Spectroscopic characterization of a low-temperature plasma ambient ionization probe operated with helium/nitrogen plasma gas mixtures

In this study, a systematic spectroscopic characterization of a low-temperature plasma (LTP) probe operated with He/N 2 gas mixtures is carried out. The in ﬂ uence of several experimental parameters ( e.g. , di ﬀ erent He/N 2 gas mixtures, discharge voltage, and gas ﬂ ow rate) on the dielectric-barrier discharge afterglow was studied. It was found that an increase of the nitrogen concentration in the helium discharge gas (to values higher than 0.5% N 2 ) leads to a signi ﬁ cant decay of N 2+ , He, and O emission intensities. At 1% N 2 and 99% He, oxygen emission bands were not detectable and intensity of He and N 2+ emission bands were reduced by approximately ﬁ ve times compared to features in a 100% He discharge. Interestingly, the opposite trend was observed for NO, OH, and N 2 species. Here, increasing the N 2 fraction in the discharge gas mixture led to an enhancement of emission intensities. Maximum emission bands intensities of OH, NO, and N 2 were detected at N 2 concentrations of 0.5, 0.6, and 1.0%, respectively. A further increase of the N 2 fraction leads to a decrease of emission intensities for all observed species (OH, NO, N 2 , He, and N 2+ ). In general, an increase in discharge gas ﬂ ow rate resulted in a signi ﬁ cant increase of NO emission band intensities for all N 2 /He mixtures tested. However, only a minor e ﬀ ect was observed for N 2 emission bands. Increasing the discharge voltage resulted in an increase of emission intensities of all detected species. Combined with spatially resolved investigations of the afterglow, these results are considered helpful to further optimize LTP performance. This is especially important for portable and direct analysis instrumentation, where LTP can be used as the ionization source and lower helium gas consumption is desirable because of cost and availability. containing 0, 33, 50, and 100% of N 2 in He. These experiments demonstrated a signi  cant in  uence of discharge gas  ow


Introduction
Plasma-based ionization sources can be used for elemental and molecular mass spectrometry depending on their discharge characteristics. One variant of a non-equilibrium plasma that is currently used in molecular mass spectrometry is the lowtemperature plasma (LTP) probe. The source is based on a dielectric-barrier discharge and was introduced by Harper et al. for its use in ambient desorption/ionization mass spectrometry (ADI-MS). 1 In this conguration, a non-equilibrium low-temperature plasma is generated at atmospheric pressure within a glass tube and extends into the ambient environment (this region of the plasma being called the aerglow). The attractive feature of the LTP probe is the ability to perform desorption and ionization with a single probe. To achieve this, the aerglow directly interacts with the sample at adjustable angles. The temperature of the surface area in contact with the plasma plume is approximately 30 C, and in turn, no signicant heating of the sample occurs. In addition, the high-voltage electrode is electrically isolated from the direct discharge region, and therefore, the sample is not liable to electric shock. These features allow direct analysis of temperature-sensitive samples or even dried chemical residues on a human nger. 1 Recently, we used a LTP successfully as an ionization source for differential ion mobility spectrometry. 2 In general, the LTP is considered a relatively so ionization source. In a previous study by Albert et al. it was found that LTP shows comparable ionization characteristics to atmospheric pressure chemical ionization (APCI) for a class of amines, amides, and aldehydes. 3 Also, LTP demonstrated higher relative ionization efficiencies compared to APCI for low molecular weight compounds and nonpolar to medium-polar compounds. Polycyclic aromatic hydrocarbons (PAHs) and imides were successfully ionized with LTP (in contrast to APCI). This general trend is attributed to the generation of several reagent ions (including O 2 + , O 2 À , N 2 + , NO + , NO 2 À , NO 3 À species and protonated water clusters) that are capable of ionizing a wide range of compounds. 4,7 An additional source of chemical reactivity stems from plasma ultraviolet (UV) photons. It was demonstrated for dielectric-barrier discharges (DBDs) that most of the UV emission comes from excited NO molecules when the discharge is sustained in He with a 10% fraction of N 2 . 5 It should also be noted that in many cases of LTP-ADI-MS application, additional heating of the sample (or sample holder) is required to enhance thermal desorption and maximize the signal (because of the low plasma temperature). However, because the probe operates at low power (<3 W) and low gas ow rates (typically between 0.08 and 1 L min À1 He) it is considered to have a great potential in portable, eld-based applications. With respect to portable and miniaturized instrumentation, the requirement to use helium may be a limitation. For example, instrumentation that is used for online (and on-site) monitoring should feature minimum helium consumption and operate with cheaper discharge gases. The use of alternative LTP discharge gases such as Ar, N 2 and air, respectively was studied in the past. 6 However, the overall sensitivity of the LTP coupled to a mass spectrometer was typically higher when helium was used as discharge gas. 7 Despite the fact that some investigations on the formation of reactive species in the LTP probe 4,7 and a DBD probe8 of similar construction operated with pure He (N 2 concentration < 600 ppm) and N 2 were performed, no systematic spectroscopic characterization of the LTP probe was carried out when the source is operated with He/N 2 gas mixtures. Since nitrogen plays an important role in the so ionization mechanism, the inuence of the nitrogen fraction on the discharge and the aerglow should be studied in detail.
The inuence of selected LTP probe parameters, such as probe geometry, electrode conguration, and tube dimensions was studied previously by Albert et al. 3 In the present study, the inuence of discharge voltage, discharge gas composition (N 2 fraction in He), and discharge gas ow rate on the dielectric barrier discharge aerglow is characterized by means of optical spectroscopy. The spatial distribution of reactive species present in the aerglow was also determined. This information is essential not only for a better understanding of the mechanisms involved but also to further optimize the LTP performance for portable and on-site analysis applications.

Low-temperature plasma probe
Conguration of the home-built LTP probe was similar to the geometry described in our previous publication. 2 In brief, an inner solid stainless-steel electrode (0.7 mm diameter, 3 mm distance to capillary exit) and an outer copper-ring electrode (6 mm length, 3 mm distance to the capillary exit) were used to generate a dielectric-barrier discharge in a glass tube (4.0 mm outer diameter, 0.8 mm inner diameter). A relatively narrow gap between the glass tube and inner electrode (ca. 0.05 mm) was selected to achieve a high gas ow velocity and fast transport of the created excited species towards the exit of the probe. A high-voltage alternating current source (3-7 kV pp at 28-35 kHz, model PVM500, Information Unlimited, Amherst, USA) was applied between the outer copper-ring electrode and the inner pin electrode to generate the plasma. Voltage was measured using a portable wide-bandwidth high-voltage probe (#60 kV, #80 MHz, 1 : 1000, PVM-1, North Star High Voltage, Marana, USA) and recorded using a digital storage oscilloscope (TDS 2012B, Tektronix, Beaverton, USA).

Experimental setup
A schematic diagram of the experimental setup is presented in Fig. 1. Helium (99.999%, water <2.0 ppm, O 2 < 2.0 ppm, N 2 < 5.0 ppm, Air Liquide, Düsseldorf, Germany) and nitrogen (99.999%, water <2.0 ppm, O 2 < 2.0 ppm, Air Liquide, Düsseldorf, Germany) were used to prepare discharge gas mixtures at ow rates between 100 and 600 mL min À1 . Gas mixtures were prepared by mixing the ow of nitrogen, controlled by a mass ow controller (MFC, GFC17, 0-500 mL min À1 N 2 , Aalborg, USA), with an additional ow of helium in the desired ratio. Helium ow rate was controlled by a mass ow controller (MFC, GFC17, 0-500 mL min À1 He, Aalborg, USA). Because of the upper ow rate limit (of 500 mL min À1 ) of the available devices (MFC), the setup was slightly modied for measurements with a desired ow rate of, e.g., 600 mL min À1 . Here, for a discharge in pure N 2 , the He-MFC was exchanged to a second N 2 mass ow controller. Similarly, the N 2 -MFC was exchanged to a second He-MFC when a ow rate of 600 mL min À1 for a discharge in pure He was required.
For measurements with N 2 fractions below 2.5% (see Fig. 3, bottom), a different MFC (GFC17, 0-20 mL min À1 N 2 , Aalborg, USA) was used for better precision. A peruoroalkoxy (PFA) Swagelok® 1 4 inch tee was used to couple the grounded stainless-steel electrode, tubing, and glass capillary. In order to minimize immediate diffusion of surrounding air into the discharge/aerglow, the outlet of the glass capillary was inserted into a quartz cuvette (approximately 23 mm to cuvette outlet; cuvette: QS 0.5 cm, Hellma Analytics, Müllheim, Germany) and sealed with PTFE-tape. The bottom of the cuvette was removed in order to provide an outlet for the discharge gas. In the current LTP setup, the concentration of surrounding oxygen and nitrogen at the capillary exit can be signicantly reduced. The estimated concentration ranges for oxygen and nitrogen vary between 0.2-4.0% and 2.4-14.5%, respectively, depending on the discharge gas ow rates (100-600 mL min À1 ). The diffusion of surrounding oxygen and nitrogen into the discharge area is considered to be negligible.

Spectroscopic measurements
Spectroscopic measurements were performed with an iStar ICCD camera (DH734-18U-03, Andor, Belfast, UK) connected to a Shamrock spectrometer (SR-303i, Andor, Belfast, UK). The readout mode was set to full vertical binning (FVB). Spectra were recorded in accumulating mode (100 accumulations) at a microchannel plate (MCP) gain level of 50, gate width of 50 ms, and an exposure time of 0.1 s. The spectrometer was equipped with a 1200.6 lines per mm grating and an entrance slit width of 100 mm. The wavelength resolution and accuracy are 0.186 nm and 0.0563 nm, respectively. Spectra could not be recorded simultaneously over the complete wavelength range of interest and were acquired in so-called step-and-glue mode (20% overlap). A UV-grade quartz optical ber (numerical aperture is 0.12) was mounted on micrometric stages and positioned perpendicularly to the plasma aerglow/jet at a distance of 1 mm to the quartz cuvette. Taking into account the calculated spatial resolution of the used system (0.42 mm) and precision of the positioning system the mapping of the discharge was performed by shiing the optical ber along the x-axis in 1 mm steps.
For each sample, three single measurements were recorded. The mean values of peak intensities and the corresponding standard deviations were calculated using Microso Excel 2010 soware.
The analyzed band heads from molecular and atomic species are as follows. The N 2 band head of 337.1 nm corresponds to transition from the second positive band head: N 2 (C 3 P u ) n 0 ¼0 / N 2 (B 3 P g ) n 0 ¼0 . 7,9,23 The N 2 + band head of 391.4 nm corresponds to transition from the rst negative band head: N 2 + (B 2 S + u ) n 0 ¼0 / N 2 + (X 2 S + g ) n¼0 . 7,9,23 The NO band head of 237 nm corresponds to transition: NO (A 2 S + ) n 0 ¼0 / NO (X 2 P) n 00 ¼1 . 10 The OH band head of 308 nm corresponds to transition: OH (A 2 S + ) n 0 ¼0 / OH (X 2 P) n 0 0 ¼0 . The He I band head of 706.5 nm corresponds to transition: 3 s 3 S 1 / 2p 3 P 0 . 7, 23 The O I band head of 777 nm corresponds to transitions: 3p 5 P 1-3 / 3p 5 S 0 2 . 7

Results and discussion
Effect of the discharge gas composition The possibility to reduce or even avoid the helium discharge gas consumption would help to reduce LTP operating costs and would make it easier to realize portable systems. Because the LTP probe is typically used as an open source, i.e. without a source housing (to be able to probe large objects etc.), diffusion of ambient air, e.g., some amount of nitrogen into the LTP discharge region is likely. Here, the inuence of the nitrogen fraction on the emission pattern of the discharge and the aerglow was studied in a rst experiment. Fig. 2 shows emission spectra (200-800 nm) measured directly at the capillary outlet (distance to capillary outlet d ¼ 0 mm). The discharge gas ow rate was 500 mL min À1 and discharge voltage was 4 kV pp . The nitrogen fraction in the discharge gas was of 0, 20, 40, 60, 80, and 100%. Strong emission bands from molecular species, such as N 2  These reactions result in the formation of NO + . Interestingly, the positive NO + ions can be also generated by He metastables via Penning ionization of NO. 13,14 Despite the fact that formation of NO + could not be conrmed spectroscopically with the current experimental setup, the formation of NO + ions was previously reported when the LTP was coupled to a mass spectrometer 15 and an ion mobility spectrometer, 16 respectively. N 2 + and NO + are considered to be important reactive species.

N 2 + is a reaction intermediate for the formation of other reagent
ions such as protonated water clusters, (H 2 O) n H + . The following ionization of analytes proceeds via proton transfer and efficiency of ionization is dependent on proton affinity of analytes. 17 NO + as a reactant ion is important for monitoring of monoaromatic volatile organic carbons (VOCs, such as benzene, toluene, and xylene (BTX)), 18 some explosive materials 19 and various other compounds. 20 The major reaction pathways observed for the NO + reagent ion are ion-molecule association, electron transfer (by analytes with IE < 9.26 eV), and hydride abstraction. 21,22 In addition, excited NO observed in the aerglow is a source of high energy photons (200-280 nm, see Fig. 2). Therefore, efficient formation and transport (within the aerglow) of these species is important for effective ionization of analytes. Addition of 20% of nitrogen to the discharge gas resulted in almost complete disappearance of He, H, O, and N 2 + emission bands. At this nitrogen concentration, NO and N 2 emission bands were dominating over other emission bands. Interestingly, the intensity of these emission bands was changed differently with respect to the corresponding values in pure helium. Intensity of N 2 emission bands was approximately 30% lower compared to that in pure helium; however, the intensity of NO emission bands increased approximately by a factor of two compared to that in pure helium (see inserts in Fig. 2). The following increase of the nitrogen fraction in discharge gas within the fractions range of 20-100% resulted in reduction of emission bands intensities of both NO and N 2 .
To investigate this phenomenon in more detail, the nitrogen fraction in the discharge gas was changed stepwise (in increments of 0.1-0.4% between 0% and 2.4% N 2 in He and in increments of 7.6-10% between 2.4% and 100% N 2 in He) and the following species were monitored: NO (237 nm), N 2 (337.1 nm), N 2 + (391.4 nm), OH (308 nm), He (706.5 nm), and O (777 nm). The corresponding data is depicted in Fig. 3. It was found that a small amount of N 2 in the discharge gas (0-0.3%) resulted only in a small increase in N 2 + emission intensity. Also, emission intensities of other species did not signicantly change as well (see Fig. 3 Increasing the nitrogen concentration in the discharge gas resulted in an increase of emission intensities. The maximal emission intensities of OH, NO, and N 2 were achieved at a nitrogen concentration of 0.5%, 0.6%, and 1.0%, respectively. The increase of nitrogen concentration in discharge gas within the fractions range from 1 to 100% leads to a decrease of a V pp is the discharge voltage, f is the discharge frequency, CS is the cross section between inner capillary wall and inner electrode, FR is the discharge gas ow rate, and GV is the gas velocity between the capillaries. GV was calculated as a ratio between FR and CS (in corresponding units).
emission intensities for all three species. At a nitrogen concentration higher than 20%, emission bands of NO and N 2 are dominating over emission from other species (emission of other species is negligible). Increasing the nitrogen fraction in the discharge gas from 20% to 100% resulted in continuous decrease of NO and N 2 emission intensities (see Fig. 3, top). These observations are in agreement with results from Olenici-Craciunescu et al., who investigated the inuence of the N 2 fraction in He on DBD characteristics (studied range 0-600 ppm N 2 in He). 23 Within the measured concentrations range, they observed an initial increase of N 2 + emission intensities with a maximum between 140 and 450 ppm followed by a slight decay within the range of 450 to 600 ppm. However, helium (706.5 nm) intensities were decreasing with increasing nitrogen concentrations. The fact that an increase of N 2 + intensities was only observed up to nitrogen concentrations of 450 ppm was explained by a limited amount of available helium metastables, which deliver the energy for the Penning ionization of nitrogen. However, the inuence of higher nitrogen concentrations was not investigated in their study. The inuence of the nitrogen content on broadband UV emission was previously studied by Lu et al. for a pulsed DBD (with a different geometry). 5 Here, it was observed that the increase of the nitrogen fraction (up to $5%) in helium resulted in an increase of broadband emission intensities between 165 nm and 310 nm. NO was found to be the main UV emission source. In our study, however, the maximum NO emission was found at a nitrogen concentration of 0.6% (see Fig. 3, bottom). The following increase of nitrogen concentration resulted in a reduction of UV emission intensity.

Effect of the discharge gas ow rate
To investigate the effect of ow rate on the emission spectra in the aerglow, experiments with ow rates within the range of 100 to 600 mL min À1 were performed. This effect was investigated for gas mixtures containing 0, 33, 50, 67, and 100% of N 2 in He. These experiments demonstrated a signicant inuence of discharge gas ow rate on the observed emission spectra in the aerglow.
At a discharge gas ow rate of 100 mL min À1 , emission bands of N 2 are dominating over emission from other species for most analysed discharge gases (33-100% N 2 ). The analysis in a pure helium at discharge gas ow rates lower than 300 mL min À1 was not possible because of an observed electrical breakdown inside the capillary. A possible reason for the absence of an electrical breakdown at helium ow rates higher than 300 mL min À1 is a slightly increased pressure within the discharge region. According to Paschen's law, an increase in pressure results in a higher breakdown voltage. An increase of the discharge gas ow rate results in a much stronger increase of NO emission as compared to the observed increase of N 2 emission. This tendency was observed for all analysed discharge gas compositions (see Fig. 4, bottom).
One of the possible reasons for this observation can be a faster transport of ions with a gas ow. When increasing the ow rate, the time that is required for the transport of species from the plasma region to the capillary outlet is shortened. Thus, plasma species created at higher ow rates have less time for the gas chemistry as compared to the experiments at lower ow rates. Increase of the carrier gas ow rate from 100 to 600 mL min À1 results in a decrease of the residence time of the species in the region between the active plasma part and the measurement position from 9.29 Â 10 À4 to 1.55 Â 10 À4 s. The calculated lifetimes for excited NO and N 2 species were determined to be approximately 1.2 Â 10 À4 and 3.0 Â 10 À4 s, respectively. The estimated lifetime of excited NO species is by 2.5 times shorter as compared to the one of excited N 2 species. Therefore, an increase of the carrier gas ow rate by six times and, thus, a decrease of the residence time by six times results in a signicant increase of excited NO species in the aerglow.
Emission spectra recorded within the wavelength range of 200-450 nm in discharge gas containing 67% of nitrogen in helium are exemplarily demonstrated in Fig. 4 (top, discharge voltage ¼ 4 kV pp ). The strongest emission of NO (237 nm) and N 2 (337.1 nm) are highlighted with inserts in Fig. 4. The change of NO (237 nm) and N 2 (337.1 nm, inset) emission intensities in different gas mixtures and ow rates (100-600 mL min À1 ) is presented in Fig. 4 (bottom).
At a helium ow rate of 300 mL min À1 , the strongest emission was observed from N 2 + . Nevertheless, the relationship between NO and N 2 emission intensity and discharge gas ow rate was similar to those observed for nitrogen containing mixtures.
Effect of the discharge voltage  Table 1). Fig. 6 shows normalized emission maps of selected molecular and atomic species observed in the LTP aerglow at different gas ow rates. In these gures, the vertical axis refers to the distance from the end of the LTP outlet (at 0 mm). In contrast to Chan et al. and Olenici-Craciunescu et al. no emission maxima outside the torch were observed for N 2 and N 2 + , even when the measurements were performed in pure helium. The reason for this is still under investigation, but based on the obtained data is difficult to discuss the mechanistic aspects of plasma ionization. Nevertheless, a few interesting observations were made. As was partially demonstrated in Fig. 4, the intensity of NO emission demonstrates a signicant dependence on the discharge gas ow rate. A ow rate decrease from 500 to 100 mL min À1 results in a decrease of NO  emission intensity by more than ve times in all discharge gases. However, the dependence of N 2 and N 2 + emission lines on the discharge gas ow rate was found to be not as severe, especially at higher helium fractions (50-100%). Emission of all observed species strongly decreases with increasing distance from the LTP outlet. At a distance of 3 mm the emission intensity of NO (237 nm) is reduced by almost 10 times as compared to its intensity measured directly at the outlet. This effect was found to be even stronger for N 2 and N 2 + emission bands (10% at a distance of 1.5-2 mm). Also, this decay was found to be relatively stronger compared to results reported earlier with similar DBD discharge parameters. 8 The reason for this fast decay is not clear today and further studies should be performed to clarify this phenomenon.

Conclusions
A low-temperature plasma probe operated with helium/ nitrogen mixtures was characterized by means of optical spectroscopy. Selected reactive species that are involved in analyte ionization processes (e.g., N 2 + , NO) were studied. A signicant decrease of emission intensity (and to a rst approximation also density) of N 2 + in the aerglow was observed for nitrogen concentrations in the discharge gas higher than 0.4%. This is important because N 2 + is generally considered to be a reaction intermediate that is also responsible for the formation of protonated water clusters. 24 Therefore, the observed decrease of N 2 + density can potentially limit the application of a N 2 /He driven LTP source for the analysis of analytes that are ionized via proton transfer reactions. In discharge gas mixtures that contained 0.4-50% nitrogen, however, the emission intensity of NO in the aerglow was higher compared to a 100% helium LTP. The maximum emission intensity was achieved in a mixture with a relatively low nitrogen fraction (0.6% N 2 in He). Here, NO emission increased approximately ten times compared to that in a 100% helium discharge. Penning ionization of NO results in formation of NO + . This ion is an important reactant ion because may offer alternative routes to analyte ionization (e.g. electron transfer, ion-molecule association, and hydride abstraction mechanisms) as compared to ionization routes which involve N 2 + intermediate (proton transfer mechanisms).
In addition, excited NO is a source of high energy photons which can provide additional plasma reactivity. Thus, the ionization mechanism can be driven by a proper choice of the LTP discharge gas. Discharge gas ow rate is another parameter, which affects the intensity (and to a rst approximation also density) of reactive species in aerglow. An increase of the discharge gas ow rate results in a much stronger increase of NO emission as compared to the observed increase of N 2 emission. Thus, it is assumed that high discharge gas ow rates will be preferable for the analysis of analytes that can be ionized via NO + species in the aerglow. However, investigation of the inuence of the discharge gas composition and the ow rate on ionization mechanisms in the aerglow was not systematically investigated in this study and, therefore, should be investigated in detail in the future.