Photofragmentation Pathways for Gas-Phase ... - ACS Publications

Oct 5, 2016 - Mary T. Berry,*,† and Qingguo Meng*,§. †. Department of Chemistry, University of South Dakota, Vermillion, South Dakota 57069, Unit...
0 downloads 0 Views 6MB Size
Article pubs.acs.org/Organometallics

Photofragmentation Pathways for Gas-Phase Lanthanide Tris(isopropylcyclopentadienyl) Complexes Yulun Han,† Dmitri S. Kilin,‡ P. Stanley May,† Mary T. Berry,*,† and Qingguo Meng*,§ †

Department of Chemistry, University of South Dakota, Vermillion, South Dakota 57069, United States Department of Chemistry and Biochemistry, North Dakota State University, Fargo, North Dakota 58108, United States § Shenyang Institute of Automation, Guangzhou, Chinese Academy of Sciences, Guangzhou 511458, China ‡

S Supporting Information *

ABSTRACT: Photofragmentation mechanisms of gas-phase lanthanide tris(isopropylcyclopentadienyl) complexes, Ln(iCp)3, were studied through experimental photoionization time-of-flight mass spectrometry (PI-TOF-MS). A DFTbased time-dependent excited-state molecular dynamics (TDESMD) algorithm, under standard approximations, was used to simulate the photofragmentation process. Two competing reaction pathways, intact ligand stripping and ligand cracking within the metal−ligand complex, were hypothesized based on experimental data. It was evident that intramolecular hydrogen, methyl, and isopropyl abstraction play an important role in the ligand-cracking reaction pathway, leading to metal carbide and metal hydrocarbide products. The TDESMD simulations also produced branching reaction pathways for ligand ejection and ligand cracking and further suggested that both pathways are initiated by ligand-to-metal charge transfer. Although the simulations reproduced several of the proposed reactions and several of the products of cracking observed in the PI-TOF mass spectra, differences between the simulation and experimental results suggest specific directions for improvement in the computational model.

1. INTRODUCTION Lanthanide oxide thin films have applications in metal-oxide semiconductor field-effect transistors (MOSFETs),1−3 protective coatings,4,5 luminescent materials,6−8 and waveguides.9,10 Lanthanide cyclopentadienyl complexes have been frequently employed to prepare lanthanide oxide thin films by both metal− organic chemical vapor deposition (MOCVD)11 and atomic layer deposition (ALD) techniques.12−15 For example, Kondo et al. deposited Pr2O3 films on silicon substrates with a high dielectric constant of 26 ± 3 using praseodymium tris(ethylcyclopentadienyl) [Pr(EtCp)3] precursors, and Blanquart et al. used erbium cyclopentadienyl complexes to deposit Er2O3 thin films by ALD.11,14 The high volatility of lanthanide cyclopentadienyl complexes also makes them potential precursors for photolytic laser-assisted metal−organic chemical vapor deposition (LCVD), allowing lower deposition temperatures. It is our goal to better understand the photofragmentation pathways of gas-phase lanthanide cyclopentadienyl complexes in order to better control the purity of thin-film deposition using LCVD. There have been many studies on the photofragmentation pathways of gas-phase molecular compounds,16−20 metal− organic complexes,21−31 and clusters.32,33 Recently, our group reported the unimolecular gas-phase photodissociation mechanisms of lanthanide tris(η5-cyclopentadienyl) complexes, Ln(Cp)3, and lanthanide tris(tetramethylcyclopentadienyl) complexes, Ln(TMCp)3, through both experimental photoionization time-of-flight mass spectrometry (PI-TOF-MS) and theoretical atomistic simulation of the fragmentation dynamics.30 © XXXX American Chemical Society

It was hypothesized that two types of reaction pathways dominate in the photofragmentation of both Ln(Cp)3 and Ln(TMCp)3 precursors: intact ligand stripping, leading to the production of bare metal, and ligand cracking within the metal−ligand complex, giving rise to metal carbide and metal hydrocarbide compounds. It was proposed that in both cases the chemistry was dominated by fragmentation of neutral molecules, whose neutral fragments were detected after subsequent photoionization. See Supporting Information. The suggested dominant intact ligand stripping pathway was proposed as hv

hv

hv

Ln(L)3 → Ln(L)2 + L → Ln(L) + 2L → Ln 0 + 3L, L = Cp or TMCp

(1)

The cracking chemistry of Ln(Cp)3 was found to be dominated by the disproportionation of the Cp radical into C2H2 and C3H3 fragments, leading directly to the major intermediates of the photoreaction. Ln(TMCp)3 showed a much richer cracking chemistry than Ln(Cp)3, with multiple photofragmentation pathways, involving intermediate metal−organic tuck-in complexes. The reaction paths illustrated the mobility of hydrogen and methyl group in the TMCp ligands and their capacity for intramolecular migration. The earliest cracking steps are Received: May 25, 2016

A

DOI: 10.1021/acs.organomet.6b00420 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

The Results and Discussion section is divided into subsections on intact ligand ejection (subsection 3.1) and ligand cracking (subsection 3.2), representing the two fundamental branches for the photoreaction. The ligand-cracking chemistry is itself highly branched, and the branching pathways are frequently cross-linked. We have divided that subsection into 3.2(i) with an overview of the mass spectrum of the cracked fragments, 3.2(ii) on the hydrogen abstraction mechanism, 3.2(iii) on the methyl abstraction, 3.2(iv) on the isopropyl abstraction and cross-links from the other pathways, and 3.2(v) on additional reactions subsequent to hydrogen abstraction. The reaction mechanisms described in 3.2(ii)−(v) all invoke interligand abstraction and thus must start with Ln(iCp)2 or, potentially, Ln(iCp)3. Section 3.2(vi) then describes reactions that follow photoexcitation of Ln(iCp) formed from previous dissociation of Ln(iCp)2. For each of the parts 3.2(ii)−(vi), a mechanistic scheme is shown for the reactions relevant to that part, along with snapshots from the TDESMD simulations that capture many of the key features of the proposed mechanism.

competing intracomplex hydrogen abstraction (eq 2) and methyl abstraction (eq 3) mechanisms identified as hv

Ln(TMCp)2 → [Ln(TMCp) − H] + HTMCp

(2)

hv

Ln(TMCp)2 → [Ln(TMCp) − CH3] + C8H10 + C2H6 (3)

where the symbol “−” should be read as “minus”. [Ln(TMCp) − H] and [Ln(TMCp) − CH3] should be read as Ln(TMCp) “minus” a hydrogen and methyl group, respectively. DFT (density functional theory)-based time-dependent excited-state molecular dynamics (TDESMD) simulations of the photoreactions of Ln(Cp)3 were consistent with the observed fragmentation in the mass spectrum. Ln(TMCp)3 presented a much larger model, demanding greater computational resources, so TDESMD was not attempted. Since the mobility and migration of hydrogen and methyl groups in TMCp ligands were thought to play a key role in the photoreactions, we were motivated to examine another lanthanide cyclopentadienyl compound that contains methyl groups to look for similar patterns and, in this case, to pursue TDESMD on the more tractable phases of the photoreaction. Here, we report the gas-phase photofragmentation of lanthanide tris(isopropylcyclopentadienyl) complexes, Ln(iCp)3, with the purpose of testing the importance of the ring substituents in the photochemistry. Photofragmentation was monitored through photoionization mass spectrometry. TDESMD calculations were performed on the entire Ln(iCp)3 model for a duration up to 2 ps, illustrating mostly intact ligand ejection with some minor cracking including hydrogen abstraction. TDESMD was also initiated starting from intermediate Ln(iCp)2 and Ln(iCp) models, from which extensive ligand cracking could be observed. Additional simulations were performed, starting from [Ln(iCp) − H] and [Ln(iCp) − CH3], analogues to the products of reactions 2 and 3, as well as for several other intermediates, as will be discussed below. Supported by TDESMD calculations, the cracking reactions are interpreted as resulting from a ligand-to-metal charge-transfer (LMCT) process, wherein reduced metal and a bound ligand radical are created in the excited state. The excited-state ligand radical shows a strong propensity for hydrogen and methyl intraligand migration and for hydrogen or methyl abstraction from adjacent ligands. The excited-state ligand radicals also show a tendency to fragment, ejecting small closed-shell molecules, e.g., C2H2n, from the metal−ligand complex. After rearrangement or ejection of the closed-shell fragment, the complex is expected to relax through metal-to-ligand back charge transfer (MLCT), restoring the negative charge and closed-shell nature to whatever radical fragment was retained. As a parallel example in the fragmentation of Ln(Cp), in previous work30 we reported

2. EXPERIMENTAL SECTION 2.1. Materials. La(iCp)3 (99.9% La), Pr(iCp)3 (99.9% Pr), and Tb(iCp)3 (99.9% Tb) were purchased from Strem Chemicals. All chemicals were used with no further purification, and were kept inside a glovebox in an argon atmosphere before use. 2.2. TOF-MS. The photofragmentation experiments were conducted using the photoionization time-of-flight mass spectrometer described in detail in previous work.30 The molecular source is an effusive beam with an estimated pressure in the interaction zone of 10−4 mTorr. A photofragmentation wavelength of 266 nm (fourth harmonic of a Continuum Surelite Nd:YAG laser with 70 mJ pulse energies, 6 ns pulse width, ∼109 W/cm2) was used. This same laser, within the same pulse, serves for photoionization of the product fragments. The system is optimized to detect fragmentation products rather than to produce clean stripping of the ligands from the metal in the precursor. 2.3. Computational Method. All calculations were done in a basis set of Kohn−Sham orbitals computed in DFT with the Perdew−Burke− Ernzerhof (PBE) functional under periodic boundary conditions in the basis of plane waves and using norm-conserving atomic pseudopotentials as implemented in the VASP software.39−42 The optimized geometries of La(iCp)3, La(iCp)2, La(iCp), [La(iCp) − H], LaCOT, [La(iCp) − CH3], LaC6H6, LaC5H5, and LaC4H4 models, which were used as the starting point for the TDESMD simulations, are shown in the Supporting Information. The absorption spectra and partial charge density distributions were generated for the unperturbed models computed by ground-state DFT. The TDESMD procedure has been derived, justified, and described in detail in previous work.31,43 It should be noted that the models cannot be interpreted as a microcanonical, canonical, or grand canonical ensemble; although the number of particles in the simulation cell is fixed, the system is driven far from thermal equilibrium and is involved in nonequilibrium dynamics. The outcome of the procedure depends on the adopted approximations, currently including (a) classical path approximation; (b) integer occupations of excited states; (c) instantaneous transitions between potential energy surfaces following the surface-hopping approach; (d) special procedure for optically driven electronic transitions; (e) absence of average over ensemble; and (f) closed-shell configuration. Although the TDESMD procedure is a first-principles atomistic technique, it uses several parameters reflecting reaction conditions and controlling precision. Briefly, in the TDESMD simulation, the system is driven by continuous laser perturbation, and the electronic states hop between the ground state and the selected excited configuration with an inverse Rabi frequency parameter, 2πΩR−1 (typically 10 fs), reflecting the laser pulse intensity and molecular properties, and with time increments for simulations of 1 fs. The total duration of the simulations is generally 1 to 2 ps. The simulation produces trajectories consisting of atomic models of several

hv ,LMCT

Ln I(Cp−) ⎯⎯⎯⎯⎯⎯⎯⎯⎯→ Ln 0(Cp•) → Ln 0(C3H 3•) + C2H 2 MLCT

⎯⎯⎯⎯⎯⎯→ Ln I(C3H3−) + C2H 2

where facile disproportionation of the cyclopentadienyl radical into acetylene and the propargyl radical (C3H3•) has been frequently observed.34−38 In fact, almost all of the photochemistry we observe in lanthanide cyclopentadienyl-type complexes can be interpreted as familiar radical chemistry. The paper is organized as follows. The Experimental Section briefly describes the metal−organic materials, the photoionization mass spectrometry, and the computational approach. B

DOI: 10.1021/acs.organomet.6b00420 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics intermediates and products formed during the course of the reaction. The interatomic bond distance is used to evaluate chemical bond breaking or forming at each step of the trajectory. The product fragments are compared with the fragments experimentally observed by PI-TOF mass spectrometry, and the computed fragmentation steps contribute to the mechanisms proposed for multiple observed cracking pathways. An important difference to note between the computational model and the physical experiment lies in the time frame in which energy is delivered to the metal−organic complex. In the computational model, energy is delivered with rapid electronic re-excitation and de-excitation of the fragmenting species, and the process is completed in times on the order of picoseconds. In the experiment, energy is delivered much more slowly, over a 6 ns period. We are assuming an approximate equivalence of simulation with a strong field over a short duration to excitation with a weak field over a long duration, based on the pulse area theorem.44 However, as a difference, the experimental conditions likely allow significant thermal relaxation of the fragmenting species between excitation steps. One strategy we employ is to start some of the simulations beginning with thermally relaxed, geometry-optimized intermediate structures, which might be more representative of the experimental conditions that exist after multiple fragmentation steps.

3. RESULTS AND DISCUSSION 3.1. Intact Ligand Ejection. The photoionization time-offlight mass spectra of Ln(iCp)3 (Ln = La, Pr, Tb) with photoexcitation at 266 nm wavelength are shown in Figure 1. The three lanthanides show similar fragmentation patterns, differing in the relative intensities of the peaks and in the mass contribution from the metal ions. The slight differences in the relative intensities are likely due to small differences in excitation power density or in molecular absorptivity. The dominant features observed are LnC2+, Ln+, and Ln2+ fragments. The LnC2+ fragments show a relatively strong intensity and are thought to result from a metalattached-ligand cracking under the influence of metal catalysis. In the high m/z region (m/z > 180), there are additional Ln-containing fragments, including Ln(iCp)3+ and Ln(iCp)2+. A weak signal for Ln(iCp)+ fragments is also observed in the mass spectra (vide infra). This suggests an intact ligand stripping reaction pathway of hv

Figure 1. PI-TOF mass spectra of (a) La(iCp)3, (b) Pr(iCp)3, and (c) Tb(iCp)3 precursors with photoexcitation at 266 nm. The reaction diagram shows the clean ligand stripping reaction pathway and subsequent postdissociation photoionization to produce charged species that can be detected by the mass spectrometer. The breaks in the top panel indicate that the spectrum was measured in three separate experiments and not as a single experiment spanning the entire m/z range.

hv

Ln(iCp)3 → Ln(iCp)2 + iCp → Ln(iCp) + 2iCp hv

→ Ln 0 + 3iCp

(4)

with neutral ligand-radical ejection, through laser excitation into the ligand-to-metal charge-transfer states, similar to the mechanism proposed for several other lanthanide complexes and as also illustrated in eq 1.25−30 The binding energies for the three dissociation steps illustrated in eq 4 are calculated as 3.8, 4.0, and 4.6 eV, respectively, for Ln = La. The intact ligand ejection pathway is reproduced by TDESMD simulations. The computed absorption spectrum, along with maps of partial charge density for the states involved in the photoexcitation, is shown in Figure 2. Snapshots of the computed trajectory of La(iCp)3 are listed in Table 1. From 0 to 75 fs, the iCp ligands are sequentially ejected from the metal. It should be noted that the ejected ligands are not entirely removed from the system due to a trapping artifact in the model. The ejected ligands are not allowed to escape from the simulation cell and thus get recaptured. However, the first 75 fs of the simulation is consistent with the dominant fragmentation path described by eq 4. 3.2. Ligand Cracking. i. Ligand Cracking Overview. The mass spectra of Ln(iCp)3 are expanded in the mass range between Ln(iCp)+ and LnC2+ in Figure 3. In this region many

Ln-containing species are observed, thought to result from a ligand cracking mechanism, in competition with the intact ligand stripping that would yield simply Ln0 and iCp. The features in Figure 3 are labeled as 1 to 6 from high to low m/z. Within each numbered group are several peaks, each corresponding to a structure with the same number of carbon atoms but differing numbers of hydrogen atoms. The species labeled as n, n′, and n″ have even numbers of hydrogen atoms and are related to each other through sequential loss of H2. The features labeled as n*, n**, and n*** have odd numbers of hydrogen atoms and are also related to each other through sequential loss of H2. In each group the features with even numbers of hydrogens dominate. Invoking the paradigm that the most likely metal−organic fragments from cracking arise through elimination of stable closed-shell molecules, it is thought that most of the fragments with even numbers of hydrogens are not derived from Ln(iCp), i.e., LnC8H11, as such reactions require the elimination of openshell fragments. Therefore, they are more likely to arise from the C

DOI: 10.1021/acs.organomet.6b00420 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

where [Ln(iCp) − H] (feature 1), corresponding to LnC8H10, should be read as Ln(iCp) “minus” a hydrogen. It should be pointed out that the isopropylcyclopentadiene molecule HiCp+ (m/z = 108) is clearly observed in the mass spectra (see Figure 1). In a previous study, the analogous [Ln(TMCp) − H]+ fragment was identified in the photodissociation of Ln(TMCp)3. However, in this study the [Ln(iCp) − H]+ feature itself is quite weak, but the product of subsequent H2 elimination (feature 1′, LnC8H8 = LnCOT) is the strongest feature in the eight-carbon group. Scheme 1 shows the proposed mechanism for the formation of structure 1 (LnC8H10) from Ln(iCp)2 and the subsequent successive H2 molecule elimination to produce structures 1′ (LnC8H8) and 1″ (LnC8H6). It should be noted that the initial hydrogen abstraction may take place at multiple sites and several different reactions may be occurring in parallel. Structure 1i-1 is calculated with a slightly higher energy than structure 1 but is thought to be an intermediate for the next reaction steps. In the second line of Scheme 1, we propose a rearrangement to give structure 1i-2, which is slightly more stable than either structure 1 or 1i-1. The elimination of a H2 molecule from structure 1i-2 then gives rise to structure 1′, LnCOT (COT2− = cyclooctatetraenyl dianion), which is featured very strongly in the mass spectra. The further elimination of H2 from LnCOT may give rise to structure 1″, LnC8H6 (C8H62− = pentalene dianion), whose structure was also calculated as the lowest in energy among other alternatives (see Supporting Information). The metal-catalyzed transannular ring closure of COT2− to form a metal pentalene complex was reported by Stone et al., who isolated ruthenium complexes of C8H62−.45,46 Cloke also reported formation of pentalenes from flash vacuum pyrolysis of substituted cyclooctatetraene.47 The propensity for hydrogen abstraction from the bis complex and dehydrogenation from LnC8H10 are both reproduced by TDESMD simulations. Snapshots of the computed trajectory of La(iCp)3 driven by the laser field in a Rabi cycle between orbitals HO−2 and LU+11 are shown in Table 2. At 1578 fs the lower right iCp ligand is ejected intact. Subsequently, a hydrogen migrates from the upper iCp ligand to the lower left iCp ligand, leading to the formation of an HiCp molecule and a [La(iCp) − H] fragment. The lower right iCp ligand radical is ultimately recaptured, but the HiCp molecule remains mostly unbound and

Figure 2. Absorption spectrum for unperturbed La(iCp)3 computed by ground-state DFT and partial charge density distribution for selected Kohn−Sham orbitals. Each orbital is labeled based on HOMO−LUMO notation. LU+n means the nth orbital above LU and HO−n means the nth orbital below HO. The transition energies of HO−1 → LU and HO−2 → LU+11 are about 3.22 eV (equivalent to λex = 385 nm) and 5.10 eV (equivalent to λex = 243 nm), respectively. These transitions are typical ligand-to-metal charge transfer and are explored in the TDESMD simulation. The silver, gray, and white spheres represent La, C, and H, respectively. Probability density isosurfaces for the electronic states involved in the excitation are shown in yellow.

bis complex Ln(iCp)2. The presence of these features with even numbers of hydrogens in the very first group of peaks (e.g., 1 = LnC8H10, 1′ = LnC8H8) and in the second (e.g., 2 = LnC7H8, 2′ = LnC7H6) suggests that ligand cracking may begin with intracomplex (interligand) hydrogen or methyl abstraction. ii. Hydrogen Abstraction. The intracomplex hydrogen abstraction from the bis complex, Ln(iCp)2, similar to eq 2 is proposed as Ln(iCp)2 → [Ln(iCp) − H] + HiCp

(5)

Table 1. Snapshots from the TDESMD Simulations Illustrating the Ligand Dissociation in eq 4

D

DOI: 10.1021/acs.organomet.6b00420 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

If we begin the simulation at the relaxed structure of LaCOT, we see the formation of the carbon skeleton for the pentalene structure following elimination of 2H, bearing great similarity to LnCOT → Ln(pentalene) + H2 or 1′ → 1″+ H2, as illustrated in Scheme 1. iii. Methyl Abstraction. Intracomplex methyl abstraction in Ln(iCp)2 would result in feature 2 of Figure 3, corresponding to LnC7H8+, which we assign to [Ln(iCp) − CH3]+ (read as Ln(iCp) “minus” a methyl group). The eliminated closed-shell fragment might potentially be isopropylmethylcyclopentadiene. However, no signal is observed at m/z = 122. So we propose instead the reaction illustrated in eq 6, Ln(iCp)2 → [Ln(iCp) − CH3] + (iCp − CH3) + C2H6 (6)

where both iCp ligands of Ln(iCp)2 lose methyl groups, with ethane, C2H6, and 6-methylfulvene, designated as (iCp − CH3) at m/z = 92, leaving the complex, where the latter species is clearly observed in the mass spectra (see Supporting Information). This is analogous to methyl abstraction in Ln(TMCp)3, where no signal for pentamethylcyclopentadiene was observed, whereas a dimethylfulvene signal was evident. A simple way to interpret the chemistry of eq 6 is to consider one ligand radical, created by LMCT, eliminating a methyl radical and directly forming 6-methylfulvene, as shown in Scheme 2. The 6-methylfulvene, neutral, relatively unreactive, and possessing high kinetic energy, would depart from the complex. The methyl radical, being highly reactive and adjacent to a bound ligand, could extract a methyl group from the bound ligand, as proposed in eq 6, or, for that matter, could extract a hydrogen or an entire isopropyl group, as will be discussed below. The top row of Table 3 captures the reaction of Scheme 2 occurring in the TDESMD simulation of photofragmentation of La(iCp)3. Scheme 3 shows the proposed mechanism for the formation of structure 2 (LnC7H8) from Ln(iCp)2 and the subsequent successive H2 elimination to produce structures 2′ (LnC7H6) and 2″ (LnC7H4). It should be noted that structure 1, resulting from hydrogen abstraction, might also give rise to structure 2′ (but not to 2) by the elimination of CH4. In the second row of Table 3, snapshots of the TDESMD simulation beginning with the energetically relaxed structure for [La(iCp) − CH3] = LaC7H8 exhibits H2 elimination, as seen for the reaction 2 → 2′ + H2, in Scheme 3. As shown in the last row of Table 3, TDESMD simulation of La(iCp)3 illustrates several features relevant to the proposed photochemistry of La(iCp)2, despite the presence of the third spectator ligand. The snapshot at 1107 fs again illustrates the lability of the iCp ligand (top) and the methyl group (lower right). In this case, both fragments are later recaptured in the simulation. The recapture results in hydrogen abstraction from the top ligand, creating HiCp in the lower left. The metalcontaining fragment is then equivalent to [La(iCp) − H](iCp), which undergoes methane elimination. If in these latter steps we consider the intact iCp on the lower right as a spectator, the methane elimination is similar to

Figure 3. PI-TOF mass spectra of (a) La(iCp)3, (b) Pr(iCp)3, and (c) Tb(iCp)3 precursors in the mass range between Ln(iCp)+ and LnC2+ with photoexcitation at 266 nm. The metal mass is subtracted from the ordinate to show the similarity of the fragmentation pattern of the ligand for the different metals. Groups are labeled as 1 to 6 from high to low m/z. Within each numbered group are several peaks, each corresponding to a structure with the same number of carbon atoms but differing numbers of hydrogen atoms.

is stable for the rest of simulation, although it cannot escape the simulation cell. Later stages of the photoreaction are often not well reproduced in the simulation because of the considerable accumulated thermal energy. One strategy is to start the simulation beginning with geometry-optimized intermediate structures. Snapshots of the computed trajectory of La(iCp) driven by the laser field in a Rabi cycle between orbitals HO−2 and LU+3 are given in the lower panels of Table 2. During the simulation, the hydrogen atoms in the iCp ring are found to be labile. We observe ejection of a single hydrogen atom, creating an intermediate analogous to the product of hydrogen abstraction as discussed above, followed by H2 elimination. In this case, the metal-containing fragment, though it illustrates the requisite ring opening, does not relax to the proposed stable LnCOT structure.

[La(iCp) − H] → LaC7H6 + CH4 or equivalently, 1 → 2′ + CH4

as illustrated in the lower line of Scheme 3 as an alternate path to LaC7H6. The path for hydrogen abstraction is thus cross-linked to the path for methyl abstraction at LaC7H6, structure 2′. iv. Isopropyl Abstraction. In addition to the intracomplex hydrogen and methyl abstraction from the bis complex, we also E

DOI: 10.1021/acs.organomet.6b00420 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics Scheme 1. Photofragmentation Mechanism from Ln(iCp)2 to Structure 1″

Table 2. Snapshots of TDESMD Simulations Related to Scheme 1

propose an analogous intracomplex isopropyl (C3H7) abstrac-

Scheme 2. Methyl Elimination from the Ligand Radical Forms 6-Methylfulvene

tion mechanism as Ln(iCp)2 → [Ln(iCp) − C3H 7] + (iCp − CH3) + C4 H10 (7) F

DOI: 10.1021/acs.organomet.6b00420 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics Table 3. Snapshots of TDESMD Simulations Related to Schemes 2 and 3

Scheme 3. Photofragmentation Mechanism from Ln(iCp)2 to Structure 2″

to LnC5H4 (feature 4), should be read as Ln(iCp) “minus” an isopropyl group. Isopropyl loss from cyclopentadienyl systems has been observed by others. For example, Sitzmann et al. studied the EI mass spectrum of the pentaisopropylcyclopentadienyl radical and proposed the loss of isopropyl to explain the spectrum.48

where one iCp ligand, created as a radical through LMCT, eliminates a methyl radical, leaving 6-methylfulvene, given as (iCp − CH3) in eq 7. The methyl radical then extracts an isopropyl group from the bound iCp ligand, leaving the closed-shell isobutane, C4H10. Note that [Ln(iCp) − C3H7], corresponding G

DOI: 10.1021/acs.organomet.6b00420 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics Table 4. Snapshots of TDESMD Simulations Related to Scheme 4

Also in Table 4, TDESMD simulation of the monoligated La(iCp) illustrates that the isopropyl group is labile. At 312 fs, in addition to the isopropyl group ejection, a C2H2 fragment is eliminated, giving rise to LaC3H2. This process is similar to the proposed reaction pathway of 4 → 6 + C2H2 in Scheme 4. In Scheme 4, the reaction pathways for hydrogen, methyl, and isopropyl abstraction are cross-linked at LnC5H4, structure 4. The last two rows of Table 4 illustrate the production of LaC5H4 from [La(iCp) − H] and [La(iCp) − CH3], respectively. v. Subsequent to Hydrogen Abstraction. Following hydrogen abstraction, structure 1, LnC8H10, is subject to further cracking, as discussed above for the reactions 1 → 4 + C3H6 and 1 → 2′ + CH4. There are additional reactions, beginning with elimination of a two-carbon fragment and ultimately leading to the prominent LnC2 product as illustrated in Scheme 5.

The ease with which an isopropyl group might be extracted is supported by TDESMD simulations that show transfer of the isopropyl group from the ring to the metal ion as illustrated in the top row of Table 4. Scheme 4 shows the proposed mechanism for the formation of structure 4 (LnC5H4) from Ln(iCp)2 and the subsequent closedshell molecule elimination to produce structures 4′ (LnC5H2), 6 (LnC3H2), and 6′ (LnC3). Note the origin of structure 4 is multifold. Structures 1 and 2 could also give rise to structure 4 by the loss of closed-shell C3H6 and C2H4 molecules, respectively. In the second row of Table 4, for TDESMD simulation of La(iCp)3, the snapshot at 1200 fs illustrates the lability of both the iCp ligand (lower right) and the isopropyl groups (lower left). Both fragments are later recaptured in the simulation. H

DOI: 10.1021/acs.organomet.6b00420 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics Scheme 4. Photofragmentation Mechanism from Ln(iCp)2 to Structure 4′ and 6′

Scheme 5. Photofragmentation Mechanism from Structure 1 to Structures 3″, 5″, and LnC2

propargyl radical, C3H3. This latter reaction, LnCOT → Ln(Cp) + C3H3, lies outside the paradigm that the most likely reactions are those involving elimination of closed-shell molecules, but represents production of a highly resonancestabilized open-shell product instead. Scott et al. studied the photodissociation of several LnCOT+ complexes and reported several fragmentation products LnCnHn (n = 2, 4, 5, 6), with the LnC5H5 product dominating.52 As shown in the first row of Table 5, TDESMD simulation of LaCOT also illustrates the propensity of C3H3 elimination. Also in Table 5, TDESMD simulation of LaC6H6 (structure 3i) shows the elimination of closed-shell H2 as well as closure of the six-membered carbon ring as in the proposed reaction pathway of 3i → 3′ + H2 in Scheme 5. In the last row of Table 5, TDESMD simulation of LaC4H4 shows H2 elimination and formation of the La(CHCCCH) species as in 5 → 5′ + H2 in Scheme 5.

Scheme 5 shows the proposed mechanism for the formation of structure 3 (LnC6H6) from structure 1 and the subsequent closed-shell molecule elimination to produce structures 3″ (LnC6H2), 5″ (LnC4), and LnC2. The dehydrogenation reaction from 3 → 3′ involves an intermediate 3i, where the metal is inserted into the ring, forming a seven-membered metalloheterocyclic ring. Clemmer et al. reported a similar insertion of the lanthanum ion into carbon clusters forming LaCn+ monocyclic rings upon annealing.49 The proposed structure of feature 3′ (LnC6H4) is a metal attached to a six-membered aromatic ring, which is similar to the product of dehydrogenation of gas-phase Ln+−benzene complexes, reported to produce LnC6H4+.50 Structure 5 (LnC4H4) corresponds to a metallocyclopentadiene complex, which is similar to the reported structure by Wesendrup and Schwarz.51 It should be noted that structure 1′ (LnC8H8 = LnCOT) could also lead to structure 3i with loss of C2H2. As illustrated in Scheme 5, structure 1′, LnC8H8, could also yield structure 4*, LnC5H5, and the I

DOI: 10.1021/acs.organomet.6b00420 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics Table 5. Snapshots of TDESMD Simulations Related to Scheme 5

Scheme 6. Photofragmentation Mechanism from Ln(iCp), Yielding Fragments with an Odd Number of Hydrogens

vi. Fragmentation of Ln(iCp). The proposed mechanisms described above are dominated by species with an even number of hydrogens, as the cracking chemistry involves the elimination, for the most part, of closed-shell molecules. To explain the appearance of those fragments with odd numbers of hydrogens, we propose that there is another cracking pathway that starts with Ln(iCp), i.e., LnC8H11, as shown in Scheme 6. The Ln(iCp) fragment is not highly stable and shows a relatively weak signal in the mass spectra, but could itself undergo elimination

reactions by losing closed-shell molecules to produce relatively stable fragments. Most clearly represented in the mass spectra are photoreactions that involve Ln(iCp) → 2* → 2** and Ln(iCp) → 4* → 4** → 4*** or 4* → 6* → 6** through the elimination of small closed-shell molecules. Another possible reaction pathway involves Ln(iCp) → 1* → 3* → 5* → 5** through C2H2n and H2 elimination. It should be noted that feature 3* (LnC6H5) fragments are not clearly observed in the mass spectra, and features 5* and 5**, though J

DOI: 10.1021/acs.organomet.6b00420 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics Table 6. Snapshots of TDESMD Simulations Related to Scheme 6

3.3. Summary. In Schemes 1−6, we propose mechanisms for the formation of photofragments from Ln(iCp)3 through the ejection of intact-ligand radicals (iCp) and through elimination of closed-shell molecules from bound ligand radicals, both types of reactions initiated by excitation of LMCT states. The most common motif in the cracking of the bound ligand radical is elimination of two-carbon fragments, C2H2n from larger fragments, as is commonly observed in the reaction of metal ions with ring systems.55 Sequential dehydrogenation, a quite common phenomenon in metal organic chemistry,56,57 is invoked for the formation of the multiple features within a given carbon group of the mass spectrum. We propose that some fragments arise through the elimination of closed-shell CH4 and C3H6 molecules, whose metal-catalyzed elimination is also commonly reported in the literature.58,59 A departure from the general rule of elimination of closed-shell species is proposed in the elimination of the relatively stable propargyl radical from LnCOT to form Ln(Cp).

apparent, are weak. The ring expansion reaction illustrated in 1* → 3* has been reported in the cracking chemistry of methylsubstituted cyclopentadienyls.53 Negru et al. reported a photofragmentation pathway of the phenyl radical leading to C4H3 by the loss of C2H2, identical to the reaction from 3* → 5*.54 As shown in Table 6, TDESMD simulation of La(iCp)2 illustrates several features relevant to the proposed photochemistry of La(iCp) despite the difference in the presence of the second ligand. We observe that the right iCp ligand undergoes cracking, while the left iCp is intact and remains bound as a nonparticipating spectator. At 399 fs, a methyl group and a hydrogen atom are lost from the right-hand iCp ligand. Barring recombination with the ring system, this would lead to elimination of CH4 and relaxation of the retained LnC7H7, equivalent to Ln(iCp) → 2* + CH4. In the simulation the eliminated methyl and hydrogen do recombine with the ring, and a new pathway is then initiated. At 485 fs, two hydrogens are lost, leaving as the closed-shell H2 molecule, similar to the proposed reaction of Ln(iCp) → 1* + H2. At 638 fs, the closed-shell C3H6 molecule is eliminated, leading to LaC5H3. It should be noted that in Scheme 6 two paths to LnC5H3 = 4** are illustrated, namely,

4. CONCLUSION Photoinduced reactions of the gas-phase lanthanide cyclopentadienyls involve a sequence of stepwise elimination of hydrocarbon fragments, resulting in metal-containing fragments of sequentially lower mass. In this study, a photofragmentation mechanism for gas-phase Ln(iCp)3 is proposed based on PI-TOF mass spectra. It is found that there are two competing reaction pathways during the photofragmentation process: intact ligand stripping, leading to the production of reduced bare metal, and ligand cracking within the metal−ligand complex, giving rise to metal carbides and hydrocarbides. The ligand cracking is easily understood as typical radical chemistry, where the ligand radical is created from the ligand anion as a transient following ligandto-metal charge transfer. Much of the cracking within the metal−ligand complex starts with the bis complex Ln(iCp)2 and

Ln(iCp) → 4* + C3H6 → 4** + H 2 + C3H6

and Ln(iCp) → 1* + H 2 → 4** + C3H6 + H 2

where the order of elimination of H2 and C3H6 is reversed for the two paths. Both paths, of course, may be active. In the last two panels of Table 6, TDESMD simulation of LaC5H5 = La(Cp) photofragmentation illustrates the elimination of the C2H2 fragment giving rise to LaC3H3, similar to the proposed reaction pathway of 4* → 6* + C2H2. K

DOI: 10.1021/acs.organomet.6b00420 Organometallics XXXX, XXX, XXX−XXX

Organometallics



hydrogen, methyl, or isopropyl interligand abstraction for the production of fragments with even numbers of hydrogens or from Ln(iCp) for the production of fragments with odd numbers of hydrogens. The elimination of small, closed-shell molecules such as CH4, C2H2n, and H2 is a common theme in the ligandcracking reaction pathway. A time-dependent excited-state molecular-dynamics algorithm, under standard approximations, with realistic parameters, was applied to the La(iCp)3 complex and to several individual photofragmentation intermediates in order to simulate the dynamics of the photofragmentation processes. Both intact ligand stripping and ligand cracking were observed in the TDESMD studies. The TDESMD simulation confirmed the central role of LMCT and MLCT in the photofragmentation of the complex and reproduced many features of the PI-TOF mass spectra. This study illustrates the utility of TDESMD in describing the photoreactions of metal−organic compounds in a strong laser field and suggests directions for future development of the method. Specifically, an improvement to the TDESMD algorithm may include a change from treating the model as a closed system, with a fixed number of particles at each step, to treating it as an open system, with changes in the number of atoms via extraction of closed-shell fragments from the computational cells, related to explicit elimination of light hydrocarbon fragments. Second, the optically driven electronic transitions are currently chosen between states indexed as HO−i and LU+j, based on the electronic structure of the original precursor. An updated recalculation of the optical properties at each time step would allow an update to the indexes, appropriate to the changing band gap and density of states of the fragmenting molecule. Third, a change from the requirement for an overall closed-shell configuration within the computational cell will allow for better treatment of open-shell intermediate radicals and will enable treatment in the future of metal−organic complexes with open shells or in high-spin configurations. The combination of PI-TOF mass spectrometry and TDESMD provides a powerful tool, with much promise for future development, for understanding the photochemistry of volatile lanthanide−organic complexes. The interpretation of the chemistry as radical reactions induced by ligand-to-metal charge transfer provides a straightforward basis for interpreting the origin of the multitude of metal-containing fragments that result from photoexcitation.



Article

AUTHOR INFORMATION

Corresponding Authors

*E-mail (M. T. Berry): [email protected]. *E-mail (Q. Meng): [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This research was supported by South Dakota Governor’s Office of Economic Development and NSF award EPS-0903804. The authors acknowledge computational resources of the USD High Performance Computational Facilities co-sponsored by South Dakota IDeA Networks of Biomedical Research Excellence NIH 2 P20RR016479 and operated by Douglas Jennewein.



REFERENCES

(1) Schlom, D. G.; Haeni, J. H. MRS Bull. 2002, 27, 198−204. (2) Leskelä, M.; Kukli, K.; Ritala, M. J. Alloys Compd. 2006, 418, 27−34. (3) Leskelä, M.; Ritala, M. J. Solid State Chem. 2003, 171, 170−174. (4) Bonnet, G.; Lachkar, M.; Colson, J. C.; Larpin, J. P. Thin Solid Films 1995, 261, 31−36. (5) Ershov, S.; Druart, M. E.; Poelman, M.; Cossement, D.; Snyders, R.; Olivier, M. G. Corros. Sci. 2013, 75, 158−168. (6) Kanarjov, P.; Reedo, V.; Oja Acik, I.; Matisen, L.; Vorobjov, A.; Kiisk, V.; Krunks, M.; Sildos, I. Phys. Solid State 2008, 50, 1727−1730. (7) Huan, J.; Hu, L.; Fang, X. ACS Appl. Mater. Interfaces 2014, 6, 1462−1469. (8) Bisson, J.-F.; Patriarche, G.; Marest, T.; Thibodeau, J. J. Mater. Sci. 2015, 50, 1267−1276. (9) Pons-Y-Moll, O.; Perriere, J.; Millon, E.; Defourneau, R. M.; Defourneau, D.; Vincent, B.; Essahlaoui, A.; Boudrioua, A.; Seiler, W. J. Appl. Phys. 2002, 92, 4885−4890. (10) de Assumpçaõ , T. A. A.; da Silva, D. M.; Del Cacho, V. D.; Kassab, L. R. P.; Alayo, M. I. J. Alloys Compd. 2014, 586, S368−S372. (11) Kondo, H.; Sakurai, S.; Sakashita, M.; Sakai, A.; Ogawa, M.; Zaima, S. Appl. Phys. Lett. 2010, 96, 012105. (12) Päiväsaari, J.; Niinistö, J.; Arstila, K.; Kukli, K.; Putkonen, M.; Niinistö, L. Chem. Vap. Deposition 2005, 11, 415−419. (13) Niinistö, J.; Putkonen, M.; Niinistö, L. Chem. Mater. 2004, 16, 2953−2958. (14) Blanquart, T.; Kaipio, M.; Niinistö, J.; Gavagnin, M.; Longo, V.; Blanquart, L.; Lansalot, C.; Noh, W.; Wanzenböck, H. D.; Ritala, M.; Leskelä, M. Chem. Vap. Deposition 2014, 20, 217−223. (15) Han, J. H.; Nyns, L.; Delabie, A.; Franquet, A.; Van Elshocht, S.; Adelmann, C. Chem. Mater. 2014, 26, 1404−1412. (16) Moreno Betancourt, A.; Bava, Y. B.; Berrueta Martínez, Y.; Erben, M. F.; Cavasso Filho, R. L.; Della Védova, C. O.; Romano, R. M. J. Phys. Chem. A 2015, 119, 8021−8030. (17) Long, J.; Wang, H.; Kvaran, Á . J. Phys. Chem. A 2014, 118, 1826− 1831. (18) Gardiner, S. H.; Lipciuc, M. L.; Karsili, T. N. V.; Ashfold, M. N. R.; Vallance, C. Phys. Chem. Chem. Phys. 2015, 17, 4096−4106. (19) Zabuga, A. V.; Kamrath, M. Z.; Boyarkin, O. V.; Rizzo, T. R. J. Chem. Phys. 2014, 141, 154309. (20) Harris, S. J.; Murdock, D.; Zhang, Y.; Oliver, T. A. A.; Grubb, M. P.; Orr-Ewing, A. J.; Greetham, G. M.; Clark, I. P.; Towrie, M.; Bradforth, S. E.; Ashfold, M. N. R. Phys. Chem. Chem. Phys. 2013, 15, 6567−6582. (21) Cheon, J.; Guile, M.; Muraoka, P.; Zink, J. I. Inorg. Chem. 1999, 38, 2238−2239. (22) Cheon, J.; Zink, J. I. Inorg. Chem. 2000, 39, 433−436. (23) Muraoka, P.; Byun, D.; Zink, J. I. Coord. Chem. Rev. 2000, 208, 193−211. (24) Bitner, T. W.; Zink, J. I. Inorg. Chem. 2002, 41, 967−972. (25) Ow, F. P.; Berry, M. T.; May, P. S.; Zink, J. I. J. Phys. Chem. A 2006, 110, 7751−7754.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.organomet.6b00420. PI-TOF mass spectra of La(iCp)3, Pr(iCp)3, and Tb(iCp)3, optimized geometry of intermediates used as the starting point for TDESMD simulations, proposed molecular structures and energies for intermediates in the mechanistic scheme for Ln(iCp) 3 , computed absorption spectra of intermediates related to TDESMD simulations, computed dissociation energies and ionization potentials for the La(iCp)n complexes, and a summary of previous arguments relating to the order of photofragmentation and photoionization (PDF) L

DOI: 10.1021/acs.organomet.6b00420 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics (26) Ow, F. P.; Berry, M. T.; May, P. S.; Zink, J. I. J. Phys. Chem. A 2007, 111, 4144−4149. (27) Nelson, B. N.; Caster, A. G.; Berry, M. T. Chem. Phys. Lett. 2004, 396, 256−260. (28) Meng, Q.; Witte, R. J.; May, P. S.; Berry, M. T. Chem. Mater. 2009, 21, 5801−5808. (29) Meng, Q.; Witte, R. J.; Gong, Y.; Day, E. L.; Chen, J.; May, P. S.; Berry, M. T. Chem. Mater. 2010, 22, 6056−6064. (30) Chen, J.; Hochstatter, A. M.; Kilin, D.; May, P. S.; Meng, Q.; Berry, M. T. Organometallics 2014, 33, 1574−1586. (31) Han, Y.; Meng, Q.; Rasulev, B.; May, P. S.; Berry, M. T.; Kilin, D. S. J. Phys. Chem. A 2015, 119, 10838−10848. (32) Martin, J. P.; Case, A. S.; Gu, Q.; Darr, J. P.; McCoy, A. B.; Lineberger, W. C. J. Chem. Phys. 2013, 139, 064315. (33) Krapf, S.; Schill, M.; Krotz, S.; Koslowski, T. Phys. Chem. Chem. Phys. 2011, 13, 14973−14983. (34) Bacskay, G. B.; Mackie, J. C. Phys. Chem. Chem. Phys. 2001, 3, 2467−2473. (35) Roy, K.; Horn, C.; Frank, P.; Slutsky, V. G.; Just, T. Symp. Combust., [Proc.] 1998, 27, 329−336. (36) Kern, R. D.; Zhang, Q.; Yao, J.; Jursic, B. S.; Tranter, R. S.; Greybill, M. A.; Kiefer, J. H. Symp. Combust., [Proc.] 1998, 27, 143−150. (37) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652. (38) Moskaleva, L. V.; Lin, M. C. J. Comput. Chem. 2000, 21, 415−425. (39) Hohenberg, P.; Kohn, W. Phys. Rev. 1964, 136, B864−B871. (40) Kresse, G.; Hafner, J. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 47, 558−561. (41) Kresse, G.; Hafner, J. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 49, 14251−14269. (42) Kresse, G.; Furthmüller, J. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 11169−11186. (43) Chen, J.; Meng, Q.; Stanley May, P.; Berry, M. T.; Kilin, D. S. Mol. Phys. 2014, 112, 508−517. (44) Shchedrin, G.; O’Brien, C.; Rostovtsev, Y.; Scully, M. O. Phys. Rev. A: At., Mol., Opt. Phys. 2015, 92, 063815. (45) Knox, S. A. R.; Stone, F. G. A. Acc. Chem. Res. 1974, 7, 321−328. (46) Maitlis, P. M.; Hill, A. F. Organometallics 2012, 31, 2489−2506. (47) Cloke, F. G. N. Pure Appl. Chem. 2001, 73, 233−238. (48) Sitzmann, H.; Bock, H.; Boese, R.; Dezember, T.; Havlas, Z.; Kaim, W.; Moscherosch, M.; Zanathy, L. J. Am. Chem. Soc. 1993, 115, 12003−12009. (49) Clemmer, D. E.; Shelimov, K. B.; Jarrold, M. F. J. Am. Chem. Soc. 1994, 116, 5971−5972. (50) Gibson, J. K. J. Phys. Chem. 1996, 100, 15688−15694. (51) Wesendrup, R.; Schwarz, H. Organometallics 1997, 16, 461−466. (52) Scott, A. C.; Foster, N. R.; Grieves, G. A.; Duncan, M. A. Int. J. Mass Spectrom. 2007, 263, 171−178. (53) Scheer, A. M.; Mukarakate, C.; Robichaud, D. J.; Ellison, G. B.; Nimlos, M. R. J. Phys. Chem. A 2010, 114, 9043−9056. (54) Negru, B.; Goncher, S. J.; Brunsvold, A. L.; Just, G. M. P.; Park, D.; Neumark, D. M. J. Chem. Phys. 2010, 133, 074302. (55) Huang, Y.; Hill, Y. D.; Sodupe, M.; Bauschlicher, C. W.; Freiser, B. S. Inorg. Chem. 1991, 30, 3822−3829. (56) Ranasinghe, Y. A.; MacMahon, T. J.; Freiser, B. S. J. Am. Chem. Soc. 1992, 114, 9112−9118. (57) Parent, D. C.; McElvany, S. W. J. Am. Chem. Soc. 1989, 111, 2393−2401. (58) Ren, J.; Cui, C.; Zhou, G.; Liu, Y.; Hu, Y.; Wang, B. Thin Solid Films 2011, 519, 3716−3721. (59) Jacobson, D. B.; Freiser, B. S. J. Am. Chem. Soc. 1985, 107, 7399− 7407.

M

DOI: 10.1021/acs.organomet.6b00420 Organometallics XXXX, XXX, XXX−XXX