Bi 2 MoO 6 Nanostrip Networks for Enhanced Visible-Light Photocatalytic Reduction of CO 2 to CH 4
Yuan Zhang,
[b, c]Liang Li,
[b, c]Qiutong Han,
[b, c]Lanqin Tang,
[b, c, d]Xingyu Chen,
[b, c]Jianqiang Hu,
[c]Zhaosheng Li,
[c, e]Yong Zhou,*
[a, b, c, e]Junming Liu,*
[b]and Zhigang Zou
[b, c, e]1. Introduction
Carbon dioxide, which leads to the greenhouse effect, is the main product obtained using fossil fuels as energy source.
While fossil fuels are not a renewable energy source, solar energy is inexhaustible. Thus, photocatalytic reduction of CO2
to hydrocarbons using solar energy is seen as hitting two birds with one stone.[1,2] Bismuth molybdates are of the general chemical formula Bi2O3·nMoO3, where n=3, 2, or 1, corre- sponding to thea,b, andgphases, respectively.[3,4]The struc- tures of theaandbphases can be considered as defective flu- orite structures, whereas the g phase is an Aurivillius type structure, which consists of alternating layers of (Bi2O2)2+ and (MoO4)2@.[3–5]Thegphase is of particular interest, since this ma-
terial is not only a good catalyst, but also shows an unusual phase transformation above 6008C to the g’phase which fur- ther transforms to the g’’ phase.[3] The g-Bi2MoO6, known as
“koechlinite”,[4] is a kind of visible-light-responsive photocata- lyst, widely being applied in the degradation of organic pollu- tants,[6–19]photocatalytic oxygen evolution,[20–23] and the photo- catalytic reduction of CO2.[24] Many g-Bi2MoO6 morphologies were reported, including nanosheets, nanoplates, flakes or nanobelts,[6–13,20] hollow boxes,[14] nanotubes,[15] hollow spher- es,[16,17]nanoparticles,[18] and flowerlike hierarchical micro- spheres.[19,24] Hierarchical architectures, building of nanorods, nanoplates, nanotubes etc., have received wide attention in photocatalysis.[24–41]Properly designed hierarchical structures (including porosity and morphology) can not only enhance light harvesting and improve molecular diffusion/transport ki- netics, but also increase the surface area and the amount of active sites, which accelerate the surface reaction kinetics.[2]
In this work,g-Bi2MoO6nanostrip networks were synthesized through a hydrothermal route using sodium oleate as a surfac- tant. This unique 3D structure exhibits enhanced photocatalyt- ic reduction of CO2into CH4, compared with its nanoplate ana- logue synthesized in the absence of sodium oleate and with the solid-state reaction. The good photocatalytic performance is assigned to its high specific surface area, high light-absorp- tion intensity, thin thickness of the nanostrip for fast charge- carrier migration, and pores for reactant transport.
2. Results and Discussion
Figure 1 shows the XRD patterns of the various Bi2MoO6mate- rials, that is, S0, synthesized via traditional solid-state reaction, and S1 and S2, obtained via hydrothermal methods in the ab- sence and presence of sodium oleate as a surfactant, respec- tively. All diffraction peaks of the three patterns can be well in- A three-dimensional Bi2MoO6 nanostrip architecture was syn-
thesized by the hydrothermal method using sodium oleate as a surfactant. The generated Bi2MoO6nanostrips intercross with each other to form a unique network structure with a band gap of 2.92 eV, corresponding to visible-light wavelength.
Time-evolution experiments reveal the formation mechanism of the Bi2MoO6 network. The photocatalytic reduction of CO2
to CH4 catalyzed by the Bi2MoO6 architecture was evaluated
and compared with the process catalyzed by a Bi2MoO6 nano- plate analogue synthesized in the absence of sodium oleate as well as with the solid-state reaction. The Bi2MoO6 nanostrips exhibit the best photocatalytic activity, which can be attributed to their high specific surface area, high light-absorption inten- sity, suitable thickness for fast charge-carrier migration, and the presence of pores for reactant transport.
[a]Prof. Y. Zhou
Key Laboratory of Modern Acoustics (MOE) Institute of Acoustics, Department of Physics Nanjing University, Nanjing 210093 (P. R. China) E-mail: [email protected]
[b]Y. Zhang, L. Li, Q. T. Han, Dr. L. Q. Tang, X. Y. Chen, Prof. Y. Zhou, Prof. J. M. Liu, Prof. Z. G. Zou
School of Physics
National Laboratory of Solid State Microstructures Collaborative Innovation Center of Advanced Microstructures Nanjing University, Nanjing 210093 (P. R. China)
E-mail: [email protected]
[c] Y. Zhang, L. Li, Q. T. Han, Dr. L. Q. Tang, X. Y. Chen, J. Q. Hu, Prof. Z. S. Li, Prof. Y. Zhou, Prof. Z. G. Zou
Eco-Materials and Renewable Energy Research Center (ERERC) Nanjing University, Nanjing 210093 (P. R. China)
[d]Dr. L. Q. Tang
College of Chemistry and Chemical Engineering Yancheng Institute of Technology
Yancheng 22401 (P. R. China)
[e]Prof. Z. S. Li, Prof. Y. Zhou, Prof. Z. G. Zou Kunshan Sunlaite New Energy Co. Ltd.
Kunshan Innovation Institute of Nanjing University Kunshan, Jiangsu 215347 (P. R. China)
An invited contribution to a Special Issue on CO2Utilisation
dexed as the orthorhombic koechlinite phase of g-Bi2MoO6
(JCPDS 21-0102), and no impurity peaks can be detected. The relatively broad diffraction peaks demonstrate the nanocrystal- line nature of S1 and S2, compared with S0, whose size is sev- eral micrometers.
Scanning electron microscopy (SEM) and transmission elec- tron microscopy (TEM) were utilized to characterize the mor- phology of the samples. S0 shows an irregular morphology with large sizes of several micrometers (Figure 2a). S1 displays thin nanoplates with thicknesses of 10–20 nm, determined through standing nanoplates (Figure 2b). S2 shows a network structure with nanostrips as building blocks (Figure 2c–f). The typical thickness of the nanostrips is measured to be about 10 nm (Figure 2e). These nanostrips aggregated and inter- crossed to generate an interconnected porous network with high specific surface area, which not only enhances the effi- ciency of light harvesting and adsorption of reactants, but also facilitates the transport of guest species to the binding sites.[2]
The TEM image further demonstrates the nanoplate mor- phology of S1 with light contrast (Figure 3a). The HRTEM image also reveals the crystal lattice of (002) of the nanoplate lying down (Figure 3b). The corresponding selective area elec- tron diffraction (SAED) pattern displays the clear diffraction spots with well alignment (Figure 3c), indicative of the single crystal structure of the nanoplates. The diffraction pattern of [010] zone axis demonstrates that the main basal facet of the nanoplate is (010) facet. This is further confirmed with the HRTEM image of the cross section of a vertically standing nanoplate, with which the spacing of lattice fringes corre- sponds to (020) facet (Figure 3d). The TEM image demon- strates that S2 displays nanostrip morphology (Figure 3e and 3 f). The HRTEM image also reveals the crystal lattice of (200) of the nanostrip along longitudinal direction (Figure 3g). The SAED pattern shows that the basal facet of the nanostrip was determined (010) facet (Figure 3h), further confirmed by the HRTEM image of the cross section of a vertically standing nanostrip (Figure 3i).
To investigate the formation mechanism of the unique Bi2MoO6structure of S2, intermediate products were gathered at different temporal stages and analyzed by FE-SEM and XRD (Figure 4). In the first 90 minutes, tiny particles grew up and/or
aggregated into large particles and formed a crystalline phase Bi3.64Mo0.36O6.55with smooth surface (Figure 4a–c). With subse- quent 30 minutes, Bi3.64Mo0.36O6.55 gradually transferred into Bi2MoO6initially formed on the surface (Figure 4d), confirmed with the corresponding XRD pattern showing several tiny peaks corresponding to Bi2MoO6 (Figure 4 f). Finally, Bi2MoO6
was completely formed with consuming the Bi3.64Mo0.36O6.55
precursor crystalline phase. In the absence of sodium oleate, generating Bi2MoO6 nanostrip prefers to grow into nanoplate morphology of S1. In contrast, with sodium oleate, the repul- sive force between hydrophobic surfactant molecules may pre- vent the nanostrip from growing into the nanoplates along [100]. Instead, these nanostrips intercross with each other to assemble into 3D network structures of S2.
The UV/Vis absorption edges of S0, S1, and S2 expand to the visible-light region (Figure 5a), indicating the visible-light activity of the photocatalysis. It is well known that the optical absorption near the band edge follows the formula ahn=
A(hn@Eg)n/2, wherea,h,n,Eg, andAare the absorption coeffi- Figure 1.XRD patterns of S0 (synthesized via traditional solid-state reaction),
S1 (synthesized via hydrothermal method), and S2 (synthesized via hydro- thermal method using sodium oleate as a surfactant).
Figure 2.SEM images of: a) S0, b) S1, c–f) S2 at different magnifications.
g) The structure model of S2.
cient, Planck constant, light frequency, band gap, and a con- stant, respectively. Among them,nis decided by the character- istics of the transition in a semiconductor.g-Bi2MoO6is a direct band gap semiconductor, meaning the valve ofnis 1.[42] Thus (ahn)2 is linear with hn. The band gaps determined from this linear relationship are 2.69 eV for S0 and 2.92 eV for both S1 and S2 (Figure 5b).
The band gaps of S1 and S2 are larger than that of S0, possi- bly due to the quantum-size effect resulting from the thin thicknesses. The absorption intensity of S2 is larger than that of S1, demonstrating the better light-absorption nature. The 3D network architecture of S2 may act as a photon trap-well to allow the multi-scattering of incidence light for the en- hancement of light absorption,[2]as schematically illustrated in Figure 2g.
The conduction band (CB) bottom and valence band (VB) top of bulkg-Bi2MoO6are estimated to be [email protected] and 2.37 V (vs. NHE), respectively.[43] Thus, the photogenerated electrons in the CB possess the ability to reduce CO2to CH4by the reac- tion CO2+8e@+8H+!CH4+2H2O (E8redox[email protected] V vs. NHE);
the photogenerated holes in the VB can oxidize water: H2O!
1/2O2+2H++2e@(Eoredox=0.82 V vs. NHE). Catalysts had been loaded carefully to exhibit their best performance.[44,45]Figure 6 shows the photocatalytic activities of S0, S1, and S2 under visi- ble light (>400 nm) irradiation of Xe lamp. S0 only yielded trace amount of CH4, possibly due to its very small specific sur- face area 0.037 m2g@1and its very big size, though it is of high light absorption intensity in the visible light region. The small specific surface area provides few reaction active sites, and big particle size lengthens the path for charge carriers to migrate to surface. Both S1 and S2 behave better photocatalytic activi- ties than S0, due to their large surface area and thin thickness- es. The higher photocatalytic activity of S2 than S1 may be as- cribed to the following three reasons: 1) S2 possesses a larger
specific surface area of 5.94 m2g@1compared to sample S1 (of 3.80 m2g@1). The larger specific surface area of S2 provides more reaction active sites and adsorbs more reactants. CO2ad- sorbed quantity for S1 is detected 1.0215 cm3g@1 (STP) and 2.711 cm3g@1 (STP) for S2. 2) The light-absorption intensity of S2 is larger than that of S1, according to the UV/Vis absorption spectra (Figure 5a), due to multi-scattering of the incident light, allowing that more electron-hole pairs can be excited in S2. The hierarchically porous core–shell and hollow photocata- lysts could increase the number of light traveling paths, which results in the enhancement of interaction time and absorption efficiency inside pores as compared to the nonporous photo- catalysts.[2] 3) The interconnected porous networks of S2 can also result in the creation of more efficient channels for the transport of reactant molecules to reactive sites present on the pore walls, which facilitate diffusion.[2] Under full-spectrum light irradiation, the photocatalytic activity of S2 can further in- Figure 3.TEM images of: a) S1, e–f) S2. HRTEM images of: b,d) S1 and g,i) S2.
SAED patterns of: c) S1 and h) S2.
Figure 4.SEM images (a–e) and XRD patterns (f) of the intermediate during the formation of S2 at 30, 60, 90, 120, and 150 min.
crease (Figure 6, line d). In addition, loading Pt as a cocatalyst behaving as electron traps also facilitates the separation of electrons and holes to intensely improve the CH4 yield (Figure 6, line e). Thus, an ideal photocatalytic activity of 1.15mmolg@1 under overall light irradiation for four hours was achieved. To directly compare the photocatalytic activities, Fig- ure 6 f illustrates the CH4 yields in the first hour. We also de- tected neglectable CO, and the yield didn’t increase with time increasing.
3. Conclusions
A 3D g-Bi2MoO6 network containing nanostrips as building blocks was synthesized via a hydrothermal method using sodium oleate as a surfactant. Its unique porous nature makes it possess a large specific surface area, high light-absorption in- tensity and improved reactant diffusion/transport. All these characteristics, combined with its thin thickness and Pt loading to facilitate charge-carrier separation, make this as-prepared hi- erarchicalg-Bi2MoO6material an ideal visible-light-driven pho- tocatalyst for the photocatalytic reduction of CO2to CH4. After four hours overall light irradiation, the best photocatalytic ac- tivity was 1.15mmolg@1.
Experimental Section
Material Preparation
All reagents used here were analytically pure and commercially available and used without further purification. The reference sample S0 was obtained by the solid-state reaction (SSR) of raw materials of Bi2O3 and MoO3 at 5508C for 40 hours.[46] To obtain sample S1, 0.5 mmol of Bi(NO3)3·5H2O and 0.25 mmol of Na- MoO4·2H2O were added to 40 mL of deionized water and stirred for 30 minutes. The solution was then transferred to a 60 mL auto- clave. The autoclave was sealed and kept at 1408C for 24 hours for hydrothermal reaction. The products were washed with deionized water and centrifuged several times. The precipitates were dried in an oven at 608C in air. While to obtain the sample S2, 0.5 mmol of Bi(NO3)3·5H2O, 0.25 mmol of NaMoO4·2H2O and 0.08 g of sodium oleate were added to 40 mL of deionized water and stirred for 30 minutes. The solution was then transferred to a 60 mL autoclave.
The autoclave was sealed and kept at 1408C for 24 hours for hy- drothermal reaction. The products were washed with ethyl alcohol and deionized water and centrifuged several times. The precipi- tates were dried in an oven at 608C in air. In addition, S2 was ob- tained by annealing the precipitates at 3508C for 8 hours. The loading of Pt was carried out by photochemical deposition with H2PtCl6·6H2O as the Pt source and CH3OH as the reductant under irradiation from a 300 W Xe arc lamp for 8 hours with stirring.
Characterization
The crystallographic phase of the samples was determined using an X-ray diffractometer (XRD) (Rigaku Ultima III, Japan) with Cu Ka radiation (l=0.154178 nm). The XRD patterns were obtained over a scanning range of 10–808at room temperature with a scan rate of 0.28s@1 at 40 kV and 40 mA. The morphology of the samples was observed by field emission scanning electron microscopy (FE- SEM, FEI NOVA NANOSEM 230). Transmission electron microscopy Figure 5.UV/Vis absorption spectra of S0, S1, and S2.
Figure 6.Photocatalytic activities of: a) S0, b) S1, and c) S2 under visible- light irradiation. d) S2 and e) Pt-loaded S2 under full-spectrum light irradia- tion.
(TEM) images, high-resolution transmission electron microscopy (HRTEM) images and selected area electron diffraction patterns (SAED) were obtained on a JEOL JEM-2100 microscope with a LaB6
filament and an accelerating voltage of 200 kV. The specific surface area of the samples was measured by nitrogen adsorption-desorp- tion at 77 K on a surface area and porosity analyzer (Micromeritics TriStar, USA) and calculated by the BET method. The CO2adsorp- tion quantity of the samples was also measured at 273 K. The UV/
Vis diffuse-reflectance spectra were recorded with a UV-vis spectro- photometer (Shimadzu UV-2550) at room temperature and trans- formed to the absorption spectrum according to the Kubelka–
Munk relationship.
Photocatalytic Tests
In the photocatalytic reduction of CO2, 0.1 g of the as-prepared Bi2MoO6sample was uniformly dispersed on a circular glass reactor with an area of 4.2 cm2. A 300 W Xenon arc lamp was used as the light source. The filter we used is 400 nm cut-off filter to acquire visible light. The volume of the reaction system was approximately 230 mL. The reaction setup was vacuum-treated several times, and then high purity CO2 gas was flowed into the reaction setup to reach ambient pressure. Then, 0.4 mL of deionized water was in- jected into the reaction system as the reducer. The as-prepared photocatalysts were allowed to equilibrate in the CO2/H2O atmos- phere for several hours to ensure that the adsorption of gas mole- cules was complete. A gas pump was used to accelerate gas diffu- sion, and circulating cooling water was used to guarantee the reac- tion was carried out at room temperature. During irradiation, ap- proximately 1 mL of gas was continually taken from the reaction cell at given time intervals for subsequent CH4concentration analy- sis using a gas chromatograph (GC-2014, Shimadzu Corp., Japan) equipped with an FID detector and a Plot Q capillary column (30 m V0.53 mmV20mm). Nitrogen gas was used as the carrier, and the temperature of the FID was approximately 2008C.
Acknowledgements
This work was supported by 973 Programs (No. 2014CB239302 and 2013CB632404), NSF of China (No. 21473091, 21773114 and 21603183), NSF of Jiangsu Province (No. BK20171246, BK2012015 and BK20130425), and Jiangsu Postdoctoral Science Foundation (1601062B).
Conflict of interest
The authors declare no conflict of interest.
Keywords: carbon dioxide · methane · nanostrip · photocatalysis·visible light
[1] B. Luo, G. Liu, L. Z. Wang,Nanoscale2016,8, 6904 – 6920.
[2] X. Li, J. G. Yu, M. Jaroniec,Chem. Soc. Rev.2016,45, 2603– 2636.
[3] L. M. Reilly, G. Sankar, C. R. A. Catlow,J. Solid State Chem. 1999, 148, 178– 185.
[4] R. Murugan, R. Gangadharan, J. Kalaiselvi, S. Sukumar, B. Palanivel, S.
Mohan,J. Phys. Condens. Matter2002,14, 4001 – 4010.
[5] Y. H. Shi, S. H. Feng, C. S. Cao,Mater. Lett.2000,44, 215 –218.
[6] H. H. Li, K. W. Li, H. Wang,Mater. Chem. Phys.2009,116, 134 –142.
[7] J. J. Mu, G. H. Zheng, Y. Q. Ma,J. Electron. Mater.2017,46, 596 –601.
[8] X. Zhao, T. G. Xu, W. Q. Yao, Y. F. Zhu,Appl. Surf. Sci.2009,255, 8036 – 8040.
[9] L. J. Xie, J. F. Ma, G. J. Xu,Mater. Chem. Phys.2008,110, 197– 200.
[10] J. H. Bi, L. Wu, J. Li, Z. H. Li, X. X. Wang, X. Z. Fu,Acta Mater.2007,55, 4699 –4705.
[11] L. W. Zhang, T. G. Xu, X. Zhao, Y. F. Zhu,Appl. Catal. B2010,98, 138 – 146.
[12] J. L. Long, S. C. Wang, H. J. Chang, B. Z. Zhao, B. T. Liu, Y. G. Zhou, W.
Wei, X. X. Wang, L. Huang, W. Huang,Small2014,10, 2791 –2795.
[13] L. Zhou, M. M. Yu, J. Yang, Y. H. Wang, C. Z. Yu,J. Phys. Chem. C2010, 114, 18812 – 18818.
[14] M. Shang, W. Z. Wang, J. Ren, S. M. Sun, L. Zhang,Nanoscale2011,3, 1474 –1476.
[15] J. Zhao, Q. F. Lu, C. Q. Wang, S. W. Liu,J. Nanopart. Res.2015,17, 189.
[16] G. H. Tian, Y. J. Chen, W. Zhou, K. Pan, Y. Z. Dong, C. G. Tian, H. G. Fu,J.
Mater. Chem.2011,21, 887– 892.
[17] Y. C. Miao, G. F. Pan, Y. N. Huo, H. X. Li,Dyes Pigm.2013,99, 382 –389.
[18] L. Zhou, W. Z. Wang, L. S. Zhang,J. Mol. Catal. A2007,268, 195 –200.
[19] N. Y. Fan, Y. J. Chen, Q. M. Feng, C. Wang, K. Pan, W. Zhou, Y. Li, H. G.
Hou, G. F. Wang,J. Mater. Res.2012,27, 1471 –1475.
[20] J. Q. Yu, A. Kudo,Chem. Lett.2005,34, 1528– 1529.
[21] Y. Shimodaira, H. Kato, H. Kobayashi, A. Kudo,J. Phys. Chem. B2006, 110, 17790 – 17797.
[22] H. Kato, M. Hori, R. Konta, Y. Shimodaira, A. Kudo,Chem. Lett.2004,33, 1348 –1349.
[23] A. Kudo, S. Hijii,Chem. Lett.1999,28, 1103– 1104.
[24] W. L. Dai, J. J. Yu, H. Xu, X. Hu, X. B. Luo, L. X. Yang, X. M. Tu,CrystEng- Comm2016,18, 3472 –3480.
[25] J. Jin, J. G. Yu, D. P. Guo, C. Cui, W. K. Ho,Small2015,11, 5262 –5271.
[26] H. Zhou, J. J. Guo, P. Li, T. X. Fan, D. Zhang, J. H. Ye,Sci. Rep.2013,3, 1667.
[27] J. S. Chen, S. Y. Qin, G. X. Song, T. Y. Xiang, F. Xin, X. H. Yin,Dalton Trans.
2013,42, 15133 –15138.
[28] B. Z. Fang, A. Bonakdarpour, K. Reilly, Y. L. Xing, F. Taghipour, D. P. Wilkin- son,ACS Appl. Mater. Interfaces2014,6, 5488 –15498.
[29] S. W. Liu, J. Q. Xia, J. G. Yu,ACS Appl. Mater. Interfaces2015,7, 8166 – 8175.
[30] M. F. Ehsan, T. He,Appl. Catal. B2015,166– 167, 345 –352.
[31] B. Z. Fang, Y. L. Xing, A. Bonakdarpour, S. C. Zhang, D. P. Wilkinson,ACS Sustainable Chem. Eng.2015,3, 2381– 2388.
[32] H. Zhou, P. Li, J. J. Guo, R. Y. Yan, T. X. Fan, D. Zhang, J. H. Ye,Nanoscale 2015,7, 113– 120.
[33] W. L. Dai, H. Xu, J. J. Yu, X. Hu, X. B. Luo, X. M. Tu, L. X. Yang,Appl. Surf.
Sci.2015,356, 173– 180.
[34] S. Q. Zhang, L. N. Li, S. G. Zhao, Z. H. Sun, M. C. Hong, J. H. Luo,J. Mater.
Chem. A2015,3, 15764 –15768.
[35] G. J. Zhang, A. T. Su, J. W. Qu, Y. Xu,Mater. Res. Bull.2014,55, 43–47.
[36] M. F. Ehsan, M. N. Ashiq, T. He,RSC Adv.2015,5, 6186 –6194.
[37] P. Renones, A. Moya, F. Fresno, L. Collado, J. J. Vilatela, V. A. de la Pe- na O’Shea,J. CO2 Util.2016,15, 24–31.
[38] Q. S. Guo, Q. H. Zhang, H. Z. Wang, Z. F. Liu, Z. Zhao,Catal. Commun.
2016,77, 118–122.
[39] H. H. Yang, Y. Bai, T. Chen, X. Shi, Y. C. Zhu,Phys. E2016,78, 100 –104.
[40] B. Pan, Y. G. Zhou, W. Y. Su, X. X. Wang,Nano Res.2017,10, 534 –545.
[41] Q. Liu, D. Wu, Y. Zhou, H. B. Su, R. Wang, C. F. Zhang, S. C. Yan, M. Xiao, Z. G. Zou,ACS Appl. Mater. Interfaces2014,6, 2356 –2361.
[42] K. R. Lai, W. Wei, Y. T. Zhu, M. Guo, Y. Dai, B. B. Huang,J. Solid State Chem.2012,187, 103– 108.
[43] M. C. Long, W. M. Cai, H. Kisch,Chem. Phys. Lett.2008,461, 102 –105.
[44] J. Albo, A. S#ez, J. Solla-Gulljn, V. Montiel, A. Irabien, Appl. Catal. B 2015,176, 709 –717.
[45] I. Merino-Garcia, J. Albo, A. Irabien,Energy Technol.2017,5, 922–928.
[46] D. H. Galv#n, S. Fuentes, M. Avalos-Borja, L. Cota-Araiza, J. Cruz-Reyes, E. A. Early, M. B. Maple,Catal. Lett.1993,18, 273 –281.
Manuscript received: June 12, 2017 Revised manuscript received: July 18, 2017 Version of record online: August 29, 2017
