands (72-116.5 GHz) are available for rotational spectroscopy observations. The receivers are equipped with 16×2.5 GHz Fast Fourier Transform spectrometers with a spectral resolution of 38.14 kHz allowing the simultaneous observation of the complete Q band and one third of the W band. The whole W band can be observed in three settings in which the Q band is always observed.

Species such as CH₃CN, OCS, and SO₂ are detected, together with many of their isotopologues and vibrationally excited states, in very short observing times. The system permits automatic overnight observations and integration times as long as 2.4×10⁵ seconds have been reached.

The chamber is equipped with a radiofrequency source to produce cold plasmas and with four ultraviolet lamps to study photochemical processes. Plasmas of CH₄, N₂, CH₃CN, NH₃, O₂, and H₂, among other species, have been generated and the molecular products easily identified by their rotational spectrum, and mass spectrometry and optical spectroscopy.

Finally, the rotational spectrum of the lowest energy conformer of CH₃CH₂NHCHO (N-Ethylformamide), a molecule previously characterized in microwave rotational spectroscopy, has been measured up to 116.5 GHz allowing the accurate determination of its rotational and distortion constants and its search in space.

1. Introduction

The study of the chemical, physical and dynamical evolution of molecular clouds, where stars and their planets form, is performed through the observation of the thermal emission or absorption produced by the rotational transitions of the molecules formed in these objects. Large single dish radiotelescopes or interferometers, equipped with very sensitive receivers, are used for these purposes.

Radio astronomical receivers permit nowadays broad instantaneous frequency coverage (several GHz) that can be fully analyzed with Fast Fourier Transform Spectrometers (FFTs) providing spectral resolutions as low as a few kHz.

Within the ERC project NANOCOSMOS¹ we proposed to use radio astronomical receivers as detectors for molecular spectroscopy and chemical reactivity experiments in a gas cell. Different experiments have been performed in the past based on similar ideas (Hüberts et al. 2006; Ren et al. 2010; Neumaier et al. 2014; Wehres et al. 2018a,b), in particular for the SWAS satellite for which an OCS gas cell was used (Tolls et al. 2004), the Herschel/HIFI receivers (Higgins et al. 2010; Higgins 2011; Teyssier et al. 2004), and the Swedish ODIN satellite that used a gas cell filled with H₂O (Frisk et al. 2003). Except the experiment of Wehres et al. (2018a,b), most of these experimental setups were focused on the pre-launch characterization of receivers onboard satellites and not on the possibility of using them for laboratory spectroscopy and chemical reactivity. A proof of concept experiment (Tanarro et al. 2018) was developed in the period 2015-2017 using a small prototype chamber placed in the beam path of a receiver covering the 42-50 GHz band installed on the 40 m ARIES radio telescope at the Yebes Observatory (Spanish National Geographic Institute; hereafter referred to as IGN). These experiments showed that radioastronomical receivers can be used to perform simulations (photochemistry, cold plasma discharges) addressing chemical compositions similar to those of interstellar clouds. Thermal emission provides, in addition, partial pressures derived from the intensity and linewidth of the observed lines in the same way radio astronomers do to derive volume and column densities when observing interstellar clouds. A comparison with other techniques, such as the chirped pulse Fourier transform microwave spectroscopy, has been done by Tanarro et al. (2018).

This paper describes the final gas cell experimental setup hereafter referred to as GACELA : GAs CEll for Laboratory Astrophysics). It has been designed to study the spectroscopy and chemistry of cold plasmas, with or without the action of ultraviolet (UV) photons, and permits to observe the Q and W bands (31.5-50 and 72-116.5 GHz) simultaneously using FFTs covering the whole Q band, and one selectable third part of the W band (the W band is divided in three different frequency ranges selected through a motorized waveguide switch). It is the first time that an instantaneous band width of 20 GHz in Q band and 3x20 GHz in W band can be used for laboratory spectroscopic experiments measuring the thermal emission of molecules. Mass spectrometry and optical spectroscopy are also performed simultaneously with the broad band rotational spectroscopy, providing a complete set of complementary measurements. Section 2 describes the whole experimental setup: the chamber, the vacuum system, the receivers, the spectrometers, the mass and optical spectrometers, the plasma source, the UV lamps, the data acquisition system, the calibration procedures, and the data analysis software. Section 3 is devoted to the results obtained for the different molecules used to characterize the GACELA experimental setup. Section 4 describes the results obtained in cold plasmas with different gas compositions. Finally, section 5.1 is devoted to the rotational characterization of N-Ethylformamide (CH₃CH₂NHCHO) up to 116.6 GHz and its search in space. Appendices A to C contain detailed descriptions of the system temperature, the data calilbration and overviews of the full spectrum of some of the species studied in this work.

2. The GACELA experimental setup

The GACELA setup is one of the experimental developments of the NANOCOSMOS project. A prototype chamber was built at the ICMM mechanical work-shop to test the concept. Results from this protoype have been published by Tanarro et al. (2018), together with detailed formulae describing the data calibration process (see also Appendix B), and how to derive physical parameters of the gas inside the chamber (partial pressures, temperatures, collisional broadening coefficients, etc.).

In this section we focus on the description of the final GACELA setup, its subsystems (chamber, vacuum pumps, broadband cryogenic HEMT receivers, spectrometers for millimeter-wave rotational spectroscopy, and optical and mass spectrometers), the analysis of the performance of the whole system, and its capability to do novel experiments on plasma physics, photochemistry and spectroscopy. We also address the spectroscopic characterization of several gases injected in the cell. The optical diagram of the GACELA setup is shown in Figure 1. Figure A.1 shows a design of the chamber. A picture showing the coil for the RF generator, which has been designed to surround the radio beam of the receivers, is shown in Figure A.2.

Fig. 1.

A diagram of the optical path of the receivers beam in the GACELA setup.

During the first commissioning experiments in February 2018 we were able to detect the rotational lines of methyl cyanide (CH₃CN) in a few seconds with a very high signal to noise ratio (S/N). The lessons learned during commissioning have been crucial to improve the system performance and to allow the second set of experiments that are described in sections 3, 4, and 5.1.

2.1. The chamber

The chamber consists of a stainless steel cylinder (304L) of 890 mm length and 490 mm diameter, and positioned with the long axis horizontally. Figure A.1 shows two views (top and bottom) of the original design of the chamber (built by SEGAINVEX²). The cylinder is closed by 30 mm thick plates that support the flanges for the microwave windows. Those flanges are tilted by 9 degrees, one in the vertical plane and the other in the horizontal plane, minimizing the possible interference fringes produced by multiple reflections between the two windows (see Figure A.1).

The flanges are of the knife-edge type, size DN 250 CF, large enough (inner diameter is 254 mm) to accomodate the microwave Gaussian beam at the lower operating frequencies. The cell has been designed with flexibility in mind, and has been fitted with multiple ports (also CF type) with sizes varying from DN 40 CF to DN 200 CF. Five ports are available on each of the top, bottom, left and right sides of the cylinder (see Figure A.1). The ports are used for different pressure/vacuum sensors, feedthroughs for electrical wiring, a manipulator for the handling of deposition substrates, radiofrecuency (RF) coupling, viewports, vacuum pumps connections, gas inlets and so on. Four more ports on the front side of the cell are provided for the coupling of ultraviolet lamps to be used in photochemistry experiments. Finally, two more additional ports are fitted in the back port. Although the pressures considered do not require Ultra high-vacuum (UHV) technology, we have chosen UHV compatible flanges for several reasons. On the one side to avoid all type of polymeric gaskets that may decompose by UV irradiation or corrosive gasses. Also, metal gaskets make easier baking out processes to clean the chamber after dirty experiments. On the other side, it helps to have less interferences with the background residual gasses, that can be easily eliminated after baking out. Finally, it makes possible to couple the chamber to new instruments or experimental set-ups, as it could be the Stardust machine (Martínez et al. 2018).

The pumping system is composed of two scroll pumps (with 30 m³/h pumping speed), one of which being used for rough pumping while the other backs a turbomolecular pump (with 950 l/s pumping speed) connected to the bottom of the cylinder through a gate valve. Typically, a final vacuum of ~ 2 × 10⁻⁷ mbar is achieved.

Several gauges are available to monitor the pressure inside the cell: Pirani/Penning gauges are used to monitor the vacuum from near-atmospheric pressure down to 10⁻⁷ mbar. When gases are loaded into the chamber a capacitive type gauge (whose reading does not depend on the gas being measured) is used, covering the range 10⁻⁵ − 10⁻¹ mbar with a nominal precision of 0.5 %. For several of the experiments shown in the present work, the chamber was internally lined with Teflon™ sheet (2mm thick), in order to decrease the molecular reactivity on the reactor walls. The gases are injected into the cell through one of the sideports to which a premixing chamber was attached. This premixing chamber has four inlets fitted with mass flow controllers, to allow for up to four different gases to be flown through the reactor. Two additional inlets can be used for liquid or solid samples. The chamber and the conductions attached to it are wrapped with heating tape, in order to decrease gas adsorption and avoid the possible condensation of low vapor pressure species on the walls. A general view of the complete experimental setup is displayed in Figure 2.

Fig. 2.

Side (left) and front (right) view of the whole GACELA system. The different components are indicated by arrows. They are described in sections 2.1, 2.2, 2.3, and 2.4. A diagram of the optics of the system is shown in Figure 1.

2.2. Cryogenic Front End Amplifiers

The cryogenic low noise Q and W band amplifiers (LNAs) of the GACELA receiver front-end are based on metamorphic GaAs millimeter-wave MMICs (Monolythic Microwave Integrated Circuits) developed by cooperation agreements of Yebes Observatory (IGN) with IAF³ and UC⁴ and fabricated with the 50 nm gate length IAF process (Leuther 2009). The Q band MMIC is a microstrip design incorporating four stages of 4×15 µm gate width HEMT devices (High Electron Mobility Transistors) which can be independently biased for optimizing the performance. The MMIC was mounted on a gold plated aluminum block containing soft substrate input and output microstrip access lines and DC bias components (Figure 3). The module can be configured either with coaxial 2.4 mm connectors or WR22 waveguide interfaces at input and output, although only waveguides were used in the final receiver version to minimize the loss. The W band MMIC incorporates four stages of 2×15 µm gate width HEMT devices with independent bias and coplanar waveguide matching networks. The MMIC is housed into a gold plated brass split-block which contains additional DC bias components (see Figure 3). The transition between the input and output WR10 waveguides and the circuit was implemented by E-plane probes optimized for the 71-116 GHz band and etched in a 50 µm quartz substrate.

Fig. 3.

Inner view of the Q band (left) cryogenic amplifier module showing the MMIC, input and output microstrip lines and bias conditioning board. MMIC size is 2×1 mm. Overall block dimensions are 40.8×25.7×10 mm including the waveguide transitions. Inner view of the W band (right) cryogenic amplifier split block showing a detail of the MMIC, quartz microstrip to waveguide transitions and DC bias protection components. MMIC size is 1.5×1 mm. Overall block dimensions are 21.5×21.5×19.1 mm.

The LNAs were thoroughly characterized in the laboratory and the bias points of the different stages were optimized for an adequate compromise between noise, reflection, gain, and flatness. Special attention was paid to the stability (absence of high frequency oscillations). Noise and gain were accurately measured using a variable temperature cryogenic waveguide load connected to the input (Malo et al. 2016), obtaining state-of-the-art average noise temperatures in the Q and W bands of 15 K and 30 K respectively. Noise and gain plots are shown in Figure 4. The main results of the measurements are summarized in Table 1.

Fig. 4.

Gain and noise temperature of the Q (top) and W (bottom) band cryogenic LNAs at 15 K physical temperature compared with the noise temperature obtained in the complete receiver.

Table 1. Measured performance of the cryogenic LNAs at 15 K physical temperature.

Amplifier Serial Number	Band (GHz)	Tn(av.) (K)	G(av.) (dB)	IRL(80%) (dB)	ORL(80%) (dB)	Pdiss (mW)
YMWN 2006	72-116	30.7	25.6	-6.3	-13.7	23.2
YMQ 2002	31-50	15.1	33.8	-5.1	-12.0	14.8

The values of noise temperature (T_n) and gain (G) are the average value in the band, while the values of the input and output reflection losses (IRL, ORL) are the worst case in 80% of the band. P_diss is the total heat power dissipation at the optimum bias value.

2.3. Optics, windows and cryostats

The dimensions of the chamber of GACELA allow the installation of corrugated Teflon™ windows with a beam clearance diameter of 240 mm. The windows have been appropriately grooved to decrease their reflectivity at both Q and W band frequencies. In the center of the cell, the optical system provides a gaussian-shaped half power beam-width of 54 mm and 40 mm for Q band and W band receivers, respectively. The beam diameter diverges from the center to the windows. Nevertheless, the goal of the optical design was to keep the theoretical truncation losses of the beam at each window below 0.1% (see Appendix A).

The optics is designed to allow an astronomical type ON/OFF observation and hot-cold load calibration. The optical system couples the radiation emitted by the background cold load, which interacts with the cell contents, to the feeds inside the cryogenic receiver. In the ON/OFF type observations the signals at the input of the receiver are a) the background cold load seen through the empty chamber (OFF-observation) and b) the background cold load seen trough the gases injected in the cell (ON-observation).

The coupling optics which guides the beams consists of four elliptical mirrors located on both sides of the gas-cell (see Figures 1 and 2). With this configuration the feed beams are perfectly terminated at the background cold load. The use of shaped mirrors allows the beams to propagate through the optical circuit with estimated spill-over losses of less than 1.5% in the whole system. An additional computer controlled mobile carriage at the receiver side allows the insertion of an ambient load or an elliptical mirror which allows re-focusing the beam into an independent cold load built into the receiver cryostat. This system is used to perform the receiver calibration. The cryogenic loads are very low reflection structures machined on ECCOSORB™ MF117 magnetically loaded epoxide material and the ambient temperature load is a piece of convoluted foam microwave absorber.

The front-end receivers and the internal cold load (hereafter referred to as ICL) are mounted in a cryostat cooled with a helium gas closed-cycle refrigerator. All components around the LNAs (polarizing grid, horn-feeds and waveguides) are cryogenically cooled down to 15 K in order to increase the radiometer sensitivity. Inside the receivers cryostat, the incoming beam is split into two orthogonal polarizations using a wire grid. One polarization is directed to the Q and the other to the W band horn feed. The background cold load (hereafter referred to as BCL) is mounted in an independent cryostat equipped with another closed-cycle refrigerator which is located at the opposite side of the chamber (see Figures 1 and 2). Most observations are performed using it as terminal load of the system (see section 2.10).

The total receiver noise temperature including the contribution of the backend and the optics but without the chamber is presented in Figure 4. This figure also shows the noise temperature contribution of the cryogenic LNAs used (see Table 1). Appendix A presents more details on the determination of the contribution of the different elements of the GACELA setup to the total system noise, T_sys, as well as the equivalent noise temperatures of the ICL and BCL. Figures A.3 and A.4 show the graphs of the these temperatures as a function of frequency for Q and W bands respectively.

2.4. The intermediate frequency downconverters and the spectrometers

The output of the receiver cryostat is connected by short waveguide sections to the ambient temperature downconverter units (Figure 5). The Q-band downconverter shifts the complete 31.5 - 50 GHz band to an intermediate frequency (IF) of 1 - 19.5 GHz by mixing with a computer controlled synthesized local oscillator (LO). Analogously, the W band downconverter accepts the complete 72 - 116.5 GHz frequency range, but it processes only one of the following sub-bands at a time: a) 72 - 90.5 GHz (WL), b) 82 - 100.5 GHz (WM) or c) 98 - 116.5 GHz (WH). A remotely controlled waveguide switch selects the specific sub-band to be converted to the 1 - 19.5 GHz IF. Image frequencies of the mixing process are effectively rejected by band pass filters. The frequency of the synthesizers used for LOs can be varied under computer control to make frequency switching type observations possible. A view of the implementation of the W band unit is shown in Figure 6.

Fig. 5.

A block diagram of the receiver system.

Fig. 6.

W-band downconverter for the GACELA receiver. The remotely controlled waveguide switch allowing the selection of one of three W sub-bands is clearly seen at the middle of the box.

The Q and W downconverter blocks provide an overall gain of 30-40 dB with noise temperatures in the 500-1500 K range. The 1-19.5 GHz IF signals from the Q and W bands are sent through low-loss coaxial cables to distribution units for further processing by an array of eight additional downconverters (see Figure 5). Each one shifts a specific 2.5 GHz slot of the 1 to 19.5 GHz input range to baseband, so the FFT spectrometers can be fed with a suitably conditioned signal. The frequencies of the LOs for the second down conversion are fixed and the gain of each unit is adjustable by 1 dB steps to optimize the power level at the digitizer input. The spectrum of the baseband signal is computed in an FFT module containing an analog to digital converter and a FPGA for real time signal processing. Each FFT module accepts a 0.1 - 2.5 GHz input signal and provides a spectrum with 65536 points which corresponds to 38.14 KHz of frequency resolution.

The FFT spectrometers were provided by Radiometer Physics, GmbH. Figure 2 shows the rack cabinet with the distribution modules (central position), the IF downconverters and the two FFTs blocks (at the top and the bottom positions). The arrangement of the modules is optimized to reduce the length of the connecting cables and to allow an adequate cooling air circulation. Note that, due to the particular system architecture, the Q band can always be observed together with one of the three selected W sub-bands. Hence, for the complete frequency coverage of the W band, the Q band will have three times more observing time than any of the three W sub-bands. The adopted blanking time to switch from one W sub-band to another is one second. All the LOs, frequency synthesizers and clocks of the FFT modules are phase locked to a 5 MHz central frequency reference standard obtained from the Yebes Observatory hydrogen maser, to ensure the accuracy of the frequency scale.

2.5. The control software

The radio receivers and backends, together with the auxiliary equipment are controlled from a central software program which mimics the one used at the 40m radiotelescope in the Yebes Observatory. The software performs two kind of operations: the control and configuration of the receiving system and data acquisition, and a first processing of the spectroscopic data and its storage. The control is done from a Linux PC using the ALMA Common Software (ACS) as the basic element, which allows the distribution of tasks and coordination of the different devices involved in the operation. This infrastructure uses CORBA as communication mechanism, implements a component-container paradigm and allows the usage of different programming languages (C++, Java and Python) and its seamless interconnection. Most of the devices are connected to a private Local Area Network at 1 Gbps although some of them are read and configured using serial and GPIB connections. In these latter cases ethernet converters are used to facilitate the operations.

The control of the mass and optical spectrometers, which are used as ancillary equipment for the detection of molecular species, are not integrated in the ACS. They are operated in an independent way from their own control and monitoring software (see sections 2.8 and 2.9).

The interface with the scientist is a command line based on IPython, fully programmable, which allows complex operations like repetitive macros, but retaining a great simplicity in its usage. The basic supported operations are the control of the receiver: selection of frequency sub-bands in the 72-116.5 GHz band, the attenuation of the IF (Intermediate Frequency) signal, the monitoring and tuning of the LOs, the calibration system (which uses hot and cold loads), the configuration of the integration time, phase composition and spectral resolution of the FFTs backends. In addition the control of the RF generator for cold plasmas, the ON/OFF switching of the UV lamps, and the readout of multiple pressure sensors of the system are also operated through this interface.

A graphical interface panel written in Python-Qt allows to visualize in real time the current status of the subsystems (receivers, loads, backends and sensors) and the running scan. This panel also allows to set scan options like the duration of the phases and its order, scan type and number of repetitions. The time evolution of some sensors readings, like pressure and temperature, can be plotted in real time.

The current system generates 16 spectra with 65536 channels each coming from 16 different FFTs which are read continuously and integrated to reduce the storage volume and the processing load. Integration times can be set from 100 ms to 5 s. The FFTs phases allow spatial observations (ON phase with the chamber full of gas and OFF phase with the chamber empty of gas) and frequency switching (FS) observations in which the LO swaps its frequency by tens of MHz (usually 25-50 MHz) typically every two seconds (see section 2.10). The control software allows a correct synchronization between the LO and the data acquisition. This latter procedure optimizes the integration time but relies on a uniform behaviour of the IF along the frequency and usually requires a post fit of the spectrum baseline.

The data are generated in CLASS format, a program from the GILDAS package⁵ developed at IRAM (Institut de Radioastronomie Millimétrique) and used in spectroscopic observations at many radio telescopes around the world. Part of the header for this spectra, which is not useful for laboratory experiments, has been adapted to store information from the different sensors that monitor the conditions of the experiment, for example, pressure and temperature at the receiver, at the chamber, and the cold calibration load. All data collected by the system can be fully processed in CLASS (data addition, baseline removal, and fitting of Gaussian line profiles among other functionalities).

All data are stored in files and sorted in folders conveniently named and tagged to allow simple operations and an easy identification. All metadata and operations are stored in a MySQL database to allow fast and simple searches.

The system permits automatic overnight observations and integration times as long as 6.05×10⁵ seconds which have been reached during a standalone overweek observation.

2.6. The plasma source

In order to study chemical processes, the cell is fitted with an internal coil that can inductively couple a plasma inside the chamber. It is formed by four turns of 6 mm diameter copper tubing, positioned in the center of the chamber avoiding any vignetting of the detector field of view. The diameter of the windings is 250 mm, and they are spaced by ~ 5 cm. One end of the coil is connected to the impedance matching unit of the radiofrequency generator, that can provide up to 300 W RF power at 13.56 MHz, while the other end is in electrical contact with the cell body and grounded. The coil is wrapped with Teflon™ tape (50 µm thick) in order to decrease the reactivity of some gases with the copper surface. The coil and its position inside the cell are shown in Figure A.2.

2.7. The UV lamps

In order to study ultraviolet (UV) induced processes, the cell is also fitted with four UV lamps that are placed symmetrically alongside its inner wall. The lamps are harboured by fused silica tube-shaped sheaths. One end of the sheath is open and bonded to a CF40 flange that attaches to the front plate of the chamber, while the inner end of the sheath is closed. In this way the lamps are operated at ambient pressure and could be eventually replaced without breaking the vacuum inside the cell. A flow of nitrogen purges the space between the lamps and their sheaths, removing atmospheric O₂ and H₂O (that, together with the formed O₃, would absorb part of the UV radiation), and also providing some cooling to the lamps. The lamps themselves are of the low-pressure mercury-arc type, with a total length of 793 mm (the discharge extends for 713 mm) and a diameter of 19 mm. They have a total output power of 13.5 W in the UV-C region according to the manufacturer, with ~ 11.7 W in the 254 nm line of Hg and ~ 1.8 W in the 185 nm line. Figure A.2 shows the chamber entrails with the UV lamps, the RF coil, and the Teflon™ lining.

The UV lamps have been successfully tested by using a flowing mixture of C₂H₂ and NH₃. The experiment was also performed in static conditions. In both cases the main products were HCN, HC₃N, HC₅N, and other compounds detected by mass spectroscopy. These results will be published elsewhere.

2.8. The mass spectrometer

In order to have more information on the composition and abundances of chemical species inside the cell, a quadrupole mass spectrometer is attached to one of the ports. The spectrometer is of the residual gas analyzer type, (Hiden, HAL 201 RGA, with a dual Faraday/Channeltron Electron Multiplier detector) and can measure masses up to 200 u. It is operated at pressures below 10⁻⁷ mbar with a dedicated vacuum system (turbomolecular plus scroll pumps), and it is connected to the gas cell via an electronically controlled needle valve. In sub-section 4.3 we show a mass spectrum taken when an RF plasma was induced in a flowing mixture of methane (CH₄) and nitrogen (N₂) inside the chamber.

2.9. The optical spectrometer

The plasma optical emission is collected through a fused silica window placed in a DN 60CF side flange of the chamber by means of a fused silica fiber and a 193 mm focal length, motorized Czerny-Turner spectrograph (Andor, Shamrock SR-193-i-A), equipped with a CCD camera (iDus DU420A-BVF). Two diffraction gratings with 1200 grooves/mm and 1800 grooves/mm provide spectral ranges of 300-1200 nm and 200-950 nm, respectively, and nominal spectral resolutions of 0.22 nm and 0.15 nm, respectively (for a 26 µm slit width). The rela tive spectral efficiency of the whole system was determined with a tungsten lamp calibrated in the optical metrology laboratories of the Institute of Optics of CSIC⁶. In section 4.3 we show the optical spectrum emitted by a CH₄ + N₂ plasma, acquired with the 1800 grooves/mm grating.

2.10. Observing Procedures

The best observing procedure to remove the instrumental response consists in performing a filled chamber observation, ON phase, minus an empty chamber observation, OFF phase. Unfortunately, switching between the ON and OFF phases takes a long time, due to the residence time of the gases in the chamber and to the time needed to refill the chamber at the same initial pressure. Hence, two alternative observing modes have been tested providing reasonable baselines for all tested pressures (10⁻⁶ − 40 × 10⁻³ mbar).

The first one consists in taking the ICL as OFF observation. Obviously, the ripples of the ICL are different from those of the chamber plus the BCL. This problem has been solved by observing, at the beginning of each experiment, a reference spectrum of the ICL and the BCL through the empty chamber. The ratio between both cold loads, R=BCL/ICL, was used during observations to generate an OFF for each ON scan. For each scan i the OFF_i was computed as R×ICL_i. The obtained baselines are rather flat (see Figure 7) but after 10-15 min it was necessary to take a new reference spectrum of the ICL and of the empty chamber plus BCL. Moreover, the switch between the positions corresponding to the chamber and the ICL takes several seconds as the movable mirror has to move around 40 cm between both optical paths. The frequent pump/filling cycles plus the dead time needed for this displacement render the observing procedure inefficient.

Fig. 7.

Observation of CH₃CN in flow at a total pressure of 25 ×10⁻³ mbar with GACELA using a load switching observing procedure. The upper panel shows the whole 72-116 GHz range as resulting from the observation of the three sub-bands of the W band receiver. The integration time is 11 minutes per sub-band. The left column shows a zoom of the J=6-5 transition of CH₃CN (upper panel), ${\text{CH}}_3^{13}\text{CN}$ (middle panel), and again ${\text{CH}}_3^{13}\text{CN}$ together with probably the ν₇, or ν₄ + ν₈, mode of CH₃CN (bottom panel). The three panels of the middle column show the J=6-5 transition of CH₃CN in the bending mode ν₈ (upper panel), and its overtones 2ν₈ (middle panel), and 3ν₈ (bottom panel). Finally, the three panels of the right column show the J=6→5 transition of the ν₄ mode of CH₃CN (upper panel), the same transition of the ground state of ¹³CH₃CN (middle panel), and of CH₃C¹⁵N (bottom panel). All the singly substituted isotopologues of methyl cyanide are in natural abundance. Intensities are expressed in K and frequencies in MHz. The kinetic temperature of the gases inside chamber is 297 K in these experiments. The difference between the observed frequencies and those measured in other laboratories, or predicted from accurate rotational constants, are less than 5 kHz for signals above 0.2 K and ~5-20 kHz for the weakest features.

The second selected procedure has been frequency switching (FS), typically with a throw of 25-50 MHz. The baselines show a considerably number of ripples but the lowest frequency one has a period of 40 MHz, which permits us to derive accurate line profiles and intensities for lines of up to ~2 MHz full width at half maximum (FWHM). This is the recommended observing procedure as the lines are observed twice and the noise gets improved by a factor square root of two. The frequency switch is fast enough (a few milliseconds) to minimize blanking times. Figure 8 shows the observed spectra in FS mode around the frequency of the J=12→11 line of HC₃N formed in a plasma of CH₄/N₂/CH₃CN at a total pressure of 25×10⁻³ mbar. The ripples with frequencies above 40 MHz can be easily eliminated by applying a FFT to each spectrum and removing the corresponding time domain, or by a polynomial baseline removal.

Fig. 8.

Cold plasma of CH₄, N₂ and CH₃CN observed in frequency Switching mode with a frequency throw of 37 MHz. Data have been folded. The upper panel shows the observed spectrum around HC₃N J=12→11 (produced in the plasma) without removing the baseline. The middle panel shows the resulting spectrum when a polynomial baseline is removed. The bottom spectrum is a zoom in frequency and intensity to show the weak lines that were detected from HC₃N ν₅ (E_up ≃1000 K) and the in-plane torsion of CH₃CH₂CN (produced in the plasma, E_up ≃320 K). The integration time for this observation is 100 sec.

3. Sensitivity and spectroscopic measurements with GACELA

3.1. Integration time and sensitivity

In order to check the longest integration time for which our system S/N performs correctly (i.e., the noise decreases with the square root of the integration time), we did, as a first experiment, an observation with very long integration time (3 days per W sub-band), with a frequency switch of 37 MHz and with the cell at 10⁻⁷ mbar. The theoretical noise, σ, for a frequency switching observation (after frequency folding) is given by

$$\sigma =\frac{\sqrt{2}{T}_{sys}}{\sqrt{\Delta \nu t}}$$ (1)

and for an observation in ON/OFF mode

$$\sigma =\frac{2T_{sys}}{\sqrt{\Delta \nu t}}$$ (2)

where σ and the system temperature, T_sys, are in K, t is the total integration time (in seconds), and Δν is the spectral resolution (in Hz).

Figure 9 shows the computed noise over a band of 40 MHz around the frequency of the J=1→0 line of HCN (note that residual gases do not emit at these frequencies and that the chamber was completely clean during the experiment). The noise, σ, decreases as the square root of time up to an observing time as long as 2×10⁵ seconds. For longer times, σ decreases slower than predicted by theory for the selected frequency range. In the three sub-bands of the W band the behaviour of σ with time is similar to that shown in Figure 9. Some frequency domains in W band, typically 50-200 MHz wide, show a poorer behaviour, with a longest integration time (following the square root dependence with time) of 1-1.5×10⁵ seconds. In the Q band the Teflon™ windows are not as well matched as in the W band and there are frequency domains where the longest integration time is 3-6× 10⁴ seconds. Taking into account the system temperatures (see Figures A.3 and A.4), these observing times imply a correct behaviour of the system down to 0.7-1.5 mK over the W band, and 1-4 mK in the Q band.

Fig. 9.

Noise as a function of time in sub-band WM for a frequency domain of 40 MHz around 88630 MHz. Each individual point represents a 1000 seconds increase in observing time.

The WIFI signal of the Observatory is observed as a broad feature, 40 MHz wide, in two FFTs of the W sub-bands and one of the Q band. Some spurious narrow (1-3 channels) features were also found in the Q and W bands. These features are easily identified and blanked (or interpolated) during data analysis.

3.2. Frequency measurements

In order to test the performance of the GACELA setup as tool for molecular aspectroscopy, we have used CH₃CN, OCS, and SO₂ to carry out different experiments. In particular we have analyzed the quality of the baseline, the appropriate working pressures for each gas to have reasonable linewidths, and the accuracy that can be reached in the measurement of the frequencies of the rotational transitions of gases inside the chamber. As commented above (see section 2.10), the sharpest ripples of our detection system have a period of ~40 MHz. Hence, it could affect lines with a FWHM larger than ~2 MHz. Methyl cyanide (CH₃CN) has been the first molecule we have observed during the commissioning of GACELA. This molecule exhibits a large self-broadening with pressure, of the order of 50 MHz/mbar. To limit the possible problems with the ripples we tested several pressures and concluded that the best operating conditions were obtained for P=0.02-0.04 mbar. Figure 7 shows the W band spectrum obtained after 660 s integration time per W sub-band at a CH₃CN pressure of 25 × 10⁻³ mbar. All the singly substituted isotopologues are detected (${\text{CH}}_3^{13}\text{CN},$ ¹³CH₃CN and CH₃C¹⁵N), together with the vibrational states ν₄, ν₈, 2ν₈, 3ν₈, and probably ν₇ and/or ν₄ + ν₈.

Several pressures were tested. At 10⁻⁵ mbar we expected to find a featureless spectrum. However, the K=0, 1 lines of the J=6→5 transition of CH₃CN are detected at 4.5σ in 90 s (see Figure 10 upper panel) and the K=3 with its hyperfine structure at 3σ. The lines become prominent at 10⁻⁴ mbar (see Figure 10 bottom panel; 100 s observing time), and very strong at 25×10⁻³ mbar (see Figure 7). The observing mode was ON/OFF in all cases. For 10⁻⁴ mbar the lines of the ν₈ mode appear clearly in 100 sec. We would like to note that at 10⁻⁵ mbar the density of CH₃CN at 300 K is ~3×10¹¹ cm⁻³, and that the column density in the cell is ~2.7×10¹³ cm⁻², i.e., three to four orders of magnitude below the column density of this molecule in Orion (Bell et al. 2014).

Fig. 10.

Lines from the J=6→5 rotational transition of CH₃CN observed at a pressure of 10⁻⁵ mbar (top panel) and 10⁻⁴ mbar (bottom panel) with integration times of 90 and 100 s, respectively. The observing mode is ON/OFF. Intensity is expressed in K and frequency in MHz. Note that intensities are proportional to pressure as collisional broadening is marginal at these pressures. The hyperfine structure of the K=3 component is clearly seen in the bottom panel. Also the K=2 component exhibits an emerging left shoulder due to the hyperfine splitting of the line.

The J=2→1 transition of CH₃CN falls in the Q band. The lines are expected to have a much weaker intensity (due to the lower Einstein coefficients) when compared with the lines in the W band (J=4→3, 5→4, and 6→5). Nevertheless, the largest integration time of this band, a factor of three compared with the whole W band (the Q band is always observed together with the WL, WM or WH sub-bands), permits us to detect the lines of the ground state and the ν₈ mode (at 362 cm⁻¹ or 521 K)(see Figure 11) during an integration of 33 min. Line broadening in Figure 11 is due to the different hyperfine components of each transition.

Fig. 11.

The J=2→1 transition of CH₃CN in the ground state and in the ν₈ mode. Integration time is 33 min, i.e., three times that of the W band. The Q band is always observed independently of the selected sub-band of the W band receiver.

The second molecule used during our tests was OCS. The pressure broadening for this molecule is less severe than that for methyl cyanide (Koshelev & Tetryakov 2009). Several pressures were tested and we selected 0.04 mbar. The whole data set in both Q and W bands is shown in Figure C.1. A polynomial baseline was removed by pieces 30 MHz wide in the Q and W bands. Figure 12 shows the observed features in a 1 GHz domain around the J=9→8 transition of OCS, which correspond mostly to vibrational excited states of this species. Figure C.2 shows the same rotational transition for selected isotopologues and vibrationally excited states. The observing time in this experiment was is 2.2 hours per W sub-band (6.6 hours in the Q band). The selected observing mode was FS with a throw of 37 MHz and a phase of 2 seconds for the frequency switch. As each rotational transition is observed both in the ON and OFF phases (shifted by 37 MHz in this case) the data have to be folded to get the final spectrum and to improve the S/N. As a consequence of the folding, each emission feature is accompanied by two negative signals at ± 37 MHz of the corresponding frequency.

Fig. 12.

Observation of OCS at a total pressure of 0.04 mbar in the chamber using a frequency switching observing procedure. The different panels show the 109050-109895 MHz frequency domain within the WH sub-band, but with different intensity scale limits (in K). Individual features are identified. The total observing time in this frequency range is 2.2 hours. The negative features seen in the different panels are produced by the folding of the frequency switched data and appear at ±37 MHz of each rotational transition (see also Figures C.1 and C.2).

From the observed intensities of the different isotopologues of OCS shown in Figure C.2 we can derive the following isotopic abundance ratios : ¹²C/¹³C=92.8±0.4 (93.5), ³²S/³⁴S=23.7±0.1 (23.5), ³²S/³³S=135.7±1.3 (133.3), ¹⁶O/¹⁸O=481.3±20.0 (487.8), ¹⁶O/¹⁷O=2636±200.0 (2630), and ¹²C׳²S/¹³C׳⁴S=2230±150.0 (2199), where the values between parentheses correspond to the natural abundances of the isotopic species (Juris et al. 2016). For lines of a given molecule observed with a S/N>30, the accuracy of the derived partial pressure is better than 3% (see also Appendix B). Hence, the quoted errorbars on these isotopic abundance ratios vary between 0.5-1.0% for the most abundant isotopes (S/N>100) to 10% for the rare isotopes (S/N>10).

The vibrational modes ν₂, 2ν₂, 3ν₂, ν₁, ν₃, and ν₂+ν₃ of OCS are detected with a high S/N (see Figure 12). The bending modes of OC³⁴S, OC³³S, and O¹³CS are also detected, together with the ground vibrational state rotational transitions of these isotopologues and those of ¹⁸OCS, ¹⁷OCS, and O¹³C³⁴S (see Figure C.2).

Finally, SO₂ was also observed at a pressure of 1.6×10⁻² mbar (linewidths at half power intensity for most lines are ~300-350 kHz; see Table C.1). Figure 13 shows the whole W band observed during this experiment. The lines corresponding to the 8_1,7-8_0,8 transition of SO₂, its isotopologues and vibrationally excited states are also shown. All line parameters for the features observed in this experiment are reported in Table C.1 and have been used to check the frequency calibration of the system and the uncertainty in deriving frequencies as a function of the data S/N. The left bottom panel of Figure 13 shows the difference between the observed frequencies and those reported in the CDMS (Müller et al. 2005), JPL (Pickett et al. 1998), or MADEX catalogs (Cernicharo 2012). Since the observed isotopologues of SO₂ have been measured previously up to 1-2 THz (i.e., up to high J and high K transitions) with high accuracy by different authors, the frequencies for their rotational transitions in these catalogs are accurate to 1 kHz, or even better, in the Q and W bands. From the whole set of observed lines, we have selected 161 features, corresponding to SO₂, ³⁴SO₂ and ³³SO₂, for which the S/N is larger than 20. We can see that 47% of the frequency differences are below 2 kHz (75 features), and this percentage becomes as high as 80% when differences of up to 4 kHz are considered. Hence, we can conclude that the frequency calibration of the system is excellent. The whole observed Q and W bands are shown in Figure C.3.

Fig. 13.

Observed spectrum of SO₂ in the W band. The integration time per W sub-band was 4.4 hours. The three upper panels show the W band with different intensity scales. The middle panels show the 8_1,7-8_0,8 transition of SO₂, ³⁴SO₂, ³³SO₂, SO₂ ν₂ and 2ν₂, and ³⁴SO₂ ν₂. The strongest hyperfine components of ³³SO₂ are nicely detected. Data for all lines observed during this experiment are given in Table C.1. The left bottom panel shows the difference, Δν(kHz), between the observed frequencies and those found in the CDMS (Müller et al. 2005), JPL (Pickett et al. 1998), and MADEX (Cernicharo 2012) catalogs. The colors correspond to different degrees of accuracy : |Δν(kHz)| ≤2 (red; 75 lines), 2< |Δν(kHz)| ≤4 (green; 53 lines), 4< |Δν(kHz)| ≤6 (blue; 17 lines), and 6< |Δν(kHz)| ≤12 (purple; 16 lines). The bottom right panels show the uncertainty (in kHz) assigned by the Gaussian fitting algorithm of the CLASS/GILDAS program to all observed lines in W band as a function of the signal to noise ratio of the peak intensity.

The bottom right panel of Figure 13 shows the uncertainty of the observed frequencies, obtained by fitting a Gaussian profile to the lines, as a function of the S/N of the line intensities. For S/N larger than 10 the uncertainty is below 7 kHz, being ~ 2 kHz for S/N~20, and ~ 1 kHz for larger S/N.

4. Cold plasmas

Cold plasmas provide very reactive media at room temperature. Fast electrons, with energies in the tens of eV, break the molecules and lead to the production of reactive ions and radicals that can initiate a complex chemistry both in the gas phase and at the vessel walls (Tanarro et al. 2011). They have been used at the IEM group for the study of species of astrophysical interest (e.g. Carrasco et al. (2012), Doménech et. al. (2013), Cueto et al. (2014), Doménech et al. (2016)). In the following, we comment on the effects of coupling a plasma on some gases and gas mixtures in our chamber. In these experiments the plasma incident power provided by the RF source was between 20 and 35 W. Approximately one half of the power was reflected.

Mass spectra were also used to observe the change in chemical composition upon plasma ignition. Additional information about the plasma products was derived from optical emission measurements.

4.1. CH₃CN

A short (80 s integration time) experiment was first performed (in ON/OFF mode) with CH₃CN, immersed in a bath at 30 °C in order to avoid fluctuations of the vapor pressure. The residual pressure in the chamber before injecting CH₃CN was 10⁻⁴ mbar, mostly due to residual air. The pressure in the chamber was 3×10⁻³ mbar at the beginning of the data acquisition

Fig. 14 displays the main rotational lines from the CH₃CN plasma in the W band. The main products observed were HCN and HC₃N. The optical emission spectrum (very similar to the one showed in Figure 18) indicated the presence of CN in the gas phase but it was not detected in rotational spectroscopy in this short observing time. The expected intensities in the Q band for the products and the precursor were too weak to be detected in 80 seconds.

Fig. 14.

Main spectral features found in the W band during the short plasma experiment of CH₃CN. The most prominent features correspond to HCN (top right panel), followed by those of the precursor (right middle and bottom panels). The four rotational lines of HC₃N falling in the W band are detected (left panels). The expected intensity of the J=2→1 transition in Q band was too low to be detected in the short observing time of the experiment (80 sec). The observing procedure was ON/OFF using the ICL as OFF.

Fig. 18.

Optical spectrum between 280 and 560 nm (top) and between 560 and 950 nm (bottom) of a plasma produced in a flowing mixture of CH₄ (1 sccm) and N₂ (1 sccm). The inserts show the CN, ${\text{N}}_2^{+}$ and N₂ bands. The corresponding mass spectrum is shown in Figure 17 and results from rotational spectroscopy during this plasma experiment are shown in Figure 16.

From the observed line intensities we derive the following partial pressures in the chamber: P(HCN)=0.39×10⁻³ mbar, P(HC₃N)=6×10⁻⁶ mbar (both corrected for their corresponding vibrational partition function at 300 K). We note that this low partial pressure of HC₃N corresponds to n(HC₃N)=1.5×10¹¹ cm⁻³, and N(HC₃N)=1.4×10¹³ cm⁻². The partial pressure of the precursor can be derived from the J=6→5 transition (see Figure 14) to be P(CH₃CN)=3×10⁻⁴ mbar (taking into account the A and E species and the contribution of the vibrational modes to the partition function). That means that only one tenth of the precursor remains unscathed in the cold plasma. With the discharge ON, the mass spectra indicated that the concentration of CH₃CN decreased by 85% (consistent with the observed intensities in the rotational spectrum) and that a large amount of hydrogen was formed in the chamber. The production of HCN was also observed by the growth of the peak at mass 27. There is also an observable decrease in the peak of background O₂ at mass 32 and a growth in the signals of CO and CO₂. Although not detectable by its rotational spectrum, N₂ was clearly visible in the mass spectrum and in the optical spectrum (see Figures 17 and 18 showing very similar results for a N₂+CH₄ plasma). From the difference of pressures between that of the precursor before the plasma was switched on, and those derived for HCN, HC₃N and the remaining CH₃CN, we conclude that the species H₂, N₂, CH₄, and perhaps C₂H₂, contribute with a pressure of 1.3×10⁻³ mbar to the total pressure in the cell. An increase of ~100 K in the vibrational temperature of CH₃CN is derived from the analysis of the intensity ratio of the rotational lines from the ground state and those from its ν₈ vibrational mode. The formation of HC₃N can proceed via the formation of C₂H₂ in the chamber (C₂H₂ + CN and CCH + HCN), or from the reaction of radicals CCN/HCCN/H₂CCN (all of them fragmentation products of CH₃CN) with CH₂ and/or CH₃. Reactions on the walls could also produce cyanoacetylene (HC₃N).

Fig. 17.

Mass spectrum taken when a plasma was produced in a flowing mixture of CH₄ (1 sccm) and N₂ (1 sccm). The optical spectrum observed during the same plasma is shown in Figure 18, and results from rotational spectroscopy during this plasma experiment are shown in Figure 16.

4.2. CH₃OH

After the CH₃CN experiment, a 100 s integration was performed in ON/OFF mode with the chamber filled with CH₃OH at a pressure of 20×10⁻³ mbar. The cell walls were not cleaned after the previous short experiment with the CH₃CN plasma. The strongest feature we found in the spectra was HCN, followed by CO, most probably as a consequence of the adsorbed molecules and sticked dust from the previous experiment. Weak lines of CH₃OH were also detected indicating that about 10% of the initial methanol molecules were able to survive in the plasma. No lines of CH₃CN were detected indicating that most of the molecules in the walls are polymers of HCN. From the observed intensities, the following partial pressures were found: P(HCN)=8.5×10⁻⁵ mbar, P(CO)=8×10⁻³ mbar, and P(CH₃OH)=5×10⁻³ mbar. With respect to the initial pressure of the precursor, 7×10⁻³ mbar are still missing, which correspond to H₂, H₂O and CO₂ as fragmentation products of CH₃OH in the plasma, as seen by mass spectrometry. The column density of HCN in the chamber was 2.0×10¹⁴ cm⁻².

In order to check our ability to introduce two liquids in a controlled way in the chamber, we performed a short observation (100 s) of a mixture of CH₃CN and CH₃OH with a total pressure of 13×10⁻³ mbar with no RF coupling. The lines of CH₃CN in the ground and ν₈, and 2ν₈ states were very strong and easily detected in this short observing time. Although this experiment was performed just after the CH₃CN and CH₃OH plasmas, no HCN was detected in this short integration time, as expected, since the plasma was switched off and therefore there was no mechanism to remove molecules from the wall.

We used the same gas composition in the chamber, CH₃CN and CH₃OH, to initiate a plasma with a total initial pressure of 30×10⁻³ mbar. The main polar products observed in the plasma gas, in addition to the precursors, are HCN, HC₃N, and CO. With respect to the pure CH₃CN plasma, the only new product is CO. In comparison with the plasma of pure CH₃OH the only new species found is HC₃N (note, however, that HCN was coming from the walls in this case). The integration time for this experiment was also 100 s.

4.3. CH₄/N₂

The next plasma studied was induced in a flowing mixture of CH₄ and N₂ with a contribution of methane and molecular nitrogen of 13.4×10⁻³ and 19×10⁻³ mbar, respectively. Since both molecules are non-polar, no rotational lines from them are observed in the Q or W bands.

In this case, the integration time was 40 min per W sub-band and the observing mode was FS with a throw of 37 MHz. The rotational spectrum shows the presence of HCN, HC₃N and CH₃CN as main products of the reactions between the fragments of the precursors in the plasma (see Figures 15 & 16 and C.4). From the observed intensities, we derive the partial pressures of 10⁻⁵ mbar, 4×10⁻⁴ mbar, and 1.5×10⁻⁵ mbar for CH₃CN, HCN, and HC₃N respectively. The corresponding column densities are 2.2×10¹³, 9.0×10¹⁴ and 3.3×10¹³ cm⁻². It is obvious that the partial pressures are far from the total pressure in the chamber (32.4×10⁻³ mbar). Most of the missing gases have to be the precursors CH₄ and N₂, with other non-polar products (H₂, C₂H₂, C₂H₆, …).

Fig. 15.

The J=1→0 line of the three isotopologues of HCN observed in the plasma of CH₄+N₂+CH₃CN (26 min of observing time per W sub-band). This frequency domain of the spectrum is observed in WL and WM sub-bands. Hence the total observing time in 52 min. Observing procedure was FS with a throw of 37 MHz. The line width at half intensity is 0.62 MHz for a total pressure of 28×10⁻³ mbar. Frequencies are relative to the strongest hyperfine component.

Fig. 16.

Selected lines of CH₃CN, HCN and HC₃N observed in the plasma of a flowing mixture of CH₄ and N₂ with a total pressure of 32.4×10⁻³ mbar. The integration time was 40 minutes in each W sub-band. The observing procedure was frequency switching with a throw of 37 MHz. The four rotational transitions of the ground state of HC₃N within the W band are shown in the middle panels. The different components of the J=12→11 rotational transition of HC₃N in its vibrationally excited states ν₆. ν₇, and 2ν₇ are shown in the bottom panels.

We monitored the variation of these species through the mass spectrometer (see Figure 17). Besides the fuel gases N₂ and CH₄, the mass spectrum with the plasma OFF showed the presence of small amounts of O₂, H₂O, CO₂ and Ar from the background gas. With the plasma ON, the mass peaks at 15 and 16 (CH₄), decrease by 50% and that of mass 28 (N₂) by about 5%. The peaks of H₂ (masses 1 and 2) increase largely and there is also an appreciable growth of the peaks at masses 26 and 27 (HCN) and 29, 30, which correspond most probably to fragments of small hydrocarbons (C_xH_y). The formation of CH₃CN and HC₃N is observed by their rotational lines and is further corroborated by the appearance of peaks at masses 41 and 51, respectively. The optical spectrum observed during this kind of plasmas is shown in Figure 18. It is dominated by N₂ bands, although signals coming from ${\text{N}}_2^{+}$ and CN were also detected.

N₂ is a very stable species and it is only marginally fragmented (less than 5%) with the power used in the RF generator for this plasma (20 W) as seen in the mass spectrum of the plasma (see Figure 17). However, about one half of CH₄ is fragmented. We have used several plasmas with different CH₄/N₂ abundance ratios to test the effect of the relative abundances in that of the products HCN and HC₃N. It is found that decreasing the methane abundance has an important impact on the abundance of HC₃N. These results, that can be of help in interpreting the chemistry of the upper layers of Titan’s atmosphere, will be published elsewhere. It seems that the limiting factor in the chemistry of these plasmas is the residence time in the gas phase of atoms (C, N, H) and radicals (CCH, CN, HCCN, CH₂, CH₃, H₂CCN) before reaching the cell walls. The power of the RF generator also plays a role in the fragmentation of the precursors. We have found that increasing the RF power from 20 to 35 W increases the production of HCN by a factor 2.

4.4. CH₄/N₂/CH₃CN

The last example of this section is that of a plasma induced in a flowing mixture of CH₄, N₂ and CH₃CN, with an RF power of 35 W and a total pressure in the chamber of 28×10⁻³ mbar. The observing time was 26 min and the observing mode was FSW with a throw of 37 MHz. The partial pressures of the three gases were identical. This plasma produces a real forest of lines (see Figure C.5), and shows very strong lines of HCN (see Figure 15) and HC₃N (see Figure C.6). In addition, species such as CH₂CHCN and CH₃CH₂CN (including some of their low energy bending and torsion modes) are detected (see Figure C.6). The vibrationally excited states of HC₃N ν₄, ν₅, ν₆, ν₇, 2ν₇, 3ν₇, 4ν₇, ν₅ + ν₇, ν₆ + ν₇ and 2ν₆ + ν₇ are detected with very good S/N. As in other plasmas studied with the GACELA setup the vibrational temperature seems to be 100-150 K above the ambient temperature (see also Tanarro et al. (2018)).

The intensity of HCN is so large that the isotopologues H¹³CN and HC¹⁵N are nicely observed (see Figure 15). Even the three single substituted ¹³C species of HC₃N are detected. The spectra shown in Figures 15, C.5 and C.6 have a high sensitivity. In fact, the HCN data from different plasmas can be used to derive the abundance of the isotopologues of HCN. We obtain HCN/H¹³CN=91.0±1.8, and HCN/HC¹⁵N=279±8, i.e., the natural abundance of ¹³C and ¹⁵N relative to ¹²C and ¹⁴N, respectively (Juris et al. 2016). These results, together with those relative to OCS and analyzed in section 3.2, show that the high accuracy in measuring molecular abundances that can be reached with GACELA. We have also detected CO, which is a low abundance residual gas in the chamber. Nevertheless, it can be also formed from CH₄ and residual O₂ and/or H₂O in the chamber. The FWHM of most features is 0.65-0.70 MHz, which indicates an average collisional broadening of 12.5 MHz/mbar (HWHM). We have estimated the following partial pressures: P(CO)= 0.4×10⁻³ mbar, P(HCN)=2.3×10⁻³ mbar, P(HC₃N)=0.6×10⁻³ mbar (corrected for the contribution of the vibrational modes to the partition function), P(CH₃CN)=7.0×10⁻³ mbar, P(CH₂CHCN)=0.02×10⁻³ mbar, and P(CH₃CH₂CN)=0.10×10⁻³ mbar. The missing pressure of 18×10⁻³ mbar is probably due to H₂, N₂, CH₄, C₂H₂, and C₂H₆. In fact, a large growth in the concentration of H₂ and a strong decrease in the mass peaks corresponding to CH₃CN (m=38-41 u) are observed in the mass spectra of the plasma. A significant increase in the signal is also observed for the mass peaks 26-30, corresponding probably to small hydrocarbons and notably to HCN (m=27 u). Much smaller signals appearing in the 50-54 mass range can be associated with CH₃CH₂CN, CH₂CHCN and HC₃N which are seen in rotational spectroscopy.

4.5. Other plasmas and cleaning of the walls

During the commissioning of GACELA, several additional plasmas were studied. Some of them quickly produced a severe deposition of material on the walls. The plasma of CH₄/N₂ and CH₄/N₂/CH₃CN produced a thin film of brown material (tholins; see Sagan & Khare (1979)) on the walls and on the copper coil of the RF generator. As commented in section 2.1 to avoid the direct contamination of the steel of the chamber, it was covered by a lining of Teflon™ 2 mm thick (see Figure A.2). The RF coil was also covered with Teflon™ (50 µm thick). The deposition of tholins and other materials was directly monitored by means of a webcam installed in one of the viewports. Some plasmas such as those of OCS or (CH₃)₃SiOSi(CH₃)₃ (Hexamethyldisiloxane; HMDSO) produced a quick deposition of sulfur and silicon compounds on the Teflon™.

The cleaning of the walls with a plasma of O₂ produces different species in the gas phase. In the case of the cleaning of the material produced by a plasma of OCS the most abundant polar species in the gas phase are SO₂, S₂O, OCS, CS, CO, and the radical SO. The observed lines of SO during these cleaning procedures are shown in Figure 19. The lines show Zeeman splitting due to the Earth’s magnetic field and the paramagnetic character of the SO molecule (³Σ⁻). The whole spectrum observed in the W band during the O₂ plasma cleaning is mainly dominated by the rotational lines of SO₂ and it is shown in Figure C.7 (to be compared with the spectrum of SO₂ shown in Figures 13 and C.3). When the chamber was subsequently submitted to a N₂ plasma, the NS radical was also found.

Fig. 19.

Rotational lines of the radical SO observed during a cleaning of the walls with a plasma of O₂ just after a plasma of OCS was performed. The lines are Zeeman splitted due to the Earth’s magnetic field.

The SO radical is produced in an efficient and continuous way when the O₂ plasma is active, probably due to the formation of SO₂ in the walls which is then easily fragmented into SO in the plasma. The radical SO reacts very slowly with O₂. The reaction has a barrier of ~2300 K (DeMore et al. 1997; Atkinson et al. 1997), and a rate at 300 K of ≃8×10⁻¹⁷ molec cm³ s⁻¹. The three body association SO+O+M is very slow at the pressures of our experiments. It seems that the reaction of SO with other gas compounds in the chamber is also rather slow (SSO, OCS, CS, and the fragments of the residual gases H₂O, N₂ and CO₂). To the best of our knowledge, there are no data on the reactivity of NS with O₂ and O. From our observations, it seems that these reactions have also low rates.

When the cleaning was performed after the plasma of HMDSO, we found the same compounds than those found when cleaning was made after a plasma of CH₄/N₂. In fact the fragmentation of HMDSO produces the radical CH₃ (Basner et al. 1998; Alexander et al. 1997), while the fragments containing silicon stick inmediately on the walls and the RF coil, and also form dust particles (Despax et al. 2016). The cleaning process did not produce any molecule containing silicon, but those typical of CH₄ plasmas, i.e., HCN, HC₃N, and CH₃CN.

5. N-Ethylformamide: CH₃CH₂NHCHO

5.1. Rotational spectroscopy

Only a few molecules in space contain the peptide group X-CONH-Y. Among them, formamide (NH₂CHO; X=H, Y=H) has atracted special attention in the last years due to its prebiotic character (see, e.g., Ceccarelli et al. (2017); Ligterink et al. (2018); Botta et al. (2018)). The next simplest species containing the peptide group is methylformamide, CH₃NHCHO, which has been characterized in the laboratory and tentatively detected in space by Belloche et al. (2017). However, it is has been searched with negative results towards IRAS 16923B by Ligterink et al. (2018). This species could be formed in ices (Ligterink et al. 2018) through the hydrogenation of methyl isocyanate, CH₃NCO, a molecule found to be very abundant in warm astrophysical environments (Cernicharo et al. 2016).

CH₃CH₂NHCHO (N-Ethylformamide, hereafter referred to as NEFA) is the next potential member of this series of peptide molecules to be present in space. Its rotational spectrum has been studied in the laboratory up to 13.4 GHz by Ohba et al. (2005), and up to 20.5 GHz by Vaquero-Vara et al. (2017). However, it is difficult to extrapolate these measurements up to the millimeter domain to search for it in the 3, 2, and 1 mm line surveys of astrophysical sources showing a rich molecular content.

In order to allow the search for this interesting species in space we have studied the rotational spectrum of NEFA in the Q and W bands using the GACELA setup. The sample of NEFA (purity >98%) was purchased from TCI Chemicals and was used without further purification. The sample was placed into a Pyrex™ vacuum Schlenk tube and it was degassed, using the common freeze-pump-thaw cycling method, through a 3 m³ h⁻¹ rotary pump (Pfeiffer Duo 3M) that allows pressure control in the sample injection line. Inlet connections from the Schlenk tube to the chamber were made using a Teflon tube of about 20 cm long. In order to avoid liquid condensation during the injection, the temperature of the sample container and the injection line were kept at 40°C using a dry heating tape and a PID temperature controller (Omega CN740). Prior to the sample introduction, the pressure inside the vacuum chamber was 2.0x10⁻⁴ mbar and during the experiment, carried out in a flow mode, the pressure was kept at 3.0x10⁻³ mbar, because higher pressures produce undesirable line broadenings. We found that the molecule has a very large self pressure broadening coefficient, of the order of 140 MHz/mbar. With the selected working pressure, the rotational lines of NEFA have a HWHM of 0.3-0.45 MHz which is well adapted to measure frequencies with high accuracy.

We encountered that NEFA partially decomposes since rotational lines of HCN, HC₃N, and CH₃CN were detected during the experiment. Only HCN, and at very low level, was present in the chamber before NEFA was injected. The observed lines of these species indicate that less than 20% of the NEFA molecules were decomposed.

The three dipole moment components of the molecule, µ_a,µ_b, and µ_c, are non-zero (Ohba et al. (2005); Vaquero-Vara et al. (2017)) which will produce a large number of rotational transitions for a gas at ambient temperature. Consequently, we performed two frequency switching observations with different frequency throws, Δν_FS, of 37 and 25 MHz. In this way, we could account for possible contamination of the emission lines by the absorption features produced during the folding process. These contaminating features are placed at ±Δν_FS of each emission feature and will, hence, appear at different frequencies in the two observations. The data with Δν_FS =37 MHz have an integration time of 29 hours per W sub-band, while the spectrum with Δν_FS =25 MHz has a total integration time of 59 hours per W sub-band and has been used as reference data for frequency measurements. It has been the longest standalone integration performed so far with the GACELA setup. The whole Q and W bands have been analyzed and a baseline has been removed by frequency ranges of 20-30 MHz depending on the density of lines. The less sensititive spectrum obtained with Δν_FS =37 MHz has been used only when the lines in the reference data were contaminated by absorption features, and also to check for line blendings, and in some cases, baseline effects.

Figure 20 shows different frequency domains of the W band spectrum. The bottom panels show two selected lines to further emphasize the high sensitivity reached in this experiment (≃1mK). NEFA exhibits indeed a very dense spectrum in the W band, with more than 100 lines per GHz. In addition to the fact that the molecule has a−, b−, and c-type transitions (Ohba et al. 2005; Vaquero-Vara et al. 2017), there are several low-lying vibrational excited states (Bermúdez et al. 2019), and other conformers, trans − ap and cis − ap, that are 137 and 482 cm⁻¹ above the trans − sc ground state conformer (Vaquero-Vara et al. 2017) that will contribute to the spectrum. The observed intensities are really low due to the large pressure broadening of the rotational lines of the molecule (which imposes a rather low working pressure of 3×10⁻³ mbar -see above), to the large values of the rotational and vibrational partition functions of the molecule (Q_rot (300K) ≃8.5×10⁴; Q_vib(300K) ≃13.1, Q_conf (300K) ≃1.47), and to its partial decomposition when in contact with the walls of the chamber. Nevertheless, the final reached noise level is ~1-2 mK with peak line intensities varying from 10 up to 80 mK, i.e., S/N>10 (see bottom right panel of Figure 20). We will focus in this paper on the ground state of NEFA. A detailed analysis of the spectrum of the molecule, its vibrationally excited states, and its conformers, is under preparation and will be published elsewhere (Bermúdez et al. 2019).

Fig. 20.

Observed spectrum of N-ethylformamide between 110.1 and 115.9 GHz. The total observing time per W sub-band was 59 hours. The observing procedure was frequency switching with a throw of 25 MHz. The pressure in the chamber was 3×10⁻³ mbar. The intensity scale is in K and the frequency one in MHz. The different panels show different zooms to the data of the upper panel. The two lines at the bottom have a linewidth of ≃0.4 MHz. The 19_1,19-18_0,18 transition shows a shoulder at lower frequencies produced by a line arising from one of the low-lying vibrationally excited states or from the conformers trans-ap or cis-ac (the expected hyperfine splitting for this transition is ≃15 kHz).

The uncertainty for most frequency determinations in the spectrum of NEFA is of a few kHz in view of the high S/N of the data (see Figure 20). However, we have assigned an uncertainty of 10 kHz for all lines showing a single component and intensity larger than 20 mK (S/N≥20). Unlike linear molecules with a nitrogen atom, for which hyperfine splittings rapidly collapse, symmetric or asymmetric tops can show significant hyperfine splitting even for high-J and high-K_a transitions. As an example, the bottom panel of Figure 10 shows the hyperfine structure of K=3 component of the J=6→5 rotational transition of CH₃CN. The K=0 and 1 components of the same transition do not show any measurable splitting, while the K=2 exhibits an emerging shoulder (clearly seen in that Figure) due to its hyperfine structure. Using the diagonal elements of the ¹⁴N quadrupole coupling tensor in NEFA (Ohba et al. 2005), we have computed the expected hyperfine splitting of all lines in the W band. Most of the lines we have identified have the hyperfine splitting concentrate around ±10 kHz of the central frequency. However, several lines will present three hyperfine components with similar intensities. Two of these components are at practically the same frequency while the third component is always at the opposite side producing a global splitting of 60-100 kHz. Hence, some of the observed lines could have a two components structure, with intensity ratio 1:2, with a strong feature accompanied by a factor two less intense shoulder, either left or right of the central frequency. Many other lines also show a complex profile due to blending with lines coming from the vibrationally excited states and the other conformers of NEFA (see, for example, the bottom right panel of Figure 20). We have fitted in most cases one single Gaussian profile to the observed lines. However, when the line profile exhibits two components we fitted two Gaussians. Depending of the degree of overlap between these components the assigned uncertainty varies between 20 and 80 kHz. When two components were obvious in the line profile, the feature assigned to NEFA was based on the expected intensity of the transition (determined with the MADEX code Cernicharo (2012)), and by comparison with the intensity of other nearby lines of NEFA.

Ohba et al. (2005) have found that the lines they observed in the microwave domain do not show any evidence for splitting due to internal rotation of the methyl group, CH₃, and derived a lower limit of 1000 cm⁻¹ to the barrier height of this internal rotation which is compatible with the estimated value derived from their ab initio calculations (1400 cm⁻¹). Hence, no internal rotation splittings are expected for the rotational transitions of this molecule.

For the assignment of the rotational lines of the ground state of NEFA we proceeded using the rotational and distortion constants derived by Ohba et al. (2005). We started the search for the strongest predicted lines having the lowest possible J and K_a within the Q and W bands. The b-type transition 19_1,19-18_0,18 was predicted a couple of MHz away from the strong line shown in the bottom right panel of Figure 20. The other strong line close to this feature (see the spectrum between 115230 and 115330 MHz in the second right panel from the bottom in Figure 20) was assigned to the b-type transition 19_0,19-18_1,18 and two features placed between both lines, and with a 50% intensity as expected, were easily assigned to the a-type transitions 19_1,19-18_1,18 and 19_0,19-18_0,18. A complete series of lines of the b−type Q branch K_a=4→3 (Δ J=0, J=14 to 39) were also assigned. These lines were fitted using the FITWAT code (Cernicharo et al. 2018) to derive a new series of rotational and distortion constants which were used to have new frequency predictions, allowing the assignment of new lines in the Q and W bands. Several sets of a−type (K_a=0-14) and b−type (up to K_a=15), were then indentified and fitted; c−type transitions are too weak to be assigned within the sensitivity of the data.

The final set of data for the ground state consists of 616 lines assigned in the Q and W bands with J(max)=54 and K_a(max)=15. Of these lines, 64 are unresolved K-doublets which gives a total of 552 independent frequency measurements. Table C.2 gives the frequencies and their uncertainties for all these lines. The inclusion of high-J lines requires the incorporation of high-order distortion constants to fit the data within the observed accuracy, which is a consequence of the highly floppy nature of the molecule (Bermúdez et al. 2019). The lines have been fitted to 24 rotational and distortion constants of a Watson Hamiltonian (Watson 1977) in the A reduction and I^r representation using the FITWAT code (Cernicharo et al. 2018) which produces a FORTRAN routine that can be directly incorporated into the MADEX code (Cernicharo 2012). The standard deviation of the fit is 30 kHz and the weighted one is 0.85; we have used also the SPFIT/SPCAT code (Pickett 1991) and obtained exactly the same results.

The derived constants with their uncertainties are given in Table 2. The agreement of our rotational and quartic distortion constants with those derived by Ohba et al. (2005) is excellent. It should be noted the improvement of the uncertainty of the present quartic distortion constants by a factor 20-30. The highest order constants are strongly correlated, but are essential to fit the observed frequencies due to the floppiness of the molecule. They permit to predict the spectrum up to 200 GHz, and even up to 300 GHz for low values of K_a. Caution has to be applied to frequency predictions for J>55, K_a>15.

Table 2. Rotational and distortion constants of CH₃CH₂NHCHO.

Molecular constants	Values MHz	Previous worka MHz
A	9904.836765 (905)	9904.8373 (6)
B	3521.099684 (209)	3521.0995 (2)
C	2984.979533 (186)	2984.9808 (2)
ΔJ	9.376237 (349) ×10–03	9.3859 (53) ×10–03
ΔJK	-6.297555 (282) ×10–02	-6.2957 (31) ×10–02
ΔK	1.659087 (181) ×10–01	1.6587 (15) ×10–01
δJ	3.223793 (197) ×10–03	3.2310 (14) ×10–03
δK	1.044634 (260) ×10–02	1.0116 (63) ×10–02
ΦJ	-2.57722 (178) ×10–07
ΦJK	3.47486 (457) ×10–06
ΦKJ	-1.82342 (324) ×10–05
ΦK	3.7193 (109) ×10–05
ϕJ	-1.06085 (209) ×10–07
ϕJK	-3.6172 (370) ×10–07
ϕK	3.8330 (475) ×10–06
LJJK	2.3014 (161) ×10–10
LKKJ	4.201 (160) ×10–09
lJ	-5.6239 (906) ×10–12
lK	1.1741 (316) ×10–08
PJJK	1.5318 (177) ×10–13
PJK	-8.165 (138) ×10–13
pJ	-4.848 (206) ×10–16
pJJK	-3.732 (121) ×10–14
pJK	2.0599 (248) ×10–12

Table Comments:

All constants are in MHz.

Numbers in parentheses represent the derived uncertainty (1 σ) of the parameter in units of the last digit.

The number of independent lines is 616 of which 64 correspond to unresolved doblets.

^aPrevious constants derived by Ohba et al. (2005)

5.2. Search for N-Ethylformamide in space

Using the rotational constants of Table 2 and the MADEX code Cernicharo (2012) we have searched for NEFA in cold and warm molecular clouds. The adopted dipole moments are those calculated by Vaquero-Vara et al. (2017) (µ_a=2.2 D, µ_b=2.9 D, and µ_c=1.0 D).

The cold dense cores TMC1-1 and B1-b have been studied in detail through systematic observations at all wavelenghts. In particular we have used our data at 3 mm (Marcelino et al. 2005, 2007, 2009, 2010; Daniel et al. 2013; Fuente et al. 2016; Cernicharo et al. 2012, 2013, 2014, 2018b) to search for NEFA towards these two objects. We have selected a couple of practically unresolved doublets, 5₅₁-4₄₀/5₅₀-4₄₁ at 92363.100 MHz and 6₆₁-5₅₀/6₆₀-5₅₁ at 112140.638 MHz, to increase the chance to detect the species. For B1-b we have adopted a kinetic temperature of 12 K (Cernicharo et al. 2012) and a linewidth at half intensity of 0.5 km s⁻¹. None of these lines are detected in our data which provides a 3σ upper limit of the column density of NEFA of ≃10¹² cm⁻². For TMC1-1 we have adopted a kinetic temperature of 10 K obtaining a similar upper limit to the column density of NEFA. These results are not totally unexpected, since formamide itself has not been detected towards these sources.

We have also searched for NEFA in two prototypical high-mass star-forming regions, Orion KL and Sgr B2, using the ALMA Science Verification (SV) data of Orion KL between 213.7 and 246.6 GHz (for observations and data reduction see, e.g., Tercero et al. 2018) and the IRAM 30m 3 mm survey of Sgr B2 provided by Belloche et al. (2013). The high angular resolution of the ALMA data ~2.00″ × 1.50″ allow us to concentrate the search in the emission peak of the related species NH₂CHO and CH₃NCO (see Cernicharo et al. 2016). Nevertheless, owing to the high frequencies of these data, we restricted the search for NEFA to lines with K_a≤5 in Orion KL to ensure the line identification.

We did not find this species above the detection limit of both sets of data. To provide upper limits to its column density, we used MADEX to derive the synthetic spectrum of NEFA for both sources assuming the same physical parameters than those derived for NH₂CHO by Cernicharo et al. (2016) and Belloche et al. (2013) in Orion KL and Sgr B2, respectively. Moreover, we searched for CH₃NHCHO in these sources to constrain the relative abundances of these related species. The results are summarized in Table 3. It is worth noting that due to the high line blending in the spectra of the two sources, together with the large partition functions of NEFA and of CH₃NHCHO, the derived upper limits correspond to relatively large values for the column density. Consequently, they are not very constraining for chemical models.

Table 3. N-Ethylformamide column densities.

	Orion KL(c) (ALMA SV)	Sgr B2(d) (IRAM 30m)
Coordinates (J2000.0)	α=5h35m14.5s δ=–05º22′133.0″	α=17h47m20.0s δ=–28º22′19.0″
HPBW(a) [″]	∼2.0×1.5	30-21
Frequency(b) [GHz]	213.7-246.6	80-115.5
vLSR [km s−1]	8.0	63.0
ΔvFWHM [km s−1]	3.0	7.0
dsou [″]	3.0	2.4
Trot [K]	150	180
N(NH2CHO) [cm−2]	3.0×1015	1.3×1018
N(CH3NHCHO) [cm−2]	≤1.0×1015	≤2.0×1017
N(NEFA) [cm−2]	≤1.0×1014	≤1.0×1017
N(NH2CHO)/N(CH3NHCHO)	≥3	≥6.5
N(NH2CHO)/N(NEFA)	≥30	≥13

^(a)HPBW (half power beam width) for observations with single dish telescope (IRAM 30m) and synthetic beam for the ALMA SV ob servations.

^(b)Range of frequencies considered in the analysis.

^(c)Physical parameters derived for the main spectral component of NH₂CHO by Cernicharo et al. (2016).

^(d)Physical parameters derived for the main spectral component of NH₂CHO by Belloche et al. (2013).

6. Conclusions

In this paper we have shown the huge potential of broad band, high spectral resolution, rotational spectroscopy using radio astronomical receivers and spectrometers as detecting devices. The sensitivity attained is similar to that obtained with the largest radio telescopes equipped with state-of-the-art receivers.

The GACELA setup can be used for a large variety of experiments. Broad band rotational spectroscopy, with its high instantaneous bandwidth, allows to perform chemical experiments in which the appearance and disappearance of molecular species can help to explain astronomical observations of molecular emission. We have shown in section 3 the enormous sensitivity we can reach in detecting polar species in a few seconds of observing time. Cold plasma experiments have been described in section 4 showing the potential to generate molecular species in the gas phase and also by interaction of atoms and radicals with the walls of the chamber (see section 4.5). The capability of the GACELA setup to perform rotational spectroscopy of interesting molecular species has been shown in section 5.1.

Broad band rotational emission spectroscopy could be used for many other experiments with GACELA, or in other reactors, in which the monitoring of the gas phase composition of polar molecules is needed. In addition, the high spectral resolution and high sensitivity provided by our detection system, could permit the identification of new molecular species produced in plasma or UV experiments, and their search in space.

The molecular thermal emission measurements performed with the GACELA setup permits us to derive accurate partial pressures for the gases filling the chamber (uncertainty below 3% for S/N>30).

We conclude that radio astronomical receivers as those of the GACELA setup can be used for Laboratory Astrophysics in a large number of novel experiments.