Potassium Hexafluoroacetylacetonate Complex with 18-Crown-6 Ether as a Volatile Precursor of Molecular and Inorganic Films: Thermal and Structural Insights

Abstract

Volatile coordination compounds are widely used as precursors for the gas phase synthesis of functional materials. However, such complexes are still very rare for alkali metals, especially for heavy representatives of this family (potassium, rubidium, cesium) due to the tendency to form polymeric structures. This work is devoted to the exploration of a potassium hexafluoroacetylacetonate complex with 18-crown-6 ether, K(18C6)(hfac), as a unique volatile precursor with an isolated molecular structure. A convenient synthesis procedure was developed, and key structural features were identified including temperature-dependent effects. The thermal properties of the complex were studied via thermogravimetry and measurements of saturated vapor pressure using the Knudsen effusion method with mass spectrometric registration of the gas phase composition. Both from solution and the gas phase, the molecular films of K(18C6)(hfac) obtained exhibit a strictly (h00) orientation, where half of the surface cations have a coordination sphere accessible to supramolecular contacts. For the first time, the possibility of producing potassium-containing films from a fluorinated precursor by metal–organic chemical vapor deposition (MOCVD) has been demonstrated. With oxygen as the reactant gas, potassium fluoride forms and interacts with the silicon substrate, while introducing water vapor significantly reduces the fluorine content, suggesting its suitability for the preparation of oxide films.

1. Introduction

Volatile metal complexes are widely used as precursors in gas phase processes to produce thin film materials and nanostructures. In particular, physical vapor deposition (PVD) methods based on the condensation of the complex vapors without any decomposition/conversion are actively applied to prepare molecular films/particles. These materials are of interest for applications such as OLEDs [1,2,3,4], organic thin film transistors [5,6], electrochemical and electrophysical sensors [7,8], and, most recently, as primary layers for the growth of metalorganic framework (MOF) films [9,10]. Methods based on chemical transformations of metal complex vapors, such as metal–organic chemical vapor deposition (MOCVD) and atomic layer deposition (ALD), are effective and scalable for the targeted formation of inorganic films and nanoparticles on a wide variety of substrates, including complex-shaped (trenches/high-porosity/3D) objects [11,12,13], insulating and polymer materials [14,15], or molecular layers [16]. The inorganic systems obtained via these methods are in high demand in various fields, from aerospace, energetic and biomedicine to micro- and optoelectronics [17,18,19]. Finally, new concepts of in situ reactions involving volatile metal–organic precursors are also being actively developed, for example, the synthesis of MOF films [20] and organic–inorganic hybrid materials [21].

Fluorinated precursors are of particular interest because they are commonly more volatile and chemically reactive than their non-fluorinated counterparts [7,8,22]. For a number of metals such as alkali/alkaline earth ones and the lanthanides, complexes with fluorinated ligands are also attractive for the vapor-phase synthesis of inorganic fluoride systems without the use of aggressive co-reagents (such as HF). Moreover, several examples demonstrate the attractive possibility of producing both fluoride and oxide materials, depending on the deposition conditions [23,24,25].

However, the chemistry of volatile fluorinated precursors for alkali metals remains poorly developed. Based on fluorinated β-diketonate ligands (L⁻ = R^FC(O)CHC(O)R), some success has been achieved recently for lithium and sodium complexes [26,27,28], while for the heavier representatives of the group starting with potassium, fluorinated precursors are mostly limited to heterometallic complexes [29]. Although such heterometallic species are suitable for the preparation of mixed-metal films (perovskites/fluoroperovskites), they do not allow for varying the stoichiometry or obtaining pure fluorides or oxides. Thus, monometallic potassium fluorinated precursors will be in demand for the production of films with improved emission properties [30,31] and mixed functional materials of new compositions [32].

For monometallic compounds to be volatile, an isolated molecular structure with weak intermolecular contacts in the crystal lattice is desirable [33]. Thus, in addition to the stabilizing anionic ligand, the coordination sphere of the potassium cation must be saturated with a neutral ligand. The latter should be multidentate due to the high K⁺ coordination numbers [34], tightly bound to the metal center to ensure stability, and with a molecular weight that is as low as possible to allow volatility. Recent studies have also emphasized the importance of the steric rigidity in the neutral ligand. In particular, the use of sterically flexible polyglymes, which have been successful for preparing volatile lithium and sodium complexes [26,28], resulted in the formation of only polymeric structures in the case of potassium [35]. In contrast, the use of 18-crown-6 ether (18C6) as a complementary ligand for potassium β-diketonates resulted in the formation of promising, in terms of volatility, structural motifs consisting of isolated K(18C6)(L) molecules linked by weak intermolecular contacts [36,37,38]. However, the properties of such complexes remain poorly studied.

Thus, the present work is devoted to a comprehensive study of the prototypical fluorinated complex of this family, the potassium hexafluoroacetylacetonate derivative, K(18C6)(hfac), as a potential volatile precursor. The research focus is on structural and thermal features, their correlations, the first clear evidence and quantitative volatility data, and its preliminary testing for the fabrication of molecular, fluoride, and oxide films.

2. Results and Discussion

2.1. Synthesis and Characterization of K(18C6)(hfac)

Although the title complex K(18C6)(hfac) has been previously known [36], it was a byproduct of the reaction of potassium metal with 18C6 and Pb(diglyme)(hfac)₂ (diglyme = (CH₃OCH₂CH₂)₂O) in toluene, characterized only by single-crystal XRD.

Thus, the synthetic approaches of this complex were explored in the current study. In particular, an effective route is the direct reaction of commonly available β-diketonate K(hfac) with crown ether similarly to [37,38,39]. The reaction could occur in a wide set of standard organic solvents, with the exception of aliphatic hydrocarbons. Compared to alcohol/ethyl acetate, the use of chloroform or toluene offers advantages in terms of visual process control while maintaining a high yield (~90%). In fact, the starting β-diketonate K(hfac) is insoluble in oxygen-free organic solvents, unlike the target K(18C6)(hfac) complex. The product was finally purified using a simple solution method, ensuring the required purity (>99 wt.%), confirmed by elemental analysis, IR, NMR, and TGA.

Spectral data such as IR and NMR for K(18C6)(hfac) are also presented herein for the first time. Overall, the IR spectrum is similar to that of the related complex K(18C6)(ptac) [38] (Figure S1a). Figure S1b shows the comparison between the IR spectra of K(18C6)(hfac) and K(hfac) in KBr pellets in the range from 400 to 4000 cm⁻¹. The spectrum of the heteroligand complex exhibits vibrational bands of both the initial potassium β-diketonate and the neutral ligand vibrations. Compared to the vibrational spectrum of K(hfac), the IR spectrum of the K(18C6)(hfac) complex exhibits intense vibrational bands in the 2700–3000 cm⁻¹ region, which are assigned to vibrations of the C-H groups (ν(C-H)) and are also clearly visible in the crown ether spectra [40,41]. The positions of the absorption bands for the β-diketonate chelate ring vibrations (ν(C=O) + ν(C=C), 1685–1450 cm⁻¹) remain comparable (Figure S1a). The characteristic bands of the crown ether ring vibrations (ν(C-O), 1150–1085 cm⁻¹) overlap with the intense absorption bands of the fluorinated groups of hfac-ligand (ν(C-F), 1300–1000 cm⁻¹). However, the local changes in this region are still observed: the intensity of the bands at 1477 and 1354 cm⁻¹, corresponding to skeletal vibrations of the macrocyclic ring [40], increases significantly (Figure S1a). The characteristic bands related to the deformations of –O–C–C– units of the 18C6 ligand (964 and 839 cm⁻¹) [41] are also clearly visible.

According to powder XRD, the polycrystalline K(18C6)(hfac) samples obtained completely correspond to the known crystal structure [36] (Figure S2a). Interestingly, the composition of this heteroligand complex is preserved when the reagent stoichiometry changes (Figure S2b), unlike the cesium analog, for which other complexes such as Cs₂(18C6)₃(hfac)₂ can be easily obtained [39].

2.2. Structural Features

The primary description of the crystal structure of K(18C6)(hfac) is given in [36]. Within the current work, we extended the structural study, focusing on intermolecular contacts, crystal packing, and temperature effects. In the further comparative analysis, we present the distances and angles at 150 K, while data for other temperatures are given in the Supplementary Materials and discussed in the corresponding section.

Briefly, the titled complex has an isolated molecular structure due to the coordination of all oxygen atoms of both types of ligands (Figure 1a). The K-O_hfac bond lengths with β-diketonate ligand (2.69–2.78 Å) are close to those of the polymeric K(hfac) derivatives [42]. The K-O_18C6 bond lengths with neutral ligand are significantly elongated and vary within the wide range 2.80–2.95 Å, which is typical for similar heteroligand potassium complexes K(18C6)(L) with coordinating anions L such as β-diketonates [37,38], carboxylates [43], and phenolates [44] (Table S1). This is explained by the fact that, although the relatively small K⁺ cation fits perfectly within the 18C6 macrocyclic cage [45], in such complexes it is significantly shifted out: the distance from K⁺ to the midplane of 18C6 reaches 0.93 Å (Table S1). In contrast, in the case of weakly or non-coordinating anions L, such as hexafluorophosphate and thiocyanate [46], the cation is localized precisely in the midplane of the crown ether, and then the bond length range decreases significantly (to 2.77–2.84 Å, Table S1). Interestingly, the corresponding triflate complex [46] occupies an intermediate position (Table S1). Further, we will focus on comparing the structure of the K(18C6)(hfac) complex with the β-diketonate analogs that contain both fluorinated (L = ptac) and non-fluorinated (L = acet) ligands (

Table 1. Selected bond lengths and angles in the structures of potassium β-diketonate complexes K(18C6)(L), where terminal substituents are also indicated for L ligands.

β-Diketonate Ligands	L (R1, R2)
	hfac (CF3, CF3)	ptac (CF3, tBu)	acet (Me, OEt)
Ref.	Current work	[38]	[37]
T, K	150(2)	150(2)	293
d(K–OL), Å	2.689(2)–2.780(2) 〈2.73(5)〉	2.648(3)–2.688(3) 〈2.67(3)〉	2.650(3)–2.732(3) 〈2.69(3)〉
d(K–O18C6), Å	2.801(2)–2.948(2) 〈2.89(3)〉	2.802(3)–3.032(3) 〈2.95(3)〉	2.828(3)–3.016(3) 〈2.93(3)〉
d(K…(18C6)), Å	0.71	0.92	0.91
d(M…F(H))intermolecular, Å	3.03–3.07	2.93	-
OLKOL chelate angle, °	64.9	66.6	68.2
Bending angle of the β-diketonate metallocycle along OL…OL line, °	12.9	12.0	24.9
Tilt of anion relative to (18C6), °	65.4	62.1	51.2
Potassium coordination number	8 + 1	8 + 1	8

).

A feature of the crystal packing of fluorinated β-diketonate complexes K(18C6)(L) is the specific intermolecular contacts “through” the crown ether ring. In particular, at L = hfac, the K…F contacts are observed to be generating chains along the a-axis (Figure 1b). The corresponding distances (3.03–3.07 Å) are comparable to those in polymeric K(hfac)-based polymeric structures (2.92–2.96 Å) [42] and K-O_18C6 bond lengths. A similar chain motif was also observed at less fluorinated L = ptac [38] due to the interaction of the cation with the tert-butyl group (d(K…H) = 2.93 Å). Thus, the coordination number of the cation in both structures is better described as 8 + 1 (Table 1). The chains are in opposite directions so that adjacent molecules are complementary (Figure 1b), and as a result, there is a noticeable tilt of the anion relative to the crown ether plane (65.4° and 62.1° for L = hfac and ptac, respectively). In this case, the β-diketonate metallocycles are relatively weakly deformed: bending angles along the O_L…O_L lines reach only 12–13° (Table 1). On the contrary, for a non-fluorinated analog (L = acet) [37], the specific intermolecular contacts are absent. In this discrete structure, the tilt of the anion is significantly reduced (51.2°) while the bending angle of the β-diketonate metallocycles is significantly increased (29.4°).

A more comprehensive analysis of the intermolecular contacts for K(18C6)(hfac) was performed by constructing the Hirschfeld surface (Figure 2). As expected, the most significant contact was the K…F contact, which forms the chain of molecules (Figure 2a). In the bc plane, there are also H…C contacts between crown ether and the carbonyl group (2.73 Å) of the complementary molecules and H…H contacts between the crown ethers (2.25 Å). In general, according to 2D fingerprint plots, the main relative contribution is formed by H…H and H…F contacts (Figure 2b). Compared with the analog for L = ptac [38], for K(18C6)(hfac), the contribution of H…H contacts is expectedly reduced (59.6 => 37.3%), while the contribution of H…F (10.2 => 20.0%) and F…H (11.2 => 20.7%) contacts is increased due to the structure of the β-diketonate ligand.

It is worth noting that the shape of the K(hfac)(18C6) molecule was found to be close to spherical (globularity index 0.74). Owing to this, a search for pseudo-periodicity in the crystal packing was performed using the translational sublattice approach [47,48,49]. The positions of the K⁺ cations approximately correspond to the geometric centers the formula units, with a maximum deviation of 0.51 Å. Therefore, a theoretical diffraction pattern was calculated considering only K⁺ cations (i.e., excluding all other atoms; Figure 3). It is important that the found subcell parameters for K(18C6)(hfac) (a_T = 9.67 Å, b_T = 8.74 Å, c_T = 7.23 Å, α_T = 88.31°, β_T = 115.79°, γ_T = 88.54°, Figure 3a) are close to those for the K(18C6)(ptac) analog (a_T = 9.25 Å, b_T = 10.08 Å, c_T = 7.33 Å, α_T = 119.12°, β_T = 88.66°, γ_T = 88.86°) [38]. Both sublattices are geometrically close to a primitive hexagonal lattice, which represents the fundamental structural motif underlying these packings. Thus, the analysis of translational sublattices is a powerful tool for identifying structurally related complexes. In particular, despite the different types of intermolecular contacts (see above), such analysis made it possible to accurately prove the relationship of the structures for K(18C6)(L) complexes (L = hfac, ptac). This is most clearly illustrated by the coincidence of the theoretical XRD patterns constructed only for the heaviest element potassium (Figure 3). In turn, for L = ptac, we showed that iso-ligand complexes of potassium and rubidium are isostructural [38], which allows us to assume a more general trend in the structural consanguinity of this family.

To identify thermal effects and expansion tensor, a SC-XRD study was conducted over a wide temperature range (100–400 K), including both cooling and heating. The primary results are presented in Tables S2 and S3 while the relative changes in parameters and the corresponding equations are summarized in Table S4.

The linear cell parameters and volume characteristically increase with increasing temperature (Figure 4a and Figure S4). The greatest relative change (4%, Table S4) is observed for the parameter a, along which the molecular chains are extended, leading to a corresponding elongated shape of the thermal expansion tensor (Figure 4b and Figure S5). Interestingly, the γ angle decreases with increasing temperature (Table S4, Figure S4), while the β angle increases up to 350 K and then decreases sharply (Table S4, Figure 4c). However, these specific changes are too small to have a significant impact on the cell volume.

In order to explain the features of the thermal expansion tensor, a preliminary assessment of the intermolecular interaction was carried out using Crystal Explorer 17.5 (CE-B3LYP/6-31G(d,p)) [50]. The obtained data on pairwise interaction potentials are presented in Figure 5. Within the bc plane, the total interactions between molecules have greater energy (6 neighbors, energy values 21.5 to 72.9 kJ/mol) while relatively weaker interactions are observed along the a-axis (2 neighbors, energy values 59.3 kJ/mol). This appears to favor the thermal expansion in the a-direction.

When discussing the influence of temperature on the parameters of a K(18C6)(hfac) molecule (Table S3), it should be noted that the K–O_hfac bond length and the bending angle of the β-diketonate metallocycle along the {O…O} line increase most noticeably (0.05 Å and 2° respectively). This is due to the increase in the displacement of the cation from the 18C6 midplane by 0.06 Å. Changes in the remaining bond lengths and angles are significantly smaller. Thus, this direct experimental study correlates well with the previous one [38] and allows for further comparison of the structure of molecules of similar complexes for the crystals studied at different temperatures.

The 2D fingerprint plots of intermolecular interactions also show only minor changes with increasing temperature, indicating that the overall packing and interaction pattern remain largely preserved. In fact, the contributions of H…H and H…F/F…H contacts slightly increase from 37.2 to 38.2% and 40.5 to 41.8%, respectively, while those of H…O/O…H and K…F/F…K decrease from 10.8 to 9.6% and from 1.8 to 1.4%, respectively. At the same time, the intermolecular contact distances change more significantly, by 0.13–0.15 Å with an increase in temperature from 100 to 400 K (Table S5). It is interesting that the greatest elongation of the key K…F contact (~0.07 Å) occurs in the range of 350–400 K, and the final value (3.165 Å) exceeds the accepted K-F bond length (3.1 Å [34,51]). Such weakening of contact can facilitate the transition of the complex into the gas phase. For example, a related sodium complex with 15-crown-5 ether, Na(15C5)(hfac), where only weak intermolecular contacts are observed, exhibited appreciable volatility (sublimation at 390 K, 5 Pa) [52].

2.3. Thermal Properties

The thermal properties of K(18C6)(hfac) in the condensed phase were studied by TGA in an inert He atmosphere (Figure 6). The wide temperature range of mass constancy up to 453 K (180 °C) indicates the thermal stability of the complex. The distinct endothermic effect on the c-DTA curve at 463 K (190 °C) corresponds to melting, as confirmed visually on a Kofler bench. The main step of mass loss is complete at approximately 573 K (300 °C), and the subsequent prolonged course of the curve indicates thermal annealing of the carbon-containing decomposition products. The residual mass corresponds to the formation of KF (exp.: 11.3%, calcd. for KF: 11.4%). A comparison with the initial potassium β-diketonate shows that the TG curves are similar, although the main mass loss for K(hfac) begins at temperatures approximately 20 degrees higher (Figure 6).

Thus, under heating at atmospheric pressure, the K(18C6)(hfac) complex decomposes and no volatilization is observed, unlike its analog K(18C6)(ptac) [38]. However, this behavior in TGA does not mean that the complex is non-volatile and unsuitable for vapor-phase deposition. In fact, MOCVD and PVD processes are typically carried out at reduced pressure, which facilitates vaporization, as clearly demonstrated in [53].

Therefore, to obtain the most complete information on the vaporization process, including vapor composition, we employed the Knudsen effusion method with mass spectrometric registration of the gas phase. The typical mass spectrum in the temperature range of 379–421 K (106–148 °C) is presented in

Table 2. Mass spectrum of the gas phase over the solid complex K(18C6)(hfac) (ionization energy 70 eV).

Ion	m/z	I, rel., 393 K
K+	39	100
(OCH2CH2O)+	60	7
CF3+	69	6
(CH2CH2)2O+	72	4
(CH2CH2O)2+	88	10
(CH2CH2O)3+	132	2
(CH2CH2O)4+	176	1
K(hfac)+	246	-
K(18C6)+	303	60

. The main intense ion current is K⁺, which is typical for volatile alkali metal compounds [54,55]. The transition of the target complex into the gas phase is confirmed by the high relative intensity of the K(18C6)⁺ ion current, while the absence of the molecular ion is most likely due to the decay of the molecule under electron impact. Moreover, the fragmentation pathway differs from that for the related heteroligand non-fluorinated complex, where all three types of metal-containing ions, K(phen)(L)⁺, K(phen)⁺, and K(L)⁺ (phen = 1,10′-phenantroline, L = β-diketonate-ion, dipivaloylmethanate), were registered [56]. The absence of the K(L)⁺ fragment in our case is most likely explained by a significant energy gain in ionization through the formation of a large cation with crown ether, K(18C6)⁺, and a neutral or negatively charged fluorinated β-diketonate ligand (hfac). Interestingly, noticeable vaporization was actually observed at temperatures above 350 K, as expected from SC-XRD data on the elongation of the key intermolecular contact (see Section 2.2).

In general, the constancy of all ion currents (Figure S6) and the absence of non-volatile residue in the effusion cell at the end of the experiment suggest a congruent sublimation process for the complex according to the phase transition:

K(18C6)(hfac) _(solid) → K(18C6)(hfac) _(vapor).

(1)

As an example, temperature dependence of the K⁺ ion current intensities is presented in Figure S7. Using these dependences for the intensities of K⁺, (CH₂CH₂O)₂⁺, and K(18C6)⁺ ion currents, the sublimation enthalpy for the K(18C6)(hfac) complex was calculated through the Clausius–Clapeyron equation and the least squares method (Table S6). It is important that the enthalpy values determined for different ions are in good agreement, which indicates the correctness and consistency of the data.

As a result, the following values of the thermodynamic parameters of sublimation processes were obtained as the arithmetic mean of 12 measurements: ∆_sH₄₀₀° = 159.0 ± 4.7 kJ/mol, ∆_sS₄₀₀° = 259 ± 10 J/(mol·K). The temperature dependence of the saturated vapor pressure for K(18C6)(hfac) is presented in Figure 7. Based on the absolute values of the partial pressure (Table S7) and the sublimation enthalpy, the following equation is obtained for the dependence of the pressure (Pa) of K(18C6)(hfac) molecules on temperature:

$$\lg p_{\mathrm{K}\left(18\mathrm{C}6\right)\left(\mathrm{h}\mathrm{f}\mathrm{a}\mathrm{c}\right)}/\mathrm{P}\mathrm{a}=-{\displaystyle \frac{8255\pm 250}{T}}+\left(19.16\pm 0.10\right),T=379-421\mathrm{K}$$

(2)

Since quantitative data on the volatility of related heteroligand fluorinated alkali metal complexes are not yet available, the vapor pressure of K(18C6)(hfac) was compared with that of the iso-ligand analog for the alkaline earth metal, Ba(18C6)(hfac)₂ [57]. The barium complex was shown to be significantly more volatile (Figure 7). This is consistent with the general trend that the vapor pressures of iso-ligand complexes of alkali metals are consistently lower than those of alkaline earth metals, as demonstrated for a number of compounds families, the most representative of which are pivalates [55,58]. This once again supports the correctness of the thermodynamic values obtained here.

Finally, it should be noted that vapor pressures sufficient for the convenient flow-type MOCVD processes (0.01–0.1 Torr) are achieved for K(18C6)(hfac) at temperatures of ~163–188 °C (436–461 K), which is acceptable for practical applications.

2.4. Molecular Film Deposition

To confirm the preservation of the vapor composition of K(18C6)(hfac) in the MOCVD reactor used, which has an elongated mass transfer path [59,60], we performed the corresponding experiments in the absence of a reactant gas and without special substrate heating. In fact, this represents thermal evaporation with vapor transport by the carrier gas, i.e., a modification of physical vapor deposition (PVD), but under moderate vacuum (1 Torr) due to the argon presence. The evaporation temperature chosen was 443 K (170 °C), which ensured the precursor partial vapor pressure of 0.02 Torr. Note that the glass substrate was slightly heated up (323–333 K, 50–60 °C) due to heat transfer inside the reactor.

The white films were thereby formed on a glass substrate, and their Raman spectra completely corresponded to those of the original complex, confirming the target molecular composition (Figure 8a). It is worth noting that the XRD pattern of such a PVD layer, in comparison with the calculated one, shows the presence of only (h00) reflections (Figure 9a). Thus, the molecular layer obtained is strictly oriented parallel to the (100) plane.

To evaluate the influence of temperature and crystallization mode (gas phase/solution) on K(18C6)(hfac) film growth, the additional experiments using the spin-coating (SC) method were carried out. In this case, a diffusion-ordered NMR spectroscopy (DOSY) was also applied to confirm the molecular composition of the complex in solution (CHCl₃). The particle size was determined based on the measured diffusion coefficient (D = 1.06 × 10⁻⁹ m²/s) using the Stokes–Einstein formula. In fact, the calculated hydrodynamic radius d_h was ~8 Å, which correlates well with the Hirschfeld surface diameter for the K(18C6)(hfac) molecule (~10 Å). The latter estimation seems to be quite relevant, since the Hirschfeld surface of this molecule has a high sphericity index (see Section 2.2). Therefore, it can be assumed that the complex does not dissociate in CHCl₃ solution. The peak position on the UV-vis spectra of the molecular film obtained by the SC technique and K(18C6)(hfac) solution (Figure 8b) also match perfectly, confirming their identical chemical composition. A typical XRD pattern presents only reflections of the (h00) type, i.e., this orientation is maintained when the deposition conditions change (Figure 9a).

To clearly confirm the molecular film texture, a more precise 2D-GIXRD study was also carried out. In particular, a three-circle diffractometer equipped with a Mo-anode source was firstly applied. Previously, such measurements had been performed either on laboratory diffractometers with Cu anode sources or at synchrotron facilities [61,62]. The diffraction center was determined directly from the primary beam position, ensuring accurate indexing of the observed reflections (Figure S8). The diffraction pattern obtained confirmed the main texture and indicated the presence of several much less intense reflections corresponding to other lattice planes (e.g., [002]; Figure S8).

To explain this feature, some aspects of crystal packing of K(18C6)(hfac) were further considered. In terms of symmetry, the relative arrangement of the molecules along the c and b axes is similar: there are two types of layers (connected by the inversion center), which are shifted relative to each other (ABAB-type, cations are arranged in a zigzag; Figure S9). On the contrary, along the a-axis, the layer arrangement is different: the layers lie without displacements (AAAA-type, the cations are located one under the other; Figure S9), i.e., it is characterized by higher symmetry. For various objects, it has been shown that as the temperature increases, more symmetrical films are formed [63,64]. Note that parameter a showed the greatest relative increase with temperature [65].

However, SC experiments showed that the molecular film texture is preserved at room temperature. In this case, it is important to note that along the a-axis, there are more contacts of donor atoms (O atoms from 18C6 and F atoms of the anion), which can potentially provide better interaction with the surface of both the substrate and the primary film islands (Figure 9b). This correlates with the explanation for the preferred orientation of films proposed in [66]. According to this theory, initially, a number of domains with different orientations are formed, and then the domain that grows fastest along a particular direction begins to displace in adjacent directions. Thus, over time, only domains of this orientation remain. This is consistent with the 2D-GIXRD results, which show “traces” of other orientations (Figure S8). Similarly, molecular layers textured perpendicular to the directions characterized by the minimal energy of intermolecular interactions were obtained by the PVD and SC methods for a volatile iron(III) β-diketonate complex [67].

It is important that with the (h00) orientation of K(18C6)(hfac) films, half the surface is occupied by the molecules oriented “crown-up,” i.e., providing K⁺ cations available for further coordination (Figure 9b, dotted line). This feature is interesting from the perspective of supramolecular chemistry. For example, the naturally occurring mineral muscovite has long attracted attention due to the unique structure of its cleaved surfaces, which are decorated with regularly arranged surface K⁺ cations exhibiting short-range order [68]. Previous studies have demonstrated that the adsorption of water molecules can induce local rearrangements of these surface cations [69]. This observation provides experimental evidence that alkali metal ions located at well-defined surface sites can participate in reversible coordination processes. In our case, the surface morphology of the textured K(18C6)(hfac) molecular films is even more regular than that of muscovite, suggesting that similar ion–molecule interactions could be exploited to design molecular sensors or to obtain hybrid MOF materials.

2.5. Inorganic Film Deposition

Having confirmed that the K(18C6)(hfac) precursor composition was maintained during vapor mass transfer in the MOCVD reactor, we carried out corresponding test experiments to deposit inorganic films onto a common silicon substrate. The deposition temperature of 773 K (500 °C) and the reagent gas O₂ were chosen based on the data regarding the MOCVD preparation of NaF films on a Si(100) substrate from the related Na(tetraglyme)(hfac) complex [28]. On the other hand, it has previously been shown that oxide materials can be produced from fluorinated precursors by introducing water vapor into the reactor [70]. Therefore, we also conducted MOCVD experiments using O₂ bubbled through distilled water (reagent atmosphere: O₂ + H₂O).

2.5.1. MOCVD in O₂ Atmosphere

When using O₂ as the reactant gas, GIXRD reveals the formation of at least three crystalline phases, namely, KF, K₂SiF₆, and K₃SiF₇ (Figure 10a). Quantitative composition assessment using the corundum number method [71] shows a roughly equal mass phase ratio. According to the EDS data, the ratio K:F ≈ 1:5 (Figure S10a), while a ratio of 1:2 was expected for an equal mass mixture of KF, K₂SiF₆, and K₃SiF₇. Taking into account the notable carbon impurity (Figure S10a), the excess fluorine can be caused by incomplete decomposition of the fluorinated fragments of the ligand. According to SEM, irregularly shaped large agglomerates of varying sizes are visible on the sample surface, with the maximum film thickness reaching 2 microns (Figure 10b).

Most likely, the multiphase composition of the resulting film is formed due to not only direct vapor-phase deposition, but also secondary reactions between K(18C6)(hfac) decomposition products and the substrate. To our knowledge, there are no examples of in situ formation of fluorinated heavier alkali metal silicates through MOCVD. However, the reactions with substrate materials to form silicates and even platinate were observed when using lithium β-diketonates, but only at a significantly higher temperature (873 K, 600 °C) [27]. Moreover, the formation of M₂SiF₆ crystals (M = K, Rb, Cs) has previously been demonstrated during etching of the Si surface with fluorine-containing reagents (HF + MHF₂), M = K, Rb, Cs [72]. In turn, the formation of K₃SiF₇ is possible via the reaction of K₂SiF₆ with KF and/or KHF₂, which occurs under temperature conditions similar to those we used (T ≥ 723 K, 450 °C) [73,74]. The developed and uneven surface relief of the sample also indirectly confirm interaction with the substrate (Figure 10b). Thus, it can be assumed that the primary product of precursor decomposition is KF, while potassium fluoride–silicate phases are formed as a result of secondary reactions involving KF and Si. Nevertheless, additional studies are necessary to fully elucidate and confirm the mechanism.

It is important to note that both K₂SiF₆ and K₃SiF₇ become unique red phosphors when doped with Mn⁴⁺ [74]. Thus, our result seems to be interesting from the point of view of the in situ formation of inorganic phosphor systems through interaction with the silicon substrate. The application of the MOCVD method could significantly expand the geometries of related devices.

2.5.2. MOCVD in O₂ + H₂O Atmosphere

When H₂O vapor is introduced into the reactor, the morphology of the resulting film changes significantly and becomes smoother, with a thickness of ~450 nm (Figure 10b). GIXRD data show the presence of K₂CO₃∙1.5H₂O phase (Figure 10a), and the corresponding stoichiometry is consistent with the EDS data (K:O ≈ 1:6, Figure S10b). It cannot be ruled out that potassium carbonate forms upon exposure of the film to air. However, previous studies of lithium-containing films have shown that films formed by the secondary carbonization reaction of the primary oxide phase typically exhibit pronounced morphological inhomogeneity [27,75]. In contrast, the film obtained in the current work is characterized by a relatively smooth surface (Figure 10b). Given the thermodynamic stability of potassium carbonate [76,77] even at temperatures as high as used in MOCVD, it can be assumed that this phase formed as a result of the primary hydrolytic decomposition of the precursor. Due to the high film growth temperature, the inclusion of water to form K₂CO₃∙1.5H₂O is most likely caused by atmospheric moisture. Importantly, the sample contains a minimal amount of fluorine, which is confirmed by the EDS method (Figure S10b).

Finally,

Table 3. The experimental conditions and the results of MOCVD application of K(18C6)(hfac) and selected volatile alkali and alkaline earth metal precursors (T_evap. = evaporation temperature; T_d = deposition temperature; p = total pressure).

Precursor *	Evaporation Condition **			Deposition Conditions			Film	Ref.
	Tevap., K	p, Pa	Gas Carrier, Sccm	Tdepos., K	Gas Reagent, Sccm	Substrate	Composition
K(18C6)(hfac)	443	1.33	Ar, 30	773	O2/O2 + H2O, 30	Si(100)	KF + K2SiF6 + K3SiF7 (O2) K2CO3∙1.5H2O (O2 + H2O),	This work
K(thd)(phen)	513	800	Ar, 167	773	Flash-CVD: O2, 200; co-agent Nb(OiPr)4(thd)	SrTiO3, MgO	KNbO3	[56]
K(thd)	403–458	666	Ar, 75	923	O2, 600	Si(111)	KNbO3	[78]
Na(tetraglyme)(hfac)	383	low	Ar, 150	773	O2, 150	Si	NaF	[28]
Li(thd)	473	1333	N2, not specified	773–848	co-agents NH3 + triethyl phosphate		N-doped Li3PO4	[79]
Li(acac)	398	333	N2, 60	773	O2 + H2O, 60; co-agent Zn(acac)2∙H2O	Si(100)	Li-doped ZnO	[80]
Li4(thd)4	443–463	1.33	N2, 30	500–600	O2, 110	Si(100), fused Si	Li2CO3 (773 K), Li4SiO4 (873 K)	[27]
Li(ptac)	473–493						LiF + Li (Si, 773 K)
Li4(LOMe)4	453					Si(100), Pt/Si, fused Si	Li2CO3 (Si, 773 K); Li2SiO3 + Li4SiO4 (Si, 873 K); Li2CO3 + Li2PtO3 (Pt/Si, 873 K)
Li(LOMeF)	443–458						Li (Si and Pt/Si, 873 K)
Li12(hfac)12(monoglyme)∙4H2O	493	933	PI-MOCVD, solvent: monoglyme	1048	co-agent Nb(thd)4 (Li:Nb = 2:1)	C-sapphire	LiNbO3	[26]
Ca(diglyme)(H2O)(hfac)2	423	133	Air, 60	623–773	Air (93 Pa) + H2O (40 Pa)	Si(100), SrTiO3	CaF2	[81]
Ba(tetraglyme)(hfac)2	418	533	Ar, 150	1073–1223	O2, 600; co-agent Ce(diglyme)(hfac)3	MgO(100)	BaCeO3	[82]
Sr(PMDETA)(thd)2	503–563	600–1333	Ar, 100	573–773	O2, 500	Pt	SrCO3	[83]

* thd = 2,2,6,6-tetramethyl-3,5-heptanedionate; acac = 2,4-pentanedionate; phen = 1,10-phenanthroline; L_OMe = 2-methoxy-2,6,6-trimethyl-3,5-heptanedionate; L_OMe^F = 1,1,1-trifluoro-5-methoxy-5-methyl-2,4-hexanedionate; monoglyme = 1,2-dimethoxyethane; diglyme = bis(2-methoxyethyl)ether; PMDETA = pentamethyldiethylenetriamine. ** PI-MOCVD = pulsed-injection metal–organic chemical vapor deposition.

presents a comparative summary of the MOCVD parameters for K(18C6)(hfac) alongside representative alkali or alkaline earth metal precursors. Thus, we demonstrate for the first time the general possibility of effectively controlling the composition of potassium-containing films by varying the MOCVD reaction atmosphere.

3. Materials and Methods

3.1. Materials

The initial potassium hexafluoroacetylacetonate K(hfac) was obtained by the standard neutralization reaction of potassium carbonate (K₂CO₃, Reakhim, Moscow, Russia, 99%) with the β-diketone Hhfac (P&M-Invest CJSC, Moscow, Russia, 99%) in diethyl ether (Kuzbassorghim, Kemerovo, Russia, 98%) [42]. 18-crown-6 ether (Alfa Aesar, Haverhill, MA, USA, 99%), chloroform (Base No. 1 of Chemical Reactants, Moscow, Russia, 99%), and hexane (Vekton CJSC, Saint Petersburg, Russia, 99%) were also used for the complex synthesis and purification.

The planar substrates of a cover glass for microscope slides (Minimed, Moscow, Russia, 0.15 × 18 × 18 mm³) and Si(100) (Telekom-STV JSC, Moscow, Russia, 0.7 × 17 × 17 mm³) were used for the molecular and inorganic film deposition, respectively.

3.2. Characterization Methods

Elemental analysis for C, H, and F was performed at the Chemical Research Center for Collective Use, Siberian Branch of the Russian Academy of Sciences (Vorozhtsov Novosibirsk Institute of Organic Chemistry SB RAS, Novosibirsk). The standard errors did not exceed 0.5 wt.% [84]. IR absorption spectra were recorded on a Scimitar FTS 2000 spectrometer (Digilab LLC, Randolph, MA, USA) in the 400–4000 cm⁻¹ range of the complex samples pressing into KBr tablets as well as for the corresponding suspensions in a fluorinated oil. Raman spectra were recorded in back scattering geometry using a LabRAM Horiba single spectrometer (488 nm line of an Ar⁺ laser, HORIBA Jobin Yvon, Montpellier, France) equipped with a microscope. UV-vis spectra were recorded using a spectrophotometer UV-VIS-3101PC Shimadzu (Shimadzu Corporation, Kyoto, Japan).

The ¹H NMR spectra were recorded for the solution of the complex in CDCl₃ using standard 5 mm NMR tubes on a Bruker Avance III 500 spectrometer with BBI detector. (CH₃)₄Si was used as a standard (TMS) (Bruker BioSpin GmbH, Rheinstetten, Germany). The ¹H DOSY NMR experiment [85,86] was performed on the same equipment with a 5 mm BBI probe with z-gradient. Diffusion coefficients were measured using the standard Bruker Ledbpgp2s sequence. Sixteen registration scans were used in the NMR diffusion experiment. The data were analyzed using the standard Bruker Dynamics Center 2.1.8 software by fitting the integral intensity decay of the corresponding signal to the Stejskal–Tanner equation. The hydrodynamic radius d_h was determined from the diffusion coefficient D for spherical objects using the Stokes–Einstein relation:

$$d_h={\displaystyle \frac{k_{B}T}{3\pi \eta D}}$$

(3)

where k_B is the Boltzmann constant, T the temperature (298 K in this study), and η the solvent viscosity (0.542 cP in this study).

3.3. K(18C6)(hfac) Synthesis

A stoichiometric amount of solid 18C6 (0.21 g, 0.78 mmol) was added to a K(hfac) suspension (0.19 g, 0.78 mmol) in 10–15 mL of chloroform. Rapid dissolution of the precipitate was observed, and the reaction mixture was then stirred for 2 h at room temperature. After evaporation of the solvent in air at room temperature, the crystals of the target product were obtained. The product was purified by precipitating from a hot chloroform solution using cold hexane. Yield 90% (0.36 g, 0.71 mmol).

Elemental analysis (wt.%): for C₁₇H₂₅F₆O₈K found: C, 40.1; H, 5.2; F, 22.1; calculated: C, 40.00; H, 4.94; F, 22.33. IR spectrum (KBr, sm⁻¹): 2910, 2868, 2831 (ν(CH)); 1672, 1558, 1531, 1477 (ν(C=O) + ν(C=C)); 1354, 1286, 1249, 1186, 1107 (ν(CF)). NMR spectrum (298 K, CDCl₃, δ, ppm): 3.65 (c, 24H, 18C6), 5.66 (c, 1H, Cα–H, hfac).

3.4. Single-Crystal, Powder, and Film XRD

The suitable crystals of K(18C6)(hfac) were easily obtained during synthesis by evaporation of the mother liquor. Single-crystal X-ray diffraction analysis (SC-XRD) was performed using a Bruker D8 Venture single-crystal diffractometer (three-circle goniometer, Incoatec IμS 3.0 microfocus source, MoKα radiation, focusing, and monochromatization with Montel multilayer mirrors, PHOTON III CPAD detector, resolution 768 × 1024, Bruker AXS GmbH, Karlsruhe, Germany). The temperature of the sample was varied from 100(2) to 400(2) K using an Oxford Cryosystems Cryostream 800 plus (Oxford Cryosystems Ltd., Oxford, UK) nitrogen-flow cryostat. Reflection intensities were measured by the ω-scan method with narrow (0.5°) frames. Absorption was corrected semi-empirically using the SADABS–2016/2 software package [87]. The structures were solved by direct methods and refined by full-matrix least squares in the anisotropic approximation for all non-hydrogen atoms using the SHELXL–2018/3 software package [88]. The positions and thermal parameters of the H atoms were refined in the rigid body model. The CF₃ groups are positionally disordered over two orientations and were refined with appropriate occupancy factors. All the primary results of the temperature study (crystallographic characteristics, selected bond lengths and angles) are presented in the Supplementary Materials (Tables S2 and S3).

2D-GIXRD patterns of molecular thin films were obtained using the same diffractometer. A thin glass substrate with a K(18C6)(hfac) layer was placed in the special sample adapter and mounted onto the goniometer head. The primary MoKα beam angle of incidence was about 0.5°.

The CrystalExplorer 17.5 software package [50] was used in this study to generate Hirshfeld surfaces at high standard resolution from the CIF input file, providing color-coded HS maps that visualize key regions of intermolecular interactions. CrystalExplorer was also employed to calculate the intermolecular interaction energies, with the wavefunctions of the molecular system obtained using the built-in Tonto program at the CE-B3LYP/6-31G(d,p) theoretical level.

An analysis of the packing features of K(18C6)(hfac) was also carried out using the translational sublattice method [47,48,49]. The relative intensities of the diffraction lines were analyzed assuming Z = 2, and triplets of strong reflections whose indices form a matrix with a determinant equal to 2 were identified. A search for all possible variants was carried out using the program described in [49]. It was found that the most symmetric sublattice is formed by the intersection of three families of crystallographic planes: {0 1 1}, {1 0 0}, {0 1 −1}. It is built on the vectors a_T = b/2 + c/2; b_T = a; c_T = b/2 − c/2, where the sublattice parameters are a_T = 9.67 Å, b_T = 8.74 Å, c_T = 7.23 Å, α_T = 88.31°, β_T = 115.79°, γ_T = 88.54°.

Powder X-ray diffraction (PXRD) was carried out using a Bruker D8 ADVANCE diffractometer (CuKα radiation, LynxEye XE-T detector, Bragg–Brentano geometry, vertical θ–θ goniometer, 2θ range 5–40°, step 0.03°, Bruker AXS GmbH, Karlsruhe, Germany) at room temperature. The samples were dry ground in an agate mortar.

3.5. Thermal Properties of K(18C6)(hfac)

A thermogravimetric study (298–973 K, 25–700 °C) was conducted on a NETZSCH TG 209 F1 Iris thermal analyzer (NETZSCH-Gerätebau GmbH, Selb, Germany) for 10 mg samples in an inert atmosphere (helium, 30 mL/min) at a constant heating rate (10 K/min) in an open Al crucible.

The vaporization process was studied using the Knudsen effusion (KE) method with mass spectral (MS) analysis of the gas phase composition on an MS 1301 instrument (Special Design Bureau for Analytical Instruments Engineering of the Academy of Sciences of the USSR, Leningrad, USSR) [89] in the temperature range of 373–421 K (100–148 °C). A molybdenum Knudsen cell with an evaporation-to-effusion area ratio of ~600 was used. Temperature was measured with a Pt/Pt-Rh thermocouple and maintained constant with an accuracy of ±1 K. During the experiments, designed to determine the temperature dependencies of the principal ion currents, temperature was cyclically increased and decreased at a rate of 0.5 K/min. This approach helped mitigate minor temperature hysteresis effects caused by the system’s thermal inertia. The uncertainties in the pressure measurements are largely due to the uncertainties of ionization cross-sections. It is generally accepted that the total uncertainty of KEMS is about 10% for substances yielding vapors of simple composition. The detailed comments for the KEMS uncertainties are provided in the IUPAC technical report [90].

3.6. Molecular Film Preparation

Thermal evaporation (PVD) experiments were performed using an original vertical flow-type MOCVD reactor [59,60]. The precursor (30 mg) was placed in a quartz boat, heated to 443 K (170 °C), and evaporated under a flow of inert carrier gas (argon, 42 mL/min) and a total pressure of 1 Torr for 30 min. Unlike the MOCVD experiments (see below), there was no dedicated heating of the substrate, so the substrate holder was heated only through the heat transfer from the evaporator. Cooling was performed under vacuum in the absence of carrier gas. Spin-coating (SC) was performed from a K(18C6)(hfac) solution in dichloromethane (~5 mg/mL) using an Elmi CM-6M centrifuge (ELMI Ltd., Riga, Latvia) (2000 rpm).

3.7. Inorganic Film Preparation and Characterization

MOCVD experiments were performed in the same reactor as PVD ones (see Section 3.6). The precursor (50 mg) was placed in a quartz boat, heated to 443 K (170 °C), and evaporated under a flow of inert carrier gas (argon, 42 mL/min) and a total pressure of 1 Torr for 40 min. In this case, heating of the substrate (silicon) to 773 K (500 °C) and a reagent gas (83 mL/min) were applied. Oxygen (O₂) and oxygen bubbled in water at room temperature (O₂ + H₂O) were used as reagents.

Scanning electron microscopy (SEM) and energy-dispersive spectroscopy (EDS) of the inorganic films obtained were carried out on a JEOL–JSM 6700F microscope coupled with an EX2300BU analyzer (JEOL Ltd., Tokyo, Japan). Film thicknesses were determined using cross-section micrographs. The grazing incidence X-ray diffraction (GIXRD) was performed a Bruker D8 ADVANCE diffractometer (see above) equipped with a Gobel mirror (CuKα radiation; θ angle of the source: 2°; equatorial Soller slits; LYNXEYE XE-T linear detector; 2θ scanning range: 10–62°; 2θ step: 0.03°; counting time per point: 20 s; sample rotation: 30 rpm).

4. Conclusions

This work presents a comprehensive study of K(18C6)(hfac) as the first volatile fluorinated monometallic precursor for potassium. A convenient synthesis procedure from commercially available reagents has been developed and reliable spectral data for identification have been provided. An additional SC-XRD study was performed over a wide temperature range (100–400 K) and a detailed analysis of the Hirschfeld surface revealed key intermolecular contacts. The translational sublattice method clearly demonstrated the structural similarity of the fluorinated β-diketonate complexes K(18C6)(L) (L = hfac, ptac). For both structures, the chains of molecules due to K…F or K…H contacts “through” the crown ether are characteristic, and thermal expansion is most pronounced in this chain direction.

For K(18C6)(hfac), the preservation of the molecular structure both in solution and in the gas phase was confirmed using NMR and mass spectrometry respectively. The volatility of the complex was quantitatively confirmed by measuring the temperature dependence of the saturated vapor pressure, which made it possible to obtain the thermodynamic parameters of sublimation. The results of thermal studies and SC-XRD correlate in the aspect that noticeable volatility is observed at temperatures where a sharp elongation of the key intermolecular contact occurs (>350 K).

The possibility of producing K(18C6)(hfac) molecular films has been demonstrated using both solution-based and gas phase deposition methods. Specifically, in both cases, the films exhibit a strong (h00) texture, providing surface potassium cations accessible for further coordination “through” the crown ether. Such a structured well-defined surface may be of interest for supramolecular applications including sensors or targeted MOF synthesis. Finally, the possibility of obtaining both fluorinated and non-fluorinated potassium-containing films from this precursor in MOCVD processes was demonstrated, depending on the reaction atmosphere.

Thus, given the structural similarity across this class of the complexes, the family of potassium/rubidium β-diketonate complexes with crown ethers, M(18C6)(L), can be considered as effective fluorinated volatile precursors for the preparation of molecular and inorganic films. In further studies, fine-tuning of the precursor’s thermal properties can be achieved by varying the β-diketonate ligand, and deposition processes can be optimized for the formation of mono- and heterometallic fluoride/oxide-functional materials.