Synthesis of 5-substituted-3H-[1,3,4]-oxadiazol-2-one derivatives: a carbon dioxide route (CDR)

Abstract

A carbon dioxide route (CDR) for making biologically important 5-substituted-3H-[1,3,4]-oxadiazol-2-ones (SHOs) has been accomplished through synthesis and cyclization of a variety of hydrazides as the key intermediates. All of these hydrazides were prepared readily in 89–97% yields by reacting acid chlorides with a hydrazine monohydrate in the initial step. Then, SHOs were obtained in high yields from hydrazides by reacting them with carbon dioxide under basic conditions. More notable than the high yields, is that the present CDR process for the first time has succeeded in providing a straightforward cyclization reaction leading to SHO formation with simple reagents in ethanol solution.

---

1. Introduction

1,3,4-Oxadiazole compounds represent a large family of biologically important 5-membered heterocyclic intermediates, which are interesting and superlative to synthetic chemists all over the world.^1–5 A huge number of 2,5-disubstituted-[1,3,4]-oxadiazole containing drugs have been studied due to their significant roles in medicinal chemistry.⁶ Specificly, the development of an effective and practical organic synthesis of 5-(aryl or alkyl)-3H-[1,3,4]-oxadiazoles is a lively subject in organic research work. In our previous work we have shown an interest in the preparation of 1,3,4-oxadiazoles due to their versatility as building blocks with a wide scope of biological activities like, anti-inflammatory, antitubercular, antibacterial, antiviral, antipyretic, anticancer, central nervous system depressing, antischistosomal, analgesic, antiemetic and anticonvulsive properties.⁷ The famous use of these heterocyclic intermediates is as scaffolds in medicinal chemistry.⁸ A number of 2,5-disubstituted-[1,3,4]-oxadiazoles are of significant interest in polymer chemistry due to their luminescence properties.⁹

Generally, in synthetic procedures established for the dehydrative cyclisation of semicarbazides, hydrazides are frequently used and corrosive reagents such as concentrated sulfuric acid or POCl₃, boron trifluoride diethyl etherate or Burgess reagent, phosphorus oxychloride, polyphosphoric acid, thionyl chlorides are involved, where these reagents result in inflexible byproducts in the reaction and inadequate yields, and have no experimental diversity.^10–14 To avoid undesirable reagents in the reaction, presently we have focused our deep interest on the improvement of novel synthetic strategies to obtain attractive functionalized oxadiazoles. In our determination to chemically link these biologically important heterocyclic molecules, we wanted to develop a modern synthetic route for the preparation of 5-substituted-3H-[1,3,4]-oxadiazol-2-one derivatives as major chemical skeletons. We have paid attention to this CDR process due to the cheaper price of reagents and efficient reaction conditions without any by-products forming or any mishaps in the handling of acid chlorides.^15–18 Moreover carbon dioxide (CO₂) is a nontoxic, abundant, non-flammable, easily available greenhouse gas and a renewable carbon reservoir. To the best of our literature awareness, limited approaches exist for the synthesis of 1,3,4-oxadiazoles with the use of CO₂ gas. Conversion of CO₂ to useful organic and inorganic products is of prominence in the scientific field.^16–19 In spite of many efforts that have been made to convert CO₂, the diversity of reactions is inadequate since CO₂ is thermodynamically stable and kinetically inactive.²⁰ The investigation of these innovative reactions for the conversion of carbon dioxide into important organic products, could support green and sustainable methods which are easier to use in industry compared to restricted reagents like morpholine²¹ and chloroform.²² The use of carbon dioxide in the cyclisation of several hydrazides to afford SHOs has been found to be just as good as various basic medium catalysts.²³ However, the synthesis of SHO derivatives and their analogues of CO₂ and hydrazides have not been well reported. Herein, we discovered for the first time that atmospheric carbon dioxide could react with hydrazides to generate SHOs without any harsh conditions (Scheme 2) under atmospheric CO₂. Prepared samples were examined using ¹H NMR, IR, ¹³C NMR and UV-visible spectroscopy.

2. Materials and methods

All commercially available reagents and solvents were obtained from commercial suppliers and used without further purification. For the cyclisation reaction, all experiments were conducted in a 100 mL round-bottomed Pyrex glass flask with continuous and constant stirring at 600 rpm. Thin layer chromatography (TLC) was conducted on silica gel plates. Flash column chromatography was performed using 300–400 mesh type silica gels. Visualisation of TLC was achieved by illumination under a UV lamp (254 nm). ¹H NMR spectra were recorded on a 400 MHz spectrometer and ¹³C NMR was recorded on a 100 MHz spectrometer, tetramethylsilane (TMS) served as an internal standard. Infrared (IR) spectra were recorded with a Perkin-Elmer Spectrum One FT-IR spectrometer. Samples were prepared as potassium bromide pellets for IR characterization. The mass spectra were analyzed on a Finnigan LTQ-Orbitrap XL instrument (ESI source) and UV-vis spectroscopy was recorded using an UV-vis spectrophotometer (UV-1601, Shimadzu, Australia). All of the substrates were prepared from the known experimental procedures with minor modification.

2.1 Preparation of butanoyl hydrazide (Scheme 1)

Reports are well-established for the synthesis of hydrazides but there is still the problem of reaction handling and experimental procedure when the synthesis involves highly reactive acid halides. Herein, we provide very useful experimental methods and procedures at an extremely lower temperature of −10 °C for these well established reactions, which could be practical and useful (Fig. 1).

Fig. 1 Experimental setup for the reactions in Schemes 1 and 2 .

2.2 A careful experimental procedure for the preparation of hydrazides

A washed and well dried 100 mL, three-necked round-bottomed flask was fitted with a Teflon-coated thermocouple and contained a magnetic paddle inside the flask. 50 mL of ethanol (98%) and sodium hydroxide (0.35 g, 0.0087 mol) were added into the flask. The magnetic paddle started to stir and continuously mixed the input chemical components to obtain a solution. An aqueous solution of hydrazine monohydrate (98%, 0.40 g, 0.008 mol) was added into the flask with standard precautions in a fume-cupboard. This reaction mixture was cooled in an ice-methanol/salt bath to attain an internal cooling temperature of −10 to 0 °C. Before adding butanoyl chloride to the reaction mixture, one neck of the flask was opened to make sure there was an outlet for the simultaneously generated HCl gas vapours to escape into cupboard. This can decrease the high pressure in the RBF, to avoid accidents. Then, benzoyl chloride (0.43 g, 0.0034 mol) was added using a long range syringe needle, the needle’s tip should be dipped in the reaction mixture through the rubber-septum correctly fitted to the other neck of the RBF to create a steady flow rate into the reaction mixture over a period of one hour, keeping the same cooling temperature (Fig. 1a). A white solid suspended product was formed within 5–10 minutes. This whole reaction mass was neutralised with 10% sodium hydroxide (15 mL) solution, the generated HCl gas was always allowed to escape through the open end of the reactor, so that the lowest quantity of base was used. The reaction mixture was further taken in the ethyl acetate (20 mL) and then separated out as an organic layer. This layer was transferred to a 100 mL, single necked flask and concentrated to a volume of ca. 10 mL by rotary evaporation (at ca. 10 mm) and finally the suspension was filtered and further concentrated to 5 mL and then 25 mL of toluene was added. The resulting liquid was taken in a double-necked round bottom flask which was equipped with a dean-stalk liquid separator and thermometer to separate out the azeotropic mixtures at the constant temperature of 110 °C (Fig. 1b). Finally, we obtained a shiny powder at the bottom of the RBF, after the complete evaporation of aqueous and organic solvents in the flask. This solid product was recrystallized from 10 mL of diethyl ether to obtain 0.38–0.395 g (89–97%) of propyl hydrazide. The completion of reaction was monitored using thin layer chromatography, infrared spectroscopy and ¹H NMR. Compounds 3a–j were prepared using this method and then further used to synthesize SHOs. Spectral data is summarized in ESI S1–S20.†

2.3 Preparation of 5-(p-methoxy)-3H-[1,3,4]-oxadiazol-2-one

Compound 3c (0.12 g, 0.0064 mol) was dissolved in ethanol (10 mL) in a 100 mL three necked round bottom flask with a magnetic paddle inside to mix the reaction mass. The solution was vigorously stirred. Then, strong base KOH (0.37 g, 0.0044 mol) was added into the reactor and stirred for 15 minutes to obtain a homogeneous solution in the flask at room temperature. Carbon dioxide was passed into the reactor through the other neck of the flask by using a gas purging glass tube while its tip was dipped in the reaction mixture and the temperature was maintained at 50 °C for six hours (Fig. 1c). Then the CO₂ inlet was removed and the reactor neck was closed and the reaction mass was refluxed for 30 minutes to obtain a crude product. This crude product was dissolved in 10–15 mL of water and quenched with dilute HCl (10%). The precipitate was filtered off, dried and the yield was 95% (m.p. 110 °C). It was further taken in ethanol and recrystallized. The completion of the reaction was monitored using thin layer chromatography and an infrared spectrophotometer.

The same procedure was applied for the selected alkanoyl hydrazides (3j) or aroyl hydrazides (3a–i) to obtain 5-substituted-1,3,4-oxadiazole-2-ones (4j) or (4a–i).

3. Results and discussion

3.1 Chemistry

The title molecules were synthesized, by ring closure reactions of various alkanoyl or aroyl hydrazides (3a–j, Scheme 1) with carbon dioxide in the presence of potassium hydroxide, in excellent yields (4a–j, Scheme 2). The substrate hydrazides (3a–j) were synthesized by dehydrochlorination of various aromatic and non-aromatic acid chlorides with hydrazine monohydrate under strong basic medium conditions.

Scheme 1

Scheme 2

3.2 Synthesis

In continuation of our effort to investigate the extent of the application scope of 5-membered cyclisation reactions, we intend to use oxadiazole compounds rather than isoxazoles.²⁴ To our surprise, when the reaction of hydrazine hydrate (2) and acid chlorides (1a–j) was carried out in the presence of sodium hydroxide at −10 °C instead of −5 to 0 °C it afforded hydrazide 3a–j within 5–10 minutes. It demonstrated that the modified procedure for the synthesis of hydrazides is more feasible and useful than the known methods with our experimentally proven conditions (Table 1). It is obvious that the basic medium plays an important role in this process. Hence, the possibility and the limitation of this procedure was explored by maintaining the temperature at −10 °C and quantity of the base at 0.0087 mol. All the selected acid chlorides (1a–j), such as alkanoyl or aroyl acid chlorides, were smoothly reacted with hydrazine hydrate and sodium hydroxide to afford the corresponding hydrazides in good to excellent yields (Table 2) under the mentioned optimized conditions in Table 1. The electron donating substituent groups on the aromatic acid chlorides slightly influenced the reaction to afford higher yields (Table 2, entries 1, 3, 5, 6 and 8). Whereas the electron withdrawing groups on the acid chlorides result in a slight decrease in the yields of the corresponding final products (Table 2, entries 2, 4, 7 and 9). Whereas in the case of entry 10 (Table 2), there was a positive inductive effect (+I) towards the nitrogen in its corresponding substrate (butanoyl chloride) resulting in the higher yield of the product. Variation of the reaction temperature from −10 °C to 0 °C shows a destructive influence on both reaction rates and yield. It is noteworthy that the yields were not improved when the reaction was carried out at 40 °C, RT, 0 °C and −5 °C, when compared to that at −10 °C.

Table 1 An optimal condition for the reaction of (p-methyl)-benzoyl chloride and hydrazine monohydrate to afford 3e

Entry	NaOH^a (mol)	Temperature (°C)	Yield (%)
a The reaction used anhydrous NaOH.
1	0.001	40	None
2	0.001	RT	10
3	0.001	0	70
4	0.0011	0	74
5	0.0018	−5	80
6	0.002	−5	83
7	0.004	−5	85
8	0.0060	−10	90
9	0.0082	−10	97
10	0.0087	−10	98

---

Table 2 Synthesis of 3a–j by using hydrazine monohydrate with alkanoyl or aroyl halides

Entry	R	Product	Conversion^a (%)	Yield^b (%)
a Conversion was based on NMR.b Isolated yields were based on NMR.
1		3a	96	94
2		3b	90	90
3		3c	98	96
4		3d	93	89
5		3e	98	96
6		3f	96	92
7		3g	95	91
8		3h	99	97
9		3i	96	93
10	Propyl	3j	97	95

---

The formation of butanoyl hydrazide was confirmed by its IR spectra (Fig. 2), ¹H NMR spectrum (Fig. S19†) and UV-spectrum(Fig. 3). The same reaction conditions were also applied fruitfully in the synthesis of other aroyl hydrazides. All hydrazides prepared for this series are compiled in Table 2.

Fig. 2 Comparative IR-spectra of hydrazine monohydrate, butanoyl chloride and butanoyl hydrazide.

Fig. 3 A comparative UV-spectra for hydrazine monohydrate, butanoyl chloride, butanoyl hydrazide and 5-propyl-1,3,4-oxadiazole.

3.3 Cyclisation of selected hydrazides into 1,3,4-oxadiazoles

In 1998, J. R. Young showed a cyclisation reaction of aryl iodides with amidoximes under carbon monoxide (CO) with palladium catalysts in triethylamine/organic solvent forming oxadiazoles.^25–27 Nevertheless, the utilized reaction conditions and chemicals were not environmentally benign and were expensive.

In this study we improved the preparation and procedures used to carry out the cyclisation of hydrazides with a greener carbon dioxide, and gas tested different reaction conditions to find the best way for the production of oxadiazoles with the lowest price and with greener reaction conditions.

After obtaining acceptable results from the synthesis of corresponding acid hydrazides, we then turned our attention to the preparation of SHOs (4a–j). To do so, we chose benzohydrazide to optimize the reaction conditions carefully (Scheme 3). The effects of different reaction conditions were investigated and are summarized in Table 3. From the literature survey, we found ethanol was the optimal solvent and potassium hydroxide the optimal base. We were deeply focussed on the quantity of potassium hydroxide and the temperature in the study of optimized conditions for the conversion of benzohydrazide to 5-phenyl-3H-[1,3,4]-oxadiazol-2-one. We found that the conventional temperature is 50 °C. The required quantity of potassium hydroxide is 0.0044 mol in ethanol in the CDR procedure (Table 3). The cyclization was carried out by purging carbon dioxide gas into the reactor, which contains the hydrazide, potassium hydroxide and ethanol to obtain the title compounds at constant temperature (50 °C) with a maximum time period of 6 hours. The scope of this simple CDR process was explored using a collection of structurally diverse organic acid chlorides and optimized with benzohydrazide (Scheme 3) and Table 3. Hydrazides reacted with carbon dioxide to form the organic intermediates 4a–j, which lost a mole ratio of potassium hydroxide to afford the closed-ring intermediate 4a–j.

Scheme 3

Table 3 Optimization of reaction conditions for the synthesis of 5-phenyl-3H-[1,3,4]-oxadiazol-2-one by using 3a

Entry	KOH^a (mol)	Solvent	Temperature (°C)	Yield (%)
a The reaction used anhydrous KOH.
1	0.001	Ethanol	0	None
2	0.001	Ethanol	RT	10
3	0.001	Ethanol	35	76
4	0.0044	Ethanol	40	73
5	0.0044	Ethanol	45	91
6	0.0044	Ethanol	50	96

---

In all of the case studies, no difficulty was found whenthe substrates were active organic halides (Table 4, entries 1, 3, 5, 6 and 8), which contained methyl or methoxy groups, or the substrates of inactive organic halides (Table 4, entries 2, 4 and 7), such as phenyl rings with halo or nitro substituents, the reactions proceeded smoothly to afford the desired SHOs (4a–j) in good to excellent yields. All the results characterization were completely matched with those of previous reports, and our technique is efficient and environmentally benign. The plausible reaction mechanism for the formation of 5-substituted-3H-[1,3,4]-oxadiazol-2-ones is proposed in Scheme 4.

Scheme 4 A plausible mechanism for the total synthesis.

Table 4 Synthesis of 5-substituted-1,3,4-oxadiazol-2-ones^a

Entry	Corresponding hydrazides (3a–j of Scheme 1)	Product	Time (hours)	Yield (%)
a Isolated yields were based proton NMR.
1	3a	4a	5	92
2	3b	4b	4	89
3	3c	4c	5	95
4	3d	4d	6	95
5	3e	4e	6	93
6	3f	4f	6	90
7	3g	4g	6	87
8	3h	4h	6	93
9	3i	4i	6	90
10	3j	4j	6	95

---

3.4 Characterization

The FTIR-spectra (Fig. 2) of unmodified hydrazine monohydrate shows absorption peaks referring to the stretching of hydroxyl (–OH) and amine (–NH) groups at wavenumbers of 3400–3500 cm⁻¹ and 3000 cm⁻¹ (shifted to lower than the hydroxyl group due to the nearby regions of the amine and hydroxyl groups) respectively. These data are consistent with those of other studies.^28–30 Whereas our selected unmodified substrate butanoyl chloride (1j) shows absorption peaks of acyl (–COCl) and aliphatic (–CH₂) groups at 1725–1700 and 2982 cm⁻¹ respectively. The product 3j shows absorption peaks that refer to stretching of the amide (–CONH), amine (–NH₂) and aliphatic (CH₂) groups at 1610 cm⁻¹, 3200 cm⁻¹ and 2980–2990 cm⁻¹ respectively. The FTIR spectra (Fig. 4) of the product 4j shows absorption peaks that were assigned to the amide group at (–CONH) 1625 cm⁻¹, aliphatic group at (CH₂) 2995 cm⁻¹, carbonyl group at (COONH) 1746 cm⁻¹, and 3267–3600 cm⁻¹ (mixed NH and any other hydroxyl groups). On the other hand the comparative studies of the UV spectrum (Fig. 3) of compound 4j (5 mg in 10 mL of ethanol) showed a distinguished split wavelength range in the visible region (200–330 nm) compared to the other compounds: hydrazine monohydrate (2) (5 mg in 10 mL of ethanol), butanoyl chloride (1j) (5 mg in 10 mL of ethanol) and butanoyl hydrazide (3j) (5 mg in 10 mL of ethanol) which show absorption below 200 nm (5 mg in 10 mL of ethanol) and were consistent with the literature.^30,31 When we observed the NMR results of 3a–j and 4a–j, peaks clearly indicating the significant chemical shift values of the amidic group (–CONH) were above 10 ppm in the ¹H NMR spectrum and 150–170 ppm in the ¹³C NMR spectrum respectively, in both cases of Scheme 1 and Scheme 2 analogues. We summarized all the distinguished chemical shift values in both the ¹H NMR and ¹³C NMR spectra in the individual figures in the ESI (Fig. S1–S40†) for all synthesized compounds.

Fig. 4 IR-spectrum of compound 4j.

3.5 Mechanism

Finally, we found that acid chlorides could be converted completely in a shorter time (5–10 minutes) in each reaction (Scheme 1, Table 2 and step-1 in Scheme 4). But a very long time i.e. 4–6 hours was essential to reach acceptable selectivity of the preferred product of (Scheme 2, Table 4 and step-2 in Scheme 4). The time (4–6 hours) required for the final products of 3a–j to 4a–j to be synthesized clearly indicates a slower reaction. This characteristic phenomenon could show that the reactant is converted into the product of an intermediate like potassium salt of 2-oxy-5-(aryl or alkyl)-1,3,4-oxadiazol-2-one as we proposed in our theoretical assumption (plausible mechanism, Scheme 4). In this work, we studied the reaction of CO₂ with hydrazides (3a–j) in detail and confirmed the products using the results of ¹H NMR, ¹³C NMR (Fig. S1–S40†) analysis and their consistency with all simulated and obtained data from some previous reports.^32,33

4. Conclusions

In summary, we have discovered a two-step synthetic process for the preparation of 5-substituted-3H-[1,3,4]-oxadiazol-2-ones (SHOs) by using a simple CDR process, where carbon dioxide can smoothly react with hydrazides to form the targeted molecules at atmospheric pressure and in basic ethanol solution. The reaction conditions are mild, isolation of products is easy and the yields of the SHOs are consistently high. Hence, this novel CDR process might have great potential for industrial applications.

Acknowledgements

The authors are thankful to the Ministry of Science and Technology, Taiwan, for the funding support (MOST 102-2221-E-005-065-MY2 & MOST 103-2811-E-005-011). One of the authors MB wants to specially thank to S.-Y. Suen and S. A. Dai.

Notes and references

1. F. Bentiss, M. Traisnel and M. Lagrenee, Corros. Sci., 2000, 42, 127–146.
2. S. G. Kucukguzel, E. E. Oruc, S. Rollas, F. Sahin and A. Ozbek, Eur. J. Med. Chem., 2002, 37, 197–206.
3. C. T. Brain, J. M. Paul, Y. Loong and P. J. Oakley, Tetrahedron Lett., 1999, 40, 3275–3278.
4. A. Almasirad, S. A. Tabatabai, M. Faizi, A. Kebriaeezadeh, N. Mehrabi, A. Dalvandi and A. Shafiee, Bioorg. Med. Chem. Lett., 2004, 14, 6057–6059.
5. A. Zarghi, S. A. Tabatabai, M. Faizi, A. Ahadian, P. Navabi, V. Zanganeh and A. Shafiee, Bioorg. Med. Chem. Lett., 2005, 15, 1863–1865.
6. M. Al-Talib, H. Tashtoush and N. Odeh, Synth. Commun., 1990, 20, 1811–1817.
7. X.-J. Zou, L.-H. Lai, G.-Y. Jin and Z.-X. Zhang, J. Agric. Food Chem., 2002, 50, 3757–3760.
8. R. Severinsen, J. P. Kilburn and J. F. Lau, Tetrahedron, 2005, 61, 5565–5575.
9. S. Hernández-Ainsa, J. Barberá, M. Marcos and J. L. Serrano, Macromolecules, 2012, 45, 1006–1015.
10. A. Souldozi, J. Chem. Res., 2015, 39, 177–179.
11. N. B. Kumar, D. M. Kuznetsov and A. G. Kutateladze, Org. Lett., 2015, 17, 438–441.
12. J. E. Sears and D. L. Boger, Acc. Chem. Res., 2015, 48, 653–662.
13. W. Guo, K. Huang, F. Ji, W. Wu and H. Jiang, Chem. Commun., 2015, 51, 8857–8860.
14. S. J. Dolman, F. Gosselin, P. D. O’Shea and I. W. Davies, J. Org. Chem., 2006, 71, 9548–9551.
15. R. A. Neves and R. M. Srivastava, Molecules, 2006, 11, 318–324.
16. R. R. Kamble and B. S. Sudha, Chin. J. Chem., 2006, 24, 79–84.
17. V. Y. Rozhkov, L. Batog and M. Struchkova, Russ. Chem. Bull., 2005, 54, 1923–1934.
18. F. Wang, Z. Qin and Q. Huang, Front. Chem. China, 2006, 1, 112–114.
19. N. Werstiuk, A. Klys and J. Warkentin, Can. J. Chem., 2006, 84, 546–554.
20. R. R. Kamble and B. S. Sudha, J. Heterocycl. Chem., 2006, 43, 345–352.
21. C. G. Levins and Z.-K. Wan, Org. Lett., 2008, 10, 1755–1758.
22. W. J. Chu, Y. Yang and C.-F. Chen, Org. Lett., 2010, 12, 3156–3159.
23. T. V. Hansen, P. Wu and V. V. Fokin, J. Org. Chem., 2005, 70, 7761–7764.
24. M. Brahmayya, B. Venkateswararao, D. Krishnarao, S. Durgarao, U. V. Prasad, T. Damodharam and R. Mishra, J. Pharm. Res., 2013, 7, 516–519.
25. S. Vodela, R. V. R. Mekala, R. R. Danda and V. Kodhati, Chin. Chem. Lett., 2013, 24, 625–628.
26. J. R. Young and R. J. DeVita, Tetrahedron Lett., 1998, 39, 3931–3934.
27. N. Aljaar, J. R. Conrad and U. Beifuss, J. Org. Chem., 2013, 78, 1045–1053.
28. P. Zoumpoulakis, C. Camoutsis, G. Pairas, M. Soković, J. Glamočlija, C. Potamitis and A. Pitsas, Bioorg. Med. Chem., 2012, 20, 1569–1583.
29. L. H. Zou, J. Mottweiler, D. L. Priebbenow, J. Wang, J. A. Stubenrauch and C. Bolm, Chem.–Eur. J., 2013, 19, 3302–3305.
30. C. P. Horwitz, in Innovations in Green Chemistry and Green Engineering, Springer, 2013, pp. 247–295.
31. F. Kurzer and M. Wilkinson, Chem. Rev., 1970, 70, 111–149.
32. V. Padmavathi, G. S. Reddy, A. V. N. Mohan and K. Mahesh, ARKIVOC, 2008, 17, 48–60.
33. J. Hu, J. Ma, Z. Zhang, Q. Zhu, H. Zhou, W. Lu and B. Han, Green Chem., 2015, 17, 1219–1225.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra08910g

‡ The authors (co-first authors) contributed equally to this work.

---