Pentacyanoiron(II) as an electron donor group for nonlinear optics: Medium-responsive properties and comparisons with related pentaammineruthenium(II) complexes

Article abstract of DOI:10.1021/ja063449m

In this article, we describe a series of complex salts in which electron-rich {Fe^II(CN)₅}^3- centers are coordinated to pyridyl ligands with electron-accepting N-methyl/aryl-pyridinium substituents. These compounds have bee

Full text of DOI:10.1021/ja063449m

Published on Web 08/24/2006
Pentacyanoiron(II) as an Electron Donor Group for Nonlinear
Optics: Medium-Responsive Properties and Comparisons
with Related Pentaammineruthenium(II) Complexes
Benjamin J. Coe,*,† Josephine L. Harries, Madeleine Helliwell, Lathe A. Jones,
†
†
†
‡
‡
§
§
Inge Asselberghs, Koen Clays, Bruce S. Brunschwig, James A. Harris,
|
|
Javier Gar ´ı n, and Jes u´ s Orduna
Contribution from the School of Chemistry, UniVersity of Manchester, Oxford Road, Manchester
M13 9PL, U.K., Department of Chemistry, UniVersity of LeuVen, Celestijnenlaan 200D,
B-3001 LeuVen, Belgium, Molecular Materials Research Center, Beckman Institute, MC 139-74,
California Institute of Technology, 1200 East California BouleVard, Pasadena, California 91125,
and Departamento de Qu ´ı mica Org a´ nica, ICMA, UniVersidad de Zaragoza-CSIC,
E-50009 Zaragoza, Spain
Received May 17, 2006; E-mail: b.coe@man.ac.uk.
II
}3- centers
Abstract: In this article, we describe a series of complex salts in which electron-rich {Fe (CN)
5
are coordinated to pyridyl ligands with electron-accepting N-methyl/aryl-pyridinium substituents. These
compounds have been characterized by using various techniques including electronic absorption
spectroscopy and cyclic voltammetry. Molecular quadratic nonlinear optical (NLO) responses have been
determined by using hyper-Rayleigh scattering (HRS) at 1064 nm, and also via Stark (electroabsorption)
spectroscopic studies on the intense, visible d f π* metal-to-ligand charge-transfer (MLCT) bands. The
0
relatively large static first hyperpolarizabilities, â , increase markedly on moving from aqueous to methanol
solutions, accompanied by large red-shifts in the MLCT transitions. Acidification of aqueous solutions allows
reversible switching of the linear and NLO properties, as shown via both HRS and Stark experiments.
Time-dependent density functional theory and finite field calculations using a polarizable continuum model
yield relatively good agreement with the experimental results and confirm the large decrease in â
0
on
II
2+
0 3 5
protonation. The Stark-derived â values are generally larger for related {Ru (NH ) }
complexes than for
II
3-
their {Fe (CN)
5
}
analogues, consistent with the HRS data in water. However, the HRS data in methanol
show that the stronger solvatochromism of the Fe complexes causes their NLO responses to surpass
II
II
those of their Ru counterparts upon changing the solvent medium.
Introduction
and cubic effects, respectively. As a subclass of molecular NLO
compounds, transition metal complexes are especially interesting
The need for new designer materials for potential uses in
sophisticated optical data processing devices has stimulated
much interest in organic compounds that display nonlinear
because they can combine all of the attractive features of organic
2
species with added multifunctionality. Noteworthy examples
illustrating the promise of metal complexes in this field are
1
optical (NLO) behavior. Quadratic (second-order) NLO effects
provided by a number of demonstrations of the use of metal-
are amenable to opto-electronic applications, while cubic (third-
order) phenomena are suited to all-optical technologies. A
diverse range of chromophores are known to display very large
molecular NLO responses, as characterized by first and second
hyperpolarizabilities (â and γ) from which originate quadratic
based redox to switch both quadratic and cubic NLO responses.3
Large values of â are typically displayed by molecules that
incorporate charge asymmetry into an extended π-conjugated
(
2) (a) Kanis, D. R.; Ratner, M. A.; Marks, T. J. Chem. ReV. 1994, 94, 195-
2
42. (b) Long, N. J. Angew. Chem., Int. Ed. Engl. 1995, 34, 21-38. (c)
Whittall, I. R.; McDonagh, A. M.; Humphrey, M. G.; Samoc, M. AdV.
Organomet. Chem. 1998, 42, 291-362. (d) Whittall, I. R.; McDonagh, A.
M.; Humphrey, M. G.; Samoc, M. AdV. Organomet. Chem. 1998, 43, 349-
†
University of Manchester.
University of Leuven.
California Institute of Technology.
Universidad de Zaragoza.
‡
405. (e) Heck, J.; Dabek, S.; Meyer-Friedrichsen, T.; Wong, H. Coord.
§
Chem. ReV. 1999, 190-192, 1217-1254. (f) Gray, G. M.; Lawson, C. M.
In Optoelectronic Properties of Inorganic Compounds; Roundhill, D. M.;
Fackler, J. P., Jr., Eds.; Plenum: New York, U.S.A., 1999; pp 1-27. (g)
Shi, S. In Optoelectronic Properties of Inorganic Compounds; Roundhill,
D. M.; Fackler, J. P., Jr., Eds.; Plenum: New York, U.S.A., 1999; pp 55-
105. (h) Le Bozec, H.; Renouard, T. Eur. J. Inorg. Chem. 2000, 229-239.
(i) Barlow, S.; Marder, S. R. Chem. Commun. 2000, 1555-1562. (j)
Lacroix, P. G. Eur. J. Inorg. Chem. 2001, 339-348. (k) Di Bella, S. Chem.
Soc. ReV. 2001, 30, 355-366. (l) Coe, B. J. In ComprehensiVe Coordination
Chemistry II; McCleverty, J. A., Meyer, T. J., Eds.; Elsevier Pergamon:
Oxford, U.K., 2004; Vol. 9, pp 621-687. (m) Coe, B. J. Acc. Chem. Res.
2006, 39, 383-393.
|
(
1) (a) Zyss, J. Molecular Nonlinear Optics: Materials, Physics and DeVices;
Academic Press: Boston, 1994. (b) Bosshard, Ch., Sutter, K., Pr eˆ tre, Ph.,
Hulliger, J., Fl o¨ rsheimer, M., Kaatz, P., G u¨ nter, P. Organic Nonlinear
Optical Materials (AdVances in Nonlinear Optics, Vol. 1.); Gordon and
Breach: Amsterdam, The Netherlands, 1995. (c) Nonlinear Optics of
Organic Molecules and Polymers; Nalwa, H. S., Miyata, S., Eds.; CRC
Press: Boca Raton, FL, 1997. (d) Marder, S. R. Chem. Commun. 2006,
1
31-134. (e) Nonlinear Optical Properties of Matter: From Molecules
to Condensed Phases; Papadopoulos, M. G., Leszczynski, J., Sadlej, A. J.,
Eds.; Kluwer: Dordrecht, 2006.
12192
9
J. AM. CHEM. SOC. 2006, 128, 12192-12204
10.1021/ja063449m CCC: $33.50 © 2006 American Chemical Society

Pentacyanoiron(II) as an Electron Donor Group
A R T I C L E S
7
5d,8
system, and such chromophores also exhibit intramolecular
charge-transfer transitions. Experimentally measured â values
are therefore enhanced to variable degrees by resonance, and it
is necessary to derive static (off-resonance) first hyperpolariz-
abilities, â0, in order to derive meaningful molecular structure-
activity relationships. In addition, since practical applications
must avoid any actual absorption of light, â0 values are the most
relevant parameter used to characterize new chromophores. The
great majority of metal-containing NLO chromophores contain
electron-rich metal center(s), so their â responses are associated
with metal-to-ligand charge-transfer (MLCT) transitions.
loss,5b outer-sphere complexation, cyclic voltammetric data,
9
formation kinetics, and the effects of inclusion into a cyclo-
dextrin cavity. The outer-sphere complexation properties and
formation kinetics of the closely related complex [Fe (CN)5-
(Mebpe )] (Mebpe ) N-methyl-4-[trans-2-(4-pyridyl)ethe-
nyl]pyridinium) have also been investigated. In the present
study, we have prepared a series of complex salts Na2[Fe -
(CN)5L] (where L is MeQ , Mebpe , or a related pyridyl
pyridinium ligand) and investigated their electronic and optical
properties by using a range of techniques. The results obtained
allow us to make detailed comparisons between the new
1
0
II
+
2-
+
7
,9
II
+
+
II
2+
compounds and related complex salts containing {Ru (NH3)5}
electron-donor groups and also to demonstrate a novel chemical
approach to switching of molecular NLO responses.
Our previous contributions to the field of organotransition
metal NLO compounds have focused primarily on the quadratic
NLO properties of a range of ruthenium ammine complexes of
2m,3a,4
pyridinium-substituted ligands.
Recently, we have also
Experimental Section
been extending our scope to include other types of complexes,
and one promising, readily accessible, and suitably electron-
II
Materials and Procedures. The compound Na
obtained from Aldrich and purified by recrystallization before use. The
3
[Fe (CN)
5
(NH
3
)] was
II
3-
rich metal center is {Fe (CN)5} . Many previous reports have
+
pro-ligand salts N-methyl-4,4′-bipyridinium chloride ([MeQ ]Cl‚
described the syntheses and properties of complexes containing
+
11
5
0.7H
2
O),4c N-phenyl-4,4′-bipyridinium chloride ([PhQ ]Cl‚2H
2
O),
this moiety coordinated to a wide variety of ligands, including
+
O),11
N-(4-acetylphenyl)-4,4′-bipyridinium chloride ([4-AcPhQ ]Cl‚2H
N-(2-pyrimidyl)-4,4′-bipyridinium chloride ([2-PymQ ]Cl), and N-meth-
yl-4-[trans-2-(4-pyridyl)ethenyl]pyridinium iodide ([Mebpe ]I) and
the compound 4-methyl-N-phenylpyridinium chloride [Phpic ]Cl were
synthesized according to published methods. We have previously
reported preparations of the intermediate compounds N-(2,4-dinitro-
phenyl)-4,4′-bipyridinium chloride ([2,4-DNPhQ
4-[trans-2-(4-pyridyl)ethenyl]pyridinium chloride ([Phbpe ]Cl‚2.25H
2
many discussions of MLCT spectroscopic properties, but no
NLO studies with such complexes have hitherto been disclosed.
Of particular relevance to our investigations, the complex
+
12
+
13
+
12
II
+
2-
+
[
Fe (CN)5(MeQ )] (MeQ ) N-methyl-4,4′-bipyridinium)
5a
was first reported by Toma and Malin, but only very limited
characterization data were presented at the time (i.e., UV-vis
absorption data). Subsequent studies with this particular complex
have included 13C NMR spectroscopy, photochemical ligand
+
]Cl)11 and N-phenyl-
+
12
2
O),
6
but considerably improved methods are included here. The complex
II
+
II
+
II
salts [Ru (NH
(
3
)
5
(MeQ )]Cl
3
, [Ru (NH
3 5
)
(PhQ )]Cl
3
, and [Ru (NH
3 5
) -
+
3a
4-AcPhQ )]Cl
3
were prepared as previously described, and the
(
3) (a) Coe, B. J.; Houbrechts, S.; Asselberghs, I.; Persoons, A. Angew. Chem.,
Int. Ed. 1999, 38, 366-369. (b) Weyland, T.; Ledoux, I.; Brasselet, S.;
Zyss, J.; Lapinte, C. Organometallics 2000, 19, 5235-5237. (c) Malaun,
M.; Reeves, Z. R.; Paul, R. L.; Jeffery, J. C.; McCleverty, J. A.; Ward, M.
D.; Asselberghs, I.; Clays, K.; Persoons, A. Chem. Commun. 2001, 49-
+
+
+
analogous compounds containing 2-PymQ , Mebpe , or Phbpe ligands
were prepared from their hexafluorophosphate counterparts by pre-
cipitation from acetone/LiCl. All other reagents were obtained com-
mercially and used as supplied. All solvents were degassed by argon
purging for 10 min prior to use, and all reactions were conducted under
an argon atmosphere in the dark. Products were dried at room
5
3
0. (d) Malaun, M.; et al., J. Chem. Soc., Dalton Trans. 2001, 3025-
038. (e) Cifuentes, M. P.; Powell, C. E.; Humphrey, M. G.; Heath, G. A.;
Samoc, M.; Luther-Davies, B. J. Phys. Chem. A 2001, 105, 9625-9627.
(f) Paul, F.; Costuas, K.; Ledoux, I.; Deveau, S.; Zyss, J.; Halet, J.-F.;
Lapinte, C. Organometallics 2002, 21, 5229-5235. (g) Powell, C. E.;
Cifuentes, M. P.; Morrall, J. P.; Stranger, R.; Humphrey, M. G.; Samoc,
M.; Luther-Davies, B.; Heath, G. A. J. Am. Chem. Soc. 2003, 125, 602-
4
temperature in a vacuum desiccator (CaSO ) for ca. 24 h prior to
characterization.
610. (h) Asselberghs, I.; Clays, K.; Persoons, A.; McDonagh, A. M.; Ward,
1
General Physical Measurements. H NMR spectra were recorded
on a Varian Gemini 200 spectrometer, and all shifts are referenced to
TMS. The fine splitting of pyridyl or phenyl ring AA′BB′ patterns is
ignored, and the signals are reported as simple doublets, with J values
referring to the two most intense peaks. Elemental and thermogravi-
metric analyses (TGA) were performed by the Microanalytical Labora-
M. D.; McCleverty, J. A. Chem. Phys. Lett. 2003, 368, 408-411. (i) Powell,
C. E.; Humphrey, M. G.; Cifuentes, M. P.; Morrall, J. P.; Samoc, M.;
Luther-Davies, B. J. Phys. Chem. A 2003, 107, 11264-11266. (j) Sporer,
C. et al., Angew. Chem., Int. Ed. 2004, 43, 5266-5268.
4) Selected recent examples: (a) Coe, B. J.; Jones, L. A.; Harris, J. A.;
Sanderson, E. E.; Brunschwig, B. S.; Asselberghs, I.; Clays, K.; Persoons,
A. Dalton Trans. 2003, 2335-2341. (b) Coe, B. J.; Jones, L. A.; Harris, J.
A.; Brunschwig, B. S.; Asselberghs, I.; Clays, K.; Persoons, A. J. Am. Chem.
Soc. 2003, 125, 862-863. (c) Coe, B. J.; Jones, L. A.; Harris, J. A.;
Brunschwig, B. S.; Asselberghs, I.; Clays, K.; Persoons, A.; Gar ´ı n, J.;
Orduna, J. J. Am. Chem. Soc. 2004, 126, 3880-3891. (d) Coe, B. J.; Harris,
J. A.; Jones, L. A.; Brunschwig, B. S.; Song, K.; Clays, K.; Gar ´ı n, J.;
Orduna, J.; Coles, S. J.; Hursthouse, M. B. J. Am. Chem. Soc. 2005, 127,
(
2
tory, University of Manchester (TGA under N , heating to 250 °C). IR
spectra were obtained as KBr disks with an ATI Mattson Genesis Series
FTIR instrument, and UV-vis spectra were recorded by using a
Hewlett-Packard 8452A diode array spectrophotometer, except for those
presented in Table 2, which were measured using an R-Helios double-
beam spectrophotometer that has a range further into the NIR region.
Cyclic voltammetric measurements were carried out with an EG&G
PAR model 283 potentiostat/galvanostat. An EG&G PAR K0264 single-
compartment microcell was used with a silver/silver chloride reference
4
845-4859. (e) Coe, B. J.; Harries, J. L.; Helliwell, M.; Brunschwig, B.
S.; Harris, J. A.; Asselberghs, I.; Hung, S.-T.; Clays, K.; Horton, P. N.;
Hursthouse, M. B. Inorg. Chem. 2006, 45, 1215-1227.
(
5) Selected examples: (a) Toma, H. E.; Malin, J. M. Inorg. Chem. 1973, 12,
1
039-1045. (b) Figard, J. E.; Petersen, J. D. Inorg. Chem. 1978, 17, 1059-
1
063. (c) Hrepic, N. V.; Malin, J. M. Inorg. Chem. 1979, 18, 409-413.
(d) Moore, K. J.; Lee, L.-S.; Mabbott, G. A.; Petersen, J. D. Inorg. Chem.
1
1
1
1
2
983, 22, 1108-1112. (e) Johnson, C. R.; Shepherd, R. E. Inorg. Chem.
983, 22, 1117-1123. (f) Johnson, C. R.; Shepherd, R. E. Inorg. Chem.
983, 22, 2439-2444. (g) Macartney, D. H. ReV. Inorg. Chem. 1988, 9,
01-151. (h) Benedix, R.; Hennig, H. Z. Anorg. Allg. Chem. 1989, 577,
3-38. (i) Estrin, D. A.; Hamra, O. Y.; Paglieri, L.; Slep, L. D.; Olabe, J.
(7) Toma, H. E. Can. J. Chem. 1979, 57, 2079-2084.
(8) Slep, L. D.; Baraldo, L. M.; Olabe, J. A. Inorg. Chem. 1996, 35, 6327-
6333.
(9) Macartney, D. H.; Warrack, L. J. Can. J. Chem. 1989, 67, 1774-1779.
(10) Ando, I.; Ujimoto, K.; Kurihara, H. Bull. Chem. Soc. Jpn. 2001, 74, 717-
721.
(11) Coe, B. J.; Harris, J. A.; Harrington, L. J.; Jeffery, J. C.; Rees, L. H.;
Houbrechts, S.; Persoons, A. Inorg. Chem. 1998, 37, 3391-3399.
(12) Coe, B. J.; Harris, J. A.; Asselberghs, I.; Persoons, A.; Jeffery, J. C.; Rees,
L. H.; Gelbrich, T.; Hursthouse, M. B. J. Chem. Soc., Dalton Trans. 1999,
3617-3625.
(13) Bergman, E. D.; Crane, F. E., Jr.; Fuoss, R. M. J. Am. Chem. Soc. 1952,
74, 5979-5982.
A. Inorg. Chem. 1996, 35, 6832-6837. (j) Waldh o¨ r, E.; Kaim, W.; Olabe,
J. A.; Slep, L. D.; Fiedler, J. Inorg. Chem. 1997, 36, 2969-2974. (k) Slep,
L. D.; Pollak, S.; Olabe, J. A. Inorg. Chem. 1999, 38, 4369-4371. (l) Parise,
A. R.; Piro, O. E.; Castellano, E. E.; Olabe, J. A. Inorg. Chim. Acta 2001,
3
19, 199-202. (m) Baraldo, L. M.; Forlano, P.; Parise, A. R.; Slep, L. D.;
Olabe, J. A. Coord. Chem. ReV. 2001, 219-221, 881-921. (n) Albor e´ s,
P.; Slep, L. D.; Weyherm u¨ ller, T.; Baraldo, L. M. Inorg. Chem. 2004, 43,
6
762-6773.
(
6) Figard, J. E.; Paukstelis, J. V.; Byrne, E. F.; Petersen, J. D. J. Am. Chem.
Soc. 1977, 99, 8417-8425.
J. AM. CHEM. SOC.
9
VOL. 128, NO. 37, 2006 12193

A R T I C L E S
Coe et al.
electrode separated by a salt bridge from a glassy carbon disk working
C
C
5
H
H
4
N), 8.23 (2 H, d, J ) 8.3 Hz, C
6
H
4
), 7.95 (2 H, d, J ) 8.2 Hz,
electrode and Pt wire auxiliary electrode. Water and methanol were
6
4
), 7.61 (2 H, d, J ) 6.4 Hz, C
H
5 4
N), 2.72 (3 H, s, Me). ν(CtN)
n
-1
distilled before use. KNO
3
and [NBu
4 6
]PF , the latter twice recrystallized
2110w, 2084m, 2048s cm . Anal. Calcd (%) for C23
H
15
N
7
FeNa O‚
2
from ethanol and dried in vacuo, were used as supporting electrolytes.
Solutions containing ca. 10 M analyte (0.1 M electrolyte) were
2
8H O: C, 42.41; H, 4.80; N, 15.05. Found: C, 42.16; H, 4.15; N, 15.56.
TGA analysis indicates the loss of ca. 7.3 molecules of water upon
heating.
-
3
deaerated by purging with N
2
. All E1/2 values were calculated from
1
(
E
pa + Epc)/2 at a scan rate of 200 mV s-
.
II
+
Synthesis of Na
2
[Fe (CN)
5
(2-PymQ )], 4. This compound was
II
II
+
Purification of Na
NH
3
[Fe (CN)
5
(NH
3
)]. Commercial Na
3
[Fe (CN)
5
-
prepared and purified in manner similar to 1 by using [2-PymQ ]Cl
(135 mg, 0.499 mmol) in place of [MeQ ]Cl‚0.7H O to afford a dark
+
(
3
)] (5.00 g) was heated to reflux in aqueous ammonia solution (100
2
mL, 70%). After 10 min under reflux, the solution was filtered while
hot to remove a dark impurity. After the solution was cooled to room
temperature, ethanol (100 mL) was added slowly with stirring. The
precipitate was filtered off, washed with ethanol then diethyl ether,
and dried to afford a light yellow solid (4.70 g).
blue/green solid: 120 mg, 39%; δ (D O) 10.01 (2 H, d, J ) 6.6 Hz,
H
2
C H N), 9.24 (2 H, d, J ) 6.0 Hz, C H N), 9.06 (2 H, d, J ) 4.8 Hz,
5
4
5
4
C H N ), 8.57 (2 H, d, J ) 6.9 Hz, C H N), 7.90 (1 H, t, J ) 11.2 Hz,
4
3
2
5
4
C H N ), 7.66 (2 H, d, J ) 6.4 Hz, C H N). ν(CtN) 2119w, 2086m,
4
3
2
5
4
-
1
2050s cm . Anal. Calcd (%) for C H N FeNa ‚8.5H O: C, 36.79;
1
9
11
9
2
2
Improved Synthesis of N-(2,4-Dinitrophenyl)-4,4′-bipyridinium
Chloride, [2,4-DNPhQ ]Cl. A solution of 2,4-dinitrochlorobenzene
H, 4.55; N, 20.32. Found: C, 36.38; H, 3.73; N, 20.95. TGA analysis
indicates the loss of ca. 7.0 molecules of water upon heating.
+
(
6.49 g, 32 mmol) and 4,4′-bipyridyl (5.00 g, 32 mmol) in ethanol (50
II
+
Synthesis of Na
2
[Fe (CN)
5
(Mebpe )], 5. This compound was
mL) was heated under reflux for 23 h. The solution was cooled to room
temperature and added to diethyl ether (500 mL), with stirring. The
golden-brown precipitate was filtered off and washed with diethyl
+
prepared and purified in manner similar to 1 by using [Mebpe ]I (162
mg, 0.500 mmol) in place of [MeQ ]Cl‚0.7H
+
2
O to afford a dark purple-
blue solid: 82 mg, 29%; δ
H
(D
N), 7.73 (2 H, d, J ) 6.9 Hz, C
.36-7.17 (2 H, m, 2CH), 7.01 (2 H, d, J ) 6.9 Hz, C
2
O) 8.86 (2 H, d, J ) 6.6 Hz, C
5
H
H
4
N),
N),
3
ether: 9.46 g, 82%; δ
H, d, J ) 6.9 Hz, C
.82 (2 H, d, J ) 6.3 Hz, C
Ph), 8.25 (1 H, d, J ) 8.6 Hz, H ), 8.01 (2 H, d, J ) 6.3 Hz, C
H
(D
2
O) 9.38 (1 H, d, J ) 2.4 Hz, H ), 9.23 (2
8
7
5 4
.48 (2 H, d, J ) 6.6 Hz, C H
5
4
5
5
H
4
N-Ph), 8.92 (1 H, dd, J ) 8.7, 2.5 Hz, H ),
N), 8.68 (2 H, d, J ) 6.8 Hz, C N-
N).
5 4
H
N), 4.24 (3
8
5
H
4
H
5 4
-1
H, s, Me). ν(CtN) 2086w, 2044s cm . Anal. Calcd (%) for C18
FeNa ‚7.5H O: C, 38.31; H, 5.00; N, 17.38. Found: C, 37.82; H, 4.37;
N, 17.03. TGA analysis indicates the loss of ca. 7.6 molecules of water
upon heating.
13 7
H N -
6
5
H
4
2
2
Note: The product is hygroscopic and prolonged contact with air causes
formation of a sticky brown solid.
Improved Synthesis of N-Phenyl-4-[E-2-(4-pyridyl)ethenyl]pyri-
dinium Chloride, [Phbpe ]Cl. To a solution of [Phpic ]Cl (100 mg,
.48 mmol) and pyridine-4-carboxaldehyde (0.05 mL, 0.45 mmol) in
ethanol (ca. 0.5 mL) was added 1 drop of pyridine. The solution was
heated at 95 °C for 3.5 h. The resulting dark brown solution was cooled
to room temperature, and the product precipitated by the dropwise
addition of diethyl ether. A dark brown solid was filtered off and washed
II
+
Synthesis of Na
2
[Fe (CN)
5
(Phbpe )], 6. This compound was
+
+
+
prepared and purified in manner similar to 1 by using [Phbpe ]Cl‚
0
+
2
.25H
a dark blue solid: 161 mg, 66%; δ
2 H, d, J ) 6.6 Hz, C N), 7.71-7.57 (6 H, m, Ph and CH), 7.36 (1
2
O (146 mg, 0.435 mmol) in place of [MeQ ]Cl‚0.7H
2
O to afford
H
(D
2
O) 8.93 (4 H, m, C
5 4
H N), 8.10
(
5 4
H
H, d, J ) 16.5 Hz, CH), 7.18 (2 H, d, J ) 6.6 Hz, C
092w, 2058s cm . Anal. Calcd (%) for C23
2.28; H, 5.09; N, 15.01. Found: C, 41.96; H, 4.44; N, 15.88. TGA
analysis indicates the loss of ca. 8.1 molecules of water upon heating.
H
5 4
N). ν(CtN)
-1
2
4
H
15
N
7
FeNa ‚9H O: C,
2
2
with diethyl ether: 119 mg, 84%; δ
H
(D
N), 8.21 (2 H, d, J ) 6.9 Hz,
N), 7.75 (1 H, d, J ) 16.2 Hz, CH), 7.69 (5 H, s, Ph), 7.63 (2 H,
N), 7.56 (1 H, d, J ) 16.4 Hz, CH). Anal. Calcd
‚2.75H O: C, 62.79; H, 6.00; N, 8.14. Found: C,
3.17; H, 5.74; N, 7.95.
2
O) 8.94 (2 H, d, J ) 6.7 Hz,
C
C
5
4 5 4
H N), 8.51 (2 H, d, J ) 6.0 Hz, C H
5
H
4
II
+
X-ray Crystallography. Crystals of the salt Na
O (1‚9H O) were obtained by diffusion of ethanol vapor into an
aqueous solution at 4 °C. A purple block crystal of approximate
2 5
[Fe (CN) (MeQ )]‚
d, J ) 6.1 Hz, C
5 4
H
9
H
2
2
(
6
%) for C18
H
15ClN
2
2
3
dimensions 0.60 × 0.60 × 0.10 mm was attached to a Hamilton
II
+
II
Synthesis of Na
NH
2
[Fe (CN)
5
(MeQ )], 1. A solution of Na
3
[Fe (CN)
5
-
Cryoloop, using fomblin oil (perfluoropolymethylisopropyl ether) and
mounted on a Bruker APEX CCD X-ray diffractometer. Cryocooling
to 100 K was carried out by using an Oxford Cryosystems 700 Series
cryostream cooler. Intensity measurements were collected by using
graphite-monochromated, Mo KR radiation from a sealed X-ray tube
with a monocapillary collimator. The intensities of reflections of a
sphere were collected with an exposure time per frame of 5 s. Data
+
(
3
)] (136 mg, 0.500 mmol) and [MeQ ]Cl‚0.7H
2
O (103 mg, 0.470
mmol) in water (5 mL) was stirred at room temperature for 6 h in the
dark. Ethanol (30 mL) was added, and the mixture stored at 4 °C in a
refrigerator overnight. The purple-blue solid was filtered off and washed
with ethanol. The product was purified by three reprecipitations from
water/ethanol to afford a dark purple-blue solid: 164 mg, 72%; δ
O) 9.15 (2 H, d, J ) 6.7 Hz, C N), 8.80 (2 H, d, J ) 6.7 Hz,
N), 7.48 (2 H, d, J ) 6.9 Hz,
H
(D
2
H
5 4
14
processing was carried out by using the Bruker SAINT software
package, and a semiempirical absorption correction was applied using
SADABS. The structure was solved by direct methods and refined
by full-matrix least-squares on all F
SHELXL-97. All non-hydrogen atoms were refined anisotropically,
with hydrogen atoms bonded to carbon included in calculated positions
using the riding method; those bonded to the water oxygen atoms were
found by difference Fourier methods and refined isotropically. All other
calculations were carried out by using the SHELXTL package.
Crystallographic data and refinement details are presented in Table S1.
C
C
5
H
H
4
N), 8.17 (2 H, d, J ) 6.7 Hz, C
N), 4.41 (3H, s, Me). ν(CtN) 2080w, 2048m, 2042s cm . Anal.
FeNa ‚4.5H O: C, 39.69; H, 4.16; N, 20.25.
5 4
H
-
1
5
4
14
Calcd (%) for C16
H
11
N
7
2
2
2
15
0
data using SHELXS-97 and
Found: C, 39.87; H, 3.70; N, 20.14. TGA analysis indicates the loss
of ca. 4.1 molecules of water upon heating.
16
II
+
Synthesis of Na
2
[Fe (CN)
5
(PhQ )], 2. This compound was prepared
+
and purified in manner similar to 1 by using [PhQ ]Cl‚2H
2
O (134 mg,
+
0
.440 mmol) in place of [MeQ ]Cl‚0.7H
2
O to afford a dark blue
N), 9.14
N), 7.73
N). ν(CtN) 2100w,
FeNa ‚7H O: C,
2.66; H, 4.60; N, 16.58. Found: C, 42.28; H, 4.18; N, 16.92. TGA
17
solid: 170 mg, 65%; δ
2 H, d, J ) 6.9 Hz, C
H
(D
N), 8.48 (2 H, d, J ) 6.6 Hz, C
5 H, m, Ph), 7.62 (2 H, d, J ) 6.3 Hz, C
2
5 4
O) 9.18 (2 H, d, J ) 6.6 Hz, C H
(
(
2
4
5
H
4
5 4
H
Hyper-Rayleigh Scattering. Details of the hyper-Rayleigh scattering
HRS) experiment have been discussed elsewhere, and the experi-
mental procedure used was as previously described. â values were
determined by using the electric-field-induced second harmonic genera-
5
H
4
18
(
-
1
080m, 2048s cm . Anal. Calcd (%) for C21
H N
13 7
2
2
19
analysis indicates the loss of ca. 6.2 molecules of water upon heating.
II
+
Synthesis of Na
2
[Fe (CN)
5
(4-AcPhQ )], 3. This compound was
(
14) SAINT (Version 6.45) and SADABS (Version 2.10); Bruker AXS, Inc.;
+
prepared and purified in manner similar to 1 by using [4-AcPhQ ]Cl‚
Madison, WI, 2003.
+
(15) Sheldrick, G. M. Acta Crystallogr., Sect. A 1990, 46, 467-473.
2
H
2
O (155 mg, 0.447 mmol) in place of [MeQ ]Cl‚0.7H
2
O to afford
(
16) Sheldrick, G. M. SHELXL 97, Program for crystal structure refinement;
a dark blue solid: 200 mg, 69%; δ
H
(D
2
O) 9.25 (2 H, d, J ) 6.3 Hz,
N), 8.50 (2 H, d, J ) 5.7 Hz,
University of G o¨ ttingen: G o¨ ttingen, Germany, 1997.
C
5
4
H N), 9.16 (2 H, d, J ) 5.9 Hz, C
H
5 4
(17) SHELXTL (Version 6.10); Bruker AXS, Inc.; Madison, WI, 2000.
1
2194 J. AM. CHEM. SOC. VOL. 128, NO. 37, 2006
9

Pentacyanoiron(II) as an Electron Donor Group
A R T I C L E S
tion â1064 for p-nitroaniline (pNA, 32.0 × 10 esu in methanol)20 as
an external reference. In the absence of a suitable reference for aqueous
solutions, the â1064 values in water were obtained by using the â1064
-30
element, Hab, for the diabatic states are given by
2
1/2
∆µ12
2
1
2
2
0
cb
)
1 -
(3)
(4)
for pNA in acetonitrile as an external reference. The absolute
magnitudes of these data are hence of questionable accuracy and should
be compared only cautiously with the data obtained in methanol. The
values for methanol solutions were determined by using the â1064 for
pNA in methanol. All measurements were performed by using the
064 nm fundamental of an injection-seeded, Q-switched Nd:YAG laser
Quanta-Ray GCR-5, 8 ns pulses, 7 mJ, 10 Hz). Dilute methanol or
2
2
[
(∆µ
)
]
1
2
+ 4µ12
E
(µ )
max 12
|
H | ) |
|
ab
2
0
∆µab
1
(
0
If the hyperpolarizability tensor, â , has only nonzero elements along
the MLCT direction, then this quantity is given by
-
5
-6
aqueous solutions (10 -10 M) were used to ensure a linear
2
dependence of I2ω/I
ω
on solute concentration, precluding the need for
2
Lambert-Beer correction factors. Samples were filtered (Millipore, 0.45
µm), and none showed any fluorescence. One-dimensional hyperpo-
larizability is assumed, i.e., â1064 ) âzzz, and a relative error of (15%
is estimated.
3∆µ (µ )
12
12
2
â0 )
(5)
(Emax)
A relative error of (20% is estimated for the â
0
values derived from
For the â switching studies, due to negligible HRS signals from the
protonated samples when using a fundamental wavelength of 1064 nm,
measurements were instead carried out by using a 800 nm Ti3+:sapphire
laser. Samples of 1-4 (5 mg) were dissolved in water (20 mL), and
portions (8 mL) of these solutions were acidified by the addition of
concentrated HCl (2 mL). Reversibility of the switching effect was
demonstrated with salt 4 as follows. Four samples (numbered 1-4) of
the stock solution (each 3 mL) were placed in 10 mL capacity
volumetric flasks. Concentrated HCl (1 drop) was added to three of
these samples (2-4), and then samples 3 and 4 were neutralized by
the addition of triethylamine until their blue colors were completely
restored. Concentrated HCl (1 drop) was then added to sample 4, and
all four samples were diluted up to 10 mL so that each contained the
same concentration of the complex species.
the Stark data and using eq 5, while experimental errors of (10% are
estimated for µ12, ∆µ12, and ∆µab, (15% for Hab, and (50% for c .
b
2
For the experiments with acidified samples, slides spin-coated with
R-cyanoacrylate were used initially because the acid etches the ITO
coating off the slides in about 30 min. However, it was subsequently
found that rapid freezing of the sample obviates the need for this
protective measure. Aliquots (50 µL) of concentrated HCl were added
to the blue (500 µL) sample solutions to give orange solutions. Upon
further acid addition, precipitates formed, likely the diprotonated neutral
II
+
3 2
complexes Fe (CN) (CNH) L (L ) MeQ , etc.). Some of the samples
precipitated immediately after addition of the first aliquot of HCl,
precluding measurements, and all samples precipitated out over the
course of a few hours. More dilute solutions, such as those originally
used to test the effects of acidification via UV-vis spectroscopy or
for the HRS experiments, do not show any precipitation.
Stark Spectroscopy. The Stark apparatus, experimental methods,
2
1,22
and data analysis procedure were as previously reported,
with the
Computational Procedures. All calculations were performed by
only modification being that a Xe arc lamp was used as the light source
24
using the Gaussian 03 program. The molecular geometries were
optimized by using the hybrid functional B3P8625 and the 6-31G* basis
set. The B3P86/6-311+G* model chemistry was used for properties
calculations. Electronic transitions were calculated by means of the
time-dependent density functional theory (TD-DFT) method, and the
excited-state dipole moments were calculated either by using the one-
particle RhoCI density or the finite-field method which involved the
evaluation of the changes in excitation energies in the presence of
external electric fields of 0.001 atomic units. The default Gaussian 03
parameters were used in every case. For the calculations on the doubly
in the place of a W filament bulb. Glycerol-water (50:50 vol %) was
used as the glassing medium, for which the local field correction, fint
,
2
2
is estimated as 1.33. The Stark spectrum for each compound was
measured a minimum of three times using different field strengths, and
the signal was always found to be quadratic in the applied field. A
two-state analysis of the MLCT transitions gives
2
2
2
∆
µab ) ∆µ12 + 4µ12
(1)
where ∆µab is the dipole moment difference between the diabatic states
corresponding with the effective (localized) electron-transfer process,
∆µ12 is the observed (adiabatic) dipole moment difference corresponding
with a delocalized electron-transfer, and µ12 is the transition dipole
moment. Analysis of the Stark spectra in terms of the Liptay treatment23
affords ∆µ12, and µ12 can be determined from the oscillator strength,
protonated complex denoted 1H
cyanide ligand and one of the cis cyanides. The X-ray crystal structure
of 1‚9H O indicates that there is greater π-back-bonding to the trans
2
, protons were placed on the trans
2
cyanide ligand (see below), suggesting that this ligand will be more
basic and hence the site of the initial protonation. Molecular orbital
contours were plotted by using Molekel 4.3.26
fos, of the transition by
Results and Discussion
-
5
1/2
|
µ | ) [f /(1.08 × 10 Emax)]
(2)
12
os
Synthetic Studies. We have previously reported syntheses
of the pro-ligand salts, [2,4-DNPhQ ]Cl and [Phbpe ]Cl,
+
+
11,12
where Emax is the energy of the MLCT maximum (in wavenumbers).
The degree of delocalization, c , and electronic coupling matrix
b
2
but include here modified and improved methods which afford
these compounds in yields two to four times higher than those
already published. The new complex salts 1-6 (Figure 1) were
prepared simply by substitution of the ammine ligand in the
(
18) (a) Clays, K.; Persoons, A. Phys. ReV. Lett. 1991, 66, 2980-2983. (b) Clays,
K.; Persoons, A. ReV. Sci. Instrum. 1992, 63, 3285-3289. (c) Hendrickx,
E.; Clays, K.; Persoons, A.; Dehu, C.; Br e´ das, J.-L. J. Am. Chem. Soc.
1
995, 117, 3547-3555. (d) Hendrickx, E.; Clays, K.; Persoons, A. Acc.
II
precursor Na3[Fe (CN)5(NH3)]. The identity and purity of the
Chem. Res. 1998, 31, 675-683.
1
(
(
(
(
(
19) Houbrechts, S.; Clays, K.; Persoons, A.; Pikramenou, Z.; Lehn, J.-M. Chem.
products are established by diagnostic H NMR spectra and
CHN elemental analyses. The latter all show the presence of
Phys. Lett. 1996, 258, 485-489.
20) St a¨ helin, M.; Burland, D. M.; Rice, J. E. Chem. Phys. Lett. 1992, 191,
2
45-250.
21) Coe, B. J.; Harris, J. A.; Brunschwig, B. S. J. Phys. Chem. A 2002, 106,
97-905.
22) Shin, Y.-g. K.; Brunschwig, B. S.; Creutz, C.; Sutin, N. J. Phys. Chem.
(24) Frisch, M. J.; et al. Gaussian 03, revision B.05; Gaussian, Inc.: Pittsburgh,
PA, 2003.
8
(25) The B3P86 Functional consists of Becke’s three-parameter hybrid functional
(Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652) with the nonlocal
correlation provided by the Perdew 86 expression: Perdew, J. P. Phys.
ReV. B 1986, 33, 8822-8824.
1
996, 100, 8157-8169.
23) (a) Liptay, W. In Excited States; Lim, E. C., Ed.; Academic Press: New
York, 1974; Vol. 1, pp 129-229. (b) Bublitz, G. U.; Boxer, S. G. Annu.
ReV. Phys. Chem. 1997, 48, 213-242.
(26) Portmann, S.; L u¨ thi, H. P. Chimia 2000, 54, 766-770.
J. AM. CHEM. SOC.
9
VOL. 128, NO. 37, 2006 12195

A R T I C L E S
Coe et al.
Within the 4,4′-bipyridinium series 1-4, the MLCT Emax
values in either solvent decrease as R changes in the order Me
>
Ph > 4-AcPh > 2-Pym. The magnitude of the total decrease
along this series is ca. 0.35 eV in both solvents. This trend
reflects the steadily increasing electron deficiency of the
pyridinium unit, as observed previously in complexes of these
II
12
II
+
ligands with Ru ammine or trans-{Ru Cl(pdma)2} [pdma
1,2-phenylenebis(dimethylarsine)]27 electron donor centers.
)
The addition of an (E)-CHdCH unit between the two pyridyl
rings in moving from 1 to 5 or from 2 to 6 causes insignificant
or very small changes in Emax, again as observed previously in
1
2,28
related compounds.
The potential for pronounced interactions between the cyanide
ligands and the solvent medium renders 1-6 very strongly
solvatochromic, with large red-sifts of ca. 0.50 eV in the MLCT
bands on moving from water to methanol (Tables 1 and 2).
These shifts are also generally accompanied by significant
increases in intensity. Decreases in the MLCT Emax indicate
Figure 1. Chemical structures of the FeII complex salts investigated.
large amounts of water of crystallization, as found in previous
investigations with related compounds,5a,5n,6 and also evidenced
by TGA studies.
II
increasing electron density at the Fe center, and the difference
Electronic Spectroscopy Studies. The UV-vis absorption
spectra of complex salts 1-6 have been measured in both water
and methanol, and the results are shown in Tables 1 and 2. These
spectra feature intense UV absorptions due to π f π*
intraligand transitions, together with intense, broad d(Fe ) f
π*(L) (L ) pyridyl ligand) visible MLCT bands.
between the two solvents may be attributed at least in part to
the greater H-bond donating ability of water acting to reduce
the basicity and enhance the π-accepting ability of the cyanide
ligands. Another factor that may be important in this context is
II
29
the Gutmann electron acceptor number of the solvent. It is
III/II
known that the MLCT Emax (and the Fe
potential, see below)
Table 1. UV-Vis Absorption and Electrochemical Data for 1-6 in Water
E, V vs. Ag/AgCl (
[FeIII/II]
∆E
p
, mV)b
λ
max, nm
E
max
(eV)a
, M cm-1)a
-
1
assignment
E
E
1/2
or Epcc
complex salt (L)
(ꢀ
1/
2
1
(MeQ+)
534 (3700)
66 (16 300)
566 (3900)
82 (18 700)
584 (4900)
84 (32 300)
618 (6400)
90 (32 800)
538 (5000)
2.32
4.66
2.19
4.40
2.12
4.37
2.01
4.28
2.30
3.90
5.34
2.21
3.71
5.34
d f π*
π f π*
d f π*
π f π*
d f π*
π f π*
d f π*
π f π*
d f π*
π f π*
π f π*
d f π*
π f π*
π f π*
0.48 (190)
0.48 (140)
0.46 (160)
0.46 (130)
0.45 (130)
-0.90 (320)
-1.42
2
2
3
4
5
(PhQ+)
-0.86 (250)
-0.74
2
(4-AcPhQ+)
(2-PymQ+)
(Mebpe+)
-0.78
2
-0.61
-0.94
2
3
2
18 (26 800)
32 (16 300)
6
(Phbpe+)
562 (6600)
0.45 (140)
-0.82
3
2
34 (37 600)
32 (21 100)
a
Solutions ca. 3-8 × 10 M. b Solutions ca. 10-3 M in 1 M aqueous KNO3 at a glassy carbon disk working electrode with a scan rate of 200 mV s-1
-5
.
c
For an irreversible reduction process.
Table 2. UV-Vis Absorption and Electrochemical Data for 1-6 in Methanol
E, V vs Ag/AgCl (
[FeIII/II]
∆E
p
, mV)b
λ
max, nm
E
max
(eV)a
, M cm-1)a
-
1
assignment
E
E
1/2
or Epcc
complex salt (L)
(ꢀ
1/
2
1
(MeQ+)
686 (5800)
70 (17 900)
738 (6600)
90 (21 800)
771 (6300)
91 (26 000)
854 (7300)
93 (24 900)
685 (5800)
1.81
4.59
1.68
4.28
1.61
4.26
1.45
4.23
1.81
3.89
5.44
1.67
3.69
5.37
d f π*
π f π*
d f π*
π f π*
d f π*
π f π*
d f π*
π f π*
d f π*
π f π*
π f π*
d f π*
π f π*
π f π*
0.28 (130)
0.29 (140)
0.31 (110)
0.30 (110)
0.24 (150)
-0.72 (140)
-0.53 (160)
-0.43 (150)
-0.23 (140)
-0.60
2
2
3
4
5
(PhQ+)
2
(4-AcPhQ+)
(2-PymQ+)
(Mebpe+)
2
2
3
2
19 (27 500)
28 (18 700)
6
(Phbpe+)
742 (6500)
0.25 (120)
-0.52
3
2
36 (35 400)
31 (18 900)
a
Solutions ca. 3-8 × 10 M. b Solutions ca. 10-3 M in 0.1 M [NBu 4]PF6 in methanol at a glassy carbon disk working electrode with a scan rate of
-5
n
-
1
c
2
00 mV s . Ferrocene internal reference E1/2 ) 0.67 V, ∆Ep ) 130 mV. For an irreversible reduction process.
12196 J. AM. CHEM. SOC.
9
VOL. 128, NO. 37, 2006

Pentacyanoiron(II) as an Electron Donor Group
A R T I C L E S
+
The Fe-C bond located trans to the MeQ ligand is shorter
than the other four Fe-C bonds by ca. 0.03 Å. This phenomenon
can be ascribed to the relative structural trans effects of the
+
31
cyanide and MeQ ligands; the latter is a weaker π-acceptor,
and therefore, more extensive π-back-bonding occurs to the trans
cyanide when compared with the cis ones. The slightly larger
relative trans Fe-C bond shortening of ca. 0.05 Å in [Et4N]3-
II
30
[
Fe (CN)5(py)] is consistent with the weaker π-acceptor ability
+
of py when compared with MeQ , and the corresponding
II
5l
difference of 0.06 Å in Na2[Fe (CN)5(NH3)]‚7H2O is in
keeping with the complete absence of π-acceptor character for
an ammonia ligand. A smaller and probably insignificant trans
Fe-C bond shortening has been observed in Na2[Fe (CN)5-
NO)]‚2H2O, consistent with the strong π-accepting ability
Figure 2. Representation of the molecular structure of 1‚9H2O (50%
probability ellipsoids).
II
increase with acceptor number for {Fe (CN)5}3- complexes,5j,m,n
II
32
(
and water has a higher acceptor number (54.8) than methanol
of a linearly coordinated nitrosyl ligand. As also observed in
(41.3). As with H-bonding, the better acceptor ability of water
II
30
[Et4N]3[Fe (CN)5(py)], the plane of the coordinated pyridyl
will also reduce the electron-donating ability and increase the
accepting ability of the cyanide ligands, thus decreasing the
ring in 1‚9H2O approximately bisects the equatorial C-Fe-C
angles.
II
electron density at the Fe center. As expected, the energies of
+
The MeQ ligand shows no significant evidence for ground-
the largely nondirectional intraligand transitions show little
solvent sensitivity.
state charge-transfer, with a dihedral angle of 24.26(4)° between
the two pyridyl rings. Reported crystal structures for other
Electrochemical Studies. The complex salts 1-6 were
studied by cyclic voltammetry in both water and methanol, and
II
+
related complexes such as trans-[Ru (NH3)4(MeQ )(PTZ)]-
33
[
[
[
PF6]3‚Me2CO (PTZ ) S-coordinated phenothiazine), trans-
Ru Cl(pdma)2(MeQ )][PF6]2, and [Re (CO)3(bpy)(MeQ )]-
PF6]2 (bpy ) 2,2′-bipyridyl) have shown that the dihedral
the results are presented in Tables 1 and 2. In all cases, reversible
II
+
27
I
+
III/II
or quasireversible Fe
waves are observed, the E1/2 values of
34
which are almost insensitive to the nature of the pyridyl ligand.
+
angle within coordinated MeQ can vary over a wide range of
ca. 8-44°. One of the sodium countercations is coordinated by
six water molecules (two shared with the other sodium cation),
and the other cation is coordinated by five water molecules and
the N atom of the trans cyanide ligand. A network of hydrogen
bonds between these water molecules and the N atoms of the
cyanide ligands links the complex units together in a centrosym-
metric structure which is unfortunately not expected to display
significant bulk NLO effects.
III/II
However, for each compound, E1/2[Fe ] decreases by ca. 200
mV on moving from water to methanol, indicating marked
destabilization of the Fe-based HOMOs. This observation can
be rationalized by following the same logic regarding solvent-
solute interactions as applied to the MLCT energies (see above).
Waves assigned to reductions of the pyridinium ligands are also
observed and, although these are less well-behaved than the
III/II
Fe
waves, increases of ca. 200 mV on moving from water
to methanol are evident. The large accompanying red-shifts in
the MLCT bands are therefore attributable to destabilization of
the Fe-based HOMOs and simultaneous stabilization of the
ligand-based LUMOs. Within the 4,4′-bipyridinium series 1-4,
the E1/2 values for pyridinium ligand-based reduction in
methanol increase as R changes in the order Me < Ph < 4-AcPh
Hyper-Rayleigh Scattering Studies. Complex salts 1-6
were studied in both water and methanol by using the HRS
18,19
technique
with a 1064 nm Nd:YAG laser fundamental.
35
Values of â0 were obtained by using the two-state model,
and the results are presented in Table 3. Although the ap-
36
plicability of this model has recently been challenged, it
<
2-Pym. This trend is consistent with the increasing electron-
remains a viable tool when considering the molecular NLO
responses of simple dipolar chromophores which possess a
accepting strength of the pyridinium unit, as observed previously
in related complexes.1
2,27
37
single major ICT band, such as those in 1-6. Furthermore,
Crystallographic Study. A single-crystal X-ray structure has
been obtained for the complex salt 1‚9H2O. A representation
of the molecular structure is shown in Figure 2, and selected
interatomic distances and angles are presented in Table S2.
II
our previous extensive studies with related Ru complexes give
ample precedent that such an analytical approach is reasonable,
and generally affords good results even when working with
colored chromophores (as verified by complementary Stark
II
3-
Although several structures of salts containing {Fe (CN)5}
2m,4,11,12,28,33
spectroscopic studies).
II
4-
moieties (mostly [Fe (CN)6] compounds) have been published
Despite the uncertainties introduced by using â1064 for pNA
in acetonitrile as an external reference for the data obtained in
water, the clearest conclusion to emerge from these HRS data
5l-n
previously,
to our knowledge this is only the second reported
structure of a compound containing a {Fe (CN)5} unit
coordinated to a pyridyl ligand; the salt [Et4N]3[Fe (CN)5(py)]
II
3-
II
3
0
(
py ) pyridine) has been described only very recently.
(31) Coe, B. J.; Glenwright, S. J. Coord. Chem. ReV. 2000, 203, 5-80.
(
32) Bottomley, F.; White, P. S. Acta Crystallogr., Sect. B 1979, 35, 2193-
2195.
(
(
(
27) Coe, B. J.; Beyer, T.; Jeffery, J. C.; Coles, S. J.; Gelbrich, T.; Hursthouse,
M. B.; Light, M. E. J. Chem. Soc., Dalton Trans. 2000, 797-803.
28) Coe, B. J.; Harries, J. L.; Harris, J. A.; Brunschwig, B. S.; Coles, S. J.;
Light, M. E.; Hursthouse, M. B. Dalton Trans. 2004, 2935-2942.
29) (a) Gutmann, V. Electrochim. Acta 1976, 21, 661-670. (b) Gutmann, V.
The Donor-acceptor Approach to Molecular Interactions; Plenum: New
York, 1978. (c) Gutmann, V.; Resch, G.; Linert, W. Coord. Chem. ReV.
(33) Coe, B. J.; Chamberlain, M. C.; Essex-Lopresti, J. P.; Gaines, S.; Jeffery,
J. C.; Houbrechts, S.; Persoons, A. Inorg. Chem. 1997, 36, 3284-3292.
(34) Busby, M.; Liard, D. J.; Motevalli, M.; Toms, H.; Vlcek, A., Jr. Inorg.
Chim. Acta 2004, 357, 167-176.
(35) (a) Oudar, J. L.; Chemla, D. S. J. Chem. Phys. 1977, 66, 2664-2668. (b)
Oudar, J. L. J. Chem. Phys. 1977, 67, 446-457.
(36) Selected examples: (a) Meshulam, G.; Berkovic, G.; Kotler, Z. Opt. Lett.
2001, 26, 30-32. (b) Woodford, J. N.; Wang, C. H.; Jen, A. K.-Y. Chem.
Phys. 2001, 271, 137-143.
1
982, 43, 133-164.
30) Chiarella, G. M.; Melgarejo, D. Y.; Koch, S. A. J. Am. Chem. Soc. 2006,
28, 1416-1417.
(
1
(37) Di Bella, S. New J. Chem. 2002, 26, 495-497.
J. AM. CHEM. SOC.
9
VOL. 128, NO. 37, 2006 12197

A R T I C L E S
Coe et al.
Table 3. MLCT Absorption and HRS Data for 1-6
water
methanol
a
b
a
b
λ
(nm)
max
â
(10
1064
30
â
0
λ
(nm)
max
â
(10
1064
30
â
0
-
(10- esu)
30
-
(10- esu)
30
complex salt (L)
esu)
esu)
+
2c
22
31
50
4
22
1
(MeQ )
534
566
584
618
538
562
311
228
218
214
239
259
686
738
771
854
685
742
330
339
728
485
363
396
128
163
380
272
140
192
+
2
3
4
5
6
(PhQ )
+
(4-AcPhQ )
(2-PymQ )
+
+
c
(Mebpe )
+
(Phbpe )
a
From 1064 nm HRS measurements at 295 K. b Derived from â1064 by using the two-state model.35 Underestimated due to proximity of λmax to 532
c
nm.
Table 4. MLCT Absorption and Stark Spectroscopic Data for 1-6a
b
c
d
e
f
g
2h
i
ab (cm-1)
â j
(10- esu)
30
complex salt (L)
λmax (nm)
E
max (eV)
f
os
µ
12 (D)
∆µ12 (D)
∆µab (D)
r
12 (Å)
r
ab (Å)
c
b
H
0
+
1
(MeQ )
533
559
581
617
557
582
2.33
2.22
2.13
2.01
2.23
2.13
0.11
0.13
0.14
0.17
0.16
0.19
3.5
4.0
4.1
4.7
4.4
4.8
22.5
23.5
26.3
26.5
25.9
29.3
23.6
24.8
27.5
28.1
27.4
30.9
4.7
4.9
5.5
5.5
5.4
6.1
4.9
5.2
5.7
5.8
5.7
6.4
0.02
0.03
0.03
0.03
0.03
0.02
2800
2900
2600
2700
2900
2700
61
88
114
166
119
174
+
2
3
4
5
6
(PhQ )
+
(4-AcPhQ )
(2-PymQ )
+
+
(Mebpe )
+
(Phbpe )
a
In glycerol-water (50:50 vol %) glasses at 77 K. Values of Emax(eV) for solutions at 295 K: 2.21 (1); 2.06 (2); 2.01 (3); 1.88 (4); 2.20 (5); 2.07 (6).
b
e
-9
2
c
d
Obtained from (4.32 × 10 M cm )A where A is the area under the absorption peak. Calculated from eq 2. Calculated from fint∆µ12 by using fint ) 1.33.
f
g
Calculated from eq 1. Delocalized electron-transfer distance calculated from ∆µ12/e. Effective (localized) electron-transfer distance calculated from
h
∆
µab/e. Calculated from eq 3. i Calculated from eq 4. j Calculated from eq 5.
is that the â0 values of the new complexes are much larger in
methanol than in water. This result is consistent with the large
red-shifts in the MLCT bands which, according to eq 5, would
be expected to lead to increased NLO responses. Also, increas-
ing the electron-accepting ability of the pyridinium group
generally causes â0 to increase within the 4,4′-bipyridinium
two-state eq 5. The general trends to emerge from these data
II
mirror those observed previously with Ru ammine or arsine
4
a,d,12,21,28
complex salts,
i.e., â0 within the 4,4′-bipyridinium
series 1-4 increases as Emax decreases, and â0 also increases
on moving from 5 to 6. N-Arylation is therefore an effective
aproach to enhancing quadratic NLO responses in the present
II
II
series 1-4, again as observed previously in related com-
Fe cyanide compounds, as in related Ru ammine or arsine
species. Furthermore, comparison of the data for the pairs 1/5
and 2/6 indicates that extension of the conjugated π-systems
leads to ca. 2-fold increases in â0.
plexes.1
2,28
The data in methanol indicate perhaps a small â0-
enhancing effect of extending the conjugated system by adding
a (E)-CHdCH unit, but the observed differences may not be
experimentally significant.
Protic Switching of Optical Properties Probed by Stark
Stark Spectroscopic Studies. We have studied complex salts
-6 by using Stark spectroscopy on the MLCT bands in
Spectroscopy. The presence of lone pairs of electrons on the
-
1
nitrogen atoms of the C-coordinated CN ligands means that
II
3-
5a
glycerol-water (50:50 vol %) glasses at 77 K, and the results
are shown in Table 4. Representative absorption spectra, Stark
spectra, and spectral components for the salts 1 and 4 are shown
in Figure S1. The parameter ∆µab refers to a transition from a
localized ground state (i.e., corrected for the effects of metal-
ligand bonding), and this quantity becomes particularly signifi-
{Fe (CN) } complexes can be reversibly protonated. Given
5
that this chemical transformation greatly changes the basicity
of these ligands, the MLCT properties are affected, and this
can be expected to alter the NLO properties. Previous reports
of protic switching of NLO behavior in purely organic and
38
metal-based chromophores have appeared, but in only one has
II
2+
cant when making comparisons with related {Ru (NH3)5}
a well-understood switching of â been experimentally demon-
3
8c
complexes (see below).
The Emax values decrease on glassing of the glycerol-water
strated. We have investigated the effects of protonation in
complex salts 1, 3, and 4 by using Stark spectroscopy, and the
results are presented in Table 5. The experimental conditions
were established by model studies at room temperature which
showed that the observed changes in the MLCT spectra stabilize
under the pH conditions used (i.e., the addition of extra
concentrated HCl to take the pH below 1.0 gives no further
spectral changes). Previous studies suggest that the species being
medium at 77 K. A similar effect has been observed previously
II
with Ru ammine or arsine complex salts on moving from
4,21,22,28
acetonitrile solutions to butyronitrile glasses.
Within the
4
,4′-bipyridinium series 1-4, fos, µ12, ∆µ12, and ∆µab all increase
steadily as Emax decreases, and the same trends are observed
on moving from 5 to 6. The latter two compounds also have
larger values for all of these parameters when compared with
5k
probed under these conditions are diprotonated, i.e., the neutral
2
II
+
their shorter analogues 1 and 2, respectively. The values of cb
complexes Fe (CN) (CNH) L (L ) MeQ , etc.). The acidified
3
2
and Hab are almost constant within the series 1-6 and indicate
only a rather limited degree of delocalization and electronic
coupling.
It is reasonable to assume that the lowest energy (i.e., MLCT)
electronic transitions dominate the NLO responses of these
chromophores, so â0 values have been derived by using the
solutions appear quite stable and neutralization with triethy-
(
38) (a) Lacroix, P. G.; Lepetit, C.; Daran, J. C. New J. Chem. 2001, 25, 451-
4
57. (b) Hurst, S. K.; et al. Organometallics 2001, 20, 4664-4675. (c)
Asselberghs, I.; Zhao, Y.-X.; Clays, K.; Persoons, A.; Comito, A.; Rubin,
Y. Chem. Phys. Lett. 2002, 364, 279-283. (d) Hurst, S. K.; Humphrey,
M. G.; Morrall, J. P.; Cifuentes, M. P.; Samoc, M.; Luther-Davies, B.;
Heath, G. A.; Willis, A. C. J. Organomet. Chem. 2003, 670, 56-65.
12198 J. AM. CHEM. SOC.
9
VOL. 128, NO. 37, 2006

Pentacyanoiron(II) as an Electron Donor Group
A R T I C L E S
Table 5. MLCT Absorption, Stark Spectroscopic and HRS Data for 1, 3, and 4 and Their Protonated Forms
a
b
b
b,c
12
∆µ12b,d
(D)
∆µabe
(D)
g
h
a,i
a,j
complex
salt
λ
max
λ
max
E
max
µ
H
ab
â
0
â
(10
800
30
â
0
2f
(cm-1)
(10- esu)
30
-
(10- esu)
30
(nm)
(nm)
(eV)
(D)
c
b
esu)
1
534
410
584
455
618
500
533
432
581
457
617
444
2.33
2.87
2.13
2.71
2.01
2.79
3.5
2.8
4.1
3.1
4.7
3.2
22.5
24.4
26.3
28.6
26.5
29.7
23.6
25.0
27.5
29.3
28.1
30.4
0.02
0.01
0.03
0.01
0.03
0.01
2800
2500
2600
2300
2700
2400
61
26
114
42
166
47
87
35
135
92
211
82
38
1
71
18
118
28
H2k
H2k
H2m
l
1
3
3
4
4
a
In water at 295 K. In glycerol-water (50:50 vol %) glasses at 77 K. Calculated from eq 2. d Calculated from fint∆µ12 by using fint ) 1.33. Calculated
b
c
e
f
g
h
i
j
from eq 1. Calculated from eq 3. Calculated from eq 4. Calculated from eq 5. From 800 nm HRS measurements at 295 K. Derived from â800 by using
3
5
k
l
the two-state model.
100 µL of concentrated HCl added to 500 µL of sample solution for Stark studies (2.0 M HCl concentration). Underestimated due
to proximity of λmax to 400 nm. m 200 µL of concentrated HCl added to 500 µL of sample solution for Stark studies (3.4 M HCl concentration).
Figure 4. HRS signals derived from the scattering curves shown in Figure
Figure 3. Effects of pH changes on the UV-vis absorption spectrum of
-
5
S2, with the value for the solvent (water) subtracted.
5
in water (ca. 10 M); original sample at neutral pH (green); acidified to
pH < 1.0 by the addition of concentrated HCl (red); basified to pH > 10.0
by the addition of triethylamine (blue).
(1H2) is grossly underestimated, all of the complexes show ca.
4
-fold decreases in their NLO responses upon protonation. The
lamine restores the original spectral profiles which show no
additional changes as the pH is increased further (Figure 3).
Protonation causes large increases in Emax of ca. 0.5-0.8 eV,
consistent with expected decreases in the electron-donating
magnitude of these changes is similar to that predicted by the
Stark measurements, although the actual â0 values derived from
the latter studies are always higher than those determined via
HRS in aqueous solutions.
II
ability of the Fe center (Figure 3). These blue-shifts are
Theoretical Studies. In previous studies on the NLO proper-
II
4c,d
accompanied by decreases in µ12 and small increases in ∆µ12
ties of dipolar Ru ammine complexes, we have demonstrated
the usefulness of density functional theory (DFT) methods along
with the effective core potential (ECP) LanL2DZ. However,
this model chemistry is inadequate for studying anionic species
which require the use of diffuse functions to reliably predict
their NLO behavior. Initial calculations were hence carried out
by using the B3P86/6-311+G* model chemistry with geometries
that were optimized in the gas phase by using the same
functional and the 6-31G* basis set. The hyperpolarizabilities
calculated in this way by using the finite-field (FF) approach,
2
and ∆µab. The values of cb and Hab show slight decreases which
are in keeping with the expectation that the less electron-rich
II
-
{
Fe (CN)3(CNH)2} moiety will engage in less effective
II
3-
π-back-bonding when compared with {Fe (CN)5} ; although
the differences for these quantities are within the experimental
errors, the consistency of the trends suggests that these may be
real. Most importantly, â0 is found to decrease upon protonation
by a factor of ca. 2-4, corresponding with a relatively efficient
switching process. This phenomenon is attributable to the
increases in Emax and decreases in band intensity (i.e., µ12) which
are partially offset by the small increases in ∆µ12.
-
30
â0 ) 104 × 10
esu for the complex anion in 1 and â0 ) 160
-
30
× 10 esu for that in 5, are in reasonable agreement with the
experimental values derived via HRS in methanol. We have
also calculated the electronic transitions for the complex anions
in these salts by using TD-DFT and the same level of theory.
The estimation of ∆µ12 values by using either the one-electron
RhoCI density or the FF method allows the calculation of â0
via eq 5. Once again, the results (Table S3) agree reasonably
well with the experimental data obtained via HRS in methanol
or via Stark spectroscopy. However, this agreement is purely
coincidental because the calculated ∆µ12 values are an order of
magnitude smaller than those obtained from Stark studies, while
the predicted values of fos and µ12 are several times larger than
the corresponding experimental data.
Protic Switching of First Hyperpolarizabilities Probed by
Hyper-Rayleigh Scattering. We have also carried out HRS
studies to directly confirm the effects of protonation on the â
3+
responses of 1-4, using a 800 nm Ti :sapphire laser due to
negligible HRS signals from the protonated samples when using
a 1064 nm fundamental. The â0 values for the neutral and
acidified solutions were estimated from the â800 data by using
the two-state model,35 and the results are presented in Table 5.
The data for 2 and 2H2 are â800(â0) ) 123(62) and 66(15) ×
-30
10
esu, respectively. Reversibility of the switching effect was
demonstrated with salt 4; the quadratic curves for samples 1-4
are shown in Figure S2, and the quadratic coefficients derived
from these curves (with the value for the solvent subtracted)
are shown in Figure 4.
As noted above, the acidified samples show large increases
in Emax, and these blue-shifts of the MLCT bands are ac-
companied by large decreases in â800 and (especially) â0. With
the exception of 1, for which the â0 of the protonated form
In the search for a model that provides better agreement
between theory and experiment, we have included solvent effects
using a polarizable continuum model (PCM). Initial single-point
PCM calculations performed on the geometries optimized in
the gas phase provide somewhat larger ∆µ12 values, but only
small changes are observed in the calculated µ12 values. For
J. AM. CHEM. SOC.
9
VOL. 128, NO. 37, 2006 12199

A R T I C L E S
Coe et al.
Figure 7. Illustrations of the 0.04 contour plots of molecular orbitals of
the complex anion in 5 calculated by PCM(H2O)-B3P86/6-311+G*.
gas phase than in solution, while those depicted as single are
longer in the gas phase. Therefore, polar media, such as water,
favor the more polar (charge-separated) structures, while non-
polar solvents favor the less-polar structures. In the present
complexes, as also observed in some purely organic chro-
3
9
mophores, the more-polar structures are aromatic while the
less-polar forms are quinoidal.
Further support for the preference of the quinoidal or aromatic
structure in different media can be obtained from the calculated
charges and also the spatial arrangement of the methyl group
Figure 5. Selected bond lengths in the complex anions in 1 (top) and 5
40
(bottom), optimized in a vacuum (black) and water (red italics). The
(
Figure S3). In the aromatic structure, the pyridyl nitrogen
crystallographic values for 1 are also shown (blue italics).
2
attached to this methyl group has sp hybridization and is
therefore planar, whereas the nitrogen in the quinoidal structure
must have sp hybridization and show a tetrahedral arrangement.
example, for the complex anion in 1, ∆µ12 increases from 1.2
to 16.7 (calculated by using the RhoCI density), while µ12
decreases from 10.2 to 9.7 (Table S3). Better agreement is
obtained when calculations are performed on geometries that
are optimized by using the PCM method (Table S3). The
calculated ∆µ12 values are now comparable to those determined
by the Stark technique, and the values of µ12 and fos, while still
larger than the experimental ones, are somewhat smaller than
those calculated on the gas-phase geometries. However, it must
be noted that PCM calculations cannot simulate the effects of
H-bonding and the results of these calculations are best
compared to the measurements in methanol which, as described
above, show Emax values that are very close to those calculated
and absorptions which are generally more intense than those in
water.
Given that molecular geometry has important effects on NLO
behavior, it is worth making a detailed comparison of the
geometries obtained in the gas phase and in solution. Figure 5
shows selected bond lengths in the optimized geometries of the
complex anions in 1 and 5 in the gas phase and in aqueous
solutions. It can be seen that passing from gas phase to the polar
solvent medium causes dramatic changes in the geometry and
that there is a close agreement between the bond lengths
calculated in water and those observed in the X-ray crystal
structure of 1‚9H2O (Figure 5 and see above).
3
As expected, the methyl group lies coplanar with the adjacent
pyridyl ring in the solution geometries but moves out of this
ring plane by ca. 21° in the gas-phase geometries.
As expected, TD-DFT shows that most of the hyperpolariz-
ability of these molecules is associated with their lowest-energy
absorptions which arise from one electron HOMO f LUMO
transitions. The topologies of the orbitals involved in these
transitions (Figures 7 and 8) show that the HOMOs are mainly
derived from iron d orbitals while the LUMOs spread over the
+
+
MeQ /Mebpe ligand, confirming the assignment of these
transitions as being MLCT character. Although the HOMO-
LUMO gap in the complex anion in 1 is larger by 0.29 eV when
compared with that in 5, the increased correlation in the more
extended complex anion in 5 reduces the difference in Emax
calculated by TD-DFT to only ca. 0.1 eV (Table S3), which is
in better agreement with the nearly identical Emax values
observed experimentally for 1 and 5. The greater length of the
complex anion in 5 when compared to that in 1 is responsible
for the larger ∆µ12 and µ12 values that give rise to a larger â0.
To study the protic switching of â in the complex anion in
1, we have also performed calculations on the monoprotonated,
diprotonated, and fully protonated complex species (denoted 1H,
1H2, and 1H5, respectively) with the same model chemistry used
for the anionic species (PCM-B3P86/6-311+G* on PCM-
B3P86/6-31G* geometries). The results of these calculations
are summarized in Table 6. TD-DFT reveals that the lowest
energy absorption corresponds in every case to a MLCT
transition, with the orbitals being those shown in Figure 8. A
Changes in geometry can be rationalized by considering the
existence of two limiting resonance structures corresponding
to the fully aromatic or quinoidal species (Figure 6). Comparison
of the bond lengths shown in Figure 5 indicates that bonds
depicted as double in the quinoidal structure are shorter in the
Figure 6. Limiting resonance structures of the complex anions in 1 (top) and 5 (bottom).
2200 J. AM. CHEM. SOC. VOL. 128, NO. 37, 2006
1
9

Pentacyanoiron(II) as an Electron Donor Group
A R T I C L E S
Table 6. Results of Theoretical Calculations on the Complex Anion in 1 and Its Protonated Forms
a
b
c
(10- esu)
30
d
(10- esu)
30
parent salt
HOMO
−LUMO gap
E
max (eV)
µ12 (D)
f
os
∆µ12 (D)
∆µ12 (D)
â
0
â
0
1
1
1
2.15
2.74
3.27
3
4.65
1.82
2.28
2.77
2.95
4.09
6.7
4.0
2.4
2.9
4.5
0.31
0.14
0.06
0.09
0.32
24.7
28.8
29.2
29.1
22.3
17.0
26.3
27.4
27.2
26.7
391
104
26
32
32
269
95
25
30
38
e
e
H
H2
.46f
1
H5
a
Calculated by using the RhoCI density. Calculated by using the FF approach. c Calculated from eq 5 by using the RhoCI ∆µ12 value. d Calculated from
b
eq 5 by using the FF ∆µ12 value. (HOMO-1) f LUMO. f (HOMO-2) f LUMO.
e
Figure 9. Chemical structures of the RuII complex salts investigated.
lowest energy transition is mainly comprised of a HOMO-2 to
LUMO excitation that also has MLCT character. The calculated
µ12 for this second excitation is similar to that of the (HOMO-
) f LUMO transition described above, but its larger µ12
produces a contribution to the hyperpolarizability that is
comparable to that of the first allowed transition despite the
lower energy of the latter (Table 6). If these two lowest energy
MLCT transitions are taken into account, then the total â0
response predicted for 1H2 by using the RhoCI ∆µ12 values in
∆
1
Figure 8. Illustrations of the 0.04 contour plots of molecular orbitals
involved in the lowest energy transition of the complex chromophores in 1
(a), 1H (b), 1H2 (c), and 1H5 (d) calculated by PCM(H2O)-B3P86/6-
3
11+G*.
comparison of these orbitals shows that protonation lowers the
-
30
eq 5 is 58 × 10
.
II
energies of the Fe -based orbitals which are located near the
Ruthenium Pentaammine Complexes: Electronic Spec-
protonation site, but the energy of the LUMO located on the
troscopy and Electrochemical Studies. Given that we have
+
MeQ ligand is less affected by the acidity of the medium. As
II
previously carried out extensive studies with Ru ammine
a consequence of these energy changes, the HOMO-LUMO
gap increases steadily on protonation from 2.15 to 4.65 eV (in
complexes of the same pyridinium ligands used here,4
a,d,12,21,28
it is of interest to draw comparisons between these two classes
of compounds. However, to do so further, new data have been
1H5) while the corresponding Emax value calculated by TD-DFT
changes from 1.82 to 4.09 eV. The calculated blue-shift of 1.05
eV passing from 1 to 1H2 is rather larger than the 0.54 eV shift
observed at 77 K in glycerol-water (50:50 vol %) (Table 5).
II
obtained for the Ru ammine complex salts under the same
II
experimental conditions as those used to study the Fe cyanide
compounds. To improve the solubility in water and alcohols,
we used the chloride salts rather than the previously studied
-
30
Furthermore, the â0 of 26 × 10 esu calculated from eq 5 for
1
H2 by using the parameters obtained from TD-DFT is strongly
4
1
hexafluorophosphates, and UV-vis and electrochemical data
for the complex salts 7-12 (Figure 9) are presented in Table
-
30
decreased with respect to the 391 × 10
esu calculated for
the anion in 1 via the same methodology, and this dramatic
drop in the hyperpolarizability mainly arises from the large blue-
shift caused by protonation.
7
.
(41) It should be noted that some uncertainty exists regarding the precise
II
3 5 3
formulations of the materials [Ru (NH ) L]Cl due to their hygroscopic
It should be noted, however, that for the neutral species 1H2
the predicted excitation energy for the second lowest energy
transition is 2.95 eV, which is only 0.18 eV above that of the
lowest energy transition, and therefore, the second excited state
is expected to have an important contribution to the overall
hyperpolarizability. According to our calculations, this second
nature coupled with the well-established difficulties in obtaining satisfactory
II
CHN elemental analyses for such Ru ammine salts. Previous studies with
7
-9 indicated that these compounds retain ca. 2.5-3.5 molecules of water
3a
of crystallization after drying under vacuum at room temperature.
However, we have repeatedly been unable to obtain satisfactory CHN
analyses for 10-12; the consistently low C and N percentages are driven
up by as much as 70% of their initial values following prolonged heating
under vacuum, with approximately constant C/N ratios, but the expected
values are never achieved. Representative data for 11 are as follows. Anal.
3 7
Calcd (%) for C13H28Cl N Ru: C, 31.88; H, 5.76; N, 20.02. Found after
(
39) Andreu, R.; Blesa, M. J.; Carrasquer, L.; Gar ´ı n, J.; Orduna, J.; Villacampa,
B.; Alcal a´ , R.; Casado, J.; Ruiz Delgado, M. C.; L o´ pez Navarrete, J. T.;
vacuum-drying at room temperature for 24 h: C, 18.87; H, 6.30; N, 10.74.
Found after vacuum-drying at 100 °C for 24 h: C, 27.23; H, 4.93; N, 15.50.
-
Allain, M. J. Am. Chem. Soc. 2005, 127, 8835-8845.
Given that 7-12 are prepared simply by anion metathesis from their PF
6
-
1
(40) The Mulliken charges (e ) calculated for the {Fe(CN)
5
} group by using
counterparts (which are confirmed to be pure by H NMR spectra and CHN
analyses), there is no reason to doubt the chromophoric purity of 7-12.
However, for the purposes of calculating the solution concentrations for
physical studies, we make the assumption that all of these samples contain
3 molar equivalents of water of crystallization.
B3P86/6-311+G* are as follows. Anion in 1, gas phase -1.844; PCM-
water) -2.859. Anion in 5, gas phase -1.826; PCM(water) -2.935. These
results are consistent with the iron atom having a formal oxidation state of
3 in the gas phase and +2 in water.
(
+
J. AM. CHEM. SOC.
9
VOL. 128, NO. 37, 2006 12201

A R T I C L E S
Coe et al.
Table 7. UV-Vis Absorption and Electrochemical Data for 7-12
E, V vs. Ag/AgCl (
[RuIII/II] or Epcd
E
1/2
∆E
p
, mV)c
, M- cm-1)a
1
E
max (eV)a
λ
max, nm (
ꢀ
, M- cm-1)b
1
E
max (eV)b
assignment
E
complex salt (L)
λ
max, nm (
ꢀ
1/
2
7
(MeQ+)
606 (9300)
62 (22 300)
622 (11 900)
83 (14 500)
638 (13 200)
87 (18 300)
677 (13 400)
87 (18 200)
592 (7700)
2.05
4.73
1.99
4.38
1.94
4.32
1.83
4.32
2.09
3.92
5.46
2.00
3.71
5.39
622 (9900)
1.99
4.68
1.89
4.32
1.85
4.26
1.73
4.34
1.98
3.90
5.44
1.89
3.67
5.37
d f π*
π f π*
d f π*
π f π*
d f π*
π f π*
d f π*
π f π*
d f π*
π f π*
π f π*
d f π*
π f π*
π f π*
0.26 (130)
0.27 (120)
0.29 (110)
0.28 (120)
0.21 (100)
-0.88 (210)
-1.33
2
265 (15 450)
655 (13 200)
287 (14 100)
672 (13 300)
291 (14 250)
715 (13 600)
286 (14 550)
625 (10 750)
318 (16 550)
228 (7250)
8
9
1
1
(PhQ+)
-0.76 (160)
2
(4-AcPhQ+)
0 (2-PymQ+)
1 (Mebpe+)
-0.71
-0.62
-0.95
2
2
3
2
16 (19 800)
27 (9500)
1
2 (Phbpe+)
620 (13 600)
657 (14 000)
337 (22 850)
231 (8600)
0.22 (110)
-0.78
3
2
34 (27 300)
30 (9800)
a
In water ca. 3-8 × 10 M. b In methanol ca. 3-8 × 10 M. All concentrations calculated by assuming that each sample contains 3 molar equivalents
-5
-5
4
1
-
of water of crystallization. Values of Emax(eV) for acetonitrile solutions of PF6 salts at 295 K: 2.10 (7); 1.97 (8); 1.90 (9); 1.84 (10); 2.08 (11); 1.97
12).
could not be obtained in methanol due to insufficient solubility of these compounds in the electrolyte solution. For an irreversible reduction process.
1
1,12
c
-3
-1
(
Solutions ca. 10 M in 1 M aqueous KNO3 at a glassy carbon disk working electrode with a scan rate of 200 mV s . Electrochemical data
d
Table 8. MLCT Absorption and HRS Data for 7-12
water
methanol
a
b
a
b
λ
(nm)
max
â
(10
1064
30
â
0
λ
(nm)
max
â
(10
1064
30
â
0
-
(10- esu)
30
-
(10- esu)
30
complex salt (L)
esu)
esu)
+
7
(MeQ )
606
622
638
677
592
620
169
255
333
402
150
212
34
62
93
148
25
50
622
655
672
715
625
657
455
493
522
497
589
522
110
158
187
220
147
170
+
8
9
1
1
1
(PhQ )
+
(4-AcPhQ )
0 (2-PymQ )
1 (Mebpe )
2 (Phbpe )
+
+
+
a
b
3
5
F
r
o
m
1
0
6
4
n
m
H
R
S
m
e
a
s
u
r
e
m
e
n
t
s
a
t
2
9
5
K
.
D
e
r
i
v
e
d
f
r
o
m
â
1
0
6
4
b
y
u
s
i
n
g
t
h
e
t
w
o
-
s
t
a
t
e
m
o
d
e
l
.
Table 9. MLCT Absorption and Stark Spectroscopic Data for 7-12a
b
c
∆µ12d(D)
e
12 f(Å)
g
2h
i
ab (cm-1)
â j
(10- esu)
30
complex salt (L)
λmax (nm)
E
max (eV)
f
os
µ
12 (D)
∆µab (D)
r
r
ab (Å)
c
b
H
0
+
7
(MeQ )
633
678
705
751
653
690
1.96
1.83
1.76
1.65
1.90
1.80
0.14
0.17
0.35
0.25
0.10
0.29
4.3
4.9
7.2
6.4
3.8
6.6
13.2
15.2
14.1
15.3
16.9
18.2
15.8
18.1
20.1
19.9
18.6
22.4
2.8
3.2
2.9
3.2
3.5
3.8
3.3
3.8
4.2
4.2
3.9
4.7
0.08
0.08
0.15
0.12
0.04
0.09
4300
4000
5000
4300
3200
4200
74
129
275
267
80
+
8
9
1
1
1
(PhQ )
+
(4-AcPhQ )
0 (2-PymQ )
1 (Mebpe )
2 (Phbpe )
+
+
+
282
a
Measured in glycerol-water (50:50 vol %) glasses at 77 K. All concentrations calculated by assuming that each sample contains 3 molar equivalents
4
1
b
-9
2
c
d
of water of crystallization.
Obtained from (4.32 × 10 M cm )A where A is the area under the absorption peak. Calculated from eq 2. Calculated
from fint∆µ12 by using fint ) 1.33. Calculated from eq 1. Calculated from ∆µ12/e. g Calculated from ∆µab/e. h Calculated from eq 3. Calculated from eq
e
f
i
j
4
.
Calculated from eq 5.
The new MLCT absorption data for 7-12 show the same
increases in â0, but the actual size of this effect is obscured by
the uncertain referencing of the data in water. Interestingly, the
effect of inserting a (E)-CHdCH unit into the conjugated bridge
appears to be different in the two solvents, i.e., decreases in â0
in water but increases in methanol, although the changes are
not large.
trends as observed previously in acetonitrile (i.e., red-shifting
with increasing acceptor strength of L),12 and the Emax values
in this solvent are similar to those in water but larger by ca. 0.1
eV when compared with those in methanol. The electrochemical
data also resemble those obtained in acetonitrile:12 the Ru
E1/2 values are smaller for 11 and 12 than for 7-10 due to the
mildly electron-donating influence of the (E)-CHdCH unit, and
the ligand reduction potentials decrease as the pyridinium group
becomes more electron deficient.
III/II
Ruthenium Pentaammine Complexes: Stark Spectro-
scopic Studies. We have also carried out Stark studies with
complex salts 7-12 in glycerol-water (50:50 vol %) glasses
at 77 K, and the results are presented in Table 9. These data
show the same general trends as observed when using buty-
Ruthenium Pentaammine Complexes: Hyper-Rayleigh
Scattering Studies. The results of HRS studies at 1064 nm with
complex salts 7-12 in both water and methanol are presented
in Table 8. As with the MLCT absorption data, these new HRS
data for 7-12 show the same trends as observed previously in
acetonitrile (i.e., increasing â0 with red-shifting of the MLCT
2
1
ronitrile glasses, although it is worth noting that these are the
first Stark results for the complex in 10 because a satisfactory
II
+
data fit could not be obtained with [Ru (NH3)5(2-PymQ )]-
2
1
[PF6]3. In particular, the now well-established correlation
between â0 and L acceptor strength is again apparent.
Comparisons between Iron Pentacyanide and Ruthenium
Pentaammine Complexes. Although these two types of chro-
1
2
band). As expected, the red-shifts in the MLCT bands on
moving from water to methanol are accompanied by large
12202 J. AM. CHEM. SOC.
9
VOL. 128, NO. 37, 2006

Pentacyanoiron(II) as an Electron Donor Group
A R T I C L E S
II
Table 10. Selected MLCT Absorption, Electrochemical, Stark Spectroscopic, and HRS Data for Complex Salts Na2[Fe (CN)5L] and
II
[
Ru (NH3)5L]Cl3
max[w]a
(eV)
E
max[m]a
(eV)
E
max[g
−
w]b
E
[MIII II]c
/
µ
d
∆µ12e
(D)
∆µab f
(D)
r
g
r
h
H
j
â
k
â
l
â
0
m
complex salt
E
1
/
2
12
12
ab
ab
0
0
2
i
(cm-1)
(10- esu)
30
(10- esu)
30
(10- esu)
30
(L)
(eV)
(V)
(D)
(Å)
(Å)
c
b
+
2n
34
22
62
31
1
7
2
8
3
9
4
1
5
1
6
1
(MeQ )
2.32
2.05
2.19
1.99
2.12
1.94
2.01
1.83
2.30
2.09
2.21
2.00
1.81
1.99
1.68
1.89
1.61
1.85
1.45
1.73
1.81
1.98
1.67
1.89
2.33
1.96
2.22
1.83
2.13
1.76
2.01
1.65
2.23
1.90
2.13
1.80
0.48
0.26
0.48
0.27
0.46
0.29
0.46
0.28
0.45
0.21
0.45
0.22
3.5
4.3
4.0
4.9
4.1
7.2
4.7
6.4
4.4
3.8
4.8
6.6
22.5
13.2
23.5
15.2
26.3
14.1
26.5
15.3
25.9
16.9
29.3
18.2
23.6
15.8
24.8
18.1
27.5
20.1
28.1
19.9
27.4
18.6
30.9
22.4
4.7
2.8
4.9
3.2
5.5
2.9
5.5
3.2
5.4
3.5
6.1
3.8
4.9
3.3
5.2
3.8
5.7
4.2
5.8
4.2
5.7
3.9
6.4
4.7
0.02
0.08
0.03
0.08
0.03
0.15
0.03
0.12
0.03
0.04
0.02
0.09
2800
4300
2900
4000
2600
5000
2700
4300
2900
3200
2700
4200
61
74
88
129
114
275
166
267
119
80
128
110
163
158
380
187
272
220
140
147
192
170
+
(MeQ )
+
(PhQ )
+
(PhQ )
+
(4-AcPhQ )
(4-AcPhQ )
(2-PymQ )
0 (2-PymQ )
+
93
50
+
+
148
4
+
n
(Mebpe )
+
1 (Mebpe )
25
22
50
+
(Phbpe )
174
282
+
2 (Phbpe )
a
b
c
-3
At 295 K; w ) water, m ) methanol. At 77 K; g-w ) glycerol-water (50:50 vol %). Reported vs Ag/AgCl reference; solutions ca. 10 M in 1
-
1
d
e
M aqueous KNO3 at a Pt disk working electrode with a scan rate of 200 mV s ; M ) Fe or Ru. Calculated from eq 2. Calculated from fint∆µ12 by using
fint ) 1.33. Calculated from eq 1. Calculated from ∆µ12/e. Calculated from ∆µab/e. Calculated from eq 3. Calculated from eq 4. k Calculated from eq
f
g
h
i
j
l
From 1064 nm HRS measurements in water. m From 1064 nm HRS measurements in methanol. n Underestimated due to proximity of λmax to 532 nm.
5
.
FeII compounds, meaning that the Emax values for the Ru
II
II
complexes are now higher than those of their Fe counterparts
by ca. 0.2-0.3 eV (Figure 10). Hence, if only MLCT energies
II
2+
are considered, then a {Ru (NH3)5} center is a stronger
II
3-
electron donor than a {Fe (CN)5}
moiety in water or
glycerol-water, but the opposite is true in methanol. The
observation that the E1/2 potentials for metal-based oxidation
in aqueous solution are lower by ca. 0.2 V for the Ru complexes
than for their Fe congeners is also consistent with the greater
electron-richness (and by inference electron-donating ability)
II
2+
of the {Ru (NH3)5} center under such conditions.
Figure 10. MLCT absorption spectra of complex salts 2 in water (green)
and methanol (blue) and 8 in water (gold) and methanol (red).
2
With the exception of 11, the derived values of µ12, cb , and
Hab are larger in the Ru complexes than in their Fe
II
II
counterparts; these observations are consistent with previous
mophore both contain strongly electron-donating d6 metal
centers, they are clearly quite different chemically, and this
might be expected to be reflected in their electronic and optical
properties. Several previous comparative studies involving
studies which conclude that π-orbital overlap is more effective
II
2+
with the {Ru (NH3)5} center. ∆µ12 relates to a transition from
a delocalized ground state, and ∆µab refers to a transition from
a localized ground state (i.e., corrected for the effects of metal-
II
3-
II
2+
5a,d-f,h
{Fe (CN)5} and {Ru (NH3)5} complexes have appeared,
2
ligand bonding). These two quantities are interrelated by cb ,
but none feature either Stark or HRS results. The general
conclusions to emerge from these earlier studies can be sum-
the degree of delocalization by eq 6.
II
∆µ12
marized as follows. (i) The MLCT bands of the Ru complexes
∆
µ )
(6)
II
ab
appear at lower energies than those of their Fe counterparts
2
1
- 2cb
II
2+
(
at least in aqueous solutions), and (ii) a {Ru (NH3)5} center
engages in more effective π-back-bonding with pyridyl ligands
In every case, the values of ∆µ12 and ∆µab decrease quite
considerably on moving from a {Fe (CN)5} complex to its
(
{
i.e., better metal-ligand orbital overlap) when compared with
Fe (CN)5} due primarily to the larger radial extension of
II
3-
II
3-
II
2+
{
Ru (NH3)5}
analogue, and hence, the electron-transfer
4
d as opposed to 3d orbitals. The back-bonding ability of the
II
distances, r12 and rab, are also smaller for the Ru complexes.
The r values indicate that in the Ru complexes the charge is
transferred onto the first pyridyl ring, whereas in their Fe
II
Fe center may also be reduced by the fact that the cyanide
co-ligands also possess π-acceptor properties. The physical
studies that we now report allow unprecedentedly detailed
comparisons to be made, and selected data are collected in Table
II
II
counterparts, the charge is transferred further toward the middle
of L. The decrease in ∆µ12 is potentially a consequence of
increased delocalization due to stronger π-back-bonding in the
10 for this purpose. Representative UV-vis spectra of complex
salts 2 and 8 in water and methanol are shown in Figure 10.
II
Ru complexes. The amount of charge transferred is given by
2
2
All of the MLCT absorption data obtained in aqueous
(1 - 2cb )e, so as the degree of delocalization, cb , increases,
less charge is transferred and ∆µ12 decreases. The difference
solutions at 295 K or in glycerol-water (50:50 vol %) glasses
II
II
at 77 K are in agreement with previous reports in that the Ru
between ∆µ12 and ∆µab also increases on replacing {Ru -
2
+
II
3-
compounds always have considerably lower Emax values when
(NH3)5} with {Fe (CN)5} . The use of ∆µab allows ‘cor-
II
II
2+
compared with their Fe counterparts. The {Ru (NH3)5} vs
{
rection’ of the dipole moment change for the manner of bonding
II
3-
22
Fe (CN)5} energy differences are ca. 0.2-0.3 and 0.3-0.4
and the extent of delocalization between the metal and L.
II
II
eV in water and glycerol-water, respectively. However, this
situation is completely reversed in methanol because the red-
shifting of the MLCT bands is much more pronounced for the
Therefore, if the only difference between the Fe and Ru
complexes were the increased π-back-bonding interaction of
II
the Ru center, then the two types of complexes should display
J. AM. CHEM. SOC.
9
VOL. 128, NO. 37, 2006 12203

A R T I C L E S
Coe et al.
similar ∆µab values. The fact that this is not the case indicates
Conclusions
We have synthesized and characterized a family of MLCT
that another effect must also be contributing to the decreased
II
∆
µ12 values observed for the Ru species.
II
3-
chromophores based on electron-donating {Fe (CN)5} centers.
HRS measurements at 1064 nm show that these complexes have
relatively large static first hyperpolarizabilities, â0, which
increase markedly on moving from aqueous to methanol
solutions. This solvent dependence of NLO responses can be
traced to large red-shifts in the MLCT transitions, which are
shown by cyclic voltammetry to be attributable to destabilization
of the Fe-based HOMOs and simultaneous stabilization of the
acceptor ligand-based LUMOs. Stark spectroscopic studies in
glycerol-water (50:50 vol %) glasses at 77 K confirm and
rationalize the large â0 values according to the two-state model.
The other effect which influences ∆µ12 is likely to be repolar-
ization of the L valence electrons in the excited state. Because
III
an MLCT state formally contains a M center, the valence elec-
trons on L will be displaced as a consequence of the increased
positive charge at the metal. This repolarization of the ligand
valence electrons causes its electron distribution to shift toward
the metal, resulting in a dipole moment change which opposes
that produced by the MLCT excitation process, thus decreasing
the observed ∆µ12.42 In the case of the Fe complexes, the π
-
electrons within the CN ligands are more easily polarized
(
when compared to NH3 which lacks π electrons) and there-
II
Acidification of solutions of the dipolar Fe complexes leads
III
fore stabilize the increased charge on the Fe center in the
to large, reversible blue-shifts in their MLCT bands due to
protonation of the cyanide ligands. These increases in MLCT
energies are accompanied by several-fold decreases in the â0
responses, as shown via both HRS and Stark experiments,
demonstrating a novel approach to protic-switching of molecular
NLO properties. TD-DFT and FF calculations using a PCM
yield relatively good agreement with the experimental results
and confirm (but overestimate) the large decrease in NLO
response on protonation. The available physical data permit
detailed comparisons with previously reported chromophores
excited state. Thus, L experiences a smaller increase in posi-
III
tive charge in the excited state when compared with the Ru
case and is not polarized to the same extent as in the Ru
complexes.
Except for 11, the â0 values derived from the Stark data by
using eq 5 are larger (and in most cases substantially so) for
II
II
the Ru complexes than for their Fe counterparts. This trend
arises because the â0-enhancing effects of decreased Emax and
increased µ12 values more than offset the decreases of ∆µ12 in
II
II
2+
the Ru complexes. The HRS experiments in water concur with
3 5
with {Ru (NH ) } electron donor centers. The Stark-derived
II
2+
II
this general result, i.e., {Ru (NH3)5} is a more effective
electron donor than {Fe (CN)5} in terms of quadratic NLO
â0 values are generally larger for the Ru complexes than for
their Fe analogues due to lower Emax and higher µ12 values
which more than offset the decreases of ∆µ12 in the Ru
II
3-
II
II
responses. However, the HRS data in methanol present an
II
opposite picture (with the exception of 5) in that the Fe center
complexes. This observation agrees with the HRS data in water,
II
becomes the better electron donor, although in most cases, the
differences are not very large. This change in relative behavior
of â0 obviously correlates with the solvatochromic properties
of the compounds under study; the very large red-shifts in the
but the HRS data in methanol show that the Fe complexes
have the larger â0 responses. The more pronounced solvato-
II
chromism of the Fe complexes means that their NLO perfor-
II
2+
mance can surpass that of their {Ru (NH3)5} counterparts
II
MLCT bands of the Fe complexes on moving from water to
upon changing the solvent medium.
II
2+
methanol mean that their {Ru (NH3)5} analogues become the
higher-energy absorbers. Although it would clearly be of interest
to also determine ∆µ12 values for 1-12 in water and methanol,
this is unfortunately precluded by the requirement of Stark
spectroscopy for suitable glassing media.
Acknowledgment. We thank the EPSRC for support (Grant
No. GR/R54293 and a Ph.D. studentship) and also the Flemish
Fund for Scientific Research (FWO-V, Grant No. G.0297.04),
the University of Leuven (GOA/2006/3), the Belgian Govern-
ment (IUAP P5/3), MCyT-FEDER (BQU2005-01368) and
Gobierno de Aragon-Fondo Social Europeo (E39). I.A. ac-
knowledges a fellowship from the FWO-V.
It is also worth noting that it is not at present possible to use
our theoretical calculations to draw comparisons between the
II
3-
II
2+
{
Fe (CN)5} and {Ru (NH3)5} complexes because they have
been calculated at different levels of theory. Meaningful com-
parisons would require a computational method that is suitable
for both types of compound (the LanL2DZ basis set is
inadequate for anions and 6-311+G* is not defined for
ruthenium).
Supporting Information Available: Tables S1-S3; Figures
S1-S3; Cartesian coordinates of theoretically optimized ge-
ometries for the anionic complexes in 1 and 5 and for the
complexes in 1H, 1H2 and 1H5; complete refs 3d, 3j, 24, and
3
8b (PDF). This material is available free of charge via the
Internet at http://pubs.acs.org.
(
42) Shin, Y.-g. K.; Brunschwig, B. S.; Creutz, C.; Sutin, N. J. Am. Chem. Soc.
1
995, 117, 8668-8669.
JA063449M
12204 J. AM. CHEM. SOC.
9
VOL. 128, NO. 37, 2006