Article pubs.acs.org/JPCB
Ionic Liquid Dependence of Triplet-Sensitized Photon Upconversion Yoichi Murakami,*,†,‡ Toshiyuki Ito,†,‡ and Akio Kawai§ †
Department of Mechanical Sciences and Engineering, Tokyo Institute of Technology, 2-12-1-I1-15 Ookayama, Meguro-ku, Tokyo 152-8552, Japan ‡ Global Edge Institute, Tokyo Institute of Technology, 2-12-1-I1-15 Ookayama, Meguro-ku, Tokyo 152-8550, Japan § Department of Chemistry, Tokyo Institute of Technology, 2-12-1-H89 Ookayama, Meguro-ku, Tokyo 152-8551, Japan S Supporting Information *
ABSTRACT: Photon upconversion (UC) is a technology used to convert wasted lower energy photons to usable higher energy photons. Triplet-sensitized UC based on the triplet−triplet annihilation (TTA) of organic molecules has recently received attention because of its applicability to noncoherent sunlight. Among the various media proposed for this UC, ionic liquids (ILs) are practically advantageous because of their nonvolatility and nonflammability. However, from previous studies, the efficiency of UC (ΦUC) has been found to depend on the ILs employed. In this article, systematic investigations were carried out on samples made using more than 10 kinds of ILs, all of which were purified before sample fabrication to enhance data reliability. Several clear tendencies were found, and they were all related to the viscosity of the ILs. We also found that the magnitude of their solvatochromic shifts did not correlate to these trends. These results show that the dynamic aspects of the molecules influence the kinetics that govern the magnitude of ΦUC. Along with related discussions and interpretations, these results should provide a guideline toward increasing the ΦUC.
■
INTRODUCTION Photon upconversion (UC) based on triplet-sensitization and triplet−triplet annihilation (TTA) of organic molecules is an emerging light conversion technology that converts presently wasted lower energy photons into usable higher energy photons.1−21 Because this UC method (TTA-UC) has the unique advantage of being applicable to low-intensity and noncoherent sunlight,2,4 it is considered to be a promising method to increase the conversion efficiencies of broad photorelated technologies such as photovoltaic12,14 and photochemical reactions.3,13 While most of the TTA-UC studies reported to date have been carried out using volatile organic solvents1−5,7−10,12−17 or hydrocarbon-based polymers,6,11,18 we have studied TTA-UC in ionic liquids (ILs),22−24 which are novel room temperature molten salts recently proposed for use in a vast number of applications because of their unique properties.25 The use of ILs provides TTA-UC with nonflammable and nonvolatile nature, which is advantageous in applications. However, it has been found that the UC quantum efficiency (ΦUC) is dependent on the ILs used.22 Therefore, to obtain a guideline to increase the ΦUC, the underlying rules as well as mechanisms that govern this IL dependence have to be understood, and this is the objective of this study. TTA-UC is carried out by combining two kinds of organic molecules: one that performs triplet sensitization (the “sensitizer”) and the other that performs TTA and the emission of an upconverted photon (the “emitter”). In this study, palladium porphyrin (PdPh4TBP, Chart 1a) and perylene (Chart 1b) were used as the sensitizer and the emitter, © 2014 American Chemical Society
Chart 1. Molecular Structures of (a) PdPh4TBP and (b) Perylene
respectively. Scheme 1 shows a qualitative energy level diagram of TTA-UC, which represents this process as follows. First, a sensitizer molecule in the ground (S0) state absorbs a lower energy photon (hν1) and becomes the lowest excited singlet (S1) state, which then undergoes intersystem crossing (ISC) to the lowest triplet (T1) state. The quantum yield of the ISC of palladium porphyrin is almost unity.26 This T1 state energy of the sensitizer molecule is then transferred to an emitter molecule in the ground (S0) state by a collision-based electronReceived: September 3, 2014 Revised: October 30, 2014 Published: November 10, 2014 14442
dx.doi.org/10.1021/jp508901d | J. Phys. Chem. B 2014, 118, 14442−14451
The Journal of Physical Chemistry B
Article
Scheme 1. Qualitative Energy Level Diagram of the Upconversion Processa
a
ISC, intersystem crossing; TET, triplet energy transfer; Phos., phosphorescence.
Table 1. List of the Samples, the Ionic Liquids Used and Their Viscosities, and Relevant Results kq/107 M−1 s−1
τUCe/ms
τTf/ms
1
3.23 3.18 3.12
1.79 1.87 1.80
4.00 4.03 3.96
28.4
0.34
8.78
0.68
1.55
Io
86.8
0.95
3.31
1.83
3.94
[C4mim][NTf2]
Io
45.7
0.59
5.91
0.96
2.05
#5
[C4mim][NTf2]
M
47.0
0.71
6.48
0.99
2.16
#6
[N3211][NTf2]
M
70.3
0.82
3.93
0.97
2.20
#7
[C3dmim][NTf2]
M
87.3
0.81
3.52
1.66
3.54
#8
[C2mim][FAP]
M
57.9
0.68
5.81
1.03
2.16
#9
[C6mim][FAP]
M
86.7
0.78
3.86
1.00
2.23
#10
[C4mpyr][NTf2]
M
71.8
0.74
3.92
1.38
3.16
#11
[C4dmim][NTf2]
M
94.7
1.09
3.35
1.78
3.77
#12
[C6mim][NTf2]
M
64.8
0.77 0.77
− 5.71
− 0.95
− 2.10
#13
[C4mpyr][FAP]
M
200
1.24 1.18 1.16
− 1.70 1.81
− 1.83 1.95
− 4.00 4.32
#14
[N8881][NTf2]
M
584
2.12 2.09
1.47 1.40
2.04 1.87
4.13 4.04
no.
ionic liquida
manufacturerb
ηc/mPa s
#1
[C4dmim][NTf2]
Io
94.5
#2
[C2mim][NTf2]
Io
#3
[C3dmim][NTf2]
#4
IUC,reld
a
[Cndmim]: 1-Alkyl-2,3-dimethylimidazolium. [Cnmim]: 1-Alkyl-3-methylimidazolium. [N3211]: Ethyldimethylpropylammonium. [C4mpyr]: 1-Butyl1-methylpyrrolidinium. [N8881]: Methyltrioctylammonium. [NTf2]: Bis(trifluoromethylsulfonyl)amide. [FAP]: Tris(pentafluoroethyl)trifluorophosphate. Water contents measured by Karl Fischer titration of the ionic liquids prepared in the same manner as the samples (see main text for details) are 29 ppm (#1), 47 ppm (#2), 19 ppm (#3), 59 ppm (#4), 26 ppm (#6), 36 ppm (#8), 18 ppm (#9), 37 ppm (#12), 119 ppm (#13), and 137 ppm (#14); the uncertainty of up to ±20−50% may exist in these values considering possible variations in sample handing and/or error in the titration measurement. bIo for IoLiTec, M for Merck. cMeasured at 26.0 ± 0.2 °C on vacuum-dried ionic liquids without solute molecules. The accuracy of the measurement was ca. ± 2%. dNormalized to that of #1. eFrom a single-exponential decay function fit to the transient UC intensity decay curve acquired in a weak excitation regime. fFrom an analytical function fit that describes the transient intensity decay of the UC fluorescence.
14443
dx.doi.org/10.1021/jp508901d | J. Phys. Chem. B 2014, 118, 14442−14451
The Journal of Physical Chemistry B
Article
#7, and #4 and #5) were made using structurally equivalent ILs from different manufacturers to confirm the validity of our results. For samples #1, #12, #13, and #14, sample fabrication and measurement were carried out more than once on different dates to ensure reproducibility, and a “sample #1” was included in each experimental batch as a standard reference. These ILs all contain either the [NTf2]− or [FAP]− anion, and these were used because our sample fabrication method requires the use of hydrophobic ILs.22 These anions are commercially wellestablished anions that impart sufficient hydrophobicity to ILs. All the ILs studied were purified before use, and the details are given in the Supporting Information. Briefly, the process started with charcoal adsorption under vacuum at 120 °C followed by the separation of the charcoal powder by centrifugation and filtration. Each IL was then repeatedly rinsed with ultrahigh purity water to remove mainly alkali metal halides, which are usually the main contaminant in ILs and can be effectively removed by water-rinsing.30 Finally, the IL was dried in two steps: First, it was predried overnight in an oven at 70 °C under atmospheric pressure and then further dried under vacuum at 120 °C for 3 to 4 h. In the latter drying process, a glass vial containing a small amount (ca. 1 mL) of the predried IL was placed in a vacuum oven (Yamato Scientific, ADP200). The oven was evacuated using an oil-free scroll pump (Edwards, nXDS15i) to ensure a clean vacuum process. The reason for this small amount is to decrease the diffusion path length to allow moisture to rapidly escape from the IL’s top surface to the vacuum. By this purification, the UV absorption by impurities (unidentified) is suppressed. Typical changes in the optical absorption are shown in Figure S1 in the Supporting Information. The samples were prepared as per our previously reported procedure,22 and the final concentrations of PdPh4TBP and perylene were 1 × 10−5 and 3 × 10−3 M, respectively. Multiple samples were prepared and measured at the same time for the same experimental batch, and a “sample #1” was included in each batch as a reference. All the samples were sealed in a square cross-section quartz tube (inner dimension: 1 mm × 1 mm) in an argon-filled glovebox. Details of the preparation procedure are described in the Supporting Information. Figure 1 shows UC fluorescence and optical absorption spectra of a typical sample made using [N8881][NTf2] (for #14). The photoemission around 475 nm is the UC fluorescence from perylene, and the absorption around 625 nm corresponds to the Q-band of PdPh4TBP. Photographs of the UC in typical samples have been shown in our previous reports.22−24
exchange process.27 This triplet energy transfer (TET) process occurs with a certain quantum efficiency ΦTET. Furthermore, when two emitter molecules in the T1 state collide by diffusion, either TTA (T1 + T1 → S1 + S0) or triplet quenching (T1 + T1 → T1 + S0) may occur. In this article, the term “TTA” is solely used for the former, which gives rise to an S1 and an S0 state based on the definition in ref 28. When this encounter results in TTA, a photon (hν2; ν2 > ν1) is emitted from the emitter in the S1 state with a fluorescence quantum yield of ΦF. As a result, one higher energy photon is generated from two or more lower-energy photons. We define ΦUC as unity when one upconverted photon is generated upon the absorption of two photons (0 ≤ ΦUC ≤ 1). ΦUC is thus expressed as the product of the efficiencies involved in Scheme 1 and is8,9 ΦUC = ΦTETΦTTA ΦF
Here, ΦTTA is the yield of TTA as follows:
ΦTTA = f2 φS
(1) 24,29
(2)
where f 2 (0 ≤ f 2 ≤ 1) denotes the ratio of the T1 emitter molecules that decay by any second-order process to those generated by TET from the sensitizer. Additionally, φS denotes the ratio of the T1 emitter molecules that decay by TTA to those that decay by any second-order process. Therefore, f 2 is 0 if the T1 → S0 decay of the emitter only occurs by a first-order spontaneous process (i.e., no second-order process), and φS is 0 if no S1 state is generated upon a collision of the two T1 emitter molecules (i.e., no TTA process). In our previous study,23 ΦTET and ΦF were found to be independent of the ILs, and thus the IL dependence of ΦUC originates from the ΦTTA in eq 1. Therefore, the kinetic parameters of the emitter such as the φS in eq 2 have to be studied. Although we showed that IL viscosity could affect φS in our previous report,24 this preliminary finding was derived from a limited number of samples made using five imidazolium ILs. Each of these ILs (ca. 400 μL held in a glass vial) was only processed by vacuum drying at 120 °C for 4 h just before sample fabrication.23,24 After the report of ref 24, we worked on developing a method of purifying hydrophobic ILs, as described in the next section. The aforementioned five ILs were investigated in our previous reports23,24 because we recognized the insignificance of impurities in those ILs, and therefore our previous conclusions have not changed. However, we noticed that the two ILs that exhibited remarkably low ΦUC (
Ionic Liquid Dependence of Triplet-Sensitized Photon Upconversion Yoichi Murakami,*,†,‡ Toshiyuki Ito,†,‡ and Akio Kawai§ †
Department of Mechanical Sciences and Engineering, Tokyo Institute of Technology, 2-12-1-I1-15 Ookayama, Meguro-ku, Tokyo 152-8552, Japan ‡ Global Edge Institute, Tokyo Institute of Technology, 2-12-1-I1-15 Ookayama, Meguro-ku, Tokyo 152-8550, Japan § Department of Chemistry, Tokyo Institute of Technology, 2-12-1-H89 Ookayama, Meguro-ku, Tokyo 152-8551, Japan S Supporting Information *
ABSTRACT: Photon upconversion (UC) is a technology used to convert wasted lower energy photons to usable higher energy photons. Triplet-sensitized UC based on the triplet−triplet annihilation (TTA) of organic molecules has recently received attention because of its applicability to noncoherent sunlight. Among the various media proposed for this UC, ionic liquids (ILs) are practically advantageous because of their nonvolatility and nonflammability. However, from previous studies, the efficiency of UC (ΦUC) has been found to depend on the ILs employed. In this article, systematic investigations were carried out on samples made using more than 10 kinds of ILs, all of which were purified before sample fabrication to enhance data reliability. Several clear tendencies were found, and they were all related to the viscosity of the ILs. We also found that the magnitude of their solvatochromic shifts did not correlate to these trends. These results show that the dynamic aspects of the molecules influence the kinetics that govern the magnitude of ΦUC. Along with related discussions and interpretations, these results should provide a guideline toward increasing the ΦUC.
■
INTRODUCTION Photon upconversion (UC) based on triplet-sensitization and triplet−triplet annihilation (TTA) of organic molecules is an emerging light conversion technology that converts presently wasted lower energy photons into usable higher energy photons.1−21 Because this UC method (TTA-UC) has the unique advantage of being applicable to low-intensity and noncoherent sunlight,2,4 it is considered to be a promising method to increase the conversion efficiencies of broad photorelated technologies such as photovoltaic12,14 and photochemical reactions.3,13 While most of the TTA-UC studies reported to date have been carried out using volatile organic solvents1−5,7−10,12−17 or hydrocarbon-based polymers,6,11,18 we have studied TTA-UC in ionic liquids (ILs),22−24 which are novel room temperature molten salts recently proposed for use in a vast number of applications because of their unique properties.25 The use of ILs provides TTA-UC with nonflammable and nonvolatile nature, which is advantageous in applications. However, it has been found that the UC quantum efficiency (ΦUC) is dependent on the ILs used.22 Therefore, to obtain a guideline to increase the ΦUC, the underlying rules as well as mechanisms that govern this IL dependence have to be understood, and this is the objective of this study. TTA-UC is carried out by combining two kinds of organic molecules: one that performs triplet sensitization (the “sensitizer”) and the other that performs TTA and the emission of an upconverted photon (the “emitter”). In this study, palladium porphyrin (PdPh4TBP, Chart 1a) and perylene (Chart 1b) were used as the sensitizer and the emitter, © 2014 American Chemical Society
Chart 1. Molecular Structures of (a) PdPh4TBP and (b) Perylene
respectively. Scheme 1 shows a qualitative energy level diagram of TTA-UC, which represents this process as follows. First, a sensitizer molecule in the ground (S0) state absorbs a lower energy photon (hν1) and becomes the lowest excited singlet (S1) state, which then undergoes intersystem crossing (ISC) to the lowest triplet (T1) state. The quantum yield of the ISC of palladium porphyrin is almost unity.26 This T1 state energy of the sensitizer molecule is then transferred to an emitter molecule in the ground (S0) state by a collision-based electronReceived: September 3, 2014 Revised: October 30, 2014 Published: November 10, 2014 14442
dx.doi.org/10.1021/jp508901d | J. Phys. Chem. B 2014, 118, 14442−14451
The Journal of Physical Chemistry B
Article
Scheme 1. Qualitative Energy Level Diagram of the Upconversion Processa
a
ISC, intersystem crossing; TET, triplet energy transfer; Phos., phosphorescence.
Table 1. List of the Samples, the Ionic Liquids Used and Their Viscosities, and Relevant Results kq/107 M−1 s−1
τUCe/ms
τTf/ms
1
3.23 3.18 3.12
1.79 1.87 1.80
4.00 4.03 3.96
28.4
0.34
8.78
0.68
1.55
Io
86.8
0.95
3.31
1.83
3.94
[C4mim][NTf2]
Io
45.7
0.59
5.91
0.96
2.05
#5
[C4mim][NTf2]
M
47.0
0.71
6.48
0.99
2.16
#6
[N3211][NTf2]
M
70.3
0.82
3.93
0.97
2.20
#7
[C3dmim][NTf2]
M
87.3
0.81
3.52
1.66
3.54
#8
[C2mim][FAP]
M
57.9
0.68
5.81
1.03
2.16
#9
[C6mim][FAP]
M
86.7
0.78
3.86
1.00
2.23
#10
[C4mpyr][NTf2]
M
71.8
0.74
3.92
1.38
3.16
#11
[C4dmim][NTf2]
M
94.7
1.09
3.35
1.78
3.77
#12
[C6mim][NTf2]
M
64.8
0.77 0.77
− 5.71
− 0.95
− 2.10
#13
[C4mpyr][FAP]
M
200
1.24 1.18 1.16
− 1.70 1.81
− 1.83 1.95
− 4.00 4.32
#14
[N8881][NTf2]
M
584
2.12 2.09
1.47 1.40
2.04 1.87
4.13 4.04
no.
ionic liquida
manufacturerb
ηc/mPa s
#1
[C4dmim][NTf2]
Io
94.5
#2
[C2mim][NTf2]
Io
#3
[C3dmim][NTf2]
#4
IUC,reld
a
[Cndmim]: 1-Alkyl-2,3-dimethylimidazolium. [Cnmim]: 1-Alkyl-3-methylimidazolium. [N3211]: Ethyldimethylpropylammonium. [C4mpyr]: 1-Butyl1-methylpyrrolidinium. [N8881]: Methyltrioctylammonium. [NTf2]: Bis(trifluoromethylsulfonyl)amide. [FAP]: Tris(pentafluoroethyl)trifluorophosphate. Water contents measured by Karl Fischer titration of the ionic liquids prepared in the same manner as the samples (see main text for details) are 29 ppm (#1), 47 ppm (#2), 19 ppm (#3), 59 ppm (#4), 26 ppm (#6), 36 ppm (#8), 18 ppm (#9), 37 ppm (#12), 119 ppm (#13), and 137 ppm (#14); the uncertainty of up to ±20−50% may exist in these values considering possible variations in sample handing and/or error in the titration measurement. bIo for IoLiTec, M for Merck. cMeasured at 26.0 ± 0.2 °C on vacuum-dried ionic liquids without solute molecules. The accuracy of the measurement was ca. ± 2%. dNormalized to that of #1. eFrom a single-exponential decay function fit to the transient UC intensity decay curve acquired in a weak excitation regime. fFrom an analytical function fit that describes the transient intensity decay of the UC fluorescence.
14443
dx.doi.org/10.1021/jp508901d | J. Phys. Chem. B 2014, 118, 14442−14451
The Journal of Physical Chemistry B
Article
#7, and #4 and #5) were made using structurally equivalent ILs from different manufacturers to confirm the validity of our results. For samples #1, #12, #13, and #14, sample fabrication and measurement were carried out more than once on different dates to ensure reproducibility, and a “sample #1” was included in each experimental batch as a standard reference. These ILs all contain either the [NTf2]− or [FAP]− anion, and these were used because our sample fabrication method requires the use of hydrophobic ILs.22 These anions are commercially wellestablished anions that impart sufficient hydrophobicity to ILs. All the ILs studied were purified before use, and the details are given in the Supporting Information. Briefly, the process started with charcoal adsorption under vacuum at 120 °C followed by the separation of the charcoal powder by centrifugation and filtration. Each IL was then repeatedly rinsed with ultrahigh purity water to remove mainly alkali metal halides, which are usually the main contaminant in ILs and can be effectively removed by water-rinsing.30 Finally, the IL was dried in two steps: First, it was predried overnight in an oven at 70 °C under atmospheric pressure and then further dried under vacuum at 120 °C for 3 to 4 h. In the latter drying process, a glass vial containing a small amount (ca. 1 mL) of the predried IL was placed in a vacuum oven (Yamato Scientific, ADP200). The oven was evacuated using an oil-free scroll pump (Edwards, nXDS15i) to ensure a clean vacuum process. The reason for this small amount is to decrease the diffusion path length to allow moisture to rapidly escape from the IL’s top surface to the vacuum. By this purification, the UV absorption by impurities (unidentified) is suppressed. Typical changes in the optical absorption are shown in Figure S1 in the Supporting Information. The samples were prepared as per our previously reported procedure,22 and the final concentrations of PdPh4TBP and perylene were 1 × 10−5 and 3 × 10−3 M, respectively. Multiple samples were prepared and measured at the same time for the same experimental batch, and a “sample #1” was included in each batch as a reference. All the samples were sealed in a square cross-section quartz tube (inner dimension: 1 mm × 1 mm) in an argon-filled glovebox. Details of the preparation procedure are described in the Supporting Information. Figure 1 shows UC fluorescence and optical absorption spectra of a typical sample made using [N8881][NTf2] (for #14). The photoemission around 475 nm is the UC fluorescence from perylene, and the absorption around 625 nm corresponds to the Q-band of PdPh4TBP. Photographs of the UC in typical samples have been shown in our previous reports.22−24
exchange process.27 This triplet energy transfer (TET) process occurs with a certain quantum efficiency ΦTET. Furthermore, when two emitter molecules in the T1 state collide by diffusion, either TTA (T1 + T1 → S1 + S0) or triplet quenching (T1 + T1 → T1 + S0) may occur. In this article, the term “TTA” is solely used for the former, which gives rise to an S1 and an S0 state based on the definition in ref 28. When this encounter results in TTA, a photon (hν2; ν2 > ν1) is emitted from the emitter in the S1 state with a fluorescence quantum yield of ΦF. As a result, one higher energy photon is generated from two or more lower-energy photons. We define ΦUC as unity when one upconverted photon is generated upon the absorption of two photons (0 ≤ ΦUC ≤ 1). ΦUC is thus expressed as the product of the efficiencies involved in Scheme 1 and is8,9 ΦUC = ΦTETΦTTA ΦF
Here, ΦTTA is the yield of TTA as follows:
ΦTTA = f2 φS
(1) 24,29
(2)
where f 2 (0 ≤ f 2 ≤ 1) denotes the ratio of the T1 emitter molecules that decay by any second-order process to those generated by TET from the sensitizer. Additionally, φS denotes the ratio of the T1 emitter molecules that decay by TTA to those that decay by any second-order process. Therefore, f 2 is 0 if the T1 → S0 decay of the emitter only occurs by a first-order spontaneous process (i.e., no second-order process), and φS is 0 if no S1 state is generated upon a collision of the two T1 emitter molecules (i.e., no TTA process). In our previous study,23 ΦTET and ΦF were found to be independent of the ILs, and thus the IL dependence of ΦUC originates from the ΦTTA in eq 1. Therefore, the kinetic parameters of the emitter such as the φS in eq 2 have to be studied. Although we showed that IL viscosity could affect φS in our previous report,24 this preliminary finding was derived from a limited number of samples made using five imidazolium ILs. Each of these ILs (ca. 400 μL held in a glass vial) was only processed by vacuum drying at 120 °C for 4 h just before sample fabrication.23,24 After the report of ref 24, we worked on developing a method of purifying hydrophobic ILs, as described in the next section. The aforementioned five ILs were investigated in our previous reports23,24 because we recognized the insignificance of impurities in those ILs, and therefore our previous conclusions have not changed. However, we noticed that the two ILs that exhibited remarkably low ΦUC (
