Panchromatic Photoproduction of H2 with Surface Plasmons - Nano

Feb 9, 2015 - ... solar spectrum of the best performing plasmonic multielectrode array was .... Broadband Hot-Electron Collection for Solar Water Spli...
2 downloads 0 Views 4MB Size
Letter pubs.acs.org/NanoLett

Panchromatic Photoproduction of H2 with Surface Plasmons Syed Mubeen,†,‡ Joun Lee,† Deyu Liu,† Galen D. Stucky,† and Martin Moskovits*,† †

Department of Chemistry and ‡Department of Chemical Engineering, University of California, Santa Barbara, California 93106, United States S Supporting Information *

ABSTRACT: The optical resonances of plasmonic nanostructures depend critically on the geometrical details of the absorber. We show that this unique property of plasmons can potentially be used to create panchromatic absorbers covering most of the useful solar spectrum, by measuring the light-tohydrogen conversion capabilities of a series multielectrode photocatalytic devices, based on functionalized gold nanorods of appropriately chosen aspect ratios. Judiciously combining nanorods of various aspect ratios almost doubles the H2 production of the device over what is optimally possible with a device using gold nanorods of a single aspect ratio (all other key parameters being equal). The estimated quantum efficiency (absorbed photons-to-hydrogen) averaged over the entire solar spectrum of the best performing plasmonic multielectrode array was approximately 0.1%, and the measured H2 production rate for all of the devices was found to be approximately proportional to the hot electron generation. The device was monitored continuously for over 200 hr of operation without measurable diminution in the rate. KEYWORDS: Plasmons, water splitting, H2 production, gold nanorods, photocatalysis

A

gold and/or silver nanostructures either as mixtures or in tandem. The fabrication method employed is simple. Nanorods of varying aspect ratios were drop cast on quartz slides forming a dense layer and converted to ensembles of “Janus” particles by coating one side of the gold nanorods with a ∼6 nm layer (mass thickness) of electron-beam deposited TiO2 that acted as a hot electron filter. This was then capped with ∼2 nm diameter platinum nanoparticles (Figure 2a) functioning as the H2 evolution catalyst in the water reduction half reaction (see Methods and Table S1 given in the Supporting Information). Photocatalysis was carried out using a stacked plasmonic device in which each element absorbed a specific portion of the solar spectrum (Figure 2b). The stacked multielectrode plasmonic photocatalytic device consisting of gold nanorods of varying aspect ratios, evolved hydrogen at a stable rate from a 1:1 water−methanol solution when illuminated with simulated sunlight (Figure 2c). We used methanol as a sacrificial reactant that acts as the electron donor. In an optimally designed artificial water splitter, H2 and O2 are produced at spatially separated electrocatalytic sites. When the loci of oxidation and reduction are in close proximity (as is true in the present case), one often uses an electron donor or an electron acceptor to optimize either the H2 production or O2 production. Passing simulated sunlight (with an AM1.5 spectral distribution) through the stack resulted in the absorption of

rtificial photosynthesis normally refers to photoredox reactions carried out using semiconductor-based photoelectrochemical cells that could potentially use sunlight to drive fuel producing reactions, such as H2 (and O2) from water,1,2 or CH4 (and O2) from CO2 and H2O.3,4 Recently it was shown that semiconductor materials are not unique in this capability; energetic charge carriers are also produced following plasmonic excitation of appropriately nanostructured metals5−8 and able to perform redox chemistries. Some of these plasmonic devices can evolve, H2 from water over significantly longer time spans, than semiconductor-based devices, although at, as yet, much lower efficiencies.9−15 Notably, since plasmon resonances depend sensitively on the nanogeometry of the system,16 plasmonic devices can potentially be prescriptively fabricated with broad and uniform absorption properties covering much of the useful portion of the solar spectrum.17 For example, the longitudinal plasmon mode of a gold nanorod can be tuned from ∼600 nm to ∼1100 nm simply by changing the rod’s aspect ratio from 1.7 to 6.8 (Figure 1). Consequently panchromatic plasmonic systems can, therefore, be fabricated, for example, by creating an absorber composed of an appropriately selected mixture of gold nanorods. Here we demonstrate such a device approach by showing that an appropriately assembled mixture of polydispersed and functionalized gold nanorod can almost double the solar-to-H2 production yield over that of the best-performing monodispersed gold nanorod sample (keeping the mass and surface area of the material approximately constant). This implies that a plasmonic absorber that absorbs light uniformly over the entire solar spectrum can be built by using appropriately architecture © XXXX American Chemical Society

Received: January 10, 2015 Revised: February 9, 2015

A

DOI: 10.1021/acs.nanolett.5b00111 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 1. Tuning the absorption of gold nanoparticles across the solar spectrum by controlling nanogeometry. UV−vis spectra of gold nanorods with aspect ratios varying from 1.7 to 6.8 and the corresponding TEM image for each are shown. The aspect ratios were chosen such that its longitudinal plasmon resonance tracks as much of the AM 1.5 solar spectrum from visible to near IR as possible. Scale bars correspond to 200 nm.

Figure 2. Strategy for assembling a panchromatic plasmonic photocatalytic device based on gold nanorods. (a) Representative TEM image of a Au− TiO2−Pt Janus particle. The hot electrons injected into Pt following plasmon decay carry out water reduction to H2; the resultant positive charges on gold are neutralized by the oxidation of methanol. (b) An idealized rendering illustration the strategy of stacking plates bearing gold nanorods with increasing aspect ratio, each stack absorbing a unique portion of the solar spectrum. Representative SEM images of Janus particles of increasing aspect ratio immobilized on quartz substrates are also shown. The 250 nm scale bar applies to all images. (c) A digital photograph of H2 bubbles rising above a tandem stack of multiple plates bearing (from left to right) Janus particles of varying aspect ratios from 1.4 to 4.1 and illuminated with white light (AM 1.5). The photocatalysis was conducted in a glass cell filled with 1:1 H2O:CH3OH solution. A rough sketch of the AM1.5 solar spectrum is drawn above the evolved bubbles to illustrate the fact that the density of H2 bubbles produced roughly tracks the solar spectrum.

Gas measurements were performed in an airtight Pyrex flask (see detailed Methods section given in the Supporting Information). The device was illuminated using a 300 W xenon lamp source fitted with an AM1.5G filter (Newport Inc.). The generated gases were sampled by syringe (Gastight) and measured by gas chromatography (GC) equipped with

most of the incident light approximately uniformly across the solar spectrum, evolving H2 from each element in approximate proportion to the fluence of the solar irradiance in its active wavelength range (Figure 2c), as indicated roughly by the density of hydrogen bubbles visible in the figure above the stack. B

DOI: 10.1021/acs.nanolett.5b00111 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters thermal conductivity detection. With white light illumination H2 was observed as the only gaseous product formed. Methanol can sequentially oxidize in stepwise two-hole reactions to formaldehyde, formic acid, and CO2. Since no CO2 was observed, it is likely that the oxidation produced only formaldehyde and/or formic acid, whose high solubility in water would result in undetectable quantities of these species in the gaseous phase above the solution (as expected from the values of their respective Henry’s constants). CO2 is also watersoluble, but to a much lower extent than the former two species, and should have been detectable by GC had it been produced to an appreciable extent. Oxygen is not expected to be produced appreciably in the presence of methanol, and, indeed, was not observed. The oxidation reactions likely occur at the exposed gold surface neutralizing the positive charges left behind in the gold nanorods that function as holes; however, gold is not an optimal catalyst. The most likely reactions taking place are summarized as follows: Reduction: 2H+(aq) + 2e− → H 2(g)

Table 1. Structural Characteristics and H2 Production Rates

Oxidation: (2)

H 2CO + 2h+ + H 2O → HCOOH + 2H+(aq)

(3)

aspect ratio(s) of the Au nanorods used

λ at maximum absorbance (nm)

H2 production rate (μmol/h)

S1 S2 S3 S4 S5 S6 S7

1.4 1.7 3.0 4.1 1.4 + 3.0 Mix 1 Mix 2

650 700 810 960 650 + 810 550 + 750 550 + 900

1.3 1.0 0.65 0.13 2.3 2.0 0.7

resonance that shifts to longer wavelengths with increasing aspect ratios; for example, the longitudinal plasmon is observed at 600 nm for the sample fabricated with gold nanorods with an aspect ratio of 1.7 and at 860 nm for samples carrying Au nanorods with an aspect ratio of 4.1. The resonance wavelength depends significantly and predictably on the dielectric constant value of the substrate on which the nanorods reside and the ambient dielectric in which they are embedded.17 This explains the variation in resonance wavelength values for samples with nominally equivalent aspect ratios but surrounded by a different medium (cf. Figure 1, bare nanorods and Figure 3a, Janus particles). For the samples consisting of plasmonic photocatalytic nanoparticles based on monodispersed gold nanorods, the hydrogen production rate decreased monotonically as the aspect ratio increased from 1.4 to 4.1 (Table 1), reflecting the increasing absorption “gap” produced by shifting the longitudinal plasmon component to longer wavelengths and away from the solar emission maximum (Figure 3a, b). The Planck radiation law function computed for an ideal emitter at 5250 K, which approximates the sun’s AM1.5 spectrum, is included in Figure 3a to illustrate this point. The H2 production performance of such samples is greatly improved when this absorption gap is “filled in” by using Au nanorods with more than one aspect ratio. For example, by using Au nanorods with aspect ratios 1.4 and 3.0 together (sample S5), one can almost double the best hydrogen production rate that a single Au nanorod sample can achieve (Figure 3d). The connection between the extinction spectrum of the device and its hydrogen production rate (assuming the reduction step to be rate-determining) can be made more quantitative as follows. Assume the H2 production rate to be proportional to the rate of hot electron generation (Ye), which, in turn, is assumed to be proportional to the integrated plasmon absorptance, scaled for the solar flux at a given wavelength, and other factors as discussed below. Accordingly,

(1)

CH3OH(aq) + 2h+ → H 2CO + 2H+(aq)

sample number

in which the electrons and holes are produced pursuant to plasmon excitation and subsequent decay. Hydrogen is the sole product of reduction (reaction 1), and methanol is oxidized to formaldehyde, and possibly, at least partially, further oxidized to formate (reaction 2). In order to better understand the mechanistic details of photocatalytic H2 production reactions summarized above, a series of reactions was carried out using stacks of quartz slides illuminated in such a way that the simulated sunlight passed through the entire stack. Each of these samples consisted of 7 slides (1.27 cm × 1.27 cm), placed in tandem. The gold nanorod coverage for these samples was approximately 30% (see Table S2 given in the Supporting Information). Two classes of sample were constructed using this protocol. In the first set (samples S1, S2, S3, and S4), each stack consisted of plates covered with nominally monodispersed gold nanorods, but with the aspect ratio varying from stack to stack, such that samples S1, S2, S3, and S4, were based, respectively, on Au nanorods with aspect ratios of 1.4, 1.7, 3.0, and 4.1. Another class of samples with more complex extinction spectra was fabricated using one of two approaches. The first (S5) was constructed by stacking plates covered with Au NPs of aspect ratio 1.4 together with plates covered with Au NPs of aspect ratio 3.1 for a total of 7 plates. In the second approach the quartz plates were covered with mixtures of Au nanorods and Au nanoparticles of varying dimensions and aspect ratios (sample S6 and S7), then selecting those plates for the 7-plate stacks that absorbed over a broader, or, at least, different portion of the solar spectrum (see Methods and Table 1). The extinction spectra and the H2 production by these samples over 12 hr are shown in Figure 3a−d. Samples S1−S4 (AuNR/TiO2/Pt, on quartz in ambient air, Figure 3a) possess two plasmon resonances, a transverse plasmon which appears at approximately 520 nm regardless of the nanoparticle’s aspect ratio and a longer-wavelength

1100

Ye ∝

∫400

S(λ) E(λ) λ(1 − η) dλ

(4)

In which S(λ) is the solar energy flux at λ, which we will approximate (for ease of computation) by Planck’s radiation function: S(λ) ∝ λ−5[e(hc/λkt) − 1]−1 using a value of 5250 K for the temperature, T; E(λ) is the measured extinction value at λ. One needs to convert the radiant energy to the number of photons by dividing by hc/λ, which accounts for the lone value of λ following E(λ) in eq 4, and (1 − η) converts the extinction, E(λ), to the absorbance as a function of λ. Since the E(λ) measurements were carried out in small wavelength intervals, the integral in eq 4 can be replaced by a sum with little error, as follows: C

DOI: 10.1021/acs.nanolett.5b00111 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 3. Connection between extinction spectra and photocatalytic activity of Au−TiO2−Pt Janus particles. (a) UV−vis spectra of four stacked photocatalytic devices, each fabricated with approximately monodispersed gold nanorods, with the nanorods’ aspect ratio varying from stack to stack as indicated in the figure. The black dotted line represents Planck’s radiation law function computed for an ideal emitter at 5250 K. (b) Photocatalytic production of H2 vs time for the four samples referred to in panel a. The H2 production rates in 1:1 H2O:CH3OH solution under simulated solar light are listed on the figure. (c−d) As in panels a and b but for three stacked devices (S5−S7) each produced by combining gold nanorods of different aspect ratio to cover a broader portion of the solar spectrum. Sample S5 is based Au nanorods with aspect ratio 1.4 and 3.0. The other two devices are made with random mixtures of Au nanoparticles and nanorods selected on the basis of their extinction spectra. Two of these devices outperform the best device fabricated with Au nanorods of a single aspect ratio. λi = 1100

Ye ∝

∑ λi = 400

S(λi) E(λi) λi(1 − η)Δλ (5)

Values of (1 − η) can be estimated from the data of Lee et al.18 who measured the quantum yield, η = (Qsca/Qext), at resonance, for nanorods of varying aspect ratios. They found that the value of η changes from 0.5 to 0.6 over the aspect ratio range 1.5−5, which encompasses the range used in this study. This translates to values of (1 − η) ranging from 0.4 to 0.5. This is a rather small variation. Accordingly, we will ignore this correction since only its variation over the aspect ratio range of the gold nanorods we used will affect values of Ye, which, being defined as a proportionality, will be insensitive to multiplicative constants. Using eq 5 a number proportional to Ye was calculated as described above for each of the seven samples. The results are plotted in Figure 4 which shows that the measured H2 production rate is approximately linear with the computed value of Ye. In principle, the relationship should have been proportional rather than merely linear; this small disagreement reflects the simplifications made in the computation. Nevertheless, the result is highly gratifying since it connects the H2 production rate directly to the plasmonic absorption properties of the various devices used. The estimated quantum efficiency (QE) spectrally averaged over the entire solar spectrum for the sample with Au nanorods of aspect ratio 1.4 and 3.0 (S5) was computed to be ∼0.1%. The quantum efficiency was defined as 100 × (2 × number of evolved H2 molecules)/(number of “lost” photons), i.e., the number of photons that fail to be transmitted. The input solar

Figure 4. Comparing hot electron generation and H2 production rates. The hydrogen production rates, for the seven types of devices investigated, are plotted against a computed number proportional to the hot electron production rate. The aspect ratios of the Au nanorods on which the devices were based, as well as the device identifier are included. The red line is the least-squares best fit.

energy absorbed was calculated as incident light intensity × (1 − transmittance); and the number of photons averaged over the entire solar spectrum for 100 mW cm−2 of solar energy absorbed was taken to be 3 × 1017 cm−2 s−1. The quantum efficiency computed above is a lower bound, since we did not correct the number of photons lost either for the substantial reflectance losses, or for the scattered (as opposed to the absorbed) fraction of the light by the device. Including such corrections would increase the quantum efficiency by at least a factor of 4. The hydrogen production performance of these samples was remarkably robust. The H2 D

DOI: 10.1021/acs.nanolett.5b00111 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

(7) Hou, W. B.; Cronin, S. B. Adv. Funct. Mater. 2013, 23, 1612− 1619. (8) Clavero, C. Nat. Photonics 2014, 8, 95−103. (9) Thomann, I.; Pinaud, B. A.; Chen, Z.; Clemens, B. M.; Jaramillo, T. F.; Brongersma, M. L. Nano Lett. 2011, 11, 3440−3446. (10) Mubeen, S.; Lee, J.; Singh, N.; Krämer, S.; Stucky, G. D.; Moskovits, M. Nat. Nanotechnol. 2013, 8, 247−251. (11) Kim, H. J.; Lee, S. H.; Upadhye, A. A.; Ro, I.; Tejedor-Tejedor, M. I.; Anderson, M. A.; Kim, W. B.; Huber, G. W. ACS Nano 2014, 8, 10756−10765. (12) Silva, C. G.; Juarez, R.; Marino, T.; Molinari, R.; Garcia, H. J. Am. Chem. Soc. 2011, 133, 595−602. (13) Yu, S.; Kim, Y. H.; Lee, S. Y.; Song, H. D.; Yi, J. Angew. Chem., Int. Ed. 2014, 53, 11203−11207. (14) Zheng, Z. K.; Tachikawa, T.; Majima, T. J. Am. Chem. Soc. 2014, 136, 6870−6873. (15) Zhong, Y.; Ueno, K.; Mori, Y.; Shi, X.; Oshikiri, T.; Murakoshi, K.; Inoue, H.; Misawa, H. Angew. Chem., Int. Ed. 2014, 53, 10350− 10354. (16) Nikoobakht, B.; El-Sayed, M. A. Chem. Mater. 2003, 15, 1957− 1962. (17) Link, S.; Mohamed, M. B.; El-Sayed, M. A. J. Phys. Chem. B 1999, 103, 3073−3077. (18) Lee, K. S.; El-Sayed, M. A. J. Phys. Chem. B 2005, 109, 20331− 20338.

evolution performance of one of these samples was tracked periodically for more than 30 days with no discernible decrease, and the frequent periodic measurements carried out over a 200 hr run are shown in Figure S1. In summary, we have described the broad-band absorptivity achievable using plasmonic absorbers whose resonance properties can be tuned by varying the system’s nanostructure, using devices comprised of stacks of plates bearing (appropriately functionalized) gold nanorods of varying aspect ratios. A similar effect would have been obtained by using samples composed of mixtures of nanorods with varying aspect ratios. We illustrate, thereby, a generalizable strategy for building panchromatic photocatalytic devices by tuning the nanogeometry of the plasmonic material of which they are constructed, with potential applications in constructing devices for carrying out unassisted solar fuel production and electrical power generation.



ASSOCIATED CONTENT

S Supporting Information *

Detailed methods, synthesis recipe, aerial coverage of gold nanrods, and long-term stability run. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Present Address

S.M.: Department of Chemical and Biochemical Engineering, University of Iowa, Iowa City, 52242, USA. Author Contributions

S.M., J.L., D.L., G.S., and M.M. contributed to the photocatalytic design concept. S.M., J.L., and D.L. performed the fabrication process and measurements. M.M. supervised the project and wrote the manuscript along with S.M. and J.L. All authors discussed the results and commented on the manuscript. S.M., J.L., and D.L. contributed equally. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We wish to acknowledge support for this work by the Institute for Collaborative Biotechnologies through contract no. W911NF-09-D-0001 from the U.S. Army Research Office. Support from the MRSEC Program of the National Science Foundation under Award No. DMR 1121053 is gratefully acknowledged. We also made extensive use of the MRL Central Facilities at UCSB, a member of the NSF-funded Materials Research Facilities Network (http://www.mrfn.org).



REFERENCES

(1) Walter, M. G.; Warren, E. L.; McKone, J. R.; Boettcher, S. W.; Mi, Q.; Santori, E. A.; Lewis, N. S. Chem. Rev. 2010, 110, 6446−6473. (2) Chen, X. B.; Shen, S. H.; Guo, L. J.; Mao, S. S. Chem. Rev. 2010, 110, 6503−6570. (3) Kohno, Y.; Hayashi, H.; Takenaka, S.; Tanaka, T.; Funabiki, T.; Yoshida, S. J. Photochem. Photobiol. A, Chem. 1999, 126, 117−123. (4) Li, Y.; Wang, W.-N.; Zhan, Z.; Woo, M.-H.; Wu, C.-Y.; Biswas, P. Appl. Catal., B 2010, 100, 386−392. (5) Linic, S.; Christopher, P.; Ingram, D. B. Nat. Mater. 2011, 10, 911−921. (6) Warren, S. C.; Thimsen, E. Energy Environ. Sci. 2012, 5, 5133− 5146. E

DOI: 10.1021/acs.nanolett.5b00111 Nano Lett. XXXX, XXX, XXX−XXX