Mass spectrometric investigation of amorphous Ga-Sb-Se thin films

Amorphous chalcogenide thin films are widely studied due to their enhanced properties and extensive applications. Here, we have studied amorphous Ga-Sb-Se chalcogenide thin films prepared by magnetron co-sputtering, via laser ablation quadrupole ion trap time-of-flight mass spectrometry. Furthermore, the stoichiometry of the generated clusters was determined which gives information about individual species present in the plasma plume originating from the interaction of amorphous chalcogenides with high energy laser pulses. Seven different compositions of thin films (Ga content 7.6–31.7 at. %, Sb content 5.2–31.2 at. %, Se content 61.2–63.3 at. %) were studied and in each case about ~50 different clusters were identified in positive and ~20–30 clusters in negative ion mode. Assuming that polymers can influence the laser desorption (laser ablation) process, we have used parafilm as a material to reduce the destruction of the amorphous network structure and/or promote the laser ablation synthesis of heavier species from those of lower mass. In this case, many new and higher mass clusters were identified. The maximum number of (40) new clusters was detected for the Ga-Sb-Se thin film containing the highest amount of antimony (31.2 at. %). This approach opens new possibilities for laser desorption ionization/laser ablation study of other materials. Finally, for selected binary and ternary clusters, their structure was calculated by using density functional theory optimization procedure.


Results and Discussion
The surface characterization of the fabricated thin films was carried out by using SEM. The morphology of the thin layers surface is of good quality as shown by an example of SEM micrograph given in Supplementary Fig. S1. The chemical composition data of thin films, as obtained by EDS when averaging at three measured positions ( Table 1), indicate that by using rf power 15-25 and 5-15 W for Ga 2 Se 3 and Sb 2 Se 3 , respectively, the co-sputtered films cover a broad range of compositions: 7.6-31.7 at. % of Ga and 5.2-31.2 at. % of Sb, while the content of Se remains fairly constant (61.2-63.3 at. %) and in good accordance with nominal composition (60 at. %). XRD patterns revealed that all the deposited films are amorphous. The thickness of the thin films is in the range 560-780 nm (±2 nm).
Initially, the Ga-Sb-Se thin films of different compositions (samples labeled as A-G, Tab. 1) fabricated via rf magnetron co-sputtering were characterized by LA TOFMS in both, positive and negative ion modes. The mass spectra were recorded by systematically increasing the laser energy and the threshold energies for the ionization were determined for each sample. The 'optimal' mass spectra with sufficient mass resolution and the highest number of clusters were recorded at 160 and 140 a.u. for positive and negative ion modes, respectively, and used for further data processing. For the positive ion mode, the threshold energy where the ionization gets started was found to be 120 a.u., while for the negative ion mode it was 110 a.u. It was observed that for parafilm coated thin films the threshold energy was higher than for the non-coated thin films.
The effect of laser fluence was studied for all the samples (A-G) by recording mass spectra at different laser energies. As mentioned above, after reaching threshold laser energy of 120 a.u. in positive ion mode, the intensity of the mass spectra peaks climbed until 160 a.u. Afterward, the intensity decreased with further increase in laser power. This behavior generally occurs due to the decomposition of the ions at very high laser energies. The stoichiometry of the plasma species was determined by comparison of theoretical and experimental isotopic patterns ( Supplementary Fig. S2). For all the thin film samples, the lowest mass ion assigned was Sb + (quadrupole ion trap does not allow the detection of ions below m/z 100, Ga + (69.723) and Se + (78.96) ions maybe formed but they were not detected in the mass spectra). In addition to this one, two other unary Ga 2 + and Se 2 + clusters were detected. Many binary Ga x Se z + and Sb y Se z + clusters were also identified. The comparison of mass spectra for all the samples in range m/z 100-400 is given in Fig. 1. In this m/z range of mass spectra mostly unary and binary together with few ternary clusters were detected. Qualitatively, the clusters obtained in this mass range are the same for all the samples; however, their intensities are varied according to the chemical composition of the sample. For all the samples, the lowest mass binary clusters identified (correspond to general formula Ga x Se z + and Sb y Se z + ) were Ga 2 Se + and SbSe + , while the highest mass clusters Ga 19  www.nature.com/scientificreports www.nature.com/scientificreports/ 800-2500 are given in Figs 2 and 3, respectively. It was observed that the intensity of the peaks corresponding to the high mass clusters is lower than those detected in the low mass range m/z 100-400. The list of clusters detected by LA TOFMS in positive ion mode for all the samples is given in Supplementary Table S1. For simplicity, the number of binary Ga x Se z + , Sb y Se z + , and ternary Ga x Sb y Se z + clusters detected during LA of each thin film (parafilm coated and non-coated) sample is plotted and shown in Supplementary Fig. S3(a-c). For non-coated thin films, the highest number of Ga x Se z + , Sb y Se z + , and ternary Ga x Sb y Se z + clusters was identified for sample B, E, and F, respectively. For samples A-D, the highest intensity clusters are gallium selenide (Ga 3 Se + or Ga 3 Se 2 + ), while for the sample E, F, and G, the antimony selenide clusters (Sb 3 Se 2 + or Sb 3 Se + ) are of highest intensity. Several low-intensity binary Ga x Se z + , Sb y Se z + , and ternary Ga x Sb y Se z + clusters were also detected, but due to the insufficient mass resolution, the stoichiometry of these species was not determined.   Secondly, the parafilm coated thin films were examined via LA at similar experimental conditions and the results were compared with the non-coated ones. Many new, especially high mass clusters were identified in case of parafilm coated films. The comparison of mass spectra obtained from parafilm coated and non-coated thin film (sample B) is given in Fig. 4. Generally, one can observe that in most of the cases the intensity of the peaks obtained for non-coated thin films is higher than for coated ones under identical experimental conditions. Similarly to the non-coated thin films, Sb + is the lowest mass species identified for all the samples, while the highest mass cluster was Ga 19 Se 28 + which is detected in samples B, D, E, and F. The highest number of Ga x Se z + (23) and Sb y Se z + (9) clusters detected for samples B and C, E, G, respectively (Supplementary Fig S3a,b). Many new ternary Ga x Sb y Se z + clusters were also generated from parafilm coated thin films especially for the samples C-G, which have larger content of antimony (≥17.5 at. %). The highest number of ternary clusters (38) was identified for the parafilm coated thin film G ( Supplementary Fig. S3c). An important part of the mass spectrum (m/z 800-5000) obtained for parafilm coated thin film G is presented in Supplementary Fig. S4. A series of peaks such as Ga 5 Se 7 , Ga 7 Se 10 , Ga 9 Se 13 , Ga 11 Se 16, etc. are prominently detected and the difference between these main peaks is Ga 2 Se 3 (c.f. Fig. 4) with the largest cluster of the series as Ga 19 Se 28 + . Considering the MS results in all m/z ranges, the highest intensity peaks for samples A and B (<17.5 at. % of Sb) are Ga 3 Se 2 + and Ga 3 Se + , while for the samples C-G (≥17.5 at. % of Sb), the most intensive are Sb 3 Se 2 + and Sb 3 Se + peaks. An overview of all the species detected for coated as well as non-coated Ga-Sb-Se thin films is summarized in Supplementary Table S1. Few other clusters were also observed in MS data, but the stoichiometry of these species was not determined due to the insufficient mass resolution.
Furthermore, for both coated and non-coated Ga-Sb-Se thin films, it was found that the intensities of most of the peaks were varied according to the chemical composition of the thin films. For example, thin films with a higher amount of Ga (samples A, B) have high intensity of Ga x Se z + clusters, while the intensities decreased as the amount of Sb increased (samples C-G) and vice-versa for Sb y Se z + clusters. To exemplify this, we have selected some clusters which are identified in all the samples and plotted their intensities versus chemical composition (Supplementary Figs S5 and S6). The intensities of the Ga 2 + and Ga 3 Se + species is the highest for samples A and B and then decreases for samples C-G as the amount of Sb particularly increased. In contrary, the intensities of Sb + and SbSe + increased; they are highest for sample G where the content of Ga is lowest (7.6 at. %) and Sb is largest (31.2 at. %). It should be noticed that no significant intensity difference was observed in the case of Se 2 + and ternary GaSbSe + clusters.
All the thin film samples (coated and non-coated) were also examined via LA TOFMS in negative ion mode; however, the spectra are not that 'rich' like in positive ion mode with respect to the number of species detected. The effect of laser energy was studied by recording mass spectra at different laser power. It was observed that the intensity of the peaks grew with the increased laser energy until 140 a.u.; with a further rise of laser energy, the intensity dropped. The 'optimal' spectra with sufficient mass resolution and the highest number of peaks were recorded at a laser energy of 140 a.u. The lowest mass species identified (at these experimental conditions) is Se 2 − in all the samples and the highest mass clusters detected was Ga 17 Se 26 − for parafilm coated thin film B. Considering the mass spectra in all m/z ranges, the highest intensity peak observed for all the thin films is GaSe 2 − .
An example of a negative ion mode mass spectra in the range of m/z 100-400 for sample E is given in Fig. 5. The complete list of negatively charged clusters identified during LA of parafilm coated and uncoated Ga-Sb-Se thin films is given in Supplementary Table S2. Most of the negatively charged clusters detected are found to be different than positively charged ones. Many new and higher mass negatively charged clusters were detected for parafilm coated thin films as compared to non-coated ones. A comparison of negative ion mode mass spectra obtained from parafilm coated and non-coated thin film F is given in Supplementary Fig. S7.
Effect of parafilm coatings of thin films on mass spectra. LA is basically a destructive method which may generally break the original structure of the studied materials. In the case of amorphous Ga-Sb-Se films, we identified many species generated during LA. As already reported in 35,37 , some of the species might originate from real glass network structure, which makes generally possible to obtain structural information for solid state materials like glasses and their thin films. In our previous study 38 , we have shown that by dispersing the powdered chalcogenide material into a parafilm solution or in polymers (polyvinylpyrrolidone, polyethylene glycol, etc.), www.nature.com/scientificreports www.nature.com/scientificreports/ resulting mass spectra contained a significantly higher number of identified species present in the plasma plume. Moreover, the parafilm came out to be the most efficient material (among investigated ones) in terms of highest number of clusters detected in mass spectra of binary arsenic chalcogenides (As 2 Ch 3 , Ch = S, Se, Te).
Therefore, in this work, we have used a parafilm to coat Ga-Sb-Se thin films. As a result, many new clusters were identified in both, positive and negative ion modes using the technique depicted above. In positive ion mode, for some thin films, more than 30 new species were detected. The largest number of new clusters (40) was identified for a film with the highest content of antimony (sample G). In negative ion mode, many new clusters were also detected with a maximum of 16 new clusters, again for film G.
One can speculate that among detected new clusters, mostly higher mass clusters can be considered as larger fragments of the original glass network structure. Therefore, one can suggest that to some extent parafilm might reduce the fragmentation of larger parts of the structures into lower fragments. As a result, some more complex structures in the form of higher mass clusters are generated when using parafilm. But, in that case, the number of low mass clusters would be decreased; nevertheless this has not been observed. Therefore, another interpretation might be that the parafilm can play some role within the reactions of highly energetic low mass species present in the plasma plume. These lower mass constituents then might undergo laser ablation synthesis (LAS) and form heavier clusters.
Both theories are equivalent for the results. Higher mass clusters are formed (i) because of lower fragmentation under the presence of parafilm or (ii) parafilm is inducing synthesis of higher mass clusters from lower ones, it is difficult to distinguish here. We have therefore studied the synthesis of binary Ga x Se z clusters also from the mixture of elements. The mass spectra obtained from Ga:Se mixture and Ga-Sb-Se thin films were compared (Fig. 6). Evidently, main clusters Ga 5 Se 7 + , Ga 7 Se 10 + , Ga 9 Se 13 + , Ga 11 Se 16 + , etc. are also detected during LAS from the mixture of elements but there are also many other clusters. One can speculate that the possible mechanism of the formation of the clusters might be step-by-step synthesis from low mass clusters reacting with selenium and/ or gallium (Fig. 7).
Indeed, the presence of parafilm probably leads to complex processes including several features resulting in the observation of many species identified in mass spectra. In the absence of the parafilm, some clusters may dissipate fast and are not observed, while under the presence of parafilm, they are stabilized and a higher number of clusters maybe detected. It cannot be excluded that under the presence of parafilm higher mass clusters are generated via LAS reactions from lower mass fragments.  www.nature.com/scientificreports www.nature.com/scientificreports/ However, there is also another possibility. As demonstrated in [39][40][41] , laser desorption ionization from the surface of parafilm leads to increased sensitivity. Concluding, all above depicted explanations might be possible, they can proceed simultaneously in lower or higher extent and it is difficult to decide which one prevails. structure of Ga x se z , sb y se z , and Ga x sb y se z clusters. TOFMS does not provide direct structural information and thus it is difficult to elucidate the structure of the generated clusters from the mass spectra itself. On the other hand, the structures of many gallium [42][43][44][45] , antimony [46][47][48] , and selenium 49,50 clusters are known. Apart from the aforementioned structures, only a few structures of the binary gallium selenide and antimony selenide clusters have been reported 51,52 . However, there are extensive laser ablation mass spectrometric results showing that in binary systems Ga-Se and Sb-Se over one hundred Ga x Se z and Sb y Se z clusters were detected 53,54 .
The structure of most of the clusters generated in this work was not previously reported. Moreover, the number of possible structural isomers highly increases with the growing number of atoms in the clusters.
The structural patterns for selected binary and ternary clusters generated during LA of Ga-Sb-Se amorphous thin films were computed by density functional theory (DFT) optimization. A few possible structural isomers which are in the ground state with local minima in the potential energy surface are shown in Figs 8 and 9. The structures with the lowest energy are shown here if more than one stable structures of the same stoichiometry were found. From the calculated data it was observed that for small Ga x Se z , Sb y Se z , and Ga x Sb y Se z (x, y, z = 1-3) clusters, the molecular geometry does not change much with the charge. The bond lengths and angles of the structure for monocationic, neutral and monoanionic entities differ by only a few percents. As we obtained extensive data in positive ion mode during LA TOFMS, we further provided the calculated molecular data for singly charged positive clusters.
The DFT optimized structures of selected binary and ternary clusters are given in Figs 8 and 9. Optimizations were done first at the lower level of theory using LDA approximation with TZ2P basis set 55 , frozen core, and scalar relativistic ZORA 56-58 methods followed by a higher level of theory with OPBE 59 functional and all electron QZ4P   62 . The frequencies are computed from the square roots of the force constants. The optimized structures correspond to the "ground state" with local minimum in the potential energy surfaces with no imaginary frequency. The structures with the lowest energy are shown here if more than one stable structures of the same stoichiometry were found. The typical distance from adjacent gallium and selenium nuclei lies between 230-280 pm, while the distance between gallium-gallium nuclei is 270-290 pm. The distance between antimony and selenium nuclei is found between 230-260 pm and for Sb-Sb nuclei it is 260-280 pm. Detailed calculations are needed to generalize the structural features of all the generated clusters; however, this is beyond the scope of this study.

Conclusions
LA of Ga-Sb-Se amorphous thin films prepared via co-sputtering method from Ga 2 Se 3 and Sb 2 Se 3 polycrystalline targets generates many unary, binary and ternary clusters (Ga x +/− , Sb y +/− , Se z +/− , Ga x Se z +/− , Sb y Se z +/− , and Ga x Sb y Se z +/− ). As deposited Ga-Sb-Se thin films with seven different chemical compositions were examined and in each case, approximately 50 clusters were identified in positive ion mode. When parafilm coated Ga-Sb-Se thin films were studied, many new and high mass clusters were identified. The highest number (40) of new clusters was detected in positive ion mode for the thin film containing the largest amount of antimony. This indicates that parafilm is playing an important role during the LA process either reducing the destruction of the original amorphous network structure during the interaction of laser pulses with amorphous material or promoting the laser ablation synthesis of heavier species from low mass ones. Also, it is possible that detection from the parafilm surface is more sensitive. To get insight into the structure of the observed species, some DFT calculations were performed.
Concluding, LA TOFMS is found to be an important and useful analytical technique to study the amorphous chalcogenide thin films in terms of identification of species present in the plasma phase when the materials are exposed to the laser pulses. The stoichiometry of the species might help for partial structural characterization of thin films. Lastly, the parafilm as a material increasing the number of identified species in the plasma can be used to widen the applicability of the LA for other materials and thin films.

Experimental
Chemicals and materials. Gallium, selenium, methanol, acetone, and acetonitrile were purchased from Sigma-Aldrich (Steinheim, Germany). Polycrystalline 2″ 99.999% Ga 2 Se 3 and Sb 2 Se 3 sputtering targets were purchased from ALB Materials, Inc., Henderson, NV, USA. Xylene (a mixture of isomers) was purchased from Mikrochem Spol. s.r.o. (Pezinok, Slovak Republic) and parafilm from Pechiney Plastic Packaging (Chicago, IL, USA). Micro-90 (cleaning agent) was from Kratos (Manchester, UK). Silicon wafers used as substrates for the thin films depositions were purchased from ON SEMICONDUCTORS (Czech Republic). Deionized water was distilled once in glass and then double distilled from a quartz apparatus Heraeus Quarzschmelze (Hanau, Germany) to produce ultrapure water. All the other reagents were of analytical grade purity.
Methods. Ga-Sb-Se thin films. Amorphous thin films from Ga-Sb-Se system were fabricated using conventional rf (13.56 MHz) magnetron co-sputtering technique. The chemical composition of the thin films was varied by changing the rf power (up to 25 W) on the two cathodes ( Table 1). The experimental conditions of the co-sputtering deposition process are described elsewhere 17 .
Ga:Se mixture. Gallium-selenium mixture was prepared as suspension (~1 mg/mL) in acetonitrile by mixing liquid gallium with selenium powder in a molar ratio of 1:1 and applying ultrasonication (15 min). This suspension was used for MS analysis. www.nature.com/scientificreports www.nature.com/scientificreports/ Scanning electron microscopy with energy dispersive X-ray spectroscopy and X-ray diffraction. A scanning electron microscope (SEM) with an energy dispersive X-ray analyzer (EDS, TESCAN VEGA 3 EasyProbe) was used for the study of surface morphology and determination of chemical composition of fabricated materials. The uncertainty of EDS measurements for the studied films is ±1 at. %. Typically, the EDS measurements were performed at 3 spots per sample and averaged. X-ray diffraction (XRD) technique (D8-Advance diffractometer, Bruker AXS, Germany) was exploited to determine the structure of co-sputtered thin films using Bragg-Brentano θ-θ geometry with CuKα radiation and a secondary graphite monochromator. The diffraction angles were measured at room temperature from 5 to 65° (2θ) within 0.02° steps.
LA TOF mass spectrometry. The cleaning of the target plate was performed according to the Shimadzu target cleaning protocol. It includes initial cleaning with water, followed by sonication in Micro-90 cleaning agent solution for 15-20 min and then rinsing several times with water, acetone, and methanol. Finally, the target was air dried and kept overnight under vacuum in MS instrument and then used for measurements.
Thin films were fixed on the target plate using an adhesive tape. A part of the thin film was coated manually with a parafilm solution prepared by dissolving a piece of parafilm (1 cm × 1 cm) in xylene (1 ml). The thickness of the parafilm layer on studied films was approximately 100 μm. The coated and non-coated parts of the thin film were then examined using LA TOF mass spectrometry. The absorption coefficient of used parafilm at the wavelength of laser exploited for LA TOFMS (337 nm) is ~95 ± 20 cm −1 as calculated from its transmission spectrum.
AXIMA Resonance mass spectrometer from Kratos Analytical Ltd. (Manchester, UK) was used to record the mass spectra. The instrument details are given elsewhere 38 . Mass spectra were acquired in both, positive and negative ion modes by accumulating the data from at least 2000 laser shots. The peaks with sufficient mass resolution and intensity values higher than 1 mV (3 sigma of noise level) are considered relevant. For both modes, each m/z range was calibrated externaly using red phosphorus clusters 63 , while the accuracy achieved was below ±20 mDa.
Theoretical isotopic patterns were calculated using Launchpad software (Kompact version 2.

Data Availability
The datasets generated and analysed during the current study are available from the corresponding author on reasonable request.