Oxygen gas sensing by luminescence quenching in crystals of Cu(xantphos)(phen)+ complexes

Article abstract of DOI:10.1021/ja103112m

We have shown that crystals of the highly emissive copper(I) compounds [Cu(POP)(dmp)]tfpb, [Cu(xantphos)(dmp)]tfpb, [Cu(xantphos)(dipp)]tfpb, and [Cu(xantphos)(dipp)]pftpb, (where POP = bis[2-(diphenylphosphino)phenyl]ether; xantphos = 4,5-bis(diphenylphosphino)-9,9-dimethylxanthene; dmp = 2,9-dimethyl-1,10-phenanthroline; dipp = 2,9-diisopropyl-1,10-phenanthroline (dipp); tfpb^- = tetrakis(bis-3,5-trifluoromethylphenylborate); and pftpb = tetrakis(pentfluorophenyl)borate) are oxygen gas sensors. The sensing ability correlates with the amount of void space calculated from the crystal structures. The compounds exhibit linear Stern-Volmer plots with reproducible K_SV constants from sample to sample; these results reinforce the observations that the sensing materials are crystalline and the sensing sites are homogeneous within the crystals. The long lifetime (～30 μs), high emission quantum yield (β = 0.66), appreciable K_SV value (5.65), and very rapid response time (51 ms for the 95% return constant) for [Cu(xantphos)(dmp)]tfpb are significantly better than those for the [Cu(NN) ₂]tfpb complexes studied previously and compare favorably with [Ru(4,7-Me2phen)₃](tfpb)₂, (K_SV = 4.76; 4,7-Me₂phen = 4,7-dimethyl-1,10- phenanthroline). The replacement of precious metals (like Ru or Pt) with copper may be technologically significant and the new compounds can be synthesized in one or two steps from commercially available starting materials. The strictly linear Stern-Volmer behavior observed for these systems and the absence of a polymer matrix that might cause variability in sensor to sensor sensitivity may allow a simple single-reference point calibration procedure, an important consideration for an inexpensive onetime limited use sensor that could be mass produced.

Full text of DOI:10.1021/ja103112m

Published on Web 09/20/2010
Oxygen Gas Sensing by Luminescence Quenching in Crystals
of Cu(xantphos)(phen)+ Complexes
Conor S. Smith,† Charles W. Branham,‡ Brian J. Marquardt,‡ and Kent R. Mann*,†
Department of Chemistry, UniVersity of Minnesota, Minneapolis, Minnesota 55455, and Applied
Physics Laboratory, UniVersity of Washington, Seattle, Washington 98105
Received April 13, 2010; E-mail: krmann@umn.edu
Abstract: We have shown that crystals of the highly emissive copper(I) compounds [Cu(POP)(dmp)]tfpb,
[Cu(xantphos)(dmp)]tfpb, [Cu(xantphos)(dipp)]tfpb, and [Cu(xantphos)(dipp)]pftpb, (where POP ) bis[2-
(diphenylphosphino)phenyl]ether; xantphos ) 4,5-bis(diphenylphosphino)-9,9-dimethylxanthene; dmp )
2,9-dimethyl-1,10-phenanthroline; dipp ) 2,9-diisopropyl-1,10-phenanthroline (dipp); tfpb- ) tetrakis(bis-
3,5-trifluoromethylphenylborate); and pftpb ) tetrakis(pentfluorophenyl)borate) are oxygen gas sensors.
The sensing ability correlates with the amount of void space calculated from the crystal structures. The
compounds exhibit linear Stern-Volmer plots with reproducible KSV constants from sample to sample; these
results reinforce the observations that the sensing materials are crystalline and the sensing sites are
homogeneous within the crystals. The long lifetime (∼30 µs), high emission quantum yield (φ ) 0.66),
appreciable KSV value (5.65), and very rapid response time (51 ms for the 95% return constant) for
[Cu(xantphos)(dmp)]tfpb are significantly better than those for the [Cu(NN)2]tfpb complexes studied previously
and compare favorably with [Ru(4,7-Me2phen)3](tfpb)2, (KSV ) 4.76; 4,7-Me2phen ) 4,7-dimethyl-1,10-
phenanthroline). The replacement of precious metals (like Ru or Pt) with copper may be technologically
significant and the new compounds can be synthesized in one or two steps from commercially available
starting materials. The strictly linear Stern-Volmer behavior observed for these systems and the absence
of a polymer matrix that might cause variability in sensor to sensor sensitivity may allow a simple single-
reference point calibration procedure, an important consideration for an inexpensive onetime limited use
sensor that could be mass produced.
Introduction
solid will have the same sensing ability because of the
crystallographic uniformity of the emission sites.
Our group has been working with crystalline compounds1-3
(such as [Ru(phen)3]tfpb2; phen ) 1,10-phenanthroline; tfpb-
) tetrakis(bis-3,5-trifluoromethylphenylborate)) that can sense
oxygen gas by luminescence quenching. These compounds
replace the typical polymer supports4-8 with a crystal lattice
that contains sufficient void space to allow diffusional oxygen
quenching of a long-lived excited state. This type of crystalline
material can mitigate technological problems such as the
degradation of the polymer support matrix or luminophor9
caused by the reactive oxygen species produced in the quenching
events. These materials can also offer the possibility of
reproducibility between sensors i.e. each sample of the pure
More recently we reported10 the synthesis and study of new
oxygen gas sensors that use a complex of Cu(I), an inexpensive,
first row transition metal as the active sensing material to
potentially decrease their price. We showed that crystalline
[Cu(NN)2]X (NN ) 2,9 alkylsubstituted 1,10-phenanthroline,
X- ) BF4-, tfpb-) complexes are photoluminescent oxygen
sensors with high photochemical and thermal stability but they
suffer from low emission quantum yields and relatively low
sensitivity.10 As a follow up of this work, our next goal was to
design new Cu(I) compounds with improved quantum yields
and lifetimes to aid detection and raise the KSV’s (the
Stern-Volmer quenching constant) for better oxygen sensing.
The improved solution phase lifetimes reported for compounds
in the [Cu(POP)(NN)]+ family11 (POP ) bis[2-(diphenylphos-
phino)phenyl]ether; see Scheme 1) suggested that we should
investigate these heteroleptic complexes as next generation
candidates. The initially selected compound [Cu(POP)(dmp)]t-
fpb (1) (dmp ) 2,9-dimethyl-1,10-phenanthroline), exhibits
reasonable oxygen sensitivity but suffered from a relatively rapid
photochemical degradation process. From this starting point we
report the synthesis and study of a second series of new
complexes of the form [Cu(xantphos)(NN)]X (xantphos ) 4,5-
† University of Minnesota.
‡ University of Washington.
(1) McGee, K. A.; Mann, K. R. J. Am. Chem. Soc. 2009, 131, 1896.
(2) McGee, K. A.; Marquardt, B. J.; Mann, K. R. Inorg. Chem. 2008, 47,
9143.
(3) McGee, K. A.; Veltkamp, D. J.; Marquardt, B. J.; Mann, K. R. J. Am.
Chem. Soc. 2007, 129, 15092.
(4) Demas, J. N.; DeGraff, B. A.; Coleman, P. B. Anal. Chem. 1999, 71,
793A.
(5) Krenske, D.; Abdo, S.; Van Damme, H.; Cruz, M.; Fripiat, J. J. J.
Phys. Chem. 1980, 84, 2447.
(6) Leventis, N.; Elder, I. A.; Rolison, D. R.; Anderson, M. L.; Merzbacher,
C. I. Chem. Mater. 1999, 11, 2837.
(7) Lin, C.-T.; Sutin, N. J. Phys. Chem. 1976, 80, 97.
(8) Shi, L.; Li, B.; Yue, S.; Fan, D. Sens. Actuators, B 2009, B137, 386.
(9) Carraway, E. R.; Demas, J. N.; DeGraff, B. A.; Bacon, J. R. Anal.
Chem. 1991, 63, 337.
(10) Smith, C. S.; Mann, K. R. Chem. Mater. 2009, 21, 5042.
(11) Cuttell, D. G.; Kuang, S.-M.; Fanwick, P. E.; McMillin, D. R.; Walton,
R. A. J. Am. Chem. Soc. 2002, 124, 6.
9
10.1021/ja103112m  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 14079–14085 14079

A R T I C L E S
Smith et al.
Scheme 1. Ligands Utilized in This Study
[Cu(xantphos)(dipp)]tfpb (3b), and [Cu(xantphos)(dipp)]pftpb (3c)
are given below.
[Cu(POP)(dmp)]tfpb (1). Natfpb (0.0797 g, 0.09 mmol) was
added to a methanol (30 mL) solution of [Cu(POP)(dmp)]BF4
(0.070 g, 0.078 mmol). The yellow solid was precipitated by
addition of water, filtered, and dried under vacuum yielding 0.083
1
g (64% yield). NMR CD2Cl2 δ 8.28 (d, 2H J ) 8.4), 7.804 (s,
2H), 7.717 (broad s, 8H), 7.553 (broad s, 4H), 7.518 (d, 2H J )
8.4), 7.343 (m, 2H), 7.242 (m, 10H), 6.996 (m, 16H), 2.437 (s,
6H) 19F NMR (CD2Cl2) δ -63.521 (s, 24F) HRESIMS (M+): calcd
for C50H40CuN2OP2, 809.1906; found 809.1995.
[Cu(xantphos)(dmp)]BF4 (2a). [Cu(CH3CN)4]BF4 (0.054 g,
0.173 mmol) was added to a CH2Cl2 (20 mL) solution of xantphos
(0.100 g, 0.173 mmol). Addition of dmp (0.036 g, 0.173 mmol)
yielded a clear yellow solution. A bright yellow solid was
precipitated upon addition of diethyl ether, filtered, and dried under
vacuum yielding 0.1128 g (70% yield). 1NMR CD2Cl2 δ 8.300 (d,
2H J ) 8.1), 7.825 (s, 2H), 7.707 (dd, 2H J ) 7.8,1.5), 7.515 (d,
2H J ) 8.4), 7.253 (m, 6H), 7.055 (m, 16H), 6.938 (m, 4H), 2.291
(s, 6H), 1.775 (s, 6H) 19F NMR (CD2Cl2) δ -154.125 (s, 4F).
[Cu(xantphos)(dmp)]tfpb (2b). Natfpb (0.0802 g, 0.09 mmol)
was added to a methanol (20 mL) solution of [Cu(xantphos)(d-
mp)]BF4 (0.0680 g, 0.073 mmol). The bright yellow solid was
precipitated by addition of water, filtered, and dried under vacuum
yielding 0.0746 g (59% yield). 1NMR CD2Cl2 δ 8.288 (d, 2H, J )
8.4), 7.760 (s, 2H), 7.718 (bs, 8H), 7.669 (dd, 2H, J ) 7.8, 1.5),
7.553 (bs, 4H), 7.458 (d, 2H, J ) 8.4), 7.214 (m, 6H), 7.017 (m,
16H), 6.902 (m, 2H), 2.255 (s, 6H), 1.738 (s, 6H) 19F NMR
(CD2Cl2) δ -63.520 (s, 24 F) HRESIMS (M+): calcd for
C53H44CuN2OP2, 849.2219; found 849.2278.
[Cu(xantphos)(dipp)]BF4 (3a). [Cu(CH3CN)4]BF4 (0.054 g,
0.173 mmol) was added to a CH2Cl2 (20 mL) solution of xantphos
(0.100 g, 0.173 mmol). Addition of dipp (0.046 g, 0.173 mmol)
yielded a clear orange-yellow solution. A bright yellow solid was
precipitated upon addition of diethyl ether, filtered, and dried under
vacuum yielding 0.1380 g (80% yield). 1NMR CD2Cl2 δ 8.388 (d,
1H J ) 8.7), 8.243 (d, 1H J ) 8.1), 7.796 (d, 2H J ) 8.4), 7.700
(m, 4H), 7.581 (d, 1H J ) 8.4), 7.519 (d, 1H J ) 8.7), 7.437 (t,
2H J ) 7.2), 7.250 (m, 12H), 7.046 (m, 4H), 6.793 (m, 4H), 6.708
(m, 4H), 3.697 (septet, 1H J ) 6.9), 2.666 (septet, 1H J ) 7.2),
2.005 (s, 3H), 1.470 (s, 3H), 1.071 (d, 6H J ) 6.9), 0.1825 (d, 6H
J ) 7.2) 19F NMR (CD2Cl2) δ -154.077 (s, 4F).
bis(diphenylphosphino)-9,9-dimethylxanthene) that exihibit higher
photochemical stability, very high emission quantum yields and
oxygen responsivity with rapid response times.
Experimental Section
General Considerations. Cu(BF4)2 ·xH2O, bis[2-(diphenylphos-
phino)phenyl]ether (POP), and 4,5-bis(diphenylphosphino)-9,9-
dimethylxanthene (xantphos) were purchased from Aldrich Chemi-
cal Co. and used as received. The solvents (dichloromethane,
methanol, diethyl ether, acetonitrile, hexanes, and toluene) were
obtained from commercial sources and either used without further
purification or after drying with molecular sieves. The ligand 2,9-
dimethyl-1,10-phenanthroline (dmp) was purchased from Aldrich,
and potassium tetrakis(pentafluorophenyl) borate (Kpftpb) was
purchased from Boulder Scientific Company. 2,9-diisopropyl-1,10-
phenanthroline (dipp) was synthesized by a literature procedure.12
Sodium(tetrakis(bis-3,5-trifluoromethyl(phenylborate))) (Natfpb)
was available from a previous study.13 [Cu(CH3CN)4]BF4 was
synthesized by a literature procedure.14
A 300 MHz Varian Unity NMR spectrometer was utilized to
1
obtain the H and 19F NMR spectra. Chemical shifts are reported
in units of ppm with an external reference to the residual proton
resonance in deuterated dichloromethane or chloroform for 1H NMR
spectra and the internal CFCl3 reference for the 19F NMR. High-
resolution mass spectrometry was carried out on a Bruker BioTOF
II mass spectrometer.
[Cu(xantphos)(dipp)]tfpb (3b). Natfpb (0.071 g, 0.08 mmol)
was added to a methanol (20 mL) solution of [Cu(xantphos)-
(dipp)]BF4 (0.060 g, 0.060 mmol). The yellow solid was precipitated
by addition of water, filtered, and dried under vacuum yielding
Synthesis and Characterization. The [Cu(POP)(dmp)]BF4
compound used in this study was synthesized as previously
reported11 by adding dmp to a solution of [Cu(CH3CN)4]BF4 and
POP in CH2Cl2 followed by precipitation of the salt with diethyl
ether. The new BF4- salts containing the xantphos ligand were made
by an analogous route. The metatheses to the tfpb- and pftpb- salts
were carried out by adding a 1.2:1 ratio of Natfpb or Kpftpb to
[Cu(diphosphine)(phen-derivative)]BF4 in MeOH. Precipitation with
water yielded pure compounds. The complete conversion by this
metathesis route was confirmed by the presence of signals for tfpb-
1
0.0457 g (46% yield). NMR CD2Cl2 δ 8.345 (d, 1H J ) 8.7),
8.210 (d, 1H J ) 8.1), 7.715 (broad s, 8H), 7.701(m, 4H), 7.566
(d, 1H J ) 8.4), 7.552 (broad s, 4H), 7.497 (d, 1H J ) 8.4), 7.437
(t, 2H J ) 6.9), 7.304 (m, 4H), 7.224 (m, 6H), 7.035 (m, 4H),
6.785 (t, 4H J ) 7.2), 6.706 (m, 4H), 3.699 (septet, 1H J ) 6.6),
2.670 (septet, 1H J ) 6.3), 2.001 (s, 3H), 1.256 (s, 3H), 1.067 (d,
6H J ) 6.6), 0.180 (d, 6H J ) 6.9) 19F NMR (CD2Cl2) δ -63.423
(s, 24F) HRESIMS (M+): calcd for C57H52CuN2OP2, 905.2845;
found 905.2842.
[Cu(xantphos)(dipp)]pftpb (3c). Kpftpb (0.0071 g, 0.01 mmol)
was added to a methanol (5 mL) solution of 3a (0.0070 g, 0.007
mmol). The light yellow solid was precipitated by addition of water,
filtered, and dried under vacuum yielding 0.010 g (90% yield).
1NMR CD2Cl2 δ 8.349 (d, 1H J ) 8.4), 8.214 (d, 1H J ) 8.4),
7.697 (m, 4H), 7.568 (d, 1H J ) 8.7), 7.500 (d, 1H J ) 8.7), 7.436
(m, 2H), 7.306 (m, 4H), 7.225 (m, 6H), 7.035 (m, 4H), 6.760 (m,
8H), 3.700 (septet, 1H J ) 6.3), 2.670 (septet, 1H J ) 6.9), 2.003
(s, 3H), 1.470 (s, 3H), 1.068 (d, 6H J ) 6.9), -0.180 (d, 6H J )
6.9) 19F NMR (CD2Cl2) δ -133.844 (s, 4F), -164.434 (t, 8F J )
20.3), -168.265 (t, 8F J ) 20.3) HRESIMS (M+): calcd for
C57H52CuN2OP2, 905.2845; found 905.2837.
or pftpb- and the absence of signals for BF4 in the 19F NMR of
-
1
the products. The unexpected complexity of the H NMR results
obtained for the [Cu(xantphos)(dipp)]X (X ) BF4, tfpb, and pftpb)
compounds is discussed in the Supporting Information. The
syntheses of [Cu(POP)(dmp)]tfpb (1), [Cu(xantphos)(dmp)]BF4
(2a), [Cu(xantphos)(dmp)]tfpb (2b), [Cu(xantphos)(dipp)]BF4 (3a),
(12) Dietrich-Buchecker, C. O.; Marnot, P. A.; Sauvage, J. P. Tetrahedron
Lett. 1982, 23, 5291.
(13) Exstrom, C. L.; Britton, D.; Mann, K. R.; Hill, M. G.; Miskowski,
V. M.; Schaefer, W. P.; Gray, H. B.; Lamanna, W. M. Inorg. Chem.
1996, 35, 549.
Single Crystal X-ray Crystallography. The crystal structures
of compounds 1, 2b, 3b, and 3c were determined in this study. All
(14) Dietrich-Buchecker, C.; Sauvage, J. P.; Kern, J. M. J. Am. Chem. Soc.
1989, 111, 7791.
9
14080 J. AM. CHEM. SOC. VOL. 132, NO. 40, 2010

Oxygen Gas Sensing
A R T I C L E S
Table 1. Crystallographic Data and Refinement Parameters
compound
1
2b
3b
3c
empirical formula
crystal color, morphology
crystal system
space group
C82H52BCuF24N2OP2
yellow, block
monoclinic
P21
C85H56BCuF24N2OP2
yellow, plate
triclinic
C89H64BCuF24N2OP2
yellow, block
triclinic
C81H52BCuF20N2OP2
yellow, plate
triclinic
j
j
j
P1
P1
P1
a, Å
b, Å
c, Å
19.127(13)
20.190(13)
19.788(14)
90
93.402(23)
90
7628(7)
4
1673.55
1.457
15.289(2)
16.539(2)
17.019(2)
102.685(2)
108.556(2)
97.253(2)
3889.9(9)
2
1713.61
1.463
123(2)
0.428
20.755(3)
21.087(3)
39.817(6)
102.684(2)
90.297(3)
106.636(2)
16246(4)
11.0860(12)
17.2010(18)
19.803(2)
76.694(2)
73.782(2)
88.210(2)
3526.4(6)
2
1585.54
1.493
173(2)
0.458
R, deg
ꢂ, deg
γ, deg
volume, Å3
Z
8
formula weight, g mol-1
density (calculated), g cm-3
temperature, K
absorption coefficient, mm-1
F(000)
1769.71
1.447
123(2)
0.413
123(2)
0.435
3384
1736
7200
1608
θ range, deg
1.03-25.07
-22 e h e 22
-24 e h e 24
-23 e l e 23
56 974
1.29-25.08
-18 e h e 18
-19 e k e 19
-20 e k e 20
38 903
1.04-25.12
-24 e k e 24
-25 e k e 25
-47 e l e 47
160 490
1.10-25.05
-13 e h e 13
-20 e l e 20
-23 e l e 23
34 822
index ranges
reflections collected
independent reflections
weighting factorsaa,b
max, min transmission
data/restraints/parameters
R1, wR2 [I > 2σ(I)]
R1, wR2 (all data)
26 506
0.0609
13 776
57 495
12 443
0.0461, 2.9848
0.94, 0.820337
13776/0/1089
0.0380, 0.0920
0.0578, 0.1052
1.015
0.1469, 32.590305
0.7452, 0.6540
57495/0/4395
0.0925, 0.2414
0.1882, 0.3005
1.057
0.0531, 2.1772
0.995, 0.872
12443/0/979
0.0455, 0.1051
0.0903, 0.1279
1.063
0.7452, 0.6435
26506/97/2113
0.0532, 0.1183
0.0949, 0.1424
1.006
GOF
largest diff. peak, hole, eÅ-3
1.443, -0.479
0.426, -0.399
2.635, -0.789
0.722, -0.519
a
2
2
w ) [σ2(Fo ) + (aP)2 + (bP)]-1, where P ) (Fo + 2Fc2)/3.
crystals were grown by slow evaporation of MeOH/H2O solutions.
Relevant crystallographic data are shown in Table 1, and details
of the structure determinations are given in the Supporting
Information. All of the data for these structure determinations were
collected at the X-ray crystallographic laboratory (Department of
Chemistry, University of Minnesota). Single crystals were attached
to glass fibers and mounted on a Bruker SMART Platform CCD
(tfpb- structures) or Siemens SMART Platform CCD (pftpb-
structure) for data collection at 123 K (tfpb- structures) or 173 K
(pftpb- structure) using graphite monochromated Mo KR radiation
(λ ) 0.71073 Å). An initial set of cell constants was calculated
from reflections harvested from three sets of 20 frames oriented
such that orthogonal wedges of reciprocal space were surveyed.
Final cell constants were calculated from at least 12443 strong
reflections from the actual data collection. Data were collected to
the extent of (1.5-2.0) hemispheres at a resolution of 0.84 Å using
ꢀ-scans. For all structures, the intensity data were corrected for
absorption and decay using SADABS.15 Solution and refinement
were performed utilizing SHELXTL-V6.12.16 The solvent acces-
sible void space fraction was calculated for these structures using
PLATON/VOID.17,18 A description of how these calculations are
performed has previously been presented.10 The refinement for
compound 3b was not completed to convergence due to the extreme
size of the unit cell that results in over 4000 parameters; however,
this level of refinement is adequate to describe the connectivity
and for void space calculations of sufficient accuracy. Further details
concerning the refinement of this structure are given in the
Supporting Information.
rod, followed by addition of a small drop of MeOH to partially
dissolve the sample. Slow evaporation of the MeOH resulted in an
adherent crystalline film. Additionally, at least one single crystal
of each compound was tested for oxygen sensing to confirm that
the films used for oxygen sensing experiments and the single
crystals used for X-ray diffraction gave similar results. In all of
these cases, the prepared samples were then placed into the
apparatus based on a 1/4′′ diameter cross swage tube fitting. A 375
nm LED source filtered through an interference filter is used for
excitation through the center fiber of a “six around one” bifurcated
fiber optic probe. The emission was collected through the six fiber
channel and sent into an Ocean Optics USB-2000 spectrometer.
Various mole fractions of oxygen in nitrogen were prepared with
a gas handling system composed of mass flow controllers. The
spectrometer, mass flow controllers, temperature and pressure
monitors were interfaced to a computer to allow control of the
unattended data acquisition via a custom Labview program.
The emission data collected were analyzed with a spreadsheet
written in Microsoft Excel. Data from three cycles of spectra
obtained at approximately 0.0, 0.10, 0.21, 0.25, 0.40, 0.50, 0.65,
0.80, 0.90, and 1.0 mol fractions (ꢁO ) of O2 in N2 were used for
2
the Stern-Volmer plots. The oxygen concentrations for the
Stern-Volmer plots (I0/I ) 1 + KSVꢁO ) reported here are expressed
2
in terms of mole fraction at the standard atmospheric pressure of
Minneapolis-St. Paul, MN of 0.97 atm. The unitless KSV values
quoted herein may be converted to partial pressure units (atm-1
)
by dividing by 0.97 atm. The acquisitions were at room temperature,
approximately 21.8 ( 1 °C. Exact gas mole fractions were
calculated by referencing the feedback voltage of the mass flow
controllers to a calibration previously performed. The emission
intensity was integrated across the entire peak and divided by the
integrated LED intensity for each spectrum to give I0 (intensity
under nitrogen) and I (intensity at a given O2 mixture). I0/I was
plotted vs oxygen mole fraction in nitrogen to yield a Stern-Volmer
plot; a linear regression model was used to calculate the slope. The
O2 Sensing Apparatus and Procedure. The crystalline films
used for gas sensing were prepared by placing small (∼1 mg)
amounts of crystalline solid on the end of a 1/4′′ diameter brass
(15) Blessing, R. H. Acta Crystallogr., A 1995, A51, 33.
(16) SHELXTL 6.12; Bruker AXS: Madison, WI, 2001.
(17) Spek, A. L. J. Appl. Crystallogr. 2003, 36, 7.
(18) Spek, A. L. PLATON, A Multipurpose Crystallographic Tool; Utrecht
University: Utrecht, The Netherlands, 2005.
9
J. AM. CHEM. SOC. VOL. 132, NO. 40, 2010 14081

A R T I C L E S
Smith et al.
Figure 1. (Left) Fiber optic probe (actual size) used for dissolved O2 sensing. (Right) Tip of fiber optic with compound at 50× magnification.
Stern-Volmer slopes from multiple runs taken on separately
prepared samples were averaged and the standard deviation was
calculated.
and the normalized intensity [(Ip - Ip0)/(I∞ - Ip0)] vs t was plotted
for compounds 2b, 3b, and 3c.
Dissolved Oxygen Sensing Procedure. A crystalline film of the
complex was deposited on the end of a fiber optic tip as shown in
Figure 1. The tip is inserted into deionized water contained in a
round-bottom flask. Illumination was gated so that the samples were
only irradiated while data were being collected, not during purging
time. The water is purged with mol fractions of O2 in N2 of 0.0,
0.10, 0.20, 0.50, and 1.0. Three cycles of spectra were taken. The
resulting data were corrected for stray light, fit to the Stern-Volmer
equation, averaged, and the standard deviation calculated. Because
these data were collected in Seattle, WA which is at a significantly
lower elevation (i.e., nominally at sea level), the unitless KSV value
that resulted was corrected so it could be directly compared with
the KSV’s calculated from data collected in Minneapolis, MN.
Solid-State Photophysical Measurements. The absolute solid-
state quantum yields of the compounds were obtained by a
Results and Discussion
Molecular Oxygen Sensing. As discussed in the introduction
we selected [Cu(POP)(dmp)]tfpb (1), as an initial candidate for
an improved Cu(I) based oxygen gas sensor. A sample of this
compound was synthesized in good yield, crystallized from
methanol/water, and found to be pure via NMR and mass
spectrometry. Crystalline films of 1 were found to sense O2,
but unfortunately the performance of these films significantly
degraded in both emission intensity and in O2 sensing ability
during the preliminary studies (∼40% after 12.5 h alternating
between 300 s exposures to N2 and O2). When not exposed to
the LED, no degradation of performance occurs under either
N2 or O2 consistent with a photoinduced origin of the degrada-
tion mode. To minimize this degradation during our measure-
ments, the LED was gated off while waiting for the cell to
equilibrate to the desired O2 percentage, and was only gated on
for the approximately 1 s required to obtain the luminescence
spectrum. Using this methodology, a KSV of 3.60(2) was found
for compound 1, which was about a 10-fold improvement over
our previous best copper centered crystalline oxygen sensor.10
Due to this encouraging result, we commenced additional
compound design, synthesis and testing to minimize the
performance degradation.
The previous demonstration and calculation that significant
distortions in the coordination sphere angles of [Cu(dppe)-
(phen)]PF6 (dppe ) 1,2-bis(diphenylphosphino)ethane) occur
in the excited state,21 suggested that a similar distortion available
to the POP ligand complex might occur and damage the surface
crystallinity of the sensor. In an attempt to restrict this distortion,
we replaced POP with the more rigid 4,5-bis(diphenylphos-
phino)-9,9-dimethylxanthene (xantphos) ligand. A number of
new compounds (2a-3c) containing xantphos were synthesized
in good yield, crystallized from methanol/water, and found to
be pure via NMR and mass spectrometry. The subsequent
screening of these compounds for stability and O2 sensing
showed a significant amount of quenching by oxygen for
compounds 2b and 3b. Luminescence spectra for a crystalline
film of 2b under several different oxygen concentrations in
nitrogen are shown in Figure 2. Additionally the stability of all
of the xantphos compounds under N2 and O2 is much improved
versus the POP compound herein reported.22 For example, we
observed an 8% decrease in intensity and sensing ability for
2b after 12.5 h of continuous illumination alternating between
nitrogen and oxygen in 300 s increments. An additional
experiment showed that the small amount of degradation
occurred during the oxygen exposures. A Stern-Volmer study
of the O2 sensing for 2b and 3b gave high KSV values of 5.65(8)
modification of the method described by Wrighton et al.19
A
Fluorilon scattering target was used instead of MgO. The methods
used to obtain and correct the data for detector response have been
described previously.3 Corrected spectra of 1, 2b, 3b and 3c are
shown in the Supporting Information. Solid-state emission lifetimes
were obtained with an in-house constructed circuit that pulses a
400 nm LED for excitation of the sample. The light was conducted
through one leg of a bifurcated fiber optic probe to the sample
housed in a sample compartment identical to the one used for the
Stern-Volmer measurements. The emission decay and scattered
light were collected by the other fiber optic leg, filtered through a
cut off filter to remove the scattered excitation light and conducted
to a Hamamatzu R928 PMT. The PMT signal was digitized by a
sampling digital oscilloscope (Phillips PM 3323) interfaced to a
computer running a LabView program similar to the one used for
control of the Stern-Volmer data collection system. The lifetimes
for compounds 1, 2b, and 3b were modeled with a single
exponential over 6 natural logarithm units; compound 3c required
a double exponential fit. More details concerning the lifetime data
acquisition and processing have been presented previously.3
The pressure jump experiments were performed utilizing the
same swage lock cross cell as described above. In this case, the
input gas stream was from a compressed air tank and the pressure
inside the cell was controlled with the regulator on the air tank.
After pressurization, the system was tested to be leak free with the
inlet and outlet valves closed. The time constant for pressure leaking
out of the cell back to 1 atm ambient was measured by pressurizing
the cell to 2.28 atm, closing the inlet valve and then monitoring
the rate of the pressure drop back to 1 atm with a computer
interfaced pressure transducer20 when the outlet valve was opened.
The full return time constant was approximately 40 ms, with a 50%
return in 15 ms. A similar method was used to measure the time
evolution of the emission spectrum except the OceanOptics
spectrometer was used to collect emission spectra before and during
the pressure release. Spectra could be collected to disk every 7
ms. The total emission intensity for each spectrum was calculated
(21) Vorontsov, I. I.; Graber, T.; Kovalevsky, A. Y.; Novozhilova, I. V.;
Gembicky, M.; Chen, Y.-S.; Coppens, P. J. Am. Chem. Soc. 2009,
131, 6566.
(19) Wrighton, M. S.; Ginley, D. S.; Morse, D. L. J. Phys. Chem. 1974,
78, 2229.
(20) Logger Pro software; Vernier Software & Technology: Beaverton, OR.
(22) Further discussion of a preliminary investigation of the performance
degradation is included in the Supporting Information.
9
14082 J. AM. CHEM. SOC. VOL. 132, NO. 40, 2010

Oxygen Gas Sensing
A R T I C L E S
-
Due to this inconsistency further studies with the BF4 salts
were not performed.
In contrast the standard deviations of the KSV measurements
for 2b, 3b, and 3c are much smaller (see Table 2). They
encompass three determinations each of three different samples
prepared and measured in a similar manner over a period of up
to 5 days. A few samples that were stored in air for several
weeks after casting were measured and found to have
Stern-Volmer plots virtually identical to those measured
directly after film casting. Samples of some of the compounds
that were stored in laboratory air for up to a year gave sensing
determinations virtually identical to samples prepared and used
immediately. The identical sensing abilities of single crystals
of each of the compounds compared with the respective
crystalline film measurements show that the films have the same
crystal structures. For compound 2b three different samples were
illuminated with a 375 nm LED for 300 s under O2 prior to
beginning the gated KSV determinations. These three samples
yielded a KSV of 5.35(2) which is lower than the 5.65(8)
measured for the nonirradiated samples, consistent with a small
amount of degradation. The small deviation in KSV values
between the three irradiated samples indicates that the degrada-
tion is consistent and reproducible. Some additional experiments
to unravel the nature of the performance degradation have been
conducted but are inconclusive. A brief discussion of these
results is given in the Supporting Information.
Figure 2. Uncorrected emission spectra for [Cu(xantphos)(dmp)]tfpb (2b)
excited at 375 nm under varying concentrations of O2 in N2. The spectra
under pure nitrogen and pure oxygen are shown in blue and red, respectively.
Mole fraction of O2 in N2 from top to bottom is 0, 0.095, 0.241, 0.511, and
1.0.
One of the possible applications of oxygen sensors of this
type is in sensing dissolved oxygen in water. To test the efficacy
of dissolved oxygen sensing a methanol film of 2b was deposited
on the tip of a fiber optic (Figure 1) and oxygen sensing studies
were performed on dissolved oxygen dilutions in deionized
water. The KSV value of 5.67(5) was determined over three
cycles and is identical within experimental error to the value
found for gaseous oxygen sensing with the same compound,
5.65(8). This indicates that compounds of this type could readily
be utilized for dissolved O2 sensing.
The gas phase time response of each xantphos Cu(I) sensor
was measured by pressure jump experiments. Compounds 2b
and 3b were of most interest and are plotted in Figure 4 for
comparison. A very fast response (95% return at 0.051(4) s and
50% return at 0.022(6) s) for 2b was observed, while compound
3b had a slower response time (95% return at 0.71(2) s and
50% return at 0.044(3) s). These results are consistent with the
presence of more void space in compound 2b than 3b (vide
infra), which leads to more rapid oxygen diffusion in the crystal
structure of compound 2b.
Photophysics. Previously the photophysics of cations of the
form [Cu(POP)(NN)]+ in dichloromethane at room temperature
have been reported.11,23 We adopt the previous assignment
(MLCT Cu(I) f phen) for the strong absorption and subsequent
emission observed for these compounds.23 The properties of
the [Cu(POP)(dmp)]+ cation have been reported as both φ )
Figure 3. Stern-Volmer plots ((I0/I) vs mole fraction of oxygen in
nitrogen) for typical samples of 1 (green diamonds), 2b (blue squares), 3b
(red circles), and 3c (orange ×); the lines are the appropriate linear regression
fits (R2 ) 0.9994, 0.9995, 0.9993, and 0.9996, respectively). Three cycles
were collected over a 15 h period for each sample; one cycle is displayed
for clarity.
and 3.41(9), respectively. Representative linear KSV plots for 1,
2b, 3b and 3c are shown in Figure 3. A much smaller KSV of
0.153(2) was found for compound 3c, a perfluoro tetraphenyl
borate salt with the same cation as 3b. We also found that the
precursor xantphos BF4- salts are oxygen sensors but they gave
inconsistent results from sample to sample, including crystals
taken from the same crystallization vial, leading to KSV values
with high standard deviations (0.7(4) for 2a and 4(2) for 3a).
Table 2. Summary of Crystallographic, Photophysical and Oxygen Sensing Data
a
KSV
compounds
λmax (nm)b
φ (N2)c
φ (O2)d
τ0 (µs, N2)e
τ (µs, O2)f
% void spaceg
[Cu(POP)(dmp)]tfpb (1)
3.60(2)
5.65(8)
3.41(9)
0.153(2)
517
540
513
514
0.88(9)
0.66(5)
0.95(5)
0.47(4)
0.19(2)
0.084(3)
0.22(1)
0.31(3)
26.0
30.2
38.5
19.5h
6.4
5.0
4.6%
3.3%
2.0%
2.0%
[Cu(xantphos)(dmp)]tfpb (2b)
[Cu(xantphos)(dipp)]tfpb (3b)
[Cu(xantphos)(dipp)]pftpb (3c)
9.6
15.6h
a Estimated limit of detection is 5 × 10-3
.
b Corrected solid-state emission peak wavelength. c Solid-state emission quantum yield under nitrogen.
d Solid-state emission quantum yield under oxygen. e Solid-state emission lifetime under nitrogen. f Solid-state emission lifetime under oxygen.
g Calculated percentage of solvent accessible void space from X-ray crystal structure. h weighted τm based on biexponential fit, see Supporting
Information for details.
9
J. AM. CHEM. SOC. VOL. 132, NO. 40, 2010 14083

A R T I C L E S
Smith et al.
Figure 4. Normalized data from pressure jump experiment performed on
[Cu(xantphos)(dmp)]tfpb (blue squares) and [Cu(xantphos)(dipp)]tfpb (red
circles) with 375 nm LED excitations. 95% return at 0.051(4)s and 50%
return at 0.022(6)s for [Cu(xantphos)(dmp)]tfpb and 95% return at 0.71(2)s
and 50% return at 0.044(3)s for [Cu(xantphos)(dipp)]tfpb. The response of
a pressure transducer is shown as a black line with 95% return at 0.035 s
and 50% return at 0.015 s.
0.15, τ ) 14.3 µs11,23 and φ ) 0.09, τ ) 4.62 µs24 in a
dichloromethane solution at room temperature and can be
compared to the solid state measurements for 1 (φ ) 0.88(9),
τ ) 26.0 µs). The larger quantum yield and longer lifetime
observed for 1 are consistent with a significant decrease in the
nonradiative decay rates for the luminescence of the cation in
the solid phase.
In addition to a large KSV value, useful oxygen sensors based
on LED excitation and photodiode detection must have a high
emission quantum yield granted by favorable photophysical
properties. Compounds 1, 2b, 3b, and 3c studied herein had
excellent quantum yields (Table 2). In particular compound 3b
has an extremely high quantum yield under N2, with a value
near unity (0.95(5)). Compared to bis-phen systems these
quantum yields under N2 are over 5 times higher10 and would
allow the construction of usable sensors with smaller amounts
of the sensor compound and less expensive and sophisticated
detectors and excitation sources.
In a previous paper we discussed the equation that determines
the magnitude of the KSV constant in terms of the lifetime and
oxygen permeability.10 For a given oxygen permeability (which
is controlled by the solid state structure) the KSV should be
directly proportional to the lifetime so increasing it is of
paramount importance. The solid-state luminescence lifetimes
(Table 1) of compounds 1, 2b, 3b, and 3c were determined as
previously described and were found to be approximately ten
times larger than those of comparable bis-phen systems.10 For
example, the best sensor compound (2b) has a lifetime of 30.2
µs under a nitrogen atmosphere. This 10-fold increase in lifetime
is consistent with the approximately 10-fold increase in KSV
observed for compounds 1, 2b, and 3b; compounds with
significant void space. A cursory examination of the effect of
oxygen on the emission lifetime was also performed (Table 1).
Representative plots of ln(I0/I) vs time for compound 2b under
pure nitrogen and pure oxygen are shown in Figure 5. All four
compounds exhibit significantly shortened lifetimes under pure
Figure 5. Plots of the natural log of the normalized emission intensity
(ln(I0/I)) vs time for crystals of [Cu(xantphos)(dmp)]tfpb (2b) under nitrogen
(blue circles, τ ) 30.2 µs) and oxygen (red squares, τ ) 5.0 µs).
oxygen, as expected. The “two point” KSV values calculated from
the lifetimes under pure nitrogen and pure oxygen are ap-
proximately 13% lower than those measured by the more
accurate emission intensity quenching method. We believe that
this diminution could result from some static quenching or from
an increase in the light penetration depth at the longer
wavelength (400 vs 385 nm) used for the lifetime experiments.
Further experiments beyond the scope of the studies reported
here will be needed to address this mechanistic detail.
Crystal Void Space and Oxygen Sensing. As outlined in the
Experimental Section the solvent accessible void space (ex-
pressed as a fraction of the unit cell volume) was calculated
for each crystal structure and is given in Table 1. Consistent
with our previous hypothesis10 that void space is a necessary
but not sufficient condition for oxygen sensing in this type of
compound, all four structures contain some void space and are
oxygen sensors but the variations in the observed KSV again
suggest that the quality of the void space is also an important
factor in determining the relative sensing ability. Closer
examination of the crystal structure of the best sensor compound
2b (illustrated in Figure 6) show that the void cavities, although
not clear channels, are lined with and separated by highly mobile
CF3 groups, some of which are still severely disordered even
at temperatures as low as 123 K. Similar situations exist for
the other two sensor compounds 1 and 3b (illustrated in the
Supporting Information), whereas the poorest oxygen sensor (3c)
contains void space restricted to distinct pockets. As in the
previous studies, we suggest that the thermal activation of the
CF3 groups cluttering potential inter void space apertures can
allow enough motion for solution like oxygen diffusion through
the crystal and subsequently enable significant oxygen quench-
ing.10
(23) Kuang, S.-M.; Cuttell, D. G.; McMillin, D. R.; Fanwick, P. E.; Walton,
R. A. Inorg. Chem. 2002, 41, 3313.
Conclusions
(24) Armaroli, N.; Accorsi, G.; Bergamini, G.; Ceroni, P.; Holler, M.;
Moudam, O.; Duhayon, C.; Delavaux-Nicot, B.; Nierengarten, J.-F.
Inorg. Chim. Acta 2007, 360, 1032.
We have demonstrated that crystals of the emissive hetero-
leptic copper compounds 1, 2b, 3b, and 3c can sense molecular
9
14084 J. AM. CHEM. SOC. VOL. 132, NO. 40, 2010

Oxygen Gas Sensing
A R T I C L E S
Figure 6. Depiction of the calculated void space as red space filling spheres for compound 2b. The fluorines of the disordered CF3 groups that abut and
clutter the void cavities are shown in green.
oxygen that is either in the gas or solution phases. The sensing
ability correlates with the amount of void space calculated from
the crystal structures. The linear Stern-Volmer plots with
reproducible KSV constants observed from sample to sample
reinforce the observations that the sensing materials are crystal-
line and the sensing sites are homogeneous within the crystals.
The long lifetimes, high quantum yields, appreciable KSV values
and very rapid response times (∼30 µs, 0.66, 5.65, and 51 ms
95% return constant, respectively, for compound 2b) for these
compounds are significantly better than our best previously
studied copper O2 sensor [Cu(dipp)2]tfpb (with a KSV value of
0.299).10 The best of the new compounds reported here is more
sensitive than the best crystalline ruthenium system ([Ru(4,7-
Me2phen)3](tfpb)2, 4.76KSV) we previously reported.3 Perhaps
of additional technological importance, copper is significantly
less expensive than ruthenium and compound 2b can be
synthesized from commercially available starting materials in
a single step. As we have previously pointed out, the strictly
linear Stern-Volmer behavior observed for this system and the
absence of a polymer matrix that causes variability in sensor to
sensor sensitivity allow, in principle, a simple single reference
point calibration procedure, an important consideration for an
inexpensive onetime limited use sensor that would be mass
produced.
would be introduced due to degradation. This approximate
calculation uses the 8% intensity decrease over 12.5 h of
continuous irradiation with alternating nitrogen and pure oxygen
discussed earlier. The high reproducibility of the degradation
could allow it to be built into the sensor’s data handling model
to further extend the useable life of the sensor by at least another
factor of 10 (50 000 measurements). For applications requiring
continuous monitoring at ten measurements per minute, this
would give a useable sensor life of approximately 3.5 days. If
the sensed gas was lower in oxygen concentration than air, even
more samples and an even longer usability parameter could be
obtained within the 1% error limit. Future studies are in progress
to obtain even more stable Cu containing materials that would
retain the other desirable sensor characteristics discussed here.
Acknowledgment. This work was supported by the Center for
Analytical Chemistry at the University of Washington and the
Initiative for Renewable Energy and the Environment at the
University of Minnesota. The authors thank Dr. Victor G. Young
Jr. of the University of Minnesota X-ray crystallography laboratory
and Dr. Kari A. McGee for assistance with some of the crystal
structures.
Supporting Information Available: A discussion of the
lifetime modeling for compound 3c, crystal structure refinement
details, a discussion of the unusual 1H NMR of the dipp
complexes, a cursory investigation of the performance degrada-
tion, data from the pressure jump experiment for compound 3c,
corrected solid-state luminescence spectra for 1, 2b, 3b, and
3c, and the void space diagrams for 1 and 3b. This material is
available free of charge via the Internet at http://pubs.acs.org.
Of final concern is the degradation in performance exhibited
by these compounds that might limit their commercial usefulness
for long-term sensing applications. For short-term disposable
applications, the degradation may not be a significant issue. For
example, if an O2 sensing device was built to utilize an LED
with similar intensity to those used for our sensing data
collection, we have calculated that a 1 s exposure time for a
single sensing event near the ambient oxygen partial pressure
in air would allow over 5000 measurements before a 1% error
JA103112M
9
J. AM. CHEM. SOC. VOL. 132, NO. 40, 2010 14085