Improving oxygen sensing performance via inner-molecular π-π stacking in a series of phosphorescent Cu(I) complexes

Article abstract of DOI:10.1016/j.saa.2020.118537

In this paper, six phosphorescent Cu(I) complexes with three diamine ligands and two phosphorous ligands were prepared. Detailed discussion was performed on these complexes, including single crystals, quantum mechanics theoretical calculation, absorption spectra, emission spectra, emission quantum yields and excited state decay dynamics. Large conjugation planes and π-π stacking were found in these complexes. Their emission was originated from MLCT excited state. Long-lived emissive center was observed due to this MLCT-based decay and the help from π-π stacking. Such long-lived emissive state and the large conjugation planes in these complexes offered enough collision probability with O₂ molecules, making themselves potential oxygen sensing probes. These six complexes were then doped into silica supporting matrix MCM-41 spheres. The resulting composite samples and their emission sensing response towards O₂ were discussed in detail. The optimal sample showed sensitivity as high as 7.80 with response time of 14 s. A careful discussion between Cu(I) complex molecular structure and sensing performance was performed. It was concluded that both a long lifetime and a large conjugation plane lead to improved sensing sensitivity since they increased the collision probability with O₂ molecules. On the other hand, it was found that both sensing response and recovery times were mainly controlled by O₂ diffusion in supporting matrix.

Full text of DOI:10.1016/j.saa.2020.118537

Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
Contents lists available at ScienceDirect
Spectrochimica Acta Part A: Molecular and Biomolecular
Spectroscopy
journal homepage: www.elsevier.com/locate/saa
Improving oxygen sensing performance via inner-molecular π-π stacking
in a series of phosphorescent Cu(I) complexes
Baohai Cheng
School of Engineering, Changchun Normal University, Changchun, Jilin 130032, China
a r t i c l e i n f o
a b s t r a c t
Article history:
In this paper, six phosphorescent Cu(I) complexes with three diamine ligands and two phosphorous ligands were
prepared. Detailed discussion was performed on these complexes, including single crystals, quantum mechanics
theoretical calculation, absorption spectra, emission spectra, emission quantum yields and excited state decay
dynamics. Large conjugation planes and π-π stacking were found in these complexes. Their emission was origi-
nated from MLCT excited state. Long-lived emissive center was observed due to this MLCT-based decay and the
help from π-π stacking. Such long-lived emissive state and the large conjugation planes in these complexes of-
fered enough collision probability with O2 molecules, making themselves potential oxygen sensing probes.
These six complexes were then doped into silica supporting matrix MCM-41 spheres. The resulting composite
samples and their emission sensing response towards O2 were discussed in detail. The optimal sample showed
sensitivity as high as 7.80 with response time of 14 s. A careful discussion between Cu(I) complex molecular
structure and sensing performance was performed. It was concluded that both a long lifetime and a large conju-
gation plane lead to improved sensing sensitivity since they increased the collision probability with O2 molecules.
On the other hand, it was found that both sensing response and recovery times were mainly controlled by O2 dif-
fusion in supporting matrix.
Received 17 March 2020
Received in revised form 12 May 2020
Accepted 22 May 2020
Available online 26 May 2020
Keywords:
Cu
MCM
MLCT
O2 sensing
© 2020 Elsevier B.V. All rights reserved.
1. Introduction
pore size/volume. These factors have been considered important to dis-
perse probe molecules uniformly, and thus to guarantee their efficient
Oxygen is one of the major elements on earth, with crust composi-
tion of 46.6% and atmosphere composition of 20.9% [1]. Gaseous oxygen
(O2) is a life-supporting material for most animals, including human be-
ings. On the other hand, industrial processing and food storage gener-
ally need low-oxygen or even oxygen-free condition [2]. These
requirements make the determination of O2 an inevitable objective for
analytical chemistry, industrial monitoring, life sciences and environ-
ment protection [3]. Optical sensing has earned its advantages in sensi-
tivity, selectivity and slim oxygen consumption during measurement,
which enlightens continuous research on optical sensing systems for
O2 [4,5]. There are generally two representative components in an opti-
cal sensing structure, which are optical probe whose optical feature
(emission, lifetime or absorbance) is sensitive to O2 concentration and
supporting matrix to assemble probe molecules, respectively [5].
Previous literatures have reported various candidates as supporting
matrix, including polymer fibers, porous silica and metal-organic frame-
works (MOFs) [4–6]. Among them, mesoporous silica-based materials,
such as MCM-41 and SBA-15, have been highly recommended due to
their stable structure, high transparence in visible region, and suitable
collision with O2 molecules. As for oxygen sensing probe, there are
more criteria to be satisfied [6]. For example, suitable excitation and
emission energy are needed to meet the optical window of supporting
matrix, a long emission lifetime is desired to ensure effective collision
between excited probe and O2 molecules. In addition, good
photostability is required to keep sensing accuracy. Various emitters
have been tried for oxygen sensing, such as luminescent transition
metal complexes, organic dyes, silicon nanoparticles and polymers
[4,5]. Among them, luminescent transition metal complexes are heavily
explored, including Pt(II) complexes, Ir(III) complexes, Ru(II) com-
plexes and Cu(I) complexes. This is because these transition metal com-
plexes have long-lived excited state which is derived from MLCT
(metal-to-ligand-charge-transfer) state, offering enough time for O2
molecules to intercept and quench their energy via a dynamic mecha-
nism of Probe* + 3O2 → Probe +1O2*. By now, the sensitivity of Pt(II)
complexes is higher than those of other transition metal complexes,
owing to their long-lived excited state and coplanar conjugation planes
in ligands [5].
There are problems to be solved, though. No need to mention their
expensive cost, most Pt(II) complexes have poor solubility in common
solvents due to the heavy Pt atom in Pt(II) complexes, leading to
phase separation and consequently compromised sensing behavior in
E-mail address: chengbh781@163.com.
https://doi.org/10.1016/j.saa.2020.118537
1386-1425/© 2020 Elsevier B.V. All rights reserved.

2
B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
corresponding sensing materials. Phosphorescent Cu(I) complexes may
be developed as an alternative option since they have long-lived MLCT-
based excited state and good solubility [7]. It has been pointed out by
Zhang and coworkers that there are inner- and inter-molecular π-π
stacking in Cu(N\\N)(P\\P) complexes, upon the presence of coplanar
conjugation planes in their ligands, showing rather long excited state
lifetime. Here N\\N and P\\P stand for diamine ligand and phosphorous
ligand, respectively [8]. It is rational to assume that a large coplanar con-
jugation in diamine ligand may trigger the π-π stacking in Cu(N\\N)
(P\\P) complexes and such long-lived excited state shall have enough
time to be quenched by O2 molecules. In addition, since excited
electrons of MLCT excited state distribute on diamine ligand coplanar
plane, a large coplanar conjugation in diamine ligand may spread ex-
cited electrons across the whole diamine ligand, offering more collision
probability with O2 molecules. Improved sensing performance can thus
be expected.
Guided by the above consideration, in this paper, three diamine li-
gands with large coplanar conjugation planes are designed. With the
help of two phosphorous ligands, POP and PPh3, corresponding Cu
(N\\N)(P\\P) complexes are synthesized, as depicted in Scheme 1.
Their geometry, electronic structure and photophysical property are
discussed. By doping them into a silica matrix MCM-41, their emission
Scheme 1. Molecular structure and preparation strategy for [Cu(N\\N)(P\\P)]BF4 complexes and [Cu(N\\N)(P\\P)]BF4/MCM-41 composite samples.

B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
3
response towards O2 is systematically explored. The correlation be-
tween ligand structure and Cu(N\\N)(P\\P) photophysical feature is
tentatively discussed as well.
2H), 9.21 (m, 4H), 8.45 (dd, 1H, J = 2.0, 9.0 Hz), 7.97 (m, 2H), 7.74
(m, 2H), 7.22 (m, 2H), 2.44 (s, 3H). MS m/z: [m]+, found 440.1.
2.3. Synthesis of [Cu(N\\N)(P\\P)]BF4 complexes (N-N=L1, L2, L3, P-
P=POP or (PPh3)2)
2. Experiment section
[Cu(N\\N)(P\\P)]BF4 complexes of this work were synthesized fol-
lowing a literature protocol [12]. A brief description is given as follows.
A CH2Cl2 solution (15 mL) containing [Cu(CH3CN)4]BF4 (10 mmol) and
P\\P ligand (POP 10 mmol or PPh3 20 mmol) was stirred at ambient
condition for 30 min. This solution was filtered before adding N\\N li-
gand (10 mmol) into it. Then this mixture was stirred for another
30 min, and ethanol (10 mL) was added. The final solution was filtered
again and allowed to vaporize naturally. Solid bulk was obtained.
2.1. General information
The synthetic route for these Cu(N\\N)(P\\P) complexes and their
oxygen sensing composites [Cu(N\\N)(P\\P)]BF4/MCM-41 is shown
in Scheme 1. Starting compounds were AR grade ones and used as re-
ceived, including Cu(CH3CN)4BF4, PPh3, bis[2-(diphenylphosphino)
phenyl] ether (POP), 1,10-phenanthroline (Phen), KBr, H2SO4, HNO3,
benzene-1,2-diamine, dipyrido[3,2-a:2′,3′-c]phenazine-11-carboxylic
acid, benzohydrazide, 4-methylbenzohydrazide and POCl3. Equipment
for sample characterization is listed as follows. NMR and MS data
were collected by a Bruker DMX 500 spectrometer and a Varian Prostar
spectrometer. Photophysical spectra were recorded using a Unico UV-
2600 spectrophotometer and a Shimadzu RF-5310PC spectrophotome-
ter. Excited state lifetime was determined via an Edberg FL920 spec-
trometer, excited by H lamp. Single crystal data were collected via a
Siemens P4 single-crystal X-ray diffractometer (Mo Kα radiation at
298 K). Powder XRD diffraction patterns were recorded using a Rigaku
X-ray diffractometer. Porous parameters were determined by a
Micromeritics ASAP2010N analyzer. Theoretical calculation was per-
formed using GAMESS at RB3LYP/SBKJC level, with single crystal struc-
ture as initial geometry.
2.3.1. [Cu(L1)(PPh3)2]BF4
1H NMR(CDCl3): δ 9.59 (d, 2H), 9.17 (d, 2H), 8.14 (d, 2H), 7.74 (d,
2H), 7.79 (m, 2H), 7.35 (t, 11H), 7.24 (m, 4H), 7.22 (m, 4H), 7.05 (m,
8H), 6.61 (m, 3H). MS m/z: [m]+, found 956.1.
2.3.2. [Cu(L2)(PPh3)2]BF4
1H NMR(CDCl3):δ 9.42 (m, 2H), 9.21 (m, 2H), 8.40 (m, 1H), 7.94 (m,
2H), 7.73 (m, 2H), 7.34 (t, 11H), 7.25 (m, 8H), 7.20 (m, 5H), 7.02 (m,
8H), 6.60 (m, 3H). MS m/z: [m]+, found 1100.2.
2.3.3. [Cu(L3)(PPh3)2]BF4
1H NMR(CDCl3):δ 9.46 (m, 2H), 9.23 (m, 4H), 8.44 (m, 1H), 7.96 (m,
2H), 7.75 (m, 2H), 7.34 (t, 11H), 7.25 (m, 4H), 7.21 (m, 6H), 7.04 (m,
8H), 6.62 (m, 2H), 2.45 (s, 3H). MS m/z: [m]+, found 1114.3.
2.2. Synthesis of ligands L1, L2 and L3
2.3.4. [Cu(L1)(POP)]BF4
1,10-phenanthroline-5,6-dione (Phen-O) was firstly synthesized
and used as a starting material in accordance with a classic method,
by treating Phen with HNO3 in the presence of concentrated H2SO4
and KBr [9]. Yellow needles were obtained. 1H NMR (CDCl3): δ 9.10
(2H, J = 5.0 Hz), 8.51 (2H, J = 8.0 Hz), 7.63 (2H, J = 5.0 Hz, 8.0 Hz).
MS m/z: [m]+, found 210.1.
1H NMR(CDCl3): δ 9.57 (d, 2H), 9.19 (d, 2H), 8.15 (d, 2H), 7.73 (d,
2H), 7.78 (m, 2H), 7.35 (t, 11H), 7.26 (m, 4H), 7.22 (m, 2H), 7.07 (m,
8H), 6.62 (m, 3H). MS m/z: [m]+, found 970.2.
2.3.5. [Cu(L2)(POP)]BF4
1H NMR(CDCl3): δ 9.41 (m, 2H), 9.21 (m, 2H), 8.43 (m, 1H), 7.95 (m,
2H), 7.73 (m, 2H), 7.35 (t, 11H), 7.23 (m, 8H), 7.22 (m, 5H), 7.02 (m,
6H), 6.61 (m, 3H). MS m/z: [m]+, found 1114.2.
2.2.1. L1
Dipyrido[3,2-a:2′,3′-c]phenazine (L1) was synthesized with Phen-O
and benzene-1,2-diamine as starting chemicals [9]. Their mixture was
heated in ethanol for 10 h. Crude product was purified in hot ethanol
to give light brown needles. 1H NMR (CDCl3): δ 9.64 (2H, J = 8.0 Hz),
9.24 (2H, J = 8.0 Hz), 8.35 (2H, J = 6.5 Hz), 7.95 (2H, J = 6.5 Hz),
7.88 (2H, J = 8.0 Hz). MS m/z: [m]+, found 282.0.
2.3.6. [Cu(L3)(POP)]BF4
1H NMR(CDCl3): δ 9.44 (m, 2H), 9.25 (m, 4H), 8.43 (m, 1H), 7.94 (m,
2H), 7.72 (m, 2H), 7.31 (t, 11H), 7.27 (m, 6H), 7.21 (m, 4H), 7.02 (m,
6H), 6.64 (m, 3H), 2.45 (s, 3H). MS m/z: [m]+, found 1128.2.
2.2.2. Phen-COOH
2.4. Synthesis of [Cu(N\\N)(P\\P)]BF4/MCM-41 (N-N = L1, L2, L3, P-
Dipyrido[3,2-a:2′,3′-c]phenazine-11-carboxylic acid (Phen-COOH)
was synthesized following a similar protocol to that for L1 [9]. However,
benzene-1,2-diamine was replaced by 3,4-diaminobenzoic acid in this
run. 1H NMR (CDCl3): δ, 9.61 (d, 2H, J = 8.0 Hz), 9.25 (d, 2H, J =
8.0 Hz), 8.22 (m, 1H), 7.94 (d, 2H, J = 6.5 Hz), 7.80 (m, 2H). MS m/z:
[m]+, found 326.1.
P=POP or (PPh3)2)
The final oxygen sensing samples, denoted as [Cu(N\\N)(P\\P)]BF4/
MCM-41, were synthesized following a literature protocol [12]. A brief
explanation is given as follows. First, [Cu(N\\N)(P\\P)]BF4 transparent
solution in CH2Cl2 (5 mL) was prepared and stirred at room tempera-
ture. Then MCM-41 spheres (1 g) were added and stirred gently for an-
other 120 min. The resulting powder sample was collected and washed
with ethanol to remove surface residue, and dried in vacuum at room
temperature.
2.2.3. L2
2-(dipyrido[3,2-a:2′,3′-c]phenazin-11-yl)-5-phenyl-1,3,4-
oxadiazole (L2) was synthesized by treating Phen-COOH and
benzohydrazide with POCl3 solution [10]. 1H NMR (CDCl3): δ, 9.45 (m,
2H), 9.24 (m, 4H), 8.43 (dd, 1H, J = 2.0, 9.0 Hz), 7.95 (m, 2H), 7.77
(m, 2H), 7.24 (m, 2H), 7.21 (m, 1H). MS m/z: [m]+, found 426.1.
2.5. Oxygen sensing operation
Each [Cu(N\\N)(P\\P)]BF4/MCM-41 powder sample was sealed in a
quartz chamber, equipped with a one-way valve. Pure N2 and pure O2
streams were controlled by gas meters and then mixed together before
injecting into this quartz chamber. Then this quartz chamber was placed
into a fluorescence spectrometer to record its emission spectra. Oxygen
sensing performance was evaluated based on steady emission spectra.
2.2.4. L3
2-(dipyrido[3,2-a:2′,3′-c]phenazin-11-yl)-5-(p-tolyl)-1,3,4-
oxadiazole (L3) was synthesized following a similar protocol to that for
L2, except that benzohydrazide was replaced by 4-
methylbenzohydrazide in this run [10]. 1H NMR (CDCl3): δ, 9.43 (m,

4
B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
2.6. Theoretical calculation details
to satisfy the requirement for coplanar conjugation plane. Such hypoth-
esis is firstly confirmed by single crystals of [Cu(L1)(PPh3)2]BF4 (CCDC
1358479), [Cu(L1)(POP)]BF4 (CCDC 1358477) and [Cu(L3)(POP)]BF4
(CCDC 1358478). As shown in Fig. 1, ligand L1 exists as a rigid coplanar
plane in both [Cu(L1)(PPh3)2]BF4 and [Cu(L1)(POP)]BF4. As for L3, re-
gardless of its free rotation of σ bond connecting 1,3,4-oxadiazole ring
and p-tolyl ring, it still exists as a rigid coplanar plane. Considering the
MLCT nature of excited [Cu(N\\N)(P\\P)]BF4 complexes, these large
coplanar conjugation planes may spread electronic distribution of ex-
cited electrons. Their collision probability with O2 molecules can thus
be increased, showing improved sensing performance.
Density functional theory calculation on [Cu(N\\N)(P\\P)]BF4 com-
plexes was finished with GAMESS at RB3LYP/LANL2DZ level, using their
single crystal structures as initial geometry. BF4 anion was deleted to
simplify calculation. Their frontier molecular orbitals were plotted by
wxMacMolPlt with contour value of 0.04.
3. Results and discussion
3.1. Characterization on [Cu(N\\N)(P\\P)]BF4 complexes
As suggested by Zhang and coworkers, inner- and inter-molecular π-
π stacking is usually observed given coplanar conjugation planes in li-
gands of [Cu(N\\N)(P\\P)]BF4 complexes, which restricts structural re-
lation of MLCT excited state, showing emission blue shift, long-lived
excited state and improved emission quantum yield [8]. Such inter-
molecular π-π stacking is observed for all three crystals, as shown in
3.1.1. Geometry and single crystal structure
As mentioned above, long-lived MLCT excited state and broad elec-
tronic distribution of MLCT excited electrons shall be resulted by di-
amine ligands with large coplanar conjugation planes and their π-π
stacking effect. The three diamine ligands, L1, L2 and L3, were designed
b
a
c
Fig. 1. A. Single crystals of [Cu(L1)(PPh3)2]BF4 (a), [Cu(L1)(POP)]BF4 (b) and [Cu(L3)(POP)]BF4 (c). B. Inter-molecular π-π stacking of [Cu(L1)(PPh3)2]BF4 (a), [Cu(L1)(POP)]BF4 (b) and [Cu
(L3)(POP)]BF4 (c).

B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
5
a
b
c
Fig. 1 (continued).
Fig. 1B. Corresponding face-to-face distance values are determined as
3.448 Å for [Cu(L1)(PPh3)2]BF4, 3.301 Å for [Cu(L1)(POP)]BF4 and
3.357 Å for [Cu(L3)(POP)]BF4, respectively, which all fall in the repre-
sentative distance for π-π interaction. Such inter-molecular π-π stack-
ing, however, is meaningless for [Cu(N\\N)(P\\P)]BF4/MCM-41
composite samples since [Cu(N\\N)(P\\P)]BF4 molecules shall be dis-
persed in MCM-41 matrix, which breaks such inter-molecular π-π
stacking apart completely. Luckily, there is still inner-molecular π-π
stacking in [Cu(L3)(POP)]BF4, as shown in Fig. 1B. It is observed that
POP ligand adjusts its phenyl rings to align parallel with L3 plane, show-
ing a closest distance of 2.812 Å to L3 surface. Such inner-molecular π-π
stacking makes the tetrahedral geometry a rigid one, suffering from lim-
ited geometry relaxation.
this work are similar to literature values (~2.3 Å), owing to the identical
coordination P atoms. There is weak interaction between the O atom of
POP ligand and Cu(I) ion in [Cu(L1)(POP)]BF4 and [Cu(L3)(POP)]BF4,
considering their Cu\\O distance of ~3 Å. It is still observed that the P-
Cu-P bite angle of [Cu(L3)(POP)]BF4 (121.06°) is obviously larger than
that of [Cu(L1)(POP)]BF4 (113.23°). This result confirms that POP ligand
is stretching itself to satisfy the inner-molecular π-π stacking between
its phenyl ring and L3 ligand. It is thus confirmed that there are coplanar
conjugation plane and π-π stacking in these [Cu(N\\N)(P\\P)]BF4
Table 1
Selected geometric parameters of [Cu(N\\N)(P\\P)]BF4 complexes.
Similar to literature case, each central Cu(I) ion is coordinated by a
bidentate diamine chelate ligand and two phosphorous ligands, forming
a typical tetrahedral coordination geometry [8]. Due to the coordination
difference between diamine and phosphorous ligands, the tetrahedral
coordination geometry in these three complexes is obviously distorted,
as indicated by their key structural parameters listed in Table 1. All
Cu\\N bonds in this work (b2.1 Å) are shorter than those (N2.1 Å)
with diamine ligands derived from five-membered ring (such as imid-
azole) [7]. The bite angles between diamine ligands and Cu(I) ion
(N81°) are found larger than those (b80°) with diamine ligands derived
from five-membered ring (such as imidazole) [7]. This is because a
Phen-derived ligand generally has a smaller coordination hindrance
than an imidazole-derived ligand. On the other hand, Cu\\P bonds in
Bond length (Å)
[Cu(L1)(PPh3)2]BF4
[Cu(L1)(POP)]BF4
[Cu(L3)(POP)]BF4
Cu-N(1)
Cu-N(2)
Cu-P(1)
Cu-P(2)
Cu…O
2.069
2.105
2.272
2.275
N/A
2.051
2.064
2.242
2.254
3.080
2.082
2.062
2.242
2.212
3.011
Bond angle (°)
[Cu(L1)(PPh3)2]BF4
[Cu(L1)(POP)]BF4
[Cu(L3)(POP)]BF4
N(1)-Cu-N(2)
N(1)-Cu-P(1)
N(2)-Cu-P(1)
N(1)-Cu-P(2)
N(2)-Cu-P(2)
P(1)-Cu-P(2)
79.36
108.62
111.23
113.59
111.15
124.09
81.16
117.21
112.82
116.75
111.51
113.23
81.06
102.99
108.19
107.10
125.56
121.06

6
B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
Table 2
Energy levels and transition energy values of [Cu(N\\N)(P\\P)]BF4 complexes.
Transition
Energy (eV)/percentage composition of transition between FMOsa
[Cu(L1)(PPh3)2]BF4
[Cu(L1)(POP)]BF4
[Cu(L3)(POP)]BF4
S0 → S1
S0 → S2
2.6263/152 → 153(97.7)
2.8209/150 → 154(79.4)&
152 → 154(18.3)
2.3978/154 → 155(98.0)
2.6607/153 → 155(89.2)
2.3430/183 → 184(97.1)
2.5970/181 → 184(79.5)&
181 → 185(14.0)
S0 → S3
S0 → S4
S0 → S5
2.8795/150 → 153(93.3)
2.8156/153 → 156(76.8)&
154 → 156(11.9)
2.8630/152 → 155(74.7)&
154 → 156(17.3)
2.8741/154 → 156(67.8)&
152 → 155(20.8)
2.7104/182 → 184(83.1)&
183 → 185(12.1)
2.7487/183 → 185(35.4)&
181 → 185(26.7)&181 → 184(14.6)
2.8038/180 → 184(88.5)
2.9073/152 → 154(66.2)&
150 → 154(13.9)
2.9196/151 → 153(88.6)
a
Molecular orbital numbers are used in Table 2, such as 152, 153 for S0 → S1 in [Cu(L1)(PPh3)2]BF4.
complexes. Given these positive factors, their oxygen sensing perfor-
mance is desired, which will discussed and confirmed below.
their energy quenching and shows emission quenching for sensing
purpose.
3.1.2. Electronic structure
3.1.3. UV–Vis absorption spectra
For most [Cu(N\\N)(P\\P)]BF4 complexes, their onset electronic
transition has a MLCT character. In other words, their occupied frontier
molecular orbital (FMO) is composed of metal character, while their un-
occupied FMO is essentially ligand π*. Their excited electrons are dis-
tributed on unoccupied FMO, waiting for energy acceptors such as O2
molecules. This is the dynamic oxygen sensing mechanism in Cu
(N\\N)(P\\P)]BF4 complexes [12]. For a tentative evaluation on the
electronic structure of these six [Cu(N\\N)(P\\P)]BF4 complexes, DFT
(density functional theory) calculation is applied. Some key parameters
of their FMOs and electronic transitions are listed in Tables 2 and 3. It is
observed that transition energy values of [Cu(L1)(PPh3)2]BF4 are higher
than corresponding values of [Cu(L1)(POP)]BF4, which means that the
oxygen-containing phosphorous ligand (POP) decreases electronic
transition energy of [Cu(N\\N)(P\\P)]BF4 complexes. Given the same
POP ligand, transition energy values of [Cu(L3)(POP)]BF4 are lower
than those of [Cu(L1)(POP)]BF4. This can be explained by the fact that
L3's large conjugation plane stabilizes [Cu(L3)(POP)]BF4 excited state
and decreases transition energy content.
As shown in Table 3, the unoccupied FMOs are essentially Ln ligands
due to their dominant contribution. While the occupied FMOs have
mixed character with contribution from three components, Cu, N\\N li-
gand and P\\P ligand. The first five electronic transitions consist of exci-
tation from occupied FMOs to unoccupied ones, and they are thus
attributed as a mixed character of metal-to-ligand-charge-transfer
(MLCT) and ligand-to-ligand-charge-transfer (LLCT). This finding is
consistent with literature reports [7]. For a visual understanding, two
representative FMOs of each complex, HOMO and LUMO, are plotted
and shown as Fig. 2. It is clear that P\\P ligand and metal center contrib-
ute a lot to HOMO, while LUMO is centralized at the conjugation plane of
N\\N ligand. In this case, it is reasonable to assume that excited elec-
trons on LUMO will be attacked by O2 molecules, which facilitates
The absorption spectra of [Cu(L1−L3)(PPh3)2]BF4 and [Cu(L1-L3)
(POP)]BF4 complexes are shown in Fig. 3. For comparison convenience,
those of ligands L1-L3, PPh3 and POP are given in Fig. 3 as well. As for
PPh3 and POP, their absorption spectra are similar to each other owing
to their rather similar molecular structures. Sharp absorption bands at
~270 nm and ~ 240 nm are observed, which are assigned as n-π* and
π-π* transitions in them. As for ligands L1-L3, their absorption spectra
are similar to each other as well since their molecular structures are
rather similar, showing strong absorption bands at ~270 nm and
~350 nm. Their electron-deficient structure endows them with a wide
energy gap, showing absorption edge at wavelength of ~420 nm. As
for [Cu(L1-L3)(PPh3)2]BF4 and [Cu(L1-L3)(POP)]BF4, their absorption
spectra generally present the absorption characteristic of their struc-
tural components, N\\N and P\\P ligands. Strong absorption bands at
250 nm, 280 nm and 350 nm are observed. These absorption spectra
are actually the absorption adduct of N\\N and P\\P ligands, with
very slight spectral shift, indicating that these absorption bands corre-
spond to ligand-to-ligand charge transfer transitions. This conclusion
is consistent with the above analysis on FMOs of [Cu(L1-L3)(PPh3)2]
BF4 and [Cu(L1-L3)(POP)]BF4.
On the other hand, the absorption spectra of [Cu(L1-L3)(PPh3)2]BF4
and [Cu(L1-L3)(POP)]BF4 extend to visible region, showing absorption
edge of ~450 nm. These weak absorption bands are not traced in absorp-
tion spectra of ligands L1-L3, PPh3 and POP. After consulting the above
DFT analysis, they are named as MLCT transitions. Their absorption
edge (450 nm) is found wider than those of most [Cu(N\\N)(P\\P)]
BF4 complexes of similar conjugation planes, which is attributed to the
electron-deficient structure in ligands L1-L3 [13–15]. On the other
hand, such electron-deficient structure decreases oscillation strength
of corresponding MLCT transitions, which explains why they have are
just weak absorption bands.
Table 3
Orbital composition of [Cu(N\\N)(PPh3)2]BF4 complexes.
[Cu(L1)(PPh3)2]BF4
[Cu(L1)(POP)]BF4
[Cu(L3)(POP)]BF4
a154
153
152
151
L1(83.3)&PPh3(13.2)
L1(94.1)
PPh3(61.8)&Cu(30.9)
PPh3(41.4)&Cu(37.6)
156
155
154
153
L1(77.3)&POP(18.9)
L1(97.9)
POP(65.9)&Cu(30.0)
Cu(40.7)&L1(37.4) &
POP(21.8)
185
184
183
182
L3(84.5)&POP(11.9)
L3(97.5)
POP(62.2)&Cu(31.3)
L3(99.3)
150
Cu(43.8)&L1(34.4) & PPh3(21.8)
152
POP(42.3)&Cu(36.7) &
L1(21.0)
181
180
L3(37.8)&Cu(37.0) &
POP(25.3)
POP(39.1)&Cu(36.4) &
L3(24.5)
a
Molecular orbital numbers are used in Table 3. HOMO and LUMO are marked with bold.

B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
7
a
c
e
b
d
f
Fig. 2. FMOs of [Cu(L1)(PPh3)2]BF4 (a, HOMO, b, LUMO), [Cu(L1)(POP)]BF4 (c, HOMO, d, LUMO) and [Cu(L3)(POP)]BF4 (e, HOMO, f, LUMO).
3.1.4. Emission spectra, quantum yields and lifetimes
(L3)(PPh3)2]BF4, respectively. There are no vibronic details or shoulder
peaks in these emission spectra, which means that their emissive state
has a charge transfer character. According to the above DFT result, the
onset electronic transition of these [Cu(L1-L3)(PPh3)2]BF4 and [Cu(L1-
L3)(POP)]BF4 complexes has a mixed character of MLCT and LLCT.
Since their emissive center is derived from such onset electronic state,
The emission spectra of [Cu(L1-L3)(PPh3)2]BF4 and [Cu(L1-L3)
(POP)]BF4 complexes are shown in Fig. 4. Broad emission bands are ob-
served for all these complexes, peaking at 620 nm for [Cu(L1)(POP)]BF4,
550 nm for [Cu(L2)(POP)]BF4, 553 nm for [Cu(L3)(POP)]BF4, 610 nm for
[Cu(L1)(PPh3)2]BF4, 547 nm for [Cu(L2)(PPh3)2]BF4, and 550 nm for [Cu
Fig. 3. Absorption spectra of [Cu(L1-L3)(PPh3)2]BF4 and [Cu(L1-L3)(POP)]BF4 complexes in CH2Cl2 (1 μM). Insets: Absorption spectra of ligands L1-L3, PPh3 and POP in CH2Cl2 (1 μM).

8
B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
Fig. 5. Emission decay dynamics of these [Cu(L1-L3)(PPh3)2]BF4 and [Cu(L1-L3)(POP)]BF4
complexes.
Fig. 4. Emission spectra of [Cu(L1-L3)(PPh3)2]BF4 and [Cu(L1-L3)(POP)]BF4 complexes.
relaxation of MLCT excited state in PPh3-based complexes and conse-
quently their low Φ values.
To confirm this hypothesis, emission decay dynamics (τ) of these
[Cu(L1-L3)(PPh3)2]BF4 and [Cu(L1-L3)(POP)]BF4 complexes are re-
corded and shown as Fig. 5, so that radiative (Kr) and non-radiative
(Knr) decay constants can be determined with Formulas 1 and 2.
it is reasonable to see a charge transfer nature of these emission spectra.
After comparing them with similar [Cu(N\\N)(P\\P)]BF4 complexes
with emission peaks around 550 nm, it is found that [Cu(L1)(POP)]BF4
and [Cu(L1)(PPh3)2]BF4 have red shifted their emission obviously
[12–15]. We attribute its causation to the large conjugation plane in li-
gand L1. As for [Cu(L2)(POP)]BF4, [Cu(L3)(POP)]BF4, [Cu(L2)(PPh3)2]
BF4 and [Cu(L3)(PPh3)2]BF4, although their N\\N ligands (L2 and L3)
have larger conjugation planes than L1 does, corresponding emission
wavelength is blue shifted to ~550 nm. This fact can be explained by
the electron-pulling oxadiazole ring in ligands L2 and L3. In addition,
it is found that the FWHM (full-width-at-half-maximum) values of
[Cu(L1)(POP)]BF4 (175 nm) and [Cu(L1)(PPh3)2]BF4 (155 nm) are obvi-
ously larger than those of [Cu(L2)(POP)]BF4 (90 nm), [Cu(L3)(POP)]BF4
(98 nm), [Cu(L2)(PPh3)2]BF4 (84 nm) and [Cu(L3)(PPh3)2]BF4 (94 nm).
Obviously, the electron-pulling oxadiazole ring restricts the emissive
state via its electron-deficient structure.
For a better understanding on these [Cu(L1-L3)(PPh3)2]BF4 and [Cu
(L1-L3)(POP)]BF4 complexes, their emission quantum yields (Φ) are
determined and listed in Table 4. It is observed that Φ values of [Cu
(L1)(POP)]BF4 (0.25) and [Cu(L1)(PPh3)2]BF4 (0.22) are lower than
those of [Cu(L2)(POP)]BF4 (0.34), [Cu(L3)(POP)]BF4 (0.31), [Cu(L2)
(PPh3)2]BF4 (0.31) and [Cu(L3)(PPh3)2]BF4 (0.30). Considering their
identical MLCT emissive state and similar molecular structures of
N\\N ligands, these improved Φ values are attributed to the electron-
pulling oxadiazole ring in L2 and L3 which restricts the emissive state
via its electron-deficient structure. This hypothesis has been above
mentioned and here confirmed. Given the same N\\N ligand, POP li-
gand tends to offer a higher Φ value for corresponding Cu(I) complex,
compared to PPh3 ligand. It is assumed that the bridge connection of O
atom between POP phenyl rings restricts the free rotation and vibration
of POP ligand, while PPh3 ligand fails to do so which endows its phenyl
rings with a high degree of freedom, leading to severe non-radiative
kr
Φ ¼
ð1Þ
ð2Þ
kr þ knr
1
τ
¼ kr þ knr
It is observed that all six complexes follow single exponential decay
pattern with lifetime at a scale of microsecond. Their lifetimes all fall in
the typical lifetime region of MLCT-based emission [15]. It has been con-
firmed by literatures that such long-lived MLCT emissive states are vul-
nerable to O2 attack, which makes them potential O2 sensors. This
hypothesis will be below discussed and confirmed. Kr and Knr values
are calculated and listed in Table 4. After comparing these values, it is
surprisingly found that the Knr value of each PPh3-based Cu
(I) complex is lower than that of each POP-based Cu(I) complex,
which conflicts with our above assumption. In this case, it is concluded
that PPh3's large steric hindrance can depress the structural relaxation
of its Cu(I) MLCT excited state, showing a low Knr value. The underde-
veloped Φ values of PPh3-based Cu(I) complexes are mainly caused by
their low Kr values.
3.2. Characterization on [Cu(N\\N)(P\\P)]BF4/MCM-41
As above mentioned, these [Cu(L1-L3)(PPh3)2]BF4 and [Cu(L1-L3)
(POP)]BF4 complexes have long-lived MLCT excited states. Since their
excited electrons are distributed on outer ligand π* orbitals, they may
be attacked and thus quenched by O2 molecules, leading to emission ab-
sence and thus sensing signal. To ensure fluent O2 gas diffusion and ho-
mogeneous quenching on Cu(I) complexes, they are usually dispersed
and embedded into a proper supporting matrix, with criteria of suitable
mechanical strength, high porosity, good transparency at the working
wavelength of sensing materials [16,17]. In this work, we intend to
adopt a silica-based supporting matrix MCM-41 owing to its promising
performance serving as a host for sensing probes of luminescent metal
complexes [18–22]. [Cu(L1-L3)(PPh3)2]BF4 and [Cu(L1-L3)(POP)]BF4
are doped into MCM-41 spheres via solution soaking method with
three doping concentrations for each complex, [Cu(L1)(PPh3)2]BF4
Table 4
Selected photophysical parameters of [Cu(N\\N)(P\\P)]BF4 complexes.
Complexes
λem (nm)
Φ
τ (μs) Kr (×104 S−1
)
Knr (×104 S−1
)
[Cu(L1)(PPh3)2]BF4 610
[Cu(L2)(PPh3)2]BF4 547
[Cu(L3)(PPh3)2]BF4 550
0.22 4.4
0.31 6.4
0.30 4.7
0.25 3.9
0.34 5.7
0.31 3.2
5.0
4.8
6.3
6.4
6.0
9.7
17.7
10.8
14.9
19.2
11.6
21.6
[Cu(L1)(POP)]BF4
[Cu(L2)(POP)]BF4
[Cu(L3)(POP)]BF4
620
550
553

B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
9
Fig. 6. SEM and TEM images of a representative sample [Cu(L2)(POP)]BF4/MCM-41 (55 mg/g).
(45 mg/g, 50 mg/g, 55 mg/g), [Cu(L2)(PPh3)2]BF4 (50 mg/g, 55 mg/g,
60 mg/g), [Cu(L3)(PPh3)2]BF4 (50 mg/g, 55 mg/g, 60 mg/g), [Cu(L1)
(POP)]BF4 (50 mg/g, 55 mg/g, 60 mg/g), [Cu(L2)(POP)]BF4 (55 mg/g,
60 mg/g, 65 mg/g) and [Cu(L3)(POP)]BF4 (55 mg/g, 60 mg/g, 65 mg/
g), aiming at performance optimization. Corresponding [Cu(N\\N)
(P\\P)]BF4/MCM-41 samples are further analyzed below.
channels of MCM-41 have been well preserved for O2 diffusion, which
favors oxygen sensing performance. This finding is consistent with the
above finding from SEM and TEM images. This observation is explained
by the fact that the molecular size of [Cu(N\\N)(P\\P)]BF4 (~1.7 nm) is
smaller than the pore size of MCM-41 (~2.5 nm).
For a better understanding on the microstructure of these [Cu
(N\\N)(P\\P)]BF4/MCM-41 samples, N2 adsorption and desorption iso-
therms of several representative samples and standard MCM-41 are re-
corded and shown as Fig. 8. As for MCM-41, standard type-IV isotherms
are observed. Its mesoporous parameters, including surface area, pore
size and pore volume, are listed in Table 5. These isotherms and meso-
porous parameters are found agree well with literature values [22,23].
After loading [Cu(N\\N)(P\\P)]BF4 complexes, each sample has pre-
served its type-IV isotherms, with shrunk mesoporous parameters
though, as shown by Table 5. It is still found that pore size and pore vol-
ume decrease with the increasing molecular weight of Cu(N\\N)
(P\\P)]BF4 complexes. This result actually confirms the successful dop-
ing of Cu(N\\N)(P\\P)]BF4 complexes in MCM-41 supporting matrix.
3.2.1. SEM, XRD and N2 adsorption/desorption
For a fast evaluation on these [Cu(N\\N)(P\\P)]BF4/MCM-41 sam-
ples, SEM and TEM images of a representative sample [Cu(L2)(POP)]
BF4/MCM-41 (55 mg/g) are shown as Fig. 6. Similar to literature reports,
these nanocrystals are hexagonal ones with length of ~100 nm [14,22].
Smooth surface and highly dispersed distribution are observed, which
are positive for O2 gas diffusion, favoring sensing performance. The
highly ordered hexagonal channels can be well detected through their
TEM images. It seems that doping Cu(I) complex molecules into
MCM-41 exerts slim effect on its microstructure.
To further confirm this hypothesis, small angle XRD curves of repre-
sentative [Cu(N\\N)(P\\P)]BF4/MCM-41 samples and standard MCM-
41 are recorded and shown as Fig. 7. Three sharp and well-resolved
Bragg reflection peaks are observed for standard MCM-41 sample,
which have been named as (d100, d110, and d200), respectively [23].
After loading [Cu(N\\N)(P\\P)]BF4 complexes, all three Bragg reflec-
tion peaks have been well preserved. The slight spectral shift of d100
can be attributed to the homogeneity difference between samples.
This observation actually suggests that loading [Cu(N\\N)(P\\P)]BF4
complexes into the hexagonal channels of MCM-41 matrix exerts slim
effect on MCM-41 backbone. In other words, the highly ordered
3.2.2. Emission spectra upon various O2 concentrations
For a direct understanding on O2 sensing response of [Cu(N\\N)
(P\\P)]BF4/MCM-41 samples, their emission spectra upon various O2
concentrations are recorded and shown as Fig. 9. It is clear that each
[Cu(N\\N)(P\\P)]BF4/MCM-41 sample decreases its emission intensity
upon increasing O2 concentration, showing emission turn-off effect to-
wards O2 gas (see Fig. S1, Supporting Information for the emission spec-
trum of blank MCM-41). This observation is consistent with literature
Fig. 7. Small angle XRD curves of representative [Cu(N\\N)(P\\P)]BF4/MCM-41 samples
Fig. 8. N2 adsorption and desorption isotherms of several representative sample and
and standard MCM-41.
standard MCM-41.

10
B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
Table 5
Mesoporous parameters of Cu(I)-doped MCM-41 samples.
Sample
Surface area (m2·g−1
)
Pore diameter (nm)
Pore volume (cm3·g−1
)
blank MCM-41
741.12
461.33
460.18
452.37
451.43
443.18
440.22
2.51
2.39
2.31
2.26
2.30
2.19
2.14
0.583
0.279
0.261
0.243
0.236
0.211
0.205
[Cu(L1)(PPh3)2]BF4 (55 mg/g)
[Cu(L2)(PPh3)2]BF4 (60 mg/g)
[Cu(L3)(PPh3)2]BF4 (60 mg/g)
[Cu(L1)(POP)]BF4 (60 mg/g)
[Cu(L2)(POP)]BF4 (65 mg/g)
[Cu(L3)(POP)]BF4 (65 mg/g)
reports on similar Cu(I)-based oxygen probes [22,23]. The emission
band shape has been well preserved upon various O2 concentrations,
suggesting that the MLCT-based emissive state remains constant. It is
observed from Table 6 that the emission of [Cu(N\\N)(P\\P)]BF4/
MCM-41 samples shows blue shift tendency, compared to that of pure
dopants mentioned in Table 4. This result is explained by
rigidochromism since the rigid microenvironment in MCM-41 channels
restricts geometric relaxation of MLCT excited state, leading to emission
blue shift [22]. It is still observed that for each [Cu(N\\N)(P\\P)]BF4
dopant, its emission shows red shift tendency with increasing doping
concentration in MCM-41. This result may be explained by the aggrega-
tion effect between dopant molecules [22].
For comparison convenience, sensing sensitivity is defined as the
emission intensity ratio of I0/I100, where I0 and I100 are the emission in-
tensity values with O2 concentrations of 0% and 100%, respectively. For
each dopant, there is an optimal doping concentration which offers
Fig. 9. A. Emission spectra of [Cu(L1-L3)(PPh3)2]BF4/MCM-41 upon various O2 concentrations. a, [Cu(L1)(PPh3)2]BF4/MCM-41 (45 mg/g), b, [Cu(L1)(PPh3)2]BF4/MCM-41 (50 mg/g), c, [Cu
(L1)(PPh3)2]BF4/MCM-41 (55 mg/g), d, [Cu(L2)(PPh3)2]BF4/MCM-41 (50 mg/g), e, [Cu(L2)(PPh3)2]BF4/MCM-41 (55 mg/g), f, [Cu(L2)(PPh3)2]BF4/MCM-41 (60 mg/g), g, [Cu(L3)(PPh3)2]
BF4/MCM-41 (50 mg/g), h, [Cu(L3)(PPh3)2]BF4/MCM-41 (55 mg/g), i, [Cu(L3)(PPh3)2]BF4/MCM-41 (60 mg/g). B. Emission spectra of [Cu(L1-L3)(POP)]BF4/MCM-41 upon various O2
concentrations. a, [Cu(L1)(POP)]BF4/MCM-41 (50 mg/g), b, [Cu(L1)(POP)]BF4/MCM-41 (55 mg/g), c, [Cu(L1)(POP)]BF4/MCM-41 (60 mg/g), d, [Cu(L2)(POP)]BF4/MCM-41 (55 mg/g), e,
[Cu(L2)(POP)]BF4/MCM-41 (60 mg/g), f, [Cu(L2)(POP)]BF4/MCM-41 (65 mg/g), g, [Cu(L3)(POP)]BF4/MCM-41 (55 mg/g), h, [Cu(L3)(POP)]BF4/MCM-41 (60 mg/g), i, [Cu(L3)(POP)]
BF4/MCM-41 (65 mg/g).

B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
11
Fig. 9 (continued).
the highest sensitivity value. Both a higher doping concentration and a
lower one compromise sensitivity, as depicted by Table 6. This result
suggests that there are at least two factors controlling sensitivity,
which shall be dopant concentration and aggregation effect between
dopant molecules. The former factor should be positive for sensing per-
formance since only with enough dopant molecules, strong emission
and efficient O2 quenching can be realized [23]. If dopant molecules
are less than enough, the weak emission from insufficient dopant mol-
ecules fails to meet O2 molecules and leads to poor sensitivity. The latter
factor, however, should be a negative one for sensing performance.
With more and more dopant molecules, their aggregation becomes se-
vere, which inhibits dopant molecules from O2 attack, leading to defi-
cient quenching and compromised sensitivity. Additionally, the self-
absorption and self-quenching at high doping concentrations may
quench dopant emission, leading to compromised sensitivity as well.
These two factors shall reach an equilibrium at certain doping concen-
tration, which is the optimal doping concentration for each [Cu(N\\N)
(P\\P)]BF4 dopant.
[Cu(N\\N)(P\\P)]BF4 dopant. For example, the sensitivity values of
[Cu(L2)(POP)]BF4-doped samples are all higher than those of [Cu(L3)
(POP)]BF4-doped samples. Similarly, the sensitivity values of [Cu(L2)
(PPh3)2]BF4-doped samples are higher than those of [Cu(L2)(POP)]
BF4-doped samples. Their dopant molecular structures and conjugation
size are similar to each other, but sensitivity values are obviously differ-
ent, which confirms the positive effect of lifetime in improving sensitiv-
ity. In addition, a large conjugation size in N\\N ligand indeed improves
sensitivity. For example, sensitivity values of [Cu(L1)(PPh3)2]BF4-based
and [Cu(L1)(POP)]BF4-based samples are all obviously lower than those
of [Cu(L2/L3)(PPh3)2]BF4-based and [Cu(L2/L3)(POP)]BF4-based sam-
ples. This is because excited electrons tend to distribute on the N\\N
π* orbitals, as mentioned in Section 3.1.2. A larger conjugation plane
in N\\N ligand means a higher collision probability with O2 molecules,
leading to efficient O2 quenching and thus improved sensitivity. As
above mentioned, a long lifetime is accompanied by a large conjugation
size in N\\N ligand, owing to the inner- and inter-molecular π-π stack-
ing within Cu(N\\N)(P\\P)]BF4 complexes [8]. This conclusion finally
explains why L2- and L3-based dopants exhibit high sensitivity values.
After comparing the optimal [Cu(N\\N)(P\\P)]BF4/MCM-41 sam-
ples and their sensitivity values, it is observed that sensitivity of [Cu
(N\\N)(P\\P)]BF4/MCM-41 is directly dominated by the emission life-
time and the conjugation plane size of [Cu(N\\N)(P\\P)]BF4 dopant.
Be more specific, a long lifetime generally gives a high sensitivity
value since a long lifetime ensures a complete quenching on emissive
3.2.3. Stern-Volmer working equation
It has been reported and confirmed by literatures that oxygen sens-
ing metal complexes, especially MLCT-based ones, such as Ru(II)-, Re
(I)- and Cu(II)-complexes, are quenched by O2 molecules via a dynamic

12
B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
Table 6
Sensing parameters of Cu(I)-doped MCM-41 samples.
Sample
λem
(nm)
I0/I
f1
f2
Ksv1
Ksv2
Tres
(s)
Trec
(s)
(O2%−1
)
(O2%−1
)
[Cu(L1)(PPh3)2]BF4 (45 mg/g)
[Cu(L1)(PPh3)2]BF4 (50 mg/g)
[Cu(L1)(PPh3)2]BF4 (55 mg/g)
[Cu(L2)(PPh3)2]BF4 (50 mg/g)
[Cu(L2)(PPh3)2]BF4 (55 mg/g)
[Cu(L2)(PPh3)2]BF4 (60 mg/g)
[Cu(L3)(PPh3)2]BF4 (50 mg/g)
[Cu(L3)(PPh3)2]BF4 (55 mg/g)
[Cu(L3)(PPh3)2]BF4 (60 mg/g)
[Cu(L1)(POP)]BF4 (50 mg/g)
[Cu(L1)(POP)]BF4 (55 mg/g)
[Cu(L1)(POP)]BF4 (60 mg/g)
[Cu(L2)(POP)]BF4 (55 mg/g)
[Cu(L2)(POP)]BF4 (60 mg/g)
[Cu(L2)(POP)]BF4 (65 mg/g)
[Cu(L3)(POP)]BF4 (55 mg/g)
[Cu(L3)(POP)]BF4 (60 mg/g)
[Cu(L3)(POP)]BF4 (65 mg/g)
588
592
593
539
542
544
545
545
546
591
593
594
541
542
547
545
547
550
3.76
4.06
3.90
6.47
7.80
6.33
4.76
5.18
4.54
3.49
3.74
3.65
5.63
7.54
7.52
4.57
5.72
4.95
0.67
0.50
0.62
0.82
0.84
0.80
0.78
0.76
0.76
0.81
0.71
0.72
0.80
0.85
0.79
0.78
0.81
0.76
0.33
0.50
0.38
0.18
0.16
0.20
0.22
0.24
0.24
0.19
0.29
0.28
0.20
0.15
0.21
0.22
0.19
0.24
0.24563
1.92056
0.23016
1.85929
2.10899
1.71925
1.20881
1.56174
1.10641
0.06368
0.50465
0.44980
0.38180
0.31633
0.39954
1.75947
1.84817
3.45966
0.00368
0.01028
0.00605
0.00166
0.00255
0.00282
0.00099
0.00244
0.00099
0.00073
0.00123
0.00088
0.00255
0.00391
0.00791
0.00041
0.00092
0.00194
13
19
19
13
14
22
18
20
20
12
13
18
22
27
28
14
13
14
91
102
121
61
65
73
98
99
101
61
61
62
74
80
81
74
87
87
collision mechanism [22]. The dynamic quenching mechanism of [Cu
(N\\N)(P\\P)]BF4/MCM-41 samples is tentatively confirmed by emis-
sion lifetime comparison of [Cu(L2)(PPh3)2]BF4/MCM-41 (55 mg/g)
under various N2 concentrations (100% N2, air and 100% O2). It is ob-
served from Fig. S2 (Supporting Information) that its emission lifetime
is decreased with increasing O2 concentrations (6.4 μs, 4.6 μs and
3.7 μs, respectively), with single exponential decay mode well pre-
served. This result confirms a dynamic collision mechanism. In this
case, emission intensity against O2 concentration variation can be de-
scribed by Stern-Volmer equation [22,23]. Literally, the intensity form
for Stern-Volmer equation can be shown as Formula 3. Here I denotes
emission intensity, I0 means the emission intensity with O2 concentra-
tion of 0%, Ksv stands for Stern-Volmer constant and [O2] is O2 concen-
tration, respectively.
all KSV1 values are much (two or even four orders of magnitude) higher
than KSV2 values, confirming that there are indeed a lot of dopant mol-
ecules with immunity to O2 quenching. The oxygen-sensitive dopant
molecules, corresponding to KSV1, are clearly attributed to the dopant
molecules trapped in MCM-41 surface tunnels, which are easier to be
attacked and quenched by O2 molecules. As for the oxygen-insensitive
dopant molecules, their immunity may be caused by two reasons.
First, geometric reason, some dopant molecules may be trapped into
the deep MCM-41 channels which are hardly penetrable by O2 mole-
cules. Second, intrinsic emission reason, as mentioned in Section 3.1.2,
[Cu(N\\N)(P\\P)]BF4 emission comes from MLCT-based excited state.
The ligand-to-ligand emission is a fast-decay component, which
makes it deficient to be quenched by O2 molecules. Its emission contrib-
utes to intrinsic emission and compromises sensitivity and linearity a
lot. In general, POP-based [Cu(N\\N)(P\\P)]BF4 complexes have
smaller f2 values than PPh3-based ones. In addition, POP-based [Cu
(N\\N)(P\\P)]BF4 complexes have smaller KSV2 values than PPh3-
based ones. This is attributed to POP's restricted geometric structure
which has a low hindrance for O2 transportation and penetration. This
result indicate4s that the effective sensing plane is the N\\N plane
while P\\P ligand here shows as a steric hindrance for O2 transportation
and penetration. To improve sensing performance, the effective sensing
plane (N\\N plane) should be increased, while the steric hindrance
plane (P\\P) plane should be decreased. In the meanwhile, the steric
hindrance plane plays another important role of preventing MLCT ex-
cited state from geometric relaxation. So, these two opposite factors
should be equally considered, aiming at large enough steric hindrance
to stop geometric relaxation and small enough negative on O2 transpor-
tation and penetration.
I0=I ¼ 1 þ KSV½O2ꢀ
ð3Þ
Theoretically, for an ideal model, its Stern-Volmer curve shall be a
linear one with slope of Ksv. This, however, is not the truth for [Cu
(N\\N)(P\\P)]BF4/MCM-41 samples. It is observed from Fig. 10 that
all Stern-Volmer curves are non-linear ones against [O2] variation. In
other words, this ideal model fails to match the practical sensing in
[Cu(N\\N)(P\\P)]BF4/MCM-41 samples. A more complicated model
will be needed here [23].
After a careful observation on the emission spectra shown in Fig. 9, it
is found that all [Cu(N\\N)(P\\P)]BF4/MCM-41 samples preserve obvi-
ous emission intensity under 100% O2, which means that there are a
number of dopant molecules whose emission intensity is immune
from O2 gas, owing to deficient O2 quenching or intrinsic emission. Bear-
ing this observation in mind, it is assumed that there are at least two
sorts of dopant molecules, efficient sensing ones and inefficient sensing
ones. Their fractional contribution to the whole sensing performance
should be quantified, and corresponding Stern-Volmer equation should
be modified as Formula 4 [22]. Here f1 and f2 denote the fractional con-
tributions of their sensing sites, KSV1 and KSV2 are Stern-Volmer
quenching constants of their sensing sites, respectively.
3.2.4. Photostability and response/recovery time
The emission intensity of each [Cu(N\\N)(P\\P)]BF4/MCM-41 sam-
ple is continuously monitored with testing atmosphere periodically
switched between 100% N2 and 100% O2, as shown in Fig. 11. It is ob-
served that all [Cu(N\\N)(P\\P)]BF4/MCM-41 samples respond to O2
quenching instantly. While, after the removal of O2, their emission can
be recovered quickly. Their emission intensity can be recovered to
their initial levels under 100% N2 atmosphere, indicating their good
photostability. Such good photostability has been considered important
for optical sensing since the quantification is based on optical signals.
We attribute this good photostability to the virtues of supporting matrix
MCM-41 spheres.
I0
I
1
¼
ð4Þ
f 1
f 2
þ
1 þ KSV1pO2 1 þ KSV2pO2
With this modified Stern-Volmer equation, all Stern-Volmer curves
in Fig. 10 are fitted and plotted. Good matching between Formula 4
and [Cu(N\\N)(P\\P)]BF4/MCM-41 Stern-Volmer plots is finally ob-
served. Corresponding fitting parameters are listed in Table 6. Clearly,
For a systematical comparison between these samples, response
time (Tres) and recovery time (Trec) are defined as the time taken by
each sample to lose or recover 95% of its final emission intensity when

B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
13
b
d
f
a
c
e
Fig. 10. Stern-Volmer plots of [Cu(L1-L3)(PPh3)2]BF4/MCM-41 and [Cu(L1-L3)(POP)]BF4/MCM-41. a, [Cu(L1)(PPh3)2]BF4/MCM-41, b, [Cu(L2)(PPh3)2]BF4/MCM-41, c, [Cu(L3)(PPh3)2]BF4/
MCM-41, d, [Cu(L1)(POP)]BF4/MCM-41, e, [Cu(L2)(POP)]BF4/MCM-41, f, [Cu(L3)(POP)]BF4/MCM-41.
testing atmosphere is switched from 100% N2 to 100% O2 (response pro-
cedure) or from 100% O2 to 100% N2 (recover procedure). Tres and Trec
values of all [Cu(N\\N)(P\\P)]BF4/MCM-41 samples are listed in
Table 6. These response and recovery time values generally are compa-
rable to literature values based on MCM-41 matrix [20–23]. It is still
found that, for each dopant, response time increases with increasing
doping concentration. This result suggests that with more and more
dopant molecules in MCM-41 tunnels, O2 transportation and diffusion
become worse, leading to long response time and long recovery time.
On the other hand, it seems that a large conjugation plane leads to a
long response time, especially for PPh3-based complexes. This fact sug-
gests that the response procedure is actually controlled by O2 diffusion.
The overloading of MCM-41 channels compromises O2 transportation
and diffusion.

14
B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
b
d
f
a
c
e
Fig. 11. Emission intensity of each [Cu(N\\N)(P\\P)]BF4/MCM-41 sample is continuously monitored with testing atmosphere periodically switched between 100% N2 and 100% O2.
For all [Cu(N\\N)(P\\P)]BF4/MCM-41 samples, their recovery time
values are much longer than their response time values. Literatures
have concluded that this fact can be explained by diffusion-controlled
recovery mechanism as well [22]. Additionally, it seems that recovery
time is prolonged by increasing doping concentration. With the increas-
ing conjugation size of N\\N ligands, the recovery time values of their
[Cu(N\\N)(P\\P)]BF4/MCM-41 samples tend to increase. This observa-
tion is similar to the case for response time. It is thus confirmed that re-
covery procedure is dominated by O2 diffusion as well. In other words,
both sensing response and recovery are controlled by O2 diffusion in
supporting matrix. The effect from sensing probe is confirmed but not
dominant.
4. Conclusion
To sum up, we prepared six [Cu(N\\N)(P\\P)]BF4 complexes and
eighteen [Cu(N\\N)(P\\P)]BF4/MCM-41 samples for oxygen sensing
application. Single crystal and DFT analysis suggested that there were

B. Cheng / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 239 (2020) 118537
15
[3] Y. He, J. He, H. Zhang, Y. Liu, B. Lei, Luminescent carbon dots assembled into meso-
porous aluminas for oxygen sensing, Opt. Mater. Express 7 (2017) 945–954.
[4] Y. Amao, Probes and polymers for optical sensing of oxygen, Microchim. Acta 143
(2003) 1–12.
[5] X. Wang, O. Wolfbeis, Optical methods for sensing and imaging oxygen: materials,
spectroscopies and applications, Chem. Soc. Rev. 43 (2014) 3666–3761.
[6] E. Hussain, C. Cheng, Y. Li, N. Niu, H. Zhou, X. Jin, J. Kong, C. Yu, Benzo[ghi]perylene &
coronene as ratiometric reversible optical oxygen nano-sensors, Sensors Actuators B
Chem. 287 (2019) 27–34.
[7] L. Zhang, Q. Zuo, A series of blue-green-yellow-red emitting Cu(I) complexes: mo-
lecular struture and photophysical performance, Spectrochim. Acta A Mol. Biomol.
Spectrosc. 223 (2019) 117280.
[8] S.M. Borisov, A.S. Vasylevska, C. Krause, O.S. Wolfbeis, Composite luminescent mate-
rial for dual sensing of oxygen and temperature, Adv. Mater. 16 (2006) 1536–1542.
[9] L. Zhang, B. Li, A series of 4,5-diazafluoren-9-one-derived ligands and their Cu
(I) complexes: synthesis, characterization and photophysical properties, Inorg.
Chim. Acta 362 (2009) 4857–4861.
[10] J. Ying, A.X. Tian, B. Li, A new blue-emitting diimine copper(I) complex: synthesis,
crystal structure and photophysical properties, Zeitschrift Für Naturforschung B
64b (2009) 287–291.
[12] L. Shi, B. Li, S. Yue, D. Fan, Synthesis, photophysical and oxygen-sensing properties of
a novel bluish-green emission Cu(I) complex, Sensors Actuators B Chem. 137
(2009) 386–392.
[13] S.M. Borisov, R. Seifner, I. Klimant, A novel planar optical sensor for simultaneous
monitoring of oxygen, carbon dioxide, pH and temperature, Anal. Bioanal. Chem.
400 (2011) 2463–2474.
large conjugation planes and π-π stacking in these complexes, which
could increase sensing collision probability with O2 molecules, improv-
ing sensing performance. Detailed oxygen sensing performance of [Cu
(N\\N)(P\\P)]BF4/MCM-41 samples was discussed, including emission
spectra, sensitivity, working equation, photostability and response/re-
covery time. It was found that these samples showed emission
quenching effect towards O2. A highest sensitivity of 7.80 was observed.
Large conjugation plane and long-lived emissive center were found pos-
itive to improve sensitivity. Non-linear working curves were observed,
showing a two-site sensing model. The non-sensitive site was attributed
to dopant molecules trapped into the deep MCM-41 channels and in-
trinsic emission from ligand-to-ligand emission. As for response time
and recovery time, they were mainly controlled by O2 diffusion in
supporting matrix. The effect from sensing probe was not dominant.
Large steric hindrance and heavy doping even compromised response/
recovery performance.
CRediT authorship contribution statement
Baohai Cheng: Conceptualization, Data curation, Formal analysis,
Writing - review & editing.
[14] X. Wang, X. Chen, Z. Xie, X. Wang, Reversible optical sensor strip for oxygen, Angew.
Chem. Int. Ed. 120 (2008) 7560–7563.
[15] A.A. Abdel-Shafia, D.R. Worrall, Mechanism of the excited singlet and triplet states
quenching by molecular oxygen in acetonitrile, J. Photochem. Photobio. A Chem.
172 (2005) 170–179.
[16] C. Chu, H. Lin, H. Liu, X. Wang, J. Wang, P. Zhang, H. Gao, C. Huang, Y. Zeng, Y. Tan, G.
Liu, X. Chen, Tumor microenvironment-triggered supramolecular system as an in
situ nanotheranostic generator for cancer phototherapy, Adv. Mater. 1605928
(2017).
[17] P.A.S. Jorgea, C. Maule, A.J. Silva, R. Benrashid, J.L. Santos, F. Farahi, Dual sensing of
oxygen and temperature using quantum dots and a ruthenium complex, Anal.
Chim. Acta 606 (2008) 223–229.
Declaration of competing interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
[18] S.M. Ramasamy, R.J. Hurtubise, Oxygen sensor via the quenching of room-
temperature phosphorescence of perdeuterated phenanthrene adsorbed on
Whatman 1PS filter paper, Talanta 47 (1998) 971–979.
Appendix A. Supplementary data
[19] A. Sharma, O.S. Wolfbeis, Unusually efficient quenching of the fluorescence of an en-
ergy transfer-based optical sensor for oxygen, Anal. Chim. Acta 212 (1988) 261–265.
[20] T. Okazaki, T. Imasaka, N. Ishibashi, Optical-fiber sensor based on the second har-
monic emission of a near-infrared semiconductor laser as light source, Anal. Chim.
Acta 209 (1988) 327–331.
[21] Y. Fujiwara, Y. Amao, An oxygen sensor based on fluorescence quenching of pyrene
chemisorbed layer onto alumina plates, Sensors Actuators B Chem. 89 (2003)
187–191.
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.saa.2020.118537.
References
[1] A. Shamirian, H.S. Afsari, A. Hassan, L.W. Miller, P.T. Snee, In vitro detection of hyp-
oxia using a ratiometric quantum dot-based oxygen sensor, ACS Sens. 1 (2016)
1244–1250.
[22] Y. Yuan, H. Peng, J. Ping, X. Wang, F. You, A Pyrene@Micelle sensor for fluorescent
oxygen sensing, Biomed. Res. Int. 245031 (2015).
[2] P. Rassu, L. Malfatti, D. Carboni, M. Casula, S. Garroni, E. Zampetti, A. Macagnano, A.
Bearzotti, P. Innocenzi, Mesoscale organization of titania thin films enables oxygen
sensing at room temperature, J. Mater. Chem. C Mater. Opt. Electron. Devices 5
(2017) 11815–11823.
[23] B.J. Basu, K.S. Rajam, Comparison of the oxygen sensor performance of some pyrene
derivatives in silicone polymer matrix, Sensors Actuators B Chem. 99 (2004)
459–467.