Synthesis and electronic structure determination of N-alkyl-substituted bis(imino)pyridine iron imides exhibiting spin crossover behavior

ARTICLE

pubs.acs.org/JACS

Synthesis and Electronic Structure Determination of

N-Alkyl-Substituted Bis(imino)pyridine Iron Imides

Exhibiting Spin Crossover Behavior

Amanda C. Bowman,^†,‡Carsten Milsmann,^†,‡,§Eckhard Bill,^‡Zo€e R. Turner,^§Emil Lobkovsky,^†

Serena DeBeer,^†,‡Karl Wieghardt,^‡and Paul J. Chirik*^,†,§

^†Department of Chemistry and Chemical Biology, Baker Laboratory, Cornell University, Ithaca, New York 14853, United States

^‡Max-Planck Institute for Bioinorganic Chemistry, Stiftstrasse 34-36, D-45470 M€ulheim an der Ruhr, Germany

^§Department of Chemistry, Princeton University, Princeton, New Jersey, 08544, United States

S Supporting Information

b

ABSTRACT: Three new N-alkyl substituted bis(imino)-

pyridine iron imide complexes, (^iPrPDI)FeNR (^iPrPDI = 2,

6-(2,6-ⁱPr₂ÀC₆H₃ÀNdCMe)₂C₅H₃N;

R = 1-adamantyl

(¹Ad), cyclooctyl (^CyOct), and 2-adamantyl (²Ad)) were

synthesized by addition of the appropriate alkyl azide to the

iron bis(dinitrogen) complex, (^iPrPDI)Fe(N₂)₂. SQUIDmagnetic

measurements on the isomeric iron imides, (^iPrPDI)FeN¹Ad

and (^iPrPDI)FeN²Ad, established spin crossover behavior with

the latter example having a more complete spin transition in the

experimentally accessible temperature range. X-ray diﬀraction on all three alkyl-substituted bis(imino)pyridine iron imides

established essentially planar compounds with relatively short FeÀN_imidebond lengths and two-electron reduction of the redox-

active bis(imino)pyridine chelate. Zero- and applied-ﬁeld M€ossbauer spectroscopic measurements indicate diamagnetic ground

states at cryogenic temperatures and established low isomer shifts consistent with highly covalent molecules. For (^iPrPDI)FeN²Ad,

M€ossbauer spectroscopy also supports spin crossover behavior and allowed extraction of thermodynamic parameters for the S = 0 to

S = 1 transition. X-ray absorption spectroscopy and computational studies were also performed to explore the electronic structure of

the bis(imino)pyridine alkyl-substituted imides. An electronic structure description with a low spin ferric center (S = 1/2)

antiferromagnetically coupled to an imidyl radical (S_imide= 1/2) and a closed-shell, dianionic bis(imino)pyridine chelate (S_PDI= 0)

is favored for the S = 0 state. An iron-centered spin transition to an intermediate spin ferric ion (S_Fe= 3/2) accounts for the S = 1 state

observed at higher temperatures. Other possibilities based on the computational and experimental data are also evaluated and

compared to the electronic structure of the bis(imino)pyridine iron N-aryl imide counterparts.

’ INTRODUCTION

replacing one of the phosphine arms with a pyrazolyl donor.⁷

Modiﬁcation of the tris-phosphine ligand architecture has since

continued, and ﬁve-coordinate iron compounds with axial silyl⁸-

and borane⁹-tethers have been synthesized to enforce trigonal

bipyramidal coordination.

The pseudotetrahedral examples exhibit rich reaction chem-

istry including NÀH bond formation and ultimately aniline

dissociation following addition of H₂.¹⁰Smith and co-workers

have also explored pseudotetrahedral, neutral iron(III) and cationic

iron(IV) imides with a tripodal tris(N-heterocyclic carbene)borato

ligand. The iron(IV) example promotes hydrogen atom transfer of

weak CÀH bonds such as those in 9,10-dihydroanthracene allow-

ing construction of a thermodynamic cycle and determination of

the NÀH bond dissociation energy as 88(5) kcal/mol.¹¹

Three-coordinate, metastable, yet isolable, iron(III) imides

bearing β-diketiminate ligands have been reported by Holland

Terminal iron imido compounds, L_nFeNR, once rare, have

attracted renewed attention due to their interesting electronic

structures, unique reactivity, and importance as model compounds

for metalloenzymes and nitrogen ﬁxation.¹These compounds,

owing to their higher d-electron counts, oﬀer opportunities for

catalytic chemistry resulting from the relative destabilization

(viz. reduction in bond order) of the ironÀnitrogen bond.^1,2

Following Lee’s preparation and characterization of a tetra-

nuclear iron cubane structure with an FedN^tBu linkage,³a host

of iron imido complexes has since been prepared and fully char-

acterized. Examples are now known with oxidation states ranging

from Fe(II) to Fe(V)⁴with a variety of spin states.

Four-coordinate, pseudotetrahedral iron imides are by far the

most prominent coordination geometry because the π-antibond-

ing orbitals with respect to the FeÀN bonds are empty. Low-spin

Fe(II)⁵and Fe(III)⁶examples with tridentate tris(phosphino)-

borate ligands have been thoroughly studied by Peters and co-

workers. A cationic iron(IV) example has also been isolated by

Received: June 20, 2011

Published: October 10, 2011

r

2011 American Chemical Society

17353

dx.doi.org/10.1021/ja205736m J. Am. Chem. Soc. 2011, 133, 17353–17369

|

Journal of the American Chemical Society

ARTICLE

Scheme 1. Preparation and Hydrogenation Chemistry of

(

2,5-^tBu-C₆H₃), furnished the N-aryl substituted imido complexes,

(^iPrPDI)FeNAr.²⁶A combination of solid-state magnetic data,

M€ossbauer spectroscopy, and X-ray diﬀraction demonstrated

that these compounds are best described as having an intermedi-

ate spin ferric center (S_Fe= 3/2) antiferromagnetically coupled

to a chelate radical anion, [^iPrPDI]^1À, with a traditional [NR]^2À

fragment, producing an overall S = 1 ground state. The solid-state

structures also revealed distorted four-coordinate geometries

where the FedNR vector was nearly linear but signiﬁcantly lifted

from the idealized iron-chelate plane. The FeÀN_imidebond

lengths of 1.7048(16) (Ar = 2,6-ⁱPr₂-C₆H₃) and 1.7165(15) Å

(Ar = 2,4,6-Me₃-C₆H₂) are some of the longest known for term-

inal iron imides, consistent with an ironÀnitrogen bond order

less than three.¹³In three cases, addition of dihydrogen to the

(^iPrPDI)FeNAr compounds (Ar = 2,6-ⁱPr₂-C₆H₃; 2,6-Et₂-C₆H₃;

2,5-^tBu₂-C₆H₃) resulted in complete hydrogenation of the

ironÀnitrogen bond and liberation of the corresponding free

aniline. Catalytic hydrogenation of the aryl azides to the anilines

was observed in the presence of (^iPrPDI)Fe(N₂)₂(Scheme 1).²⁶

This reactivity is inspiring for hypothetical cycles involving homo-

geneous N₂hydrogenation²⁷and suggests that, if suﬃcient imido

character can be imparted to an iron dinitrogen complex, release

of hydrogenated products may indeed be plausible.

^iPrPDI)FedNAr Compounds

and co-workers.^12,13Spectroscopic, structural, and computa-

tional characterization have established S = 3/2 ground states

with short FedNR bonds of 1.68 Å and a formal FeÀN bond

order of 2.5.¹²Addition of a fourth ligand such as tert-butyl

pyridine resulted in CÀH bond activation chemistry by hydro-

gen atom transfer.^12,14,15Other notable reactivity includes reduc-

tive coupling to form an unusual diiron(II) hexazene compound¹⁶

and catalytic group transfer reactions with isocyanides and carbon

monoxide.¹⁷Using a related but weaker ﬁeld monoanionic,

bidentate dipyrromethane ligand scaﬀold, Betley and co-workers

have also explored the synthesis and reactivity of the correspond-

ing iron imide compounds.^18,19Inclusion of a chloride ligand

renders the formal oxidation state one unit higher than the ana-

logous four-coordinate β-diketiminate examples. The original

examples promote intramolecular CÀH activation of a ligand

substituent,¹⁸and modiﬁcation of the dipyrromethane and imide

architectures has resulted in iron compounds that are active for

the catalytic, intermolecular amination of a benzylic CÀH bond

of toluene and the aziridination of styrene.¹⁹Spectroscopic,

structural, and computational studies support ferric compounds

with an imido-based radical, a bonding motif ﬁrst positively iden-

tiﬁed in chromium chemistry.²⁰Hydrogen atom abstraction by

a putative Fe(IV) imide has also been postulated by Borovik

and co-workers following treatment of a four-coordinate ureate/

amidate complex with aryl azides.²¹

Our laboratory has been investigating the chemistry of re-

duced iron and cobalt compounds in the context of base-metal

catalysis^22,23and studying fundamental transformations relevant

to small-molecule activation with redox-active bis(imino)pyridine

ligands.^24,25Because this class of chelate is tridentate and prefers

meridional ligation, we sought to explore the synthesis and elec-

tronic structure of square-planar iron imido complexes, a geometry

and ligand ﬁeld distinct from the other classes of known L_nFeNR

compounds. These complexes may ultimately prove more re-

active, as dπ* orbitals with respect to the FeÀN bond will likely

be populated. Treatment of the bis(imino)pyridine iron bis-

(dinitrogen) complex, (^iPrPDI)Fe(N₂)₂(^iPrPDI = 2,6-(2,6-ⁱPr₂À

C₆H₃ÀNdCMe)₂C₅H₃N),²²with sterically hindered aryl azides,

ArN₃(Ar = 2,6-ⁱPr₂-C₆H₃, 2,6-Et₂-C₆H₃, 2,4,6-Me₃-C₆H₂, or

In this context, eﬀorts have also been devoted toward the pre-

paration of the bis(imino)pyridine iron parent imido derivative,

[(^iPrPDI)FedNH], by treatment of (^iPrPDI)Fe(N₂)₂with the

[NH] group transfer agent, Hdbabh (dbabh =2,3:5,6-dibenzo-7-

azabicyclo[2.2.1]hepta-2,5-diene).²⁸Instead of the desired iron

imido product, a 3:1 mixture of the iron amide and iron ammonia

compounds, (^iPrPDI)Fe(dbabh) and (^iPrPDI)FeNH₃, respectively,

was isolated (Scheme 2). Deuterium-labeling studies were con-

sistent with initial [NH] group transfer to form the desired parent

iron imide, [(^iPrPDI)FedNH], followed by NÀH activation of

the remaining Hdbabh to yield the observed products.

The structural distortions, electronic structure, and hydroge-

nation activity of the (^iPrPDI)FeNAr compounds and the pro-

posed NÀH bond activation chemistry of the putative parent

imide, (^iPrPDI)FeNH, prompted exploration of other bis(imino-

pyridine iron imido complexes with diﬀerent imido nitrogen

substituents. By manipulating both the steric and electronic pro-

perties of this group, we sought to determine the origin and impor-

tance of the deviation of the [NR] ligand from planarity and the

impact of this distortion on the electronic structure and ulti-

mately the reactivity of the molecule. Similar to the bis(imino)-

pyridine chelate, imido ligands are also redox-active^19,20render-

ing overall electronic structure determination of (^iPrPDI)FeNR

compounds challenging. Here we describe the synthesis and

spectroscopic characterization of three new bis(imino)pyridine

iron imido compounds bearing N-alkyl imido substituents and

establish spin crossover between S = 0 and S = 1 states. This

behavior is distinct from the N-arylated variants and highlights

the redox ﬂexibility of both the bis(imino)pyridine and the imido

fragments.

’ RESULTS AND DISCUSSION

Synthesis and Characterization of Bis(imino)pyridine Iron

N-Alkyl Imido Complexes. The synthetic method used to pre-

pare the bis(imino)pyridine iron N-aryl imido compounds²⁶was

extended to three new N-alkyl derivatives. Slow addition of a pentane

solution of either ¹AdN₃, ²AdN₃, or ^CyOctN₃(¹Ad = 1-adamantyl;

²Ad = 2-adamantyl; ^CyOct = cyclooctyl) to a pentane solution of

17354

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

Scheme 2. Preparation of Transient [(^iPrPDI)FeNH] by [NH] Group Transfer

(^iPrPDI)Fe(N₂)₂resulted in rapid effervescence of nitrogen gas

and furnished the desired iron N-alkyl imide complexes,

(^iPrPDI)FeN¹Ad and (^iPrPDI)FeN^CyOct, as dark purple crystals

in 60 and 86% yields, respectively, and (^iPrPDI)FeN²Ad as black

crystals in 73% yield following recrystallization (eq 1). Slow addi-

tion of the azide is required to avoid formation of iron tetrazole

compounds, (^iPrPDI)Fe[(R)N₄(R)],²⁹the synthesis and full char-

acterization of which will be reported elsewhere.³⁰We note that

until the data for the electronic structures of the compounds

are presented, the molecules will be drawn with the bis(imino)-

pyridine ligand in its neutral form.

Figure 1. Variable-temperature SQUID magnetization data for (^iPrPDI)-

FeN¹Ad. Experimental data are represented by open circles; data are

corrected for underlying diamagnetism. The simulation (red line) is a

superposition of a paramagnetic impurity with CurieÀWeiss behavior

(8% S = 1, Θ_W= À8 K, blue line) and an S = 0 to S = 1 phase transition,

for which the high-spin fraction is given by x = 1/[1 + exp{(nΔH/R)-

(1/T À 1/T_c)}] according to Sorai’s domain model,^32,33with a transi-

tion temperature T_c= 482 K and an enthalpy term nΔH = 1442 cm^À1for

domain size n, which determines the width of the transition.

The two isomeric iron imides, (^iPrPDI)FeN¹Ad and (^iPrPDI)-

FeN²Ad, exhibit diﬀerent NMR spectral features in benzene-d₆

solution at 23 °C. In contrast to the previously reported N-aryl

substituted bis(imino)pyridine iron imido compounds with S = 1

ground states,²⁶(^iPrPDI)FeN¹Ad exhibits sharp and readily assign-

able ¹H and ¹³C NMR spectra with observable J couplings at 23 °C,

suggestive of an S = 0 ground state. Similar spectra are observed

for (^iPrPDI)FeN^CyOct. Both compounds exhibit the number of

peaks consistent with C_2vsymmetric molecules in solution.

Despite the appearance of S = 0 ground states, widely dispersed

chemical shifts, especially for the in-plane chelate hydrogens,

were observed (Figure S1, Supporting Information [SI]). For

example, the imine methyl groups appear at À15.25 and À16.36

ppm for (^iPrPDI)FeN¹Ad and (^iPrPDI)FeN^CyOct, respectively.

Likewise, the meta- and para-pyridine hydrogens of (^iPrPDI)-

FeN¹Ad were located at 16.41 and 22.30 ppm. For (^iPrPDI)-

FeN^CyOct, these hydrogens were located at 18.02 and 20.64 ppm,

respectively.

shift window. For example, the imine methyl group was located

at À121 ppm and the most downﬁeld resonance, either the

p-pyridine or a 2-adamantyl hydrogen, appears at 94.01 ppm.

A benzene-d₆solution magnetic moment of 2.8 μ_Bwas measured

at 23 °C, consistent with an S = 1 state at this temperature.

In order to determine whether the unusual shifts are due to

thermal population of a low-lying triplet (S = 1) excited state, a

phenomenon observed by Budzelaar and co-workers in related

bis(imino)pyridine cobalt alkyl compounds,³¹variable temperature

NMR data in toluene-d₈solutions were collected for (^iPrPDI)-

FeN¹Ad. Cooling the sample resulted in a contraction of the

chemical shift dispersion observed for the compound (Figure S2, SI),

thereby ruling out temperature-independent paramagnetism.²⁴

For example, the resonance for the imine methyl group shifts

from À15.25 ppm at 23 °C to 0.06 ppm at À80 °C, the lowest

temperature experimentally accessible. Likewise, the p-pyridine reso-

nance moves upﬁeld from 22.30 ppm at 23 °C to 9.27 ppm at À80 °C.

By contrast, the benzene-d₆¹H NMR spectrum of (^iPrPDI)-

FeN²Ad at 23 °C is reminiscent of the S = 1 (^iPrPDI)FeNAr

compounds with resonances located over a 200 ppm chemical

17355

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

strength, such as thiocyanate (NCS^À).³⁴Spin crossover behavior

is also well-documented for iron complexes with terdentate ligands,

such as the terpyridine derivatives 2,4-bis-(2-pyridyl)thiazole and

2-(2-pyridylamino)-4-(2-pyridyl)thiazole.³⁵

The eﬀective magnetic moment was also measured for a

polycrystalline sample of (^iPrPDI)FeN²Ad between 2 and 300 K

(Figure 2). Although the NMR spectra for (^iPrPDI)FeN²Ad at

23 °C are similar to those observed for the aryl-substituted imido

complexes at the same temperature, the solid-state magnetic data

are markedly diﬀerent from the simple S = 1 behavior observed

for (^iPrPDI)FeNAr. Analogous to (^iPrPDI)FeN¹Ad, the solid-

state magnetic data demonstrate spin crossover from an S = 0 to

an S = 1 state. However, for (^iPrPDI)FeN²Ad, the data exhibit a

more pronounced spin transition and demonstrate that the S = 1

state is dominant at room temperature. This spin state is con-

sistent with the NMR data where paramagnetically broadened

and shifted resonances are observed at 23 °C. As with (^iPrPDI)-

FeN¹Ad, the low temperature plateau is higher than expected for

an S = 0 state. However, in the case of (^iPrPDI)FeN²Ad this

results from the presence of some fraction of the high-spin species

even at low temperatures. This was conﬁrmed by M€ossbauer

spectroscopy and will be discussed in a later section of the article.

X-ray Crystallographic Studies. The solid-state structures of

all three bis(imino)pyridine iron imides prepared in this work

were determined by single-crystal X-ray diffraction. Representa-

tions of the molecules are presented in Figures 3À5 and selected

metrical parameters are presented in Tables 1 and 2. Also

included in Table 1 are the corresponding distances and angles

for the previously structurally characterized S = 1 bis(imino)-

pyridine iron N-aryl substituted imido complexes, (^iPrPDI)-

FeNDipp and (^iPrPDI)FeNMes (Dipp =2,6-ⁱPr₂-C₆H₃; Mes =2,

4,6-Me₃-C₆H₂).²⁶

Figure 2. Variable-temperature SQUID magnetization data for (^iPrPDI)-

FeN²Ad. Experimental data are represented by open circles; data are

corrected for underlying diamagnetism. The simulation (red line) is an

S = 0 to S = 1 phase transition, for which the high-spin fraction is given by

x = 1/[1 + exp{(nΔH/R)(1/T À 1/T_c)}] according to Sorai’s domain

model,^32,33with a transition temperature T_c= 185 K and an enthalpy

term nΔH = 738 cm^À1for domain size n, which determines the width of

the transition.

The unusual NMR shifts and behavior of the three new iron

imide compounds prompted a more thorough investigation of

their magnetic properties. The two isomeric compounds, (^iPrPDI)-

FeN¹Ad and (^iPrPDI)FeN²Ad, were selected as representative

examples. The eﬀective magnetic moment was measured for a

polycrystalline sample of (^iPrPDI)FeN¹Ad between 2 and 300 K

by SQUID magnetometry. The data (Figure 1, open circles)

exhibit a steep increase with increasing temperature, followed by

a plateau at 0.8 μ_Band a second steep increase in eﬀective mag-

netic moment. The plateau is higher than the magnetic suscept-

ibility value expected for an S = 0 state; however, this low-

temperature region was successfully modeled using 8% of an S = 1

paramagnetic compound (Figure 1, blue line). This component

was also observed by M€ossbauer spectroscopy (vide infra) and is

derived from intramolecular CÀH bond activation of the bis-

(imino)pyridine ligand by (^iPrPDI)FeN¹Ad, the details of which

will be presented elsewhere.³⁰The increase of μ_eﬀ(T) above

150 K, however, could not be reasonably modeled using Boltzmann

population of a low-lying triplet state; the onset temperature and

the slope of the curve would not be consistent (Figure S3, SI).

Instead, the high temperature increase is consistent with spin

crossover (SCO) from an S = 0 to an S = 1 ground state, although

complete transition to the S = 1 state was not observed at the

temperatures studied. The data were ﬁt using the domain model

proposed by Sorai and Seki for a SCO transition^32,33with a

transition temperature of 483 K; thus, observation of a complete

transition is not expected (the high-spin fraction at 300 K is

6.8%). In addition, the change in chemical shift observed for the

methyl backbone resonance of (^iPrPDI)FeN¹Ad by variable-

Consistent with the idealized C_2vsymmetry observed in solu-

tion by NMR spectroscopy, the solid-state structure of (^iPrPDI)-

FeN¹Ad has a crystallographically imposed mirror plane that

contains the N(2)ÀFe(1)ÀN(3) plane and renders the two

halves of the molecule equivalent. As is typical in compounds of

this type, the aryl groups are oriented such that the isopropyl

substituents are above and below the metalÀchelate plane. The

iron imide group is slightly lifted from the idealized square plane

of the molecule with an N_pyÀFeÀN_imideangle of 168.10(8)°.

This distortion is signiﬁcantly smaller than those observed with

the N-aryl substituted imides, (^iPrPDI)FeNAr, whereN_pyÀFeÀN_imide

angles of 138.79(7)° (Ar = Dipp) and 154.75(7)° (Ar = Mes)

were observed. In (^iPrPDI)FeN¹Ad, the iron imide is slightly bent

with an FeÀN_imideÀC_imideangle of 159.87(15)°.

The solid state structure of (^iPrPDI)FeN^CyOct was also

determined by single crystal X-ray diﬀraction. Only small, weakly

diﬀracting crystals were obtained and as a result the data were

collected at 293 K using synchrotron (λ = 0.9782 Å) radiation.

The overall molecular geometry of the compound is similar to

(^iPrPDI)FeN¹Ad (Figure 4). Two molecules are present in the

unit cell, and metrical parameters for both are presented in

Table 1. Given the similarity between the data, only values from

one will be cited in the subsequent discussion. An idealized (but

not crystallographically imposed) C_2vgeometry is observed with

the orientation of the cyclooctyl substituent responsible for the

descent in symmetry. The iron imide group is the most planar

observed in the bis(imino)pyridine series of compounds with an

N_pyÀFeÀN_imideangle of 175.14(16)°. As with other iron imides

reported in Table 1, a slight bend is observed at the imide nitro-

gen with an FeÀN_imideÀC_imideangle of 160.9(3)°.

1

temperature H NMR could be ﬁt using a similar SCO model

for independent molecules in solution, yielding a somewhat

higher transition temperature (865 K) than the solid-state mag-

netization data (Figure S4, SI). Spin crossover behavior is commonly

observed for iron(II) and iron(III) compounds, especially for

complexes where strong ﬁeld ligands such as 1,10-phenanthro-

line or bipyridine are combined with ligands of intermediate ﬁeld

17356

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

Figure 3. Solid-state structure of (^iPrPDI)FeN¹Ad (left) and core of the molecule (right) at 30% probability ellipsoids. Hydrogen atoms omitted for

clarity.

Figure 4. Solid-state structure of (^iPrPDI)FeN^CyOct (left) and core of the molecule (right) at 30% probability ellipsoids. Hydrogen atoms omitted for

clarity.

Figure 5. Solid-state structure of (^iPrPDI)FeN²Ad (left) and core of the molecule (right) at 30% probability ellipsoids. Hydrogen atoms omitted for

clarity. Data collected at 253 K.

In both (^iPrPDI)FeN¹Ad and (^iPrPDI)FeN^CyOct, the FeÀN_imide

distances of 1.6481(15) and 1.648(3) Å are statistically indis-

tinguishable from each other but are signiﬁcantly shorter than the

FeÀN_imidebond lengths of 1.705(2) (Ar = Dipp) and 1.717(2) Å

(Ar = Mes) reported for the (^iPrPDI)FeNAr compounds. The

shorter FeÀN_imidebonds in the N-alkyl substituted compounds

are in the range of 1.625(4) to 1.6578(2) Å typically observed

in neutral, monomeric iron derivatives with terminal iron imide

ligands.¹The metrical parameters of the bis(imino)pyridine

chelates in both (^iPrPDI)FeN¹Ad and (^iPrPDI)FeN^CyOct are within

17357

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

Table 1. Selected Bond Distances (Å) and Angles (deg) for (^iPrPDI)FeN¹Ad, (^iPrPDI)FeN^CyOct, and (^iPrPDI)FeNAr^a

(

^iPrPDI)FeN¹Ad

(

^iPrPDI)FeN^CyOct

(

^iPrPDI)FeN^CyOct

(

^iPrPDI)FeNDipp

(

^iPrPDI)FeNMes

FeÀN_imide

FeÀN_py

FeÀN_imine

1.648(2)

1.895(2)

1.933(1)

1.648(3)

1.870(3)

1.901(3)

1.908(3)

1.658(3)

1.884(3)

1.917(3)

1.705(2)

1.840(2)

2.034(2)

1.987(2)

1.717(1)

1.872(1)

2.011(1)

1.979(1)

N_imineÀC_imine

C_imineÀC_ipso

1.340(2)

1.418(2)

1.345(5)

1.355(5)

1.405(5)

1.406(5)

1.347(5)

1.409(5)

1.413(6)

1.320(2)

1.331(2)

1.437(3)

1.431(3)

1.321(2)

1.334(2)

1.436(2)

1.426(2)

N_pyÀFeÀN_imide

FeÀN_imideÀC_ipso

168.10(8)

175.14(16)

160.9(3)

172.78(15)

157.7(4))

138.79(7)

154.75(7)

159.87(15)

165.68(15)

159.00(13)

^aData for the (^iPrPDI)FeNAr compounds taken from reference 26.

the range established for two-electron reduction (Table 1).^24,25

Notably, the distortions to the chelate are more pronounced in

these compounds than those observed for the (^iPrPDI)FeNAr

derivatives where one-electron bis(imino)pyridine reduction was

assigned.²⁶Thus, the position of the terminal iron imide inﬂu-

ences the degree of bis(imino)pyridine ligand reduction and

hence electronic conﬁguration at the metal.

Table 2. Selected Bond Distances (Å) and Angles (deg) for

^iPrPDI)FeN²Ad at 253 K, 220 K, and 100 K

(

^iPrPDI)FeN²Ad

(253 K)

(

^iPrPDI)FeN²Ad

(220 K)

(

^iPrPDI)FeN²Ad

(100 K)

FeÀN_imide

FeÀN_py

FeÀN_imine

1.680(3)

1.899(2)

1.983(2)

1.964(2)

1.675(4)

1.898(2)

1.978(2)

1.960(2)

1.666(10)

1.880(9)

1.971(9)

1.929(9)

X-ray quality crystals of (^iPrPDI)FeN²Ad were obtained by

cooling a concentrated pentane solution of the compound to

À35 °C. Because of the spin crossover behavior of the com-

pound, data were collected at three temperatures: 100, 220, and

253 K. These temperatures were selected to obtain metrical data

on the low- and high-spin forms as well as at an intermediate

temperature. Unfortunately, cooling the sample to temperatures

below 220 K consistently results in cracking of the crystal.

Accordingly, the data at 100 K are of relatively poor quality.

For example, high-quality data were collected at 253 K, then

cooling the sample resulted in the presence of multiple crystal-

lographic domains and weakening of the diﬀraction strength. A

representation of the solid-state structure determined at 253 K is

presented in Figure 5. Selected metrical parameters from all three

data sets are presented in Table 2.

N_imineÀC_imine

C_imineÀC_ipso

1.323(4)

1.330(4)

1.434(4)

1.415(4)

1.337(4)

1.339(4)

1.434(4)

1.414(4)

1.367(12)

1.367(14)

1.431(13)

1.387(15)

N_pyÀFeÀN_imide

FeÀN_imideÀC_ipso

176.28(15)

155.6(4)

175.64(13)

158.7(4)

172.4(4)

155.6(15)

The C_imineÀN_iminebond lengths are elongated to 1.323(4) Å and

1.330(4) Å, while the C_imineÀC_ipsodistances are contracted to

1.434(4) Å and 1.415(4) Å. These distortions are similar to those

observed with (^iPrPDI)FeN¹Ad and (^iPrPDI)FeN^CyOct and

are diagnostic of two-electron reduction. The FeÀN_pydistance

of 1.899(2) Å and the FeÀN_iminedistances of 1.983(2) Å and

1.964(2) Å are signiﬁcantly longer than the ironÀligand dis-

tances exhibited by the molecular structures of (^iPrPDI)FeN¹Ad

and (^iPrPDI)FeN^CyOct. As demonstrated by an overlay of the

molecular structures of isomeric (^iPrPDI)FeN¹Ad and (^iPrPDI)-

FeN²Ad (Figure S5, SI), the imide nitrogen and N-adamantyl

substituent of (^iPrPDI)FeN¹Ad are lifted out of the plane of the

chelate at an angle of 168.10(8)°, which is signiﬁcantly larger than

the angle of 176.28(15)° observed for (^iPrPDI)FeN²Ad. However,

the adamantyl groups overlay quite well for both complexes. In

addition, the shorter FeÀN_imidebond length of 1.6481(15) Å is

apparent in (^iPrPDI)FeN¹Ad.

The solid-state structure (Figure 5) establishes a C_2vsym-

metric molecule with idealized planar geometry with the sum of

the angles around iron equal to 360.08(9)°. The N_pyÀFeÀN_imide

angle of 176.28(15)° is nearly identical to that observed for

(^iPrPDI)FeN^CyOct, indicating that the imide substituent is not

signiﬁcantly lifted out of the idealized plane of the chelate

(Table 2). It should be noted that there is disorder and thermal

motion in the adamantyl group, resulting in elongation of the

N(4) ellipsoids. While this possibly indicates a deviation from

planarity, this distortion is small and not on the order of the

deviation seen with the aryl-substituted iron imides.

The FeÀN_imidebond length of 1.680(3) Å observed for

(^iPrPDI)FeN²Ad is slightly longer than the FeÀN_imidedistances

for (^iPrPDI)FeN¹Ad and (^iPrPDI)FeN^CyOct. The FeÀN_imide

distance for (^iPrPDI)FeN²Ad is also shorter than the FeÀN_imide

distances for (^iPrPDI)FeNDipp and (^iPrPDI)FeNMes. Consis-

tent with all other structurally characterized bis(imino)pyridine

iron imide complexes, a modest bend is observed at the imide nitro-

gen with an FeÀN_imideÀC angle of 155.6(4)°. Examination of

the metrical parameters of the bis(imino)pyridine chelate estab-

lish elongation of the C_imineÀN_iminebonds and contraction of the

C_imineÀC_ipsobonds that are consistent with two electron reduction.

€

Mossbauer Spectroscopic Studies. The electronic structures

of the bis(imino)pyridine iron imide complexes were also investi-

gated by both zero- and applied-field ⁵⁷Fe M€ossbauer spectro-

scopies. Representative zero-field spectra for (^iPrPDI)FeN¹Ad

and (^iPrPDI)FeN^CyOct are presented in Figure 6, and the data

are summarized in Table 3. The zero-field spectra of (^iPrPDI)-

FeN²Ad are temperature dependent and are more complex.

The discussion of these data will be presented separately.

17358

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

Figure 6. Zero-ﬁeld ⁵⁷Fe M€ossbauer spectra of (^iPrPDI)FeN¹Ad (left) and (^iPrPDI)FeN^CyOct (right) recorded at 80 K.

The N-alkyl substituted imides, (^iPrPDI)FeN¹Ad and (^iPrPDI)-

Table 3. ⁵⁷Fe Mo€ssbauer parameters of bis(imino)pyridine

iron imide compounds recorded at 80 K

FeN^CyOct, have isomer shifts of 0.04 and À0.02 mm s^À1

,

3

respectively, signiﬁcantly lower than the values of 0.30 mm s^À1

compound

δ (mm s^À1

)

ΔE_Q(mm s^À1

)

3

measured for both (^iPrPDI)FeNAr compounds (Table 3).²⁶

3

The N-alkyl substituted imides, (^iPrPDI)FeN¹Ad and (^iPrPDI)-

(

^iPrPDI)FeN¹Ad

^iPrPDI)FeN²Ad

^iPrPDI)FeN^CyOct

^iPrPDI)FeNDipp^a

^iPrPDI)FeNMes^b

0.04

0.01

À2.38

À2.26

2.38^c

FeN^CyOct also exhibit quadrupole splittings of |2.38| mm s^À1

,

3

larger than the absolute values of 1.08 and 1.34 mm s^À1deter-

À0.02

0.30

3

mined for (^iPrPDI)FeNDipp and (^iPrPDI)FeNMes, respectively.

Because some degree of spin transition was observed in the

SQUID magnetization data, the M€ossbauer spectrum of (^iPrPDI)-

FeN¹Ad was also recorded at room temperature (Figure S7, SI).

However, there is no observable diﬀerence from the 80 K spec-

trum, as is consistent with the incomplete spin crossover and the

resulting small amount of the high-spin fraction at 298 K.

Because complete spin crossover behavior for (^iPrPDI)-

FeN²Ad was established by SQUID magnetometry, zero-ﬁeld

⁵⁷Fe M€ossbauer spectra were recorded over a range of tempera-

tures (Figure 7). The spectrum obtained at 15 K exhibits a single

+1.08

1.34^c

0.30

^aData taken from reference 26. ^bSee Figure S6, SI. ^cSign not

determined.

quadrupole splitting decrease with increasing temperature, as is

observed for (^iPrPDI)FeN²Ad.

The presence of two temperature-dependent species in the

zero-ﬁeld M€ossbauer spectrum is consistent with spin crossover

behavior, and the relative amounts of the two iron species at a

given temperature are approximately consistent with the vari-

able-temperature magnetic data (Figure 2). Using the concen-

trations of each species observed at each temperature by

M€ossbauer spectroscopy, a van’t Hoﬀ plot was constructed

(Figure S8, SI). The ﬁt of the data yielded the following

M€ossbauer doublet with an isomer shift of 0.01 mm s^À1and

3

a quadrupole splitting of |2.26| mm s^À1, parameters that are

3

similar to the other alkyl iron imides (Table 3). However, upon

warming the sample to 80 K, a second quadrupole doublet was

observed that increased in intensity with additional warming.

The intensity of the doublet observed at 15 K decreased until it

was no longer observed at 200 K (Figure 7). The parameters of

thermodynamic parameters, ΔH° = 4.10(25) kJ mol^À1and

3

ΔS° = 31.6(1.9) J K^À1mol^À1, consistent with a low-spin

3

compound converting to a higher-spin species at higher tem-

peratures. The spin transition was also observed visually, with a

color change from black to purple occurring upon cooling a

benzene solution to 77 K.³⁶

the second quadrupole doublet (δ = 0.19 mm s^À1and ΔE_Q=

3

0.75 mm s^À1) are signiﬁcantly diﬀerent than the parameters of

3

the doublet observed at 15 K. Cooling the sample again to 15 K

regenerated the initial spectrum observed at that temperature,

indicating the spin crossover behavior is reversible. A small de-

crease in magnitude for both the isomer shift and the quadrupole

splitting of the S = 1 doublet was observed with increasing

temperature. Temperature-dependent M€ossbauer parameters

for a given doublet have been observed due to population of close-

lying excited states, where relaxation on the experimental time

scale is fast compared to the excited-state lifetime such that the

observed doublet represents an average of the ground state and

any populated excited states.³⁵Generally the isomer shift and

The van’t Hoﬀ relationship established for the spin transition

in (^iPrPDI)FeN²Ad by variable temperature M€ossbauer spec-

troscopy and the ﬁtting of the NMR data for (^iPrPDI)FeN¹Ad

allow comparison of the relative enthalpies for conversion from

the low spin species to the high spin species for each compound.

For (^iPrPDI)FeN²Ad, a value of ΔH = 342 cm^À1was obtained

while for the isomeric (^iPrPDI)FeN¹Ad derivative a value ΔH =

1083 cm^À1was determined.

Applied-ﬁeld M€ossbauer spectra were collected on examples

of an N-alkyl and an N-aryl substituted bis(imino)pyridine imide

17359

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

Figure 8. Applied ﬁeld M€ossbauer spectrum of (^iPrPDI)FeN¹Ad re-

corded at 7 T and 4.2 K.

Figure 9. Applied ﬁeld M€ossbauer spectrum of (^iPrPDI)FeN²Ad re-

corded at 7 T and 4.2 K.

(^iPrPDI)FeN¹Ad, (^iPrPDI)FeN²Ad, and (^iPrPDI)FeNDipp are

presented in Figures 8, 9 and 10, respectively. The data for

(^iPrPDI)FeN¹Ad and (^iPrPDI)FeN²Ad were recorded at 7 T,

both at 4.2 and 15 K. The data for (^iPrPDI)FeN¹Ad were sim-

ulated using the δ and ΔE_Qparameters from the zero-ﬁeld

measurements and indicate an S = 0 ground state for the mole-

cule. This is also supported by the observation that there is no

discernible diﬀerence between the applied-ﬁeld spectra at diﬀer-

ent temperatures, as is expected with a diamagnetic ground state.

From the ﬁt of the data, the sign and magnitude of the quadru-

Figure 7. Variable temperature zero-ﬁeld ⁵⁷Fe M€ossbauer spectra of

(

^iPrPDI)FeN²Ad.

to gain additional insight into the electronic structure diﬀerences

in the two classes of compounds. (^iPrPDI)FeN¹Ad was chosen as

a representative example of an N-alkyl substituted imido com-

plex while (^iPrPDI)FeNDipp was chosen as a representative

example for the N-aryl class of imides. The applied-ﬁeld spec-

trum of (^iPrPDI)FeN²Ad was also obtained to examine the S = 0

state observed at low temperature. The experimental data for

pole splitting was determined to be À2.38 mm s^À1with an

3

asymmetry parameter η = 0.28. Because an identical value of

|2.38| mm s^À1was measured for (^iPrPDI)FeN^CyOct, and the

3

two compounds otherwise have nearly identical spectroscopic

17360

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

Figure 11. Comparison of the normalized Fe K-edge XAS spectra for

(

^iPrPDI)Fe(DMAP), (^iPrPDI)FeNDipp, and (^iPrPDI)FeN¹Ad. Data

collected at 10 K.

intermediate spin iron(III) assignment.²⁶In addition, the ﬁt of

the data produced an Euler angle, β, of 49.34° where the β angle

is deﬁned as the diﬀerence between the magnetic quantization

axis (z with respect to the orbitals in a crystal ﬁeld picture with d_z₂

along the spin axis) and the main component, V_max, of the electric

ﬁeld gradient tensor. This misalignment of the principal coordi-

nate systems for the magnetic hyperﬁne tensor and the electric

ﬁeld gradient suggests that these two phenomena are dominated

by diﬀerent interactions, which may be correlated to the distorted

geometry of the molecule. Note that the angle β can be related to

the angle Θ (β ≈ 90 À Θ), which describes the bending of the

imide ligand with respect to the (PDI)Fe plane. A more detailed

analysis will be presented in Computational Studies.

X-ray Absorption Spectroscopic Studies. To gain additional

insight into the iron oxidation states in each class of iron imide,

two representative examples, (^iPrPDI)FeNDipp and (^iPrPDI)-

FeN¹Ad, were studied by X-ray absorption spectroscopy. Also

included in the study is (^iPrPDI)Fe(DMAP), an established iron(II)

compound with a dianionic bis(imino)pyridine chelate.²⁶A com-

parison of the normalized Fe K-edge XAS spectra for (^iPrPDI)-

Fe(DMAP), (^iPrPDI)FeNDipp, and (^iPrPDI)FeN¹Ad is shown

in Figure 11. The 1s to 3d pre-edge features appear at 7111.2,

7112.3, and 7112.4 eV for (^iPrPDI)Fe(DMAP), (^iPrPDI)FeNDipp,

and (^iPrPDI)FeN¹Ad, respectively. This is consistent with an Fe(II)

assignment for (^iPrPDI)Fe(DMAP)²⁴and an Fe(III) assignment

for (^iPrPDI)FeNDipp and (^iPrPDI)FeN¹Ad. Similar shifts are ob-

served in the rising edge inflection points, again indicating that an

overall ferric oxidation state assignment is most appropriate for

(^iPrPDI)FeN¹Ad. Additional features are present at ∼7115À

7117 eV, which are tentatively attributed to Fe 1s to PDI ligand

π* transitions. The scans were monitored for photoreduction

throughout the course of data collection, and no evidence for

beam-dependent reduction was observed. A detailed investiga-

tion of the energies and intensities of these features and their

correlation to the degree of ligand reduction will be the subject of

future studies.

Figure 10. Applied ﬁeld M€ossbauer spectra of (^iPrPDI)FeNDipp

recorded at 1 T, 4.2 K (bottom); 4 T, 4.2 K; 7 T, 4.2 K; 7 T, 60 K (top).

parameters, it is almost certain that the sign of ΔE_Qfor (^iPrPDI)-

FeN^CyOct is also negative.

The data for (^iPrPDI)FeN²Ad were also simulated using the δ

and ΔE_Qparameters from the zero-ﬁeld measurement. Similar

to (^iPrPDI)FeN¹Ad, a quadrupole splitting of À2.26 mm s^À1

3

and an asymmetry parameter η = 0.26 were determined for

(^iPrPDI)FeN²Ad from the ﬁt of the applied ﬁeld data. The data

were successfully ﬁt using an S = 0 model, and there was no

evident change in the applied-ﬁeld spectra at diﬀerent tempera-

tures. This indicates an S = 0 ground state for (^iPrPDI)FeN²Ad at

the temperatures measured (15 K and below), consistent with

the solid-state magnetic data (Figure 2).

Applied-ﬁeld measurements were also carried out on

(^iPrPDI)FeNDipp at 4.2 K and at magnetic ﬁelds of 1 T, 4 T,

7 T, as well as at 60 K with a magnetic ﬁeld of 7 T. The data were

simulated using the δ and ΔE_Qparameters obtained from the

zero-ﬁeld spectrum and an S = 1 spin state for the molecule

(Figure 10). The ﬁt of the data for (^iPrPDI)FeNDipp established

a positive quadrupole splitting of +1.08 mm s^À1. While the

3

applied ﬁeld measurement on (^iPrPDI)FeNMes was not carried

out, the similarity of this compound to (^iPrPDI)FeNDipp

suggests that the sign of |1.34| for ΔE_Qobtained from the

zero-ﬁeld measurement is also positive. In addition, the ﬁt of the

data yielded hyperﬁne coupling constants of A_xx/g_Nβ_N

=

À81.95, A_yy/g_Nβ_N= À8.69, and A_zz/g_Nβ_N= 145.60 kG, and

an asymmetry parameter, η, of 0.39. The obtained A values

indicate an internal magnetic ﬁeld around the iron nucleus where

the majority of spin is on the metal center, consistent with an

Computational Studies. The electronic structures of the

bis(imino)pyridine iron imido complexes were also studied with

DFT calculations. Unless otherwise noted, geometry optimiza-

tions were initiated from the crystallographically determined

17361

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

structures, and calculations were performed using the ORCA

program on the full molecules without truncations. Two exam-

ples, (^iPrPDI)FeNDipp and (^iPrPDI)FeN¹Ad, were chosen as

representative for the two distinct classes of imides. Both com-

pounds were structurally characterized and studied by an array of

spectroscopic techniques providing numerous experimental bench-

marks against which to calibrate the computational results. Recall,

a valid computational description must account for the following

experimental differences (^iPrPDI)FeNDipp and the alkyl-substituted

imides, (^iPrPDI)FeN¹Ad, respectively:

(i) S = 1 and S = 0 ground states at low temperature

(ii) [PDI]^Àversus [PDI]^2Àchelate reduction

Figure 12. Spin density plot for (^iPrPDI)FeNDipp obtained from a

Mulliken population analysis.

(iii) bent versus linear N_pyÀFeÀN_imideangle

(iv) long (1.705(2) Å) versus short (1.648(2) Å) FeÀN_imide

bonds

(v) high (δ = 0.30 mm s^À1) versus low (δ = 0.04 mm s^À1

)

3

Table 4. Comparison of Selected Experimental and

Computed Bond Distances (Å) and Angles (deg) for

M€ossbauer isomer shifts and quadrupole splitting (ΔE_Q)

of opposite sign

(

^iPrPDI)FeNDipp

(vi) similar Fe K-edge spectra, including pre-edge energies

(vii) ultimately electronic structure descriptions of both the

low- and high-spin states of the SCO alkyl-substituted

imides

experimental

BP86

B3LYP

FeÀN_imide

FeÀN_py

FeÀN_imine

1.705(2)

1.840(2)

2.034(2)

1.987(2)

1.710

1.824

1.998

1.977

1.795

2.009

2.481

2.129

In our initial communication, the electronic structure of the

aryl-substitued iron imido complex, (^iPrPDI)FeNDipp, was as-

signed solely on the basis of the experimental data as an inter-

mediate spin ferric compound (S_Fe= 3/2) antiferromagnetically

coupled to a bis(imino)pyridine radical (S_PDI= 1/2).²⁶Perform-

ing a geometry optimization on the compound with a triplet

ground state using the BP86 functional followed by an unrest-

ricted single-point calculation at the B3LYP level resulted in

spontaneous symmetry breaking with three unpaired electrons

on the [FedNAr] fragment and one electron with opposite spin

on the bis(imino)pyridine chelate. This solution is denoted

“broken symmetry (3,1)” or “BS(3,1)”.^37À39Examination of

the Mulliken spin density plot, presented in Figure 12, reveals

a more complex bonding picture. Considerable electron density

is located on the imide nitrogen and the 2,6-diisopropylphenyl

ring, indicating signiﬁcant covalency in the FeÀN_imidebond.

Accordingly, oxidation state assignments based on traditional

and overly simplistic formalisms become challenging and possi-

bly meaningless. This situation is reminiscent of S = 1 iron(IV)

oxo complexes where the covalent interaction between the metal

center and the oxo ligand places considerable spin density on the

formally O^2Àligand.^40,41In both the iron imido and oxo

complexes, it is important to note that the spin density on the

imido/oxo ligands is the same sign as the spin on the iron,

indicating considerable electron sharing, e.g. covalency, between

the metal and multiply bonded ligand. For (^iPrPDI)FeNDipp, the

DFT studies suggest that even with the ambiguities that arise

from covalency, the most reasonable electronic structure descrip-

tion is an intermediate spin ferric compound (S_Fe= 3/2) anti-

ferromagnetically coupled to a bis(imino)pyridine radical

(S_PDI= 1/2), e.g. (^iPrPDI^1À)Fe^IIINDipp, as we proposed

previously.²⁶It is important to note that, in this bonding des-

cription and hence iron oxidation state assignment, the imido ligand

is considered as a traditional, closed-shell dianionic fragment,

N_imineÀC_imine

C_imineÀC_ipso

1.320(2)

1.331(2)

1.437(3)

1.431(3)

1.347

1.349

1.436

1.435

1.290

1.330

1.439

1.482

N_pyÀFeÀN_imide

FeÀN_imideÀC_ipso

138.79(7)

138.58

166.60

143.85

170.11

165.68(15)

are also in good agreement with the experimental values. The

calculated isomer shift of 0.22 mm s^À1reproduces the experi-

_À3 ₁

mental measurement of 0.30 mm s within the accepted error

3

range of 0.10 mm s

.

^{À1 42,43}The small, positive calculated quad-

3

rupole splitting of +0.77 mm s^À1is also in excellent agreement

3

with the experimentally observed sign and value of +1.08 mm s^À1

.

3

Analysis of the calculated electric ﬁeld gradient revealed that its

orientation is dominated by the short, covalent FeÀN_imidebond

(Figure 13). The bent (out of the iron chelate plane) coordina-

tion of the imido ligand rationalizes the misalignment of the

principal coordinate systems of the electric ﬁeld gradient and the

magnetic hyperﬁne interaction observed in applied ﬁeld M€ossbauer

spectroscopy because the latter interaction is most likely domi-

nated by the antiferromagnetic coupling of the iron center with

the bis(imino)pyridine radical. The correct alignment of the

electric ﬁeld gradient further substantiates the validity of the

calculated electronic structure.

The geometry optimization of (^iPrPDI)FeNDipp at B3LYP

level of DFT produced a structure that exhibited signiﬁcant devia-

tions from the crystallographically determined parameters (Table 4).

Most signiﬁcantly, the FeÀN_iminebonds are overestimated by

more than 0.2 Å, while the FeÀN_pybond is elongated by 0.1 Å.

Analysis of the computed electronic structure revealed that the

overall S = 1 ground state arises from a high spin ferrous complex

(S_Fe= 2) antiferromagnetically coupled both to a bis(imino)-

pyridine radical anion (S_PDI= 1/2) and an imidyl radical,

[^•NAr]^1À. The computed M€ossbauer parameters for this elec-

tronic structure (δ = 0.64 mm s^À1, ΔE_Q= 1.52 mm s^À1), are

[NAr]^2À

.

The structural parameters obtained from the BS(3,1) solution

for (^iPrPDI)FeNDipp at the BP86 level are in good agreement

with the data obtained from X-ray diﬀraction (Table 4). The

computed M€ossbauer parameters based on this optimized structure

17362

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

Figure 13. Depiction of the electric ﬁeld gradient of (^iPrPDI)-

FeNDipp obtained from B3LYP DFT calculations.

Figure 14. Computed states for (^HPDI)FeNMe. All energies are

relative to the lowest-lying state (S = 1), and energies were obtained

from single-point calculations using the B3LYP functional on the

idealized, truncated molecule shown on the right.

typical for a high-spin ferrous complex but disagree with the

experimental data. As a result, this electronic structure was

discarded. However, this in silico experiment demonstrates the

sensitivity of the resulting electronic structure description to the

computational method and overall molecular geometry. It also

highlights that comparison of the computational output with experi-

mentally determined spectroscopic data is crucial for evaluating

the validity of the calculations. As will be shown in a subsequent

section, this sensitivity is magniﬁed in the essentially planar

imides represented by (^iPrPDI)FeN¹Ad.

At the B3LYP level of theory, several diﬀerent, energetically

plausible spin states were obtained. An ordering of these com-

puted states along with a depiction of the truncated molecule

used for these studies is presented in Figure 14. The lowest-

energy solution did not correspond to the experimentally deter-

mined low temperature S = 0 ground state but rather to a triplet

(S = 1) conﬁguration arising from spontaneous symmetry break-

ing and a BS(3,1) electronic structure description. Despite the

diﬀerence in geometry (idealized planar versus a lifted FeNR

fragment), this electronic structure is nearly identical to the one

described for (^iPrPDI)FeNDipp, where the iron is best described

as intermediate spin ferric (S_Fe= 3/2) with strong covalent

bonding between the iron center and the [NMe] fragment that

results in signiﬁcant spin delocalization onto the imido ligand.

Antiferromagnetic coupling between the half-occupied d_xzorbi-

tal and the bis(imino)pyridine b₂SOMO²⁴gives rise to the

overall S = 1 state. It is important to note that in this solution, the

bis(imino)pyridine chelate is in its monoreduced form, e.g.

[PDI]^1À, which is inconsistent with the experimentally obtained

crystallographic data for all three alkyl-substituted ironÀimido

complexes (Tables 1 and 2).

A closed-shell singlet state was obtained using a restricted

DFT approach and was found to be 13.2 kcal/mol higher in

energy than the BS(3,1) triplet state and is denoted RKS (S = 0)

in Figure 14. In this electronic structure, the bis(imino)pyridine

ligand is doubly reduced, e.g. [PDI]^2À(S_PDI= 0), with a doubly

occupied b₂orbital. If the imide ligand is considered in its typical

[NMe]^2Àform then the metal is best described as Fe(IV) with

doubly occupied d_yzand d_z₂orbitals (Figure 15). Once again, the

bonding between the iron and imide nitrogen is highly covalent

thereby complicating the oxidation state assignment. Examina-

tion of the doubly occupied molecular orbitals reveals a low

energy linear combination of the iron d_xyand imide nitrogen p_y

orbitals that is 40% metal in character. Thus, an ambiguity arises

where the metal may be assigned the +4 oxidation state as stated

above or as an alternative, low-spin Fe(II) compound with a neutral

nitrene ligand. The most accurate depiction of the compound

is likely a resonance structure between these two extremes with

a highly delocalized [FeNMe] fragment, akin to the bonding in

metal nitrosyls.⁴⁵Importantly, the XAS measurements on the

The electronic structure of (^iPrPDI)FeN¹Ad was also explored

computationally as a representative alkyl-substituted iron imide

that exhibits spin crossover behavior. Recall that both the applied

ﬁeld M€ossbauer experiments and magnetic susceptibility mea-

surements established a diamagnetic ground state at cryogenic

temperatures. At higher temperatures, however, there is conver-

sion to a triplet ground state, which is further substantiated by the

1

unusual H NMR shifts observed at room temperature. The

accurate theoretical treatment of SCO complexes is a challenge

for modern quantum chemical methods due to the presence of

two (or more) states close in energy. Even for simple mononuclear

coordination compounds the prediction of the correct ground

state by DFT is often challenging.⁴⁴The present case is further

complicated by the presence of potentially redox-active bis-

(imino)pyridine and imido ligands.

As an initial step, a variety of diﬀerent electronic structures and

spin states was investigated. To improve computational expe-

diency, these initial studies were performed without geometry

optimization on the simpliﬁed model complex, (^HPDI)FeNMe,

where the 2,6-diisopropyl substituents on the aryl rings were

replaced by hydrogen and the N_imideadamantyl substituent was

replaced by a methyl group. The computational input was also

constrained such that the imido fragment was forced into the

idealized ironÀchelate plane with a perfectly linear FeÀN_imideÀCH₃

bond and hence overall C_2vsymmetry. The phenyl rings on the

imines were also ﬁxed orthogonally to the metal-chelate plane

and the bond distances in the molecule constrained to the values

determined by X-ray diﬀraction. The results obtained from this

model system were then used as a benchmark for the full mole-

cule calculations. Computed M€ossbauer parameters for various

electronic structure descriptions of the model complex, (^HPDI)-

FeNMe, are reported in Table S1, SI.

17363

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

Figure 15. Molecular orbital diagram obtained from a restricted closed-shell calculation for (^HPDI)FeNMe (RKS S = 0).

experimental compound, (^iPrPDI)FeN¹Ad, are consistent with

an average of these two extremes and give the appearance of an

Fe(III) compound.

overlap values of 0.81 and 0.77 for the d_yzand d_xyorbitals with

their corresponding orbitals on the ligands imply high covalency

and lead to reduced spin density values.

In addition to the closed-shell state, three singlet states arising

from broken symmetry solutions were also evaluated (Figure 14).

The BS(1,1) and BS(2,2) calculations converge to the same

solution, corresponding to a BS(2,2) state, arising from two un-

paired electrons on an intermediate spin ferrous center (S_Fe= 1)

Additional unrestricted DFT calculations revealed two other

paramagnetic states. The lower in energy of the two is a quintet

(S = 2) state that is nearly degenerate with the closed-shell, RKS

(S = 0) solution, and is 13.3 kcal/mol above the lowest-energy

triplet (Figure 14). Analysis of the orbital picture from this state

revealed a similar electronic structure to the low-energy triplet

with the principal diﬀerence arising from ferromagnetic rather

than antiferromagnetic coupling between the unpaired spins on

the iron and the bis(imino)pyridine radical. The second para-

magnetic excited state is a septet (UKS, S = 3, Figure 14) and is

well separated by 31.9 kcal/mol from the low-energy triplet. The

Mulliken spin density indicates two unpaired spins on the

bis(imino)pyridine ligand, e.g. [PDI]^2À, S_PDI= 1, three unpaired

spins on the iron and one additional spin on the imide fragment.

In summary, the initial investigations based on the truncated

model system with enforced C_2vsymmetry reveal that a number

of spin states may be accessible in a relatively narrow energy

range. In contrast to the experimental data, the lowest-energy

state is a triplet with a monoanionic bis(imino)pyridine. How-

ever, it has been shown that modern hybrid DFT functionals like

B3LYP are slightly biased in favor of the high-spin state for SCO

compounds and spin state alone should not be the only measure

of the validity of the computational results.⁴⁴

antiferromagnetically coupled to a [PDI]^1Àradical anion (S_PDI

=

1/2) and an [^•NMe]^1Àimidyl radical (S_imide= 1/2). The d_yzand

d_xyorbitals on the iron are doubly occupied while the d_x₂_Ày₂

orbital is empty. The remaining d_xzand d_z₂orbitals are singly

occupied. While the d_xzorbital has the appropriate symmetry to

overlap with the singly occupied b₂orbital of the bis(imino)-

pyridine ligand resulting in antiferromagnetic coupling, the d_z₂

orbital cannot interact with the unpaired electron of the imidyl

radical, which is predominantly located in the p_yorbital of the

imide nitrogen. Despite this unfavorable magnetic interaction,

the BS(2,2) open-shell singlet solution is slightly lower in energy

than the closed-shell singlet. Itiscomputed to be only 9.5 kcal/mol

higher in energy than the low-lying triplet solution. However, the

computed singly reduced bis(imino)pyridine is inconsistent with

the experimentally determined bond distances and is therefore

discarded.

Finally, a BS(3,3) calculation yielded the lowest-energy singlet

solution, lying only 7.6 kcal/mol above the low-energy triplet.

Analysis of the resulting orbitals demonstrates that the iron center

is best described as intermediate spin ferric (S_Fe=3/2) with a

doubly occupied d_z₂orbital, three singly occupied orbitals (d_xy,

d_xz, d_yz) and an empty d_x₂_Ày₂orbital. Antiferromagnetic coupling

of the electrons in the d_xzand d_yzorbitals with the b₂and a₂

SOMO orbitals of the triplet (S_PDI= 1) bis(imino)pyridine

diradical, respectively, and of the d_xyorbital with an imidyl radical

(S_imide= 1/2) gives rise to the overall S = 0 state. However, high

The computational results obtained from the model complex

inspired the three approaches used for the full molecule: a

restricted closed-shell singlet (RKS), an open-shell singlet BS-

(3,3), and an open-shell triplet BS(3,1) calculation. Consistent

with the calculations for (^iPrPDI)FeNDipp, geometry optimiza-

tions for (^iPrPDI)FeN¹Ad were initially carried out at the BP86

level of DFT without imposing geometric constraints. However,

all attempts to obtain an open-shell singlet solution failed with

17364

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

Table 5. Comparison of Selected Experimental and

Computed Bond Distances (Å) and Angles (deg) for

(

Table 6. Comparison of Experimental and Computed

Mo€ssbauer Parameters for (^iPrPDI)FeN¹Ad

^iPrPDI)FeN¹Ad

experimental

BS(3,1)

BS(2,2)

RKS

experimental

BS(3,1)

BS(2,2)

RKS

δ/ mm s^À1

0.04

0.09

1.29

0.04

0.01

3

ΔE_Q/ mm s^À1

À2.38

À2.03

À4.08

FeÀN_imide

FeÀN_py

1.648(2)

1.895(2)

1.933(1)

1.727

1.914

2.053

1.774

1.896

2.072

1.629

1.900

2.002

3

FeÀN_imine

while it is in good agreement for the broken-symmetry, BS(2,2)

description.

N_imineÀC_imine

C_imineÀC_ipso

1.340(2)

1.418(2)

1.325

1.443

1.324

1.442

1.331

1.428

Unfortunately, the computational results described above do

not provide a conclusive description of the electronic ground

state of (^iPrPDI)FeN¹Ad. Of all investigated models for the full

molecule, the restricted closed-shell (RKS) solution best repro-

duces the experimental criteria listed at the beginning of the

computational section. Most importantly, and in contrast to the

two alternative descriptions, it reproduces the bond distances of

the doubly reduced bis(imino)pyridine ligand. Furthermore, the

diamagnetic low-spin conﬁguration in both resonance structures

N_pyÀFeÀN_imide

168.10(8)

171.21

160.29

169.35

155.82

166.23

168.50

FeÀN_imideÀC_ipso159.87(15)

this functional. Instead, the calculations converged back to a

closed-shell singlet solution. This is a well-documented phenom-

enon for pure gradient functionals like BP86, which overestimate

the stability of pure low-spin states.⁴⁶Accordingly, the B3LYP

functional was used for geometry optimizations for all three

approaches. Note that in contrast to the calculations for (^iPrPDI)-

FeNDipp, the obtained geometries for the BS(3,1) and the RKS

approach were almost identical for both functionals in this case.

Although the electronic structures obtained from the closed-shell

singlet and the triplet computations were qualitatively identical

to those obtained for the idealized planar model system, all

attempts to generate a BS(3,3) solution failed. Instead, the mole-

cule converged to a BS(2,2) solution that is diﬀerent from the

BS(2,2) solution described above. In contrast to the truncated

model system, the two unpaired electrons on the iron(II) center

are located in the d_xzand the d_xyorbitals, maximizing the overlap

with the singly occupied b₂orbital of the bis(imino)pyridine

radical and the imidyl radical ligand.

Comparing the total energies of the three electronic structures,

the triplet solution was the lowest in energy while the open-shell

singlet (BS(2,2)) and the closed-shell singlet, were 7.1 and 18.1

kcal/mol higher in energy, respectively. This is again in agree-

ment with the slight bias of modern hybrid DFT functionals

toward high-spin states in SCO complexes.⁴⁴A comparison of

the computed structural parameters for each solution is reported

in Table 5. In agreement with their electronic structures, the

metrical parameters obtained from the triplet BS(3,1) and open-

shell singlet BS(2,2) calculations exhibit bis(imino)pyridine

bond distances indicative of a monoreduced [PDI]^1Àradical

ligand. This is in conﬂict with the experimental bond distances,

which indicate a doubly reduced [PDI]^2Àligand. Furthermore,

the FeÀN_imideand FeÀN_iminebond lengths are signiﬁcantly

overestimated by both solutions. In contrast, the closed-shell

singlet solution (denoted RKS) exhibits much better agreement

with the experimental structure (Table 5).

(low-spin Fe(II), neutral nitrene T low-spin Fe(IV), [NR]^2À

)

readily explains the very short FeÀN_imineand FeÀN_imidebonds

and leads to a higher formal bond order for the FeÀN_imidebond

as compared to the aryl-substituted iron imides. In contrast to the

S = 1 and BS(2,2) solutions, the closed-shell RKS solution is in

agreement with the experimentally observed diamagnetic ground

state (at cryogenic temperatures) and the metal oxidation state

observed by XAS (also recorded at cryogenic temperatures),

respectively. Nevertheless, the relatively high energy of this

solution and the deviation of the computed M€ossbauer quadru-

pole splitting are problematic. Because none of the employed

computational approaches consistently reproduces the structural

and spectroscopic observables of the compound, it seems likely

that the complexity of the electronic structure of (^iPrPDI)-

FeN¹Ad is beyond the scope of the current, single-determinant

DFT methodology applicable to systems of the given size.

Evaluation of Electronic Structure Possibilities. From the

combination of spectroscopic data, metrical parameters from X-ray

diffraction, magnetochemistry, and computational studies, a clear

description of the electronic structure the aryl-substituted bis-

(imino)pyridine compounds has emerged. This class of mol-

ecules is best described as intermediate spin ferric compounds

(S_Fe= 3/2) antiferromagnetically coupled to a bis(imino)pyridine

radical anion with a traditional closed-shell imido fragment,

[NAr]^2À. In this bonding description, four iron d-orbitals are

populated giving rise to an FeÀN_imidebond order of 2 with two

2-center 3-electron π bonds, consistent with the long FeÀN

bond distances of 1.705(2) (Ar = Dipp) and 1.717(2) Å (Ar =Mes)

observed experimentally.

The electronic structure of the alkyl-substituted bis(imino)-

pyridine iron imides is not as straightforward to interpret despite

the same breadth of experimental and computational data. The

combination of spin crossover behavior and two potentially

redox active ligands, the bis(imino)pyridine and the imide frag-

ment, complicate oxidation state assignment. Possibilities for the

S = 0 ground states will be initially evaluated. Recall that the metrical

parameters from X-ray diﬀraction establish two-electron chelate

reduction in all cases, and therefore only electronic structures

derived from these possibilities will be considered (Figure 16).

The most plausible electronic structure description for the S = 0

ground states of the bis(imino)pyridine iron N-alkyl-substituted

imido complexes is a resonance structure between Fe(II) (A)

and Fe(IV) (B). This depiction is supported computationally

To further assay the validity of each solution, M€ossbauer para-

meters were computed and compared to the experimentally

determined values for (^iPrPDI)FeN¹Ad. These values are re-

ported in Table 6. All three solutions successfully reproduce the

experimental isomer shift, but the calculated quadrupole split-

tings vary widely. Most notably, the triplet BS(3,1) calculation

yielded a relatively narrow and positive ΔE_Q, in disagreement

with the experimental data. In fact, this value more closely

resembles the values for the S = 1 aryl-substituted iron imides,

(^iPrPDI)FeNAr. By contrast, the two singlet solutions reproduce

the negative quadrupole splitting. However, the magnitude of

ΔE_Qis severely overestimated by the closed-shell (RKS) model,

17365

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

Figure 16. Electronic structure possibilities for N-alkyl-substituted bis(imino)pyridine imido compounds.

(Figure 15) and indicates a high degree of covalency between the

iron and the bis(imino)pyridine and imide ligands. Such a bond-

ing situation, often encountered with metal nitrosyls,⁴⁵renders

a single oxidation state assignment physically meaningless. With

the present compounds, this is apparent from M€ossbauer and

X-ray data that are consistent with an Fe(IV) assignment (or low-

spin Fe(II) and XAS measurements that detect the “hybrid” ferric

form). These low-spin assignments, either Fe(II) or Fe(IV),

account for the contracted FeÀN_imidedistances as well as the

relatively short metal-chelate bond lengths. Two other Fe(III)

possibilities, structures C and D, were also considered. Possibility

C, with an intermediate spin ferric center (S_Fe= 3/2) antiferro-

magnetically coupled to both a imidyl radical and a triplet bis-

(imino)pyridine diradical, has been eliminated, as the M€ossbauer

isomer shift is too low for an intermediate spin Fe(III) com-

pound. By contrast, possibility D, with a low-spin Fe(III) center

antiferromangetically coupled to an imidyl radical is consistent

with all of the experimental data and is a plausible electronic

structure description for this class of compounds. We note that

such an electronic structure description could not be reproduced

computationally.

supports these possibilities. Given all of the experimental and

computational data, this electronic structure, along with structure

D in Figure 16, are the preferred electronic structures for

(^iPrPDI)FeN²Ad.

One ﬁnal question is why do the electronic structures of the

bis(imino)pyridine iron imides vary as a function of the N-imido

substituent? The current data are most consistent with a steric

eﬀect that in turn triggers electronic structure diﬀerences. Both

the FeÀN_imidedistance and vector (in-plane versus out-of-plane)

are determined by the size of the imido substituent, where the

larger 2,6-disubstituted aryl groups lift out of the ironÀchelate

plane to avoid steric interactions with the bis(imino)pyridine aryl

substituents. Accordingly, the ironÀnitrogen overlap decreases,

resulting in an overall weaker ﬁeld interaction giving rise to a one-

electron reduced bis(imino)pyridine and higher spin state at

iron. For the N-alkyl substituted cases, the FeÀNR fragment can

lie between the chelate aryl substituents and increase orbital

overlap with the metal, increasing its covalency and, accordingly,

its ﬁeld strength, resulting in a two-electron reduced chelate and a

low-spin iron center. These eﬀects highlight the true redox

ﬂexibility of the bis(imino)pyridine chelate and the ability to

accept electron density in response to the ﬁeld strength of a trans

(or pseudo-trans) ligand.

What is the electronic structure description for the S = 1 state

of (^iPrPDI)FeN²Ad? At ﬁrst glance, one might conclude this

compound has the same electronic structure as the S = 1

(^iPrPDI^1À)Fe^IIINAr class of molecules. While much of the high

temperature (T > 200 K) experimental data for (^iPrPDI)FeN²Ad

begins to approach the values obtained for the (^iPrPDI^1À)-

Fe^IIINAr derivatives, in many cases it remains suﬃciently dis-

similar, supporting yet another unique electronic structure. Re-

call that the metrical parameters from the X-ray data, collected at

253 K where the molecule is completely in its S = 1 form,

establish a two-electron reduced bis(imino)pyridine chelate and

a nearly planar molecule, features distinct from the N-aryl

substituted imides. In addition, the M€ossbauer parameters (δ =

0.19 mm s^À1and ΔE_Q= |0.75| mm s^À1) are lower than the

’ EXPERIMENTAL SECTION

General Considerations. All air- and moisture-sensitive manip-

ulations were carried out using standard vacuum line, Schlenk, and

cannula techniques or in an MBraun inert atmosphere drybox contain-

ing an atmosphere of purified nitrogen. Solvents for air- and moisture-

sensitive manipulations were initially dried and deoxygenated using

literature procedures. Benzene-d₆was purchased from Cambridge Iso-

tope Laboratories and dried over 4 Å molecular sieves. The iron com-

plexes, (^iPrPDI)Fe(N₂)₂,²²(^iPrPDI)FeNDipp,²⁶and (^iPrPDI)FeNMes,²⁶

were prepared as described previously.

¹H NMR spectra were recorded on Varian Mercury 300, Inova 400,

and 500 spectrometers operating at 299.76, 399.78, and 500.62 MHz,

respectively. All ¹H chemical shifts are reported relative to SiMe₄using

the ¹H (residual) shift of the solvent as a secondary standard. Variable

temperature (2À300 K) magnetization data were recorded in a 1 T

magnetic ﬁeld on a MPMS Quantum Design SQUID magnetometer.

The experimental magnetic susceptibility data were corrected under-

lying diamagnetism using tabulated Pascal’s constants.⁴⁸Elemental

analyses were performed at Robertson Microlit Laboratories, Inc., in

Madison, NJ.

3

aryl compounds (δ = 0.30 mm s^À1and ΔE_Q= +1.08 and

3

|1.34| mm s^À1). Unfortunately, due to the limitations of applied

3

ﬁeld M€ossbauer spectroscopy the sign of the quadrupole splitting

for the S = 1 state of (^iPrPDI)FeN²Ad cannot be experimentally

determined. One possibility for the S = 1 state of

(^iPrPDI)FeN²Ad is an intermediate spin Fe(IV) (or intermediate

spin Fe(II)) compound with closed-shell dianionic, [PDI]^2Àand

[N²Ad]^2À, ligands. An alternative and equally plausible possibi-

lity is an intermediate spin ferric compound (S_Fe= 3/2)

antiferromagnetically coupled to an imidyl radical (S_imide

1/2) with a closed-shell dianionic chelate. The observation

of a higher M€ossbauer isomer shift for the S = 1 spin isomer

=

X-ray crystallographic data were collected at 173 K on a Bruker X8

APEX2 diﬀractometer diﬀractometer using graphite monochromated

Mo Kα radiation (λ = 0.71073 Å), 100 K on a Bruker D8 Avance

17366

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

diﬀractometer using graphite monochromated Mo Kα or Cu Kα radiation

(λ = 1.54178 Å). A hemisphere routine was used for data collection and

determination of lattice constants. The space group was identiﬁed and the

data were processed using the Bruker SAINT+ program and corrected for

absorption using SADABS. The structures were solved using direct methods

completed by subsequent Fourier synthesis and reﬁned by full-matrix least-

squares procedures. Unless otherwise stated, all non-hydrogen atoms were

reﬁned with anisotropic displacement parameters and hydrogen atoms were

placed using a riding model and reﬁned with ﬁxed isotropic displacement

parameters. The crystal structure of (^iPrPDI)FeN^CyOct was of poor data

quality and was collected at the Cornell High Energy Synchroton Source

(CHESS) at λ = 0.97820 Å at 293 K.

Preparation of (^iPrPDI)FeN¹Ad. A 20-mL scintillation vial was

charged with 0.094 g (0.174 mmol) of (^iPrPDI)Fe(N₂)₂and approxi-

mately 15 mL of pentane forming a dark green solution. A solution of

0.028 g (0.174 mmol) 1-adamantyl azide in 3 mL of pentane was

prepared and added dropwise to the vial, resulting in a gradual color

change of the solution from green to purple. The resulting solution was

stirred for 15 min then filtered through a glass frit. The purple solution

was cooled to À35 °C to yield 0.065 g (60%) of purple diamond-shaped

crystals identified as (^iPrPDI)FeN¹Ad. Analysis for C₄₃H₅₈N₄Fe: Calc.

C, 75.20; H, 8.51; N, 8.16. Found C, 74.82; H, 8.49; N, 7.80. ¹H NMR

(benzene-d₆, 20 °C): δ = À15.25 (s, 6H, C(CH₃)), À6.72 (br s, 4H,

CH(CH₃)₂), À1.23 (br s, 12H, CH(CH₃)₂), À0.48 (br s, 12H,

CH(CH₃)₂), 2.54 (br s, 6H, adamantyl CH₂), 3.05 (d, 12 Hz, 3H,

adamantyl CH₂), 3.88 (d, 12 Hz, 3H, adamantyl CH₂), 6.03 (d, 8 Hz,

4H, m-aryl), 7.27 (t, 8 Hz, 2H, p-aryl), 9.27 (br s, 3H, adamantyl CH₂),

Zero-ﬁeld ⁵⁷Fe M€ossbauer data were recorded on a SEE Co.

M€ossbauer spectrometer (MS4) at 80 K in constant acceleration mode,

using ⁵⁷Co/Rh as the radiation source. The minimum experimental line

width was 0.24 mm s^À1(full width at half-height). Zero-ﬁeld M€ossbauer

spectra were simulated with the program WMOSS (least-squares ﬁtting

to Lorentzian peaks). The temperature of the samples was controlled by

a Janis Research Co. CCS-850 He/N₂cryostat within an accuracy of

(0.3 K. Applied ﬁeld M€ossbauer data were collected using an Oxford

Instruments M€ossbauer-Spectromag cryostat, where the split-pair super-

conducting magnet system allows applied ﬁelds up to 8 T and the

temperature of the samples can be varied in the range of 1.5À250 K. The

ﬁeld at the sample is perpendicular to the γ beam. Magnetic M€ossbauer

spectra were simulated with the program MX (by Eckhard Bill). Isomer

shifts for all spectra were determined relative to α-iron at 298 K.

XAS data were recorded at the Stanford Synchrotron Radiation

Laboratory (SSRL) on focused beamline 9-3, under ring conditions of

3 GeV and 60À100 mA. A Si(220) double-crystal monochromator was

used for energy selection, and a Rh-coated mirror (set to an energy cutoﬀ

of 10 keV) was used for harmonic rejection. Internal energy calibration

was performed by assigning the ﬁrst inﬂection point of the Fe foil spec-

trum to 7111.2 eV. Samples were prepared by dilution in boron nitride,

pressed into a pellet and sealed between 38 μm Kapton tape windows in

a 1-mm aluminum spacer. All samples were maintained at 10 K during

data collection using an Oxford Instruments CF1208 continuous ﬂow

liquid helium cryostat. Data were measured in transmission using nitrogen-

ﬁlled ionization chambers. The data were calibrated and averaged using

EXAFSPAK.⁴⁹Normalization of the data was achieved by subtracting

the spline and normalizing the postedge region to 1.

1

16.41 (br s, 2H, m-pyridine), 22.30 (br s, 1H, p-pyridine). H NMR

(toluene-d₈, À80 °C): À0.06 (s, 6H, C(CH₃)), 0.56 (br s, 12H,

CH(CH₃)₂), 0.61 (br s, 12H, CH(CH₃)₂), 1.44 (br s, 6H, adamantyl

CH₂), 1.65 (br s, 4H, CH(CH₃)₂), 1.82 (br s, 3H, adamantyl CH₂), 2.45

(br s, 3H, adamantyl CH₂), 2.75 (br s, 3H, adamantyl CH₂), 6.76 (br s,

4H, m-aryl), 7.10 (br s, 2H, p-aryl), 8.75 (br s, 2H, m-pyridine), 9.27 (br s,

1H, p-pyridine). ¹³C {¹H} NMR (benzene-d₆): δ = 20.61, 20.78, 29.93,

30.11, 33.39, 42.66, 42.93, 71.40, 95.52, 100.41, 117.73, 124.05, 124.23,

140.92, 178.98, 202.17.

Preparation of (^iPrPDI)FeN^CyOct. This compound was prepared

in the same manner as (^iPrPDI)FeN¹Ad with 0.244 g (0.411 mmol) of

(

^iPrPDI)Fe(N₂)₂and 0.057 g (0.372 mmol) of cyclooctyl azide and

yielded 0.243 g (86%) of dark-purple crystals identified as (^iPrPDI)-

FeN^CyOct. Analysis for C₄₁H₅₈N₄Fe: Calc. C, 74.30; H, 8.82; N, 8.45.

Found C, 74.67; H, 8.80; N, 7.97. ¹H NMR (benzene-d₆): δ = À16.36

(s, 6H, C(CH₃)), À6.31 (br s, 4H, CH(CH₃)₂), À0.96 (br s, 12H,

CH(CH₃)₂), À0.24 (br s, 12H, CH(CH₃)₂), 3.26 (m, 1H, cyclooctyl

CH), 3.90 (m, 8H, cyclooctyl CH₂), 4.07 (m, 2H, cyclooctyl CH₂), 4.58

(m, 2H, cyclooctyl CH₂), 5.48 (m, 2H, cyclooctyl CH₂), 6.17 (d, 7.5 Hz,

4H, m-aryl), 7.40 (t, 7.5 Hz, 2H, p-aryl), 18.02 (br s, 2H, m-pyridine),

20.64 (br s, 1H, p-pyridine). {¹H}¹³C NMR (benzene-d₆): δ = 20.31,

31.84, 33.94, 34.37, 39.96, 72.71, 76.49, 106.04, 118.36, 124.22, 126.73,

136.45, 166.14, 189.59, 210.32.

Preparation of (^iPrPDI)FeN²Ad. This compound was prepared in

the same manner as (^iPrPDI)FeN¹Ad with 0.103 g (0.174 mmol) of

(

^iPrPDI)Fe(N₂)₂, approximately 15 mL of pentane, and 0.031 g (0.174

Computational Details. All DFT calculations were performed

with the ORCA program package⁵⁰using the BP86^51À53or B3LYP^51,54,55

functionals.⁵⁶The all-electron Gaussian basis sets were those developed

by the Ahlrichs group.^57À59Triple-ζ quality basis sets def2-TZVP with

one set of polarization functions on iron and on the atoms directly coor-

dinated to the metal center were used. For the carbon and hydrogen

atoms, slightly smaller polarized split-valence def2-SV(P) basis sets were

used, that were of double-ζ quality in the valence region and contained a

polarizing set of d-functions on the non-hydrogen atoms. Auxiliary basis

sets were chosen to match the orbital basis.^60À62The RIJCOSX^63À65

approximation was used to accelerate the calculations.

mmol) of 2-adamantyl azide and yielded 0.087 g (73%) of black crystals

identified as (^iPrPDI)FeN²Ad. Analysis for C₄₃H₅₈N₄Fe: Calc. C, 75.20;

H, 8.51; N, 8.16. Found C, 74.78; H, 7.75; N, 7.83. Magnetic suscept-

ibility (benzene-d₆, 293 K): μ_eff= 2.8(1) μ_B. ¹H NMR (benzene-d₆): δ =

À121.05 (152.0 Hz, 6H, C(CH₃)), À56.34 (273.0 Hz, 4H, CH(CH₃)₂),

À12.11 (91.7 Hz, 12H, CH(CH₃)₂), À7.30 (44.6 Hz, 12H, CH-

(CH₃)₂), À0.03 (47.8 Hz, 4H), 7.87 (40.0 Hz, 2H), 12.07 (36.3 Hz,

2H), 14.07 (77.5 Hz, 4H), 21.68 (101.0 Hz, 4H), 21.95 (156.2 Hz, 2H),

60.74 (204.7 Hz, 2H), 72.80 (110.4 Hz, 2H, m-pyridine), 94.01 (104.3 Hz,

1H, adamantyl or p-pyridine), one resonance not located.

Throughout this paper we describe our computational results by using the

broken-symmetry (BS) approach by Ginsberg⁶⁶and Noodleman.⁶⁷Because

several broken symmetry solutions to the spin-unrestricted KohnÀSham

equations may be obtained, the general notation BS(m,n)⁶⁸has been

adopted, where m(n) denotes the number of spin-up (spin-down) electrons

at the two interacting fragments. Canonical and corresponding⁶⁹orbitals,

as well as spin density plots, were generated with the program Molekel.⁷⁰

Nonrelativistic single-point calculations on the optimized geo-

metry were carried out to predict M€ossbauer spectral parameters

(isomer shifts and quadrupole splittings). These calculations employed

the CP(PPP) basis set for iron.⁷¹The M€ossbauer isomer shifts were

calculated from the computed electron densities at the iron centers as

previously described.^72,73

’ ASSOCIATED CONTENT

S

Supporting Information. Additional variable-temperature

b

NMR data, Eyring plot, computational results, and crystallographic

data for (^iPrPDI)FeN¹Ad, (^iPrPDI)FeN^CyOct, and (^iPrPDI)-

FeN²Ad in CIF format. This material is available free of charge

via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION

Corresponding Author

pchirik@princeton.edu

17367

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

’ ACKNOWLEDGMENT

(28) Bowman, A. C.; Bart, S. C.; Heinemann, F. W.; Meyer, K.;

Chirik, P. J. Inorg. Chem. 2009, 48, 5587.

We thank U.S. National Science Foundation and Deutsche

Forschungsgemeinschaft for a Cooperative Activities in Chem-

istry between U.S. and German Investigators grant. S.D. thanks

the Department of Chemistry and Chemical Biology at Cornell

University for ﬁnancial support. SSRL operations are funded by

the Department of Energy, Oﬃce of Basic Energy Sciences. The

Structural Molecular Biology program is supported by the

National Institutes of Health, National Center for Research

Resources, Biomedical Technology Program and by the Depart-

ment of Energy, Oﬃce of Biological and Environmental Re-

search. We also thank Jon Darmon for preparation of the Table of

Contents graphic.

(29) Cowley, R. E.; Bill, E.; Neese, F.; Brennessel, W. W.; Holland,

P. L. Inorg. Chem. 2009, 48, 4828.

(30) Bowman, A. C. Ph. D. Thesis, Cornell University 2010.

(31) Knijnenburg, Q.; Hetterscheid, D.; Kooistra, T. M.; Budzelaar,

P. H. M. Eur. J. Inorg. Chem. 2004, 1204.

(32) Kahn, O. Molecular Magnetism; VCH: New York, 1993.

(33) Sorai, M.; Seki, S. J. Phys. Chem. Solids 1974, 35, 555.

(34) Long, G. J. M€ossbauer spectroscopy as a structural probe. In

M€ossbauer Spectroscopy; Dickson, D. P. E., Berry, F. J., Eds.; Cambridge

University Press: Cambridge, 1986; p 70.

(35) G€utlich, P. Spin crossover in iron(II)-complexes. In Metal

Complexes; Clarke, M. J., Hemmerich, P., Jørgensen, C. K., Reinen, D.,

Williams, R. J. P., Eds.; Springer-Verlag: Berlin, 1981; Vol. 44, p 83.

(36) G€utlich, P.; Goodwin, H. A. Spin Crossover À An Overall

Perspective. In Spin Crossover in Transition Metal Compounds I; G€utlich,

P., Goodwin, H. A., Eds.; Springer-Verlag: Berlin, 2004; Vol. 233; p 1.

(37) Ginsberg, A. P. J. Am. Chem. Soc. 1980, 102, 111.

(38) Noodleman, L.; Peng, C. Y.; Case, D. A.; Mouesca, J. M. Coord.

Chem. Rev. 1995, 144, 199.

’ REFERENCES

(1) Mehn, M. P.; Peters, J. C. Inorg. Biochem. 2006, 100, 634.

(2) Berry, J. F. Prog. Inorg. Chem. 2009, 30, 28.

(3) Verma, A. K.; Nazif, T. N.; Achim, C.; Lee, S. C. J. Am. Chem. Soc.

2000, 122, 11013.

(39) Kirchner, B.; Wennmohs, F.; Ye, S.; Neese, F. Curr. Opin. Chem.

Biol. 2007, 11, 134.

(4) Ni, C.; Fettinger, J. C.; Long, G. J.; Brynda, M.; Power, P. P.

Chem. Commun. 2008, 6045.

(40) Berry, J. F.; DeBeer George, S.; Neese, F. Phys. Chem. Chem.

Phys. 2008, 10, 4361.

(41) de Visser, S. P. J. Am. Chem. Soc. 2007, 129, 6168.

(42) Sinnecker, S.; Slep, L. D.; Bill, E.; Neese, F. Inorg. Chem. 2005,

44, 2245.

(43) R€omelt, M.; Ye, S.; Neese, F. Inorg. Chem. 2009, 48, 784.

(44) Ye, S. F.; Neese, F. Inorg. Chem. 2010, 49, 772.

(45) Ye, S. F.; Neese, F. J. Am. Chem. Soc. 2010, 132, 3646.

(46) (a) Rong, C.; Lian, S.; Yin, D.; Shen, B.; Zhong, A.; Liu, S.

J. Chem. Phys. 2006, 125, 174102. (b) Rong, C.; Lian, S.; Yin, D.; Zhong,

A.; Zhang, R.; Liu, S. Chem. Phys. Lett. 2007, 434, 149.

(47) Alvarez, S. G.; Alvarez, M. T. Synthesis 1997, 413–414.

(48) Bain, G. A.; Berry, J. F. J. Chem. Ed. 2008, 85, 532.

(49) George, G. N. EXAFSPAK, Stanford Synchrotron Radiation

Laboratory, Stanford Linear Accelerator Center, Stanford University:

Stanford, CA.

(50) Neese, F., Orca - an ab initio, DFT and Semiempirical Electro-

nic Structure Package, Version 2.8, Revision 2287; Institut f€ur Physika-

lische und Theoretische Chemie, Universit€at Bonn, Bonn (Germany),

November 2010.

(51) Becke, A. D. J. Chem. Phys. 1986, 84, 4524.

(52) Perdew, J. P.; Yue, W. Phys. Rev. B 1986, 33, 8800.

(53) Perdew, J. P. Phys. Rev. B 1986, 33, 8822.

(5) Brown, S. D.; Peters, J. C. J. Am. Chem. Soc. 2005, 127, 1913.

(6) Brown, S. D.; Betley, T. A.; Peters, J. C. J. Am. Chem. Soc. 2003,

125, 322.

(7) Thomas, C. M.; Mankad, N. P.; Peters, J. C. J. Am. Chem. Soc.

2006, 128, 4956.

(8) Mankand, N. P.; M€uller, P.; Peters, J. C. J. Am. Chem. Soc. 2010,

132, 4083.

(9) Moret, M.-E.; Peters, J. C. Angew. Chem., Int. Ed. 2011, 50, 2063.

(10) Brown, S. D.; Peters, J. C. J. Am. Chem. Soc. 2004, 126, 4538.

(11) Nieto, I.; Ding, F.; Bontchev, R. P.; Wang, H.; Smith, J. M. J. Am.

Chem. Soc. 2008, 130, 2716.

(12) Eckert, N. A.; Vaddadi, S.; Stoian, S.; Lachicotte, R. J.; Cundari,

T. R.; Holland, P. L. Angew. Chem., Int. Ed. 2006, 45, 6868.

(13) Cowley, R. E.; DeYonker, N. J.; Eckert, N. A.; Cundari, T. R.;

DeBeer, S.; Bill, E.; Ottenwaelder, X.; Flaschenriem, C.; Holland, P. L.

Inorg. Chem. 2010, 49, 6172.

(14) Cowley, R. E.; Holland, P. L. Inorg. Chim. Acta 2011, 369, 40.

(15) Cowley, R. E.; Eckert, N. A.; Vaddadi, S.; Figg, T. M.; Cundari,

T. R.; Holland, P. L. J. Am. Chem. Soc. 2011, 133, 9796.

€

(16) Cowley, R. E.; Elhaik, J.; Eckert, N. A.; Brennessel, W. W.; Bill,

E.; Holland, P. L. J. Am. Chem. Soc. 2008, 130, 6074.

€

(17) Cowley, R. E.; Eckert, N. A.; Elhaik, J.; Holland, P. L. Chem.

(54) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.

Commun. 2009, 1760.

(55) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. Rev. B 1988, 37, 785.

(56) The BP86 functional is a pure DFT functional, which includes

the generalized gradient approximation (GGA). Generally, this func-

tional has been shown to give good results for optimized geometries

at relatively low computational cost. Pure GGA-functionals, however,

suﬀer from diﬃculties in expressing the exchange part of the total energy,

which is particularly problematic for exchange-coupled systems as

presented in this work. Therefore, all electronic structure descriptions

and calculations of spectroscopic parameters presented in this work were

obtained from computations using the B3LYP functional. This hybrid

DFT functional includes a component of exact exchange energy cal-

culated by HartreeÀFock theory. For all geometry optimizations with

BP86 we carefully checked that the electronic structure obtained by this

functional was qualitatively identical to the B3LYP solution.

(57) Sch€afer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97, 2571.

(58) Sch€afer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100,

5829.

(18) King, E. R.; Betley, T. A. Inorg. Chem. 2009, 48, 2361.

(19) King, E. R.; Hennessey, E. T.; Betley, T. A. J. Am. Chem. Soc.

2011, 133, 4917.

(20) Lu, C. C.; George, S. D.; Weyherm€uller, T.; Bill, E.; Bothe, E.;

Wieghardt, K. Angew. Chem., Int. Ed. 2008, 47, 6384.

(21) Lucas, R. L.; Powell, D. R.; Borovik, A. S. J. Am. Chem. Soc. 2005,

127, 11596.

(22) Bart, S. C.; Lobkovsky, E.; Chirik, P. J. J. Am. Chem. Soc. 2004,

126, 13794.

(23) Chirik, P. J.; Wieghardt, K. Science 2010, 327, 794.

(24) Bart, S. C.; Chlopek, K.; Bill, E.; Bouwkamp, M. W.; Lobkovsky, E.;

Neese, F.; Wieghardt, K.; Chirik, P. J. J. Am. Chem. Soc. 2006, 128, 13901.

(25) Knijnenburg, Q.; Gambarotta, S.; Budzelaar, P. H. M. Dalton

Trans. 2006, 5442.

(26) Bart, S. C.; Lobkovsky, E.; Bill, E.; Chirik, P. J. J. Am. Chem. Soc.

2006, 128, 5302–5303.

(27) (a) Fryzuk, M. D.; Love, J. B.; Rettig, S. J.; Young, V. G. Science

1997, 275, 1445. (b) Pool, J. A.; Lobkovsky, E.; Chirik, P. J. Nature 2004,

427, 527. (c) Pool, J. A.; Bernskoetter, W. H.; Chirik, P. J. J. Am. Chem.

Soc. 2004, 126, 14326. (d) Bernskoetter, W. H.; Lobkovsky, E.; Chirik,

P. J. J. Am. Chem. Soc. 2005, 127, 14051.

(59) Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7,

3297.

(60) Eichkorn, K.; Weigend, F.; Treutler, O.; Ahlrichs, R. Theory

Chem. Acc. 1997, 97, 119.

17368

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Journal of the American Chemical Society

ARTICLE

€

(61) Eichkorn, K.; Treutler, O.; Ohm, H.; H€aser, M.; Ahlrichs, R.

Chem. Phys. Lett. 1995, 240, 283.

(62) Eichkorn, K.; Treutler, O.; Ohm, H.; H€aser, M.; Ahlrichs, R.

€

Chem. Phys. Lett. 1995, 242, 652.

(63) Neese, F.; Wennmohs, F.; Hansen, A.; Becker, U. Chem. Phys.

2009, 356, 98.

(64) Kossmann, S.; Neese, F. Chem. Phys. Lett. 2009, 481, 240.

(65) Neese, F. J. Comput. Chem. 2003, 24, 1740.

(66) Ginsberg, A. P. J. Am. Chem. Soc. 1980, 102, 111.

(67) Noodleman, L.; Peng, C. Y.; Case, D. A.; Mouesca, J. M. Coord.

Chem. Rev. 1995, 144, 199.

(68) Kirchner, B.; Wennmohs, F.; Ye, S.; Neese, F. Curr. Opin. Chem.

Biol. 2007, 11, 134.

(69) Neese, F. J. Phys. Chem. Solids 2004, 65, 781.

(70) Molekel, Advanced Interactive 3D-Graphics for Molecular

Sciences, available under http://www.cscs.ch/molkel/.

(71) Neese, F. Inorg. Chim. Acta 2002, 337, 181.

(72) Sinnecker, S.; Slep, L. D.; Bill, E.; Neese, F. Inorg. Chem. 2005,

44, 2245.

(73) R€omelt, M.; Ye, S.; Neese, F. Inorg. Chem. 2009, 48, 784.

17369

dx.doi.org/10.1021/ja205736m |J. Am. Chem. Soc. 2011, 133, 17353–17369

Article Doi

Synthesis and electronic structure determination of N-alkyl-substituted bis(imino)pyridine iron imides exhibiting spin crossover behavior

DOI: 10.1021/ja205736m

Source and publish data:

Authors:

Article abstract of DOI:10.1021/ja205736m

Full text of DOI:10.1021/ja205736m

Products guided by the article

R&D Labs maybe for 1338601-62-9

Relevant to this article

Hot Product