fac-Tris(2-phenylpyridine) iridium [Ir(ppy)3] has been investigated by means of soft desorption/ionization induced by neutral SO2 clusters in combination with mass spectrometry. Desorption of intact Ir(ppy)3 was observed. Further analysis of the isotopic pattern revealed two forms of ionization, either by uptake of a proton or by electron abstraction. The relative contribution of the two processes depends on measurement time and H2O partial pressure, as well as preparation scheme and surface morphology of the samples.
I. INTRODUCTION
Organic light-emitting diodes (OLEDs) based on fac-tris(2-phenylpyridine) iridium [Ir(ppy)3] [compare Fig. 1(a)] as the phosphorescent component have been shown to exhibit an internal quantum efficiency close to 100%.1–6 Ir(ppy)3 proved to be a reliable guest-emitter in guest-host OLED-systems with the basic structure of Ir(ppy)3 providing a basis for further ligand functionalization and thus tuning of the electronic behavior.7–11 With this versatility, organic iridium-complexes may also play an important role in the future development of OLEDs. At all stages of OLED research and development, materials’ characterization is crucial. Mass spectrometry (MS) as one method of choice requires the transfer of the OLED components from the solid state into the gas phase. In commonly applied methods such as secondary-ion mass spectrometry (SIMS) and matrix-assisted laser desorption/ionization (MALDI), this transfer often goes along with some or even predominant fragmentation of the analyte molecules.12–18 On the other hand, OLED materials such as iridium complexes with fragile substituents at its ligands, in particular, require soft and fragmentation-free desorption.19,20 In the case of MALDI, the application to metal complexes is further complicated as the employed matrix can even replace the ligand.19
Desorption/ionization induced by neutral SO2 clusters (DINeC) has been proven as a method for soft desorption of organic analytes such as peptides, dyes, phospholipids, ionic liquids, and polymers.21–26 In this study, we investigate the desorption and ionization of intact Ir(ppy)3 molecules by means of DINeC-MS with emphasis on the ionization mechanisms operative during cluster-induced desorption/ionization. Both, ionization by proton uptake (chemical ionization, leading to [Ir(ppy)3+H]+) as well as ionization by the abstraction of an electron5 (electronic ionization, leading to [Ir(ppy)3]+) were observed. The predominant ionization mechanism was found to depend on the duration of the measurement and the H2O partial pressure, as well as sample preparation and morphology.
II. EXPERIMENTAL
fac-Tris(2-phenylpyridine) iridium (purity ) was purchased from Carbosynth Ltd., Compton, UK and subsequently used without further purification. Three different types of samples were investigated: (1) samples prepared by physical vapor deposition (PVD) consisting of a thermally evaporated Ir(ppy)3 layer of 20 nm in thickness as described in Ref. 27. (2) Drop-cast samples on SiO2 which were prepared from Ir(ppy)3 dissolved either in chloroform (CHCl3) or chlorobenzene (MCB). The concentration of Ir(ppy)3 in CHCl3 was mol/l; for the solution of Ir(ppy)3 in MCB, the concentration was mol/l. Prior to preparation of the drop cast samples, all Si substrates with their natural oxide layer were cleaned in an ultrasonic bath in ethanol and acetone subsequently;28 all solvents were of HPLC grade (Carl Roth GmbH + Co. KG, Karlsruhe) and used as received without any further purification. (3) “Powder” samples of undissolved Ir(ppy)3 powder were prepared by directly pressing the Ir(ppy)3 powder as purchased into a piece of cleaned indium foil ( cm2). All samples were stored under ambient conditions prior to the measurements.
For DINeC measurements, neutral SO2 clusters were generated via adiabatic expansion of a 3% SO2/97% He gas mixture through a pulsed nozzle which was operated at 2 Hz with an effective opening time of 0.5 ms. With a pressure of 15 bar of the gas mixture in the nozzle, the supersonic expansion into the vacuum chamber (base pressure mbar) results in SO2 clusters with a mean size of – SO2 molecules and a mean velocity of m/s.29 The SO2 clusters hit the sample surface and desorb/ionize a fraction of the analyte molecules; the desorption proceeds via dissolution of the analyte in the shattering cluster and its fragments which serve as a transient matrix.21,30 In the further course of the desorption process, most of the SO2 molecules attached to the desorbed analyte are evaporated; this can lead to the bare analyte molecule in the gas phase.31 Depending on the analyte and the exact beam parameters, some of the remainders of the cluster fragments in which the analyte was dissolved can be detected as adducts to the analyte molecule itself if the measurement is performed with a time-of-flight spectrometer.21,32 In the present study, the desorbed and ionized analyte molecules were analyzed in a commercial ion trap mass spectrometer (amaZon SL, Bruker Daltonic GmbH, Bremen, Germany) which is equipped with a custom-built DINeC-source.22 In the ion trap, loosely bound SO2 adducts from the shattered cluster, if still present, are removed as the analyte ions undergo multiple collisions with the cooling gas in the ion trap. DINeC-MS measurements were performed in dark.
III. RESULTS AND DISCUSSION
Figure 1(a) shows a mass spectrum as obtained from an Ir(ppy)3 sample which was prepared via PVD; the spectrum was averaged over the first 30 min of the measurement. The obtained isotopic pattern [Fig. 1(b)] is dominated by the 191Ir and 193Ir isotopes and can be mainly assigned to [Ir(ppy)3]+. Additional peak groups, which are shifted by to higher values [Fig. 1(c)], are also observed. They are attributed to the oxidation of one or more phenylpyrdine ligands to 2-phenylpyridine 1-oxide.33,34 A small peak around ( % of the intensity of the main peak) is attributed to [Ir(ppy)2]+, which might result from sample preparation or the desorption process itself. No further fragments such as [Ir(ppy)1]+ or [Ir]+ were observed (compare Figs. S1 and S235). Whereas the spectrum in Fig. 1 was integrated over 30 min, three different time intervals of the measurement are shown in Figs. 2(b)–2(d). These spectra are directly compared to simulated spectra of chemically ionized and electronically ionized Ir(ppy)3, which are shown in Figs. 2(a) ([Ir(ppy)3+H]+) and 2(e) ([Ir(ppy)3]+), respectively. In Figs. 2(b) to 2(d), the measured data (depicted as red line) are fitted with a superposition of the isotope pattern for [Ir(ppy)3+H]+ (contribution shown as blue-shaded area) and [Ir(ppy)3]+ (contribution shown as plain green area). For the first minute, the spectrum is dominated by chemically ionized [Ir(ppy)3+H]+; almost no electronically ionized [Ir(ppy)3]+ is observed. Within the first 5 min, the spectrum changes from predominantly chemically ionized [Ir(ppy)3+H]+ towards electronically ionized [Ir(ppy)3]+. After a measurement time of about 15 min, the spectrum is dominated by electronically ionized [Ir(ppy)3]+; this trend further continues with longer measuring time. When compared with typical erosion rates measured for DINeC,36 about 1 nm is expected to be eroded during 10 min of measurement.
Both ionization mechanisms, proton uptake and electron abstraction, have been observed in DINeC-MS before. Proton uptake is the predominant ionization mechanism when basic functional groups are present in the respective molecules; this process is promoted by the presence of H O in the sample which, in the interaction with SO from the impacting clusters, forms sulfurous acid, the latter acting as an efficient proton supply.32,37,38 In the case of Ir(ppy)3 , which does not exhibit an explicit basic functional group, proton attachment can occur at one of the C atoms originally bound to Ir in analogy to acid-induced protonation in solution.39
Electronic ionization in terms of electron abstraction has been previously reported for DINeC-MS, e.g., for porphyrins with center metal ions;38 but also for bare alkali atoms, the separation of cation and electron in the shattering cluster has been observed.40,41 The creported values for the ionization energy of Ir(ppy)3 (experiment: 5.2–5.3 eV, condensed phase;42,43 theory: 5.9 eV, gas phase44,45), are indeed comparable to the range of the ionization energies of the systems for which electron abstraction induced by DINeC has been previously observed (cobalt(II)tetraphenylporphine, gas phase: 4.9 eV;46 alkali atoms: between 3.9 and 5.4 eV for Cs to Li atoms in the gas phase47). Density functional calculations indicate electron abstraction from the 5d-related orbitals in Ir(ppy)344,45 which formally can be seen as an Ir(III) 5d complex.44 Given the low electron affinity of SO of 1.1 eV,48 electron abstraction from Ir(ppy)3 is a strongly endothermic process. On the other hand, the generated ions are energetically stabilized by the polar SO molecules of the cluster fragments in which they are dissolved during cluster surface impact.30,32 According to our results, the total process can then be activated under the conditions of cluster surface impact with its high transient local temperatures of up to several thousand Kelvin.21
Figure 3 shows the absolute value of the integrated intensity for each contribution to the fit as a function of time: In this example, the contribution of [Ir(ppy)3+H]+ decreases to a similar extent as [Ir(ppy)3]+ increases and the overall intensity stays almost constant. Whereas the initial ratio of [Ir(ppy)3+H]+ and [Ir(ppy)3]+ does not change much from sample to sample, the strength and speed of the increase of the [Ir(ppy)3]+ signal varies. As an example, a slightly slower and less pronounced increase of the [Ir(ppy)3]+ signal intensity with increasing measuring time is observed in Fig. S3.35
In any case, the change of the type of ionization can be reversed by taking the sample out of the vacuum chamber: a measurement with a sample that was exposed to air after a first measurement shows the same initial contribution of [Ir(ppy)3]+ and [Ir(ppy)3 + H]+ as a freshly prepared sample. Furthermore, the observed change of ionization mechanism with measurement time is also reset in part by exposing a sample to a much lower H2O vapor pressure in the vacuum chamber: Fig. 4 shows the relative contribution of [Ir(ppy)3+H]+ and [Ir(ppy)3]+ during two measurement cycles, which were separated by dosing H2O for 60 min at a partial pressure of mbar. After exposing the sample to H2O vapor, the initial relative contributions of the different ionization mechanisms are again closer to the values measured at the beginning of the first measurement and the overall evolution of the relative contributions is also similar to the first measurement but the final saturation values are reached on a shorter time scale.
The decrease of the [Ir(ppy)3+H]+ signal and the recovery of chemical ionization with exposure to H2O can be explained on the basis of the importance of H2O for the chemical ionization process: traces of water, e.g., in the form of a molecular film on the analyte surface, were previously shown to interact with SO2 from the SO2 clusters forming sulphurous acid which acts as a proton supply and facilitates chemical ionization by means of proton uptake.32,37,38 As more and more H2O molecules are desorbed from the surface with increasing measurement time, also the concentration of available protons is decreased, leading to a decrease of chemically ionized Ir(ppy)3 species. Once water is resupplied either via exposure of the sample to a controlled dose of H O vapor pressure or to ambient conditions, the chemical ionization is recovered.
The increase of electronically ionized Ir(ppy)3 species with measurement time also has to be explained by the reduction of H O on the surface as it is reset by exposure to water as well. However, in the case of electronic ionization, there is no straight-forward explanation how surface-adsorbed water reduces the ionization efficiency. Possible explanations may include that the surface-adsorbed water layer changes the dynamics of the cluster surface impact and with this the probability for the strongly activated process of electron abstraction. With our experiments, we cannot give a definite answer on the origin of the increase of the [Ir(ppy)3]+ signal with time; from our experimental results, we only can conclude that surface-adsorbed water plays a role also in this case. For the PVD samples, the combined desorption/ionization efficiency is comparable for chemical and electronic ionization.
This situation indeed changes when samples which were prepared by different methods are compared. Samples consisting of Ir(ppy)3 powder pressed into indium foil showed an initial ionization behavior as well as a dependence on measurement time, which were very similar to those of the PVD samples (Fig. 5). Although the final value of the ratio between [Ir(ppy)3+H]+ and [Ir(ppy)3]+ is slightly different for the two types of samples, the overall trend is very similar. In contrast, the two samples prepared by drop casting from solutions based on organic solvents, here MCB and CHCl3, show very different behavior when compared to the powder and PVD samples: The ionization process in the former samples was found to be dominated by electronically ionized Ir(ppy)3 from the very beginning of the measurement; the final distribution shows more than 95% of electronically ionized Ir(ppy)3.
The initial absolute intensity of [Ir(ppy)3+H]+ and its decrease with time [Fig. S4(b)35] is comparable to the PVD and powder samples, and the observation is thus interpreted along the same line. The much higher initial absolute value of the [Ir(ppy)3]+ signal (Figs. 5 and S435) is then correlated to the different morphology of the samples investigated: When prepared from solutions under ambient conditions, relatively large microcrystals on the m-scale are formed, whereas the PVD and powder samples show structures on a sub-100-nm-scale (compare Fig. 6). With respect to the higher desorption/ionization efficiency of the [Ir(ppy)3]+ species observed for the drop-cast samples, this might be discussed, e.g., in terms of a change of electronic structure with sample morphology. With larger crystal size and less defect density, a change in the work function is expected. For electron abstraction as a strongly activated process, even a slight change may lead to a pronounced difference in the ionization efficiency. The further increase in intensity of the [Ir(ppy)3]+ signal with measurement time, which can be again reset by exposure to ambient conditions/increased partial pressure of water (Fig. S535), has then to be discussed in analogy to the PVD samples.
IV. CONCLUSIONS
DINeC mass spectrometry was demonstrated to lead to almost fragmentation-free desorption of Ir(ppy)3 from solid samples which were prepared by various methods. Two different ionization mechanisms were observed, i.e., chemical ionization via protonation resulting in [Ir(ppy)3+H]+ and electronic ionization by electron abstraction leading to [Ir(ppy)3]+. The relative contribution of these two ionization mechanisms of Ir(ppy)3 was further found to change with measuring time; regardless the initial ratio of the two contributions, we always observed a shift toward [Ir(ppy)3]+ with time. The decrease of the [Ir(ppy)3+H]+ signal with time was explained taking into account the interplay between SO from the cluster beam and surface-adsorbed water which promotes chemical ionization by proton supply but depends on the availability of water on the surface. With increasing measuring time, less water is available on the surface necessary for this process. The [Ir(ppy)3]+ signal was found to strongly depend on the sample preparation scheme with much higher intensity for drop-cast samples with larger crystalline structures.
As a result of the complex nature of the ionization process, sample preparation and the detailed experimental parameters influence the obtained signal intensity measured by means of DINeC-MS from Ir(ppy)3, the latter serving as an example for active OLED material. From a practical point of view, this might restrict the applicability of DINeC-MS as an analytical tool for this kind of materials. Nonetheless, in particular, the fragmentation-free desorption process can be of advantage when it comes, e.g., to the investigation of degradation processes in such materials, which then can be identified in terms of the fragments generated during these processes.24,49
ACKNOWLEDGMENTS
The authors acknowledge financial support from BMBF through Grant No. 05K19RG1.
AUTHOR DECLARATIONS
Conflict of Interest
The authors have no conflicts to disclose.
Author Contributions
Philip Keller: Conceptualization (equal); Data curation (lead); Formal analysis (lead); Visualization (lead); Writing – original draft (lead); Writing – review & editing (equal). Satoka Aoyagi: Conceptualization (equal); Investigation (supporting); Resources (lead); Writing – review & editing (supporting). Michael Dürr: Conceptualization (lead); Funding acquisition (lead); Project administration (lead); Resources (lead); Supervision (lead); Writing – original draft (equal); Writing – review & editing (lead).
DATA AVAILABILITY
The data that support the findings of this study are available from the corresponding author upon reasonable request.