Scale-up of a Luminescent Solar Concentrator ... - ACS Publications

Subscriber access provided by READING UNIV

Article

Scale up of a luminescent solar concentrator based photomicroreactor via numbering-up Fang Zhao, Dario Cambié, Jeroen Janse, Eric Wieland, Koen P. L. Kuijpers, Volker Hessel, Michael G. Debije, and Timothy Noël ACS Sustainable Chem. Eng., Just Accepted Manuscript • DOI: 10.1021/ acssuschemeng.7b02687 • Publication Date (Web): 07 Nov 2017 Downloaded from http://pubs.acs.org on November 11, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Sustainable Chemistry & Engineering is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

Scale up of a luminescent solar concentrator based photomicroreactor via numbering-up Fang Zhao,a Dario Cambié,a Jeroen Janse,a Eric W. Wieland,a Koen P. L. Kuijpers,a Volker Hessel,a Michael G. Debijeb and Timothy Noëla,* a

Micro Flow Chemistry and Process Technology, Department of Chemical Engineering and

Chemistry, Eindhoven University of Technology, Den Dolech 2, 5600 MB, Eindhoven, the Netherlands. b

Functional Organic Materials & Devices, Department of Chemical Engineering and Chemistry,

Eindhoven University of Technology, Den Dolech 2, 5600 MB, Eindhoven, the Netherlands. *Corresponding author. E-mail: [email protected]

Keywords: Photomicroreactor, Luminescent solar concentrator, Solar energy, Numbering-up, Photochemistry.

Abstract: The use of solar energy to power chemical reactions is a long-standing dream of the chemical community. Recently, visible-light mediated photoredox catalysis has been recognized as the ideal catalytic transformation to convert solar energy into chemical bonds. However, scaling photochemical transformations has been extremely challenging due to Bouguer-Lambert-

ACS Paragon Plus Environment

1


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 29

Beer law. Recently, we have pioneered the development of luminescent solar concentrator photomicroreactors (LSC-PM), which display an excellent energy efficiency. These devices harvest solar energy, convert the broad solar energy spectrum to a narrow wavelength region and subsequently waveguide the re-emitted photons to the reaction channels. Herein, we report on the scalability of such LSC-PMs via a numbering up strategy. Paramount in our work was the use of molds which were fabricated via 3D printing. This allowed us to rapidly produce many different prototypes and to optimize experimentally key design aspects in a time-efficient fashion. Reactors up to 32 parallel channels have been fabricated, which display an excellent flow distribution using a bifurcated flow distributor (standard deviations below 10%). This excellent flow distribution was crucial to scale up a model reaction efficiently, displaying yields comparable to those obtained in a single-channel device. We also found that inter-channel spacing is an important and unique design parameter for numbered-up LSC-PMs, which influences greatly the photon flux experienced within the reaction channels.

Introduction Sunlight is the most abundant source of renewable energy on our planet. In one hour, more solar energy strikes the Earth than we currently consume on a yearly basis. It is envisioned that solar energy will become the primary energy source in the future. Efficient use of solar energy will allow us to meet the increasing energy demand while stabilizing the atmospheric CO2 concentration at an acceptable level.1, 2


2

Page 3 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


An enormous number of studies have been performed to capture, convert and store solar energy. Throughout the years, researchers have unraveled the mechanism of light absorption and the subsequent energy transfer in the natural photosynthetic process.3, 4 These insights have inspired the scientific community to create ingenious and ‘Nature-style’ solutions to harvest solar energy.5,

6

One particular field of research focusses on the conversion of solar energy into

electrical energy. This field has benefitted from insights in the natural light-harvesting system to produce improved solar cells. For example, dye-sensitized solar cells, which separate the charge generation and charge transport processes by mimicking photosynthesis, have achieved AM 1.5G (Air Mass 1.5-Global Illumination) power conversion efficiencies up to 13%,7 becoming a promising alternative to the conventional P-N junction devices.8, 9 Another discipline has devoted its efforts towards the conversion of solar energy into chemical bond energy: many ingenious photocatalysts have been developed to produce solar fuels by imitating either the morphology or the mechanism of Nature’s tree leaf.10,

11

These ‘artificial leaves’ are composed of a light

harvester and water splitter consisting of an anode for oxygen evolution and a cathode for hydrogen evolution.12, 13 Solar energy can also be utilized to enable the synthesis of complex molecules, e.g. via visible light photocatalysis14-17. Hereto, so-called solar reactors, such as parabolic troughs, parabolic dishes, solar furnaces and flatbed reactors, can be used to increase the efficiency of the photochemical process.18 However, these reactors suffer from a number of disadvantages, including outdoor use of large amounts of hazardous solvents, need for sun trackers and reflective mirrors, high cost, low efficiency and difficulty in scaling-up. Moreover, such solutions are less cost-efficient in sun-poor regions.19, 20 In order to develop reactor solutions which can operate at relatively high latitudes, where light is less abundant and more diffuse (e.g. by clouds and buildings), we have recently developed so-


3


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 29

called Luminescent Solar Concentrator based Photomicroreactors (LSC-PMs).21 These reactors constitute a synergistic merger between luminescent solar concentrators22 and micro flow chemistry23-25. The LSC-PM is based on fluorescent dye-doped PDMS material and, similar to Nature’s biomachinery, it allows harvesting of solar energy and its downconversion to a narrow wavelength window which matches with the absorption maximum of the photocatalytic reaction. The re-emitted photons are subsequently lightguided to the embedded reaction channels. LSCPMs display all the advantages which are associated with microreactors, such as enhanced mass transfer, uniform irradiation of the reaction mixture and increased safety26, 27. However, like microreactor technology, it suffers from its small scale which is not sufficient to meet the productivity requirements of large scale chemical processes. Two strategies are often employed for the scale-up of microreactors: sizing-up and numbering-up. The former involves a dimension-enlarging strategy, in which the reactor size is gradually increased whilst keeping the mass and heat transport phenomena stable.28, 29 However, due to the light attenuation effect, this strategy is not compatible with photochemical reactions. A potential solution to scale up LSCPMs is to use a numbering-up strategy. In analogy to a tree, which has many leaves through which water and other nutrients are pumped, several LSC-PM devices can be placed in parallel to increase the overall throughput. The challenge for efficient numbering-up is to ensure an equal distribution of the fluid flow over the different reactors. To achieve equipartitioning among the parallel microchannels, various flow distributors/collectors with different structures and geometries have been developed.30-32 Many methods have been reported to characterize the flow distribution, including experimental methods33, 34, modelling35, CFD simulations36, micro-PIV (micro-particle image velocimetry)37 and flow visualization techniques38. For chip- or plate-


4

Page 5 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


based microreactors, the most frequently used distributor/collector designs are manifold-type and bifurcation-type, as shown in Figure 1 A and B.

Figure 1. Three distributor/collector designs for flow equipartition. (A) Manifold-type design with a lateral inlet and outlet. (B) Bifurcation/bifurcation design. Bifurcation structure is used for both the distributor and the collector. (C) Bifurcation/chamber design. Bifurcation structure is used in the distributor while a collecting chamber is used for the collector.

For the manifold-type design, the flow manifold (or chamber) can have different inlet/outlet positions. The design shown in Figure 1A has a right inlet and left outlet which makes that the total residence time is the same for each channel. The inlet and outlet can also be both located at the center of the distributing and collecting manifolds, respectively.36 Moreover, the flow manifold can have varying cross sections to ensure a uniform pressure drop over the parallel reaction channels.39


5


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 29

The bifurcation structure is a symmetrical design and typically provides an excellent distribution as shown by both simulations and experiments,40-42 with flow variations below 5% for singlephase flow43. However, in the context of this work, there are two main disadvantages associated with the bifurcation design: the large footprint of the distributor/collection section compared to the reactor section and the relatively high pressure drop due to the splitters. This is especially true when the total number of parallel reaction channels is high. A good compromise is the combination of a bifurcation distributor design with a collecting chamber.38, 44 This numberedup design ensures a good flow distribution, while minimizing the overall footprint and the pressure drop (Figure 1C). In this work we describe a numbered-up version of our solar driven LSC-PM which allows us to scale-up solar photochemistry in an efficient fashion. The developed design provides high flow uniformity in devices containing up to 32 parallel channels. A crucial parameter in our LSC-PM numbered-up design is the inter-channel spacing which is equivalent to the solar energy harvesting area.

Experimental Materials Ethanol (purity (V/V) ≥ 99.5%) and diethyl ether (technical) were purchased from VWR Chemicals. Acetonitrile (AR, Biosolve Chimie SARL) and methylene blue were bought from Merck KGaA. Lumogen F red 305 and 9,10-diphenylanthracene (> 95.0%) were acquired from BASF and TCI, respectively. Sylgard® 184 silicone elastomer (base + curing agent) was purchased from Dow Corning.


6

Page 7 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


For the 3D-printing of the reactor molds: high-impact polystyrene filament (ø1.75 mm, neutral, Apollo series) was obtained from webshop 123-3D.nl; polystyrene plates (black, 270 length ×260 width ×3 thickness mm3) were provided by S-POLYTEC GmbH.

Device design and fabrication In this paper, polydimethylsiloxane (PDMS) is used to construct the LSC-PM. The PDMS is doped with Lumogen F red 305 (LR305) as the fluorescent dye. The numbered-up LSC-PM can be fabricated by combining the entire design (distributor, reaction channels and collector) in one single PDMS slab. This scaled-up device has only a single entrance and exit port, thus only a single pumping system is required to deliver the feeds. Hence, our scaled-up device consists of five primary parts (Figure 2): inlet zone, distributing zone, reaction zone, collecting zone and outlet zone. The microchannels of these five zones, all with a height of 1 mm, are embedded in a 3 mm-thick PDMS slab, centrally located in the vertical cross-section.

Figure 2. A schematic representation of our scaled-up device. The dashed lines do not represent the actual flow path.


7


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 29

In the reaction zone, identical parallel microchannels (0.35 width × 1 height × 150 length mm3) are distributed with constant inter-channel distance. The inter-channel spacing is defined as the spacing between two adjacent parallel channels. To equalize the flow among the parallel channels, the bifurcation design is employed in the distributing and collecting zones (Figure 1B). For a system with a high pressure drop, or just for simplification of the device, a collecting chamber can be employed as a substitute for the bifurcation structure in the collecting zone (Figure 1C). The main process for manufacturing our LSC-PM device is presented in Figure 3. Two molds are printed on polystyrene plates with a 3D printer (FELIX, Pro 1): one mold is for the 2 mm-thick layer wherein the microchannels are embedded and the other mold allows to produce the 1 mmthick covering layer. The main advantage of this 3D-printed mold strategy is that it enables rapid prototyping, which makes it possible to screen multiple designs in a time-efficient fashion.45, 46 Next, the pre-polymer mixture (silicone elastomer base + curing agent) with or without the dye LR305 was prepared and poured into the molds, followed by a degassing procedure in a vacuum oven (Heraeus, VTR 5036). For all the dye-doped devices in this paper, the dye loading was 200 ppm, which is the optimal concentration for the LSC-PM.21 After the PDMS is completely cured, the two layers can be peeled off the molds and are subsequently treated with oxygen plasma (Plasma asher Tantec, SpotTEC), and bonded together to form the 3 mm-thick device.


8

Page 9 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3. A schematic diagram describing the basic steps of the LSC-PM fabrication process.

Characterization of flow distribution The devices used to evaluate the flow distribution did not contain a collecting zone, as shown in Figure 4. Instead, ETFE capillaries (IDEX, 1.6 mm OD × 1 mm ID, 2 cm length) were inserted and allowed us to collect the effluent of individual channels in separate vials. Ethanol was used as the fluid and pumped through the device using an HPLC pump (Shimadzu, LC-20AD).

Figure 4. Sketch of the numbered-up device used to investigate the flow distribution. The dark blue channels in the red waveguide represent the microchannels and the light blue outside the red waveguide are the capillaries which are inserted in the microchannel outlets.


9


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 29

Effluents were collected individually during a given time in a 1.5 mL vial and weighed. Next, the flow rate in the reaction channel i, Fi (mL/min), was calculated based on the mass of the collected effluent. As a measure for the uniformity of the flow distribution, the relative standard deviation RSD (%) of the flow rates in N parallel reaction channels was used, which is given by 1 1 RSD% = − − 1

(1)

where 1 =

(2)

Characterization of reactor performance The [4+2] cycloaddition of 9,10-diphenylanthracene (DPA) with singlet oxygen, generated via photosensitization in the presence of methylene blue (MB), was used as the benchmark reaction to evaluate the performance of the scaled-up LSC-PMs (Figure 5).

Figure 5. The singlet oxygen mediated cycloaddition of 9,10-diphenylanthracene using methylene blue as the photocatalyst and acetonitrile as the solvent at room temperature.


10

Page 11 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 6 displays the microfluidic set-up, which is used to characterize the performance of the numbered-up reactor. Feed solutions of 0.2 mM DPA and 0.02 mM MB were pumped by two HPLC pumps (Shimadzu, LC-20AD). The two streams were merged in a PEEK T-micromixer (IDEX, P-714, inner diameter 1 mm) and the combined reaction mixture was introduced into the device, which was illuminated by a metal halide solar simulator (KHS Solar Constant 1200). The conversion was measured by directing the effluent through the micro flow cell (Avantes, Micro flow Z-cell-10) of an inline spectrophotometer (Avantes, AvaSpec-2048L) equipped with a balanced deuterium light source (Avantes, AvaLight-DH-S-BAL). The conversion was measured in real time by monitoring the disappearance of the DPA peak at 372 nm.

Figure 6. Experimental set-up for the characterization of reactor performance. (A) An overall schematic of the experimental system. 1- Feed for MB solution; 2- Feed for DPA solution; 3HPLC pump; 4- T-micromixer; 5- Scaled-up LSC-PM device; 6- micro flow cell; 7- waste flask; 8- Deuterium light source; 9- Spectrometer; 10- Computer. (B) A schematic diagram for the


11


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 29

scaled-up LSC-PM device. Black paper is used to cover the non-reaction zones to avoid the occurrence of any reaction outside the reaction zone. The reaction zone is illuminated with light emitted by a solar simulator, which mimics the solar spectrum.

The conversion X and space-time yield Y were calculated using the obtained DPA concentration C as follows: =

=

−

∙ ∙

(3) (4)

where C0 is the initial DPA concentration, FT is the total volumetric flow rate and VLH is the light-harvesting volume of the device (= the total device volume exposed to the light – the total reaction volume). The space-time yield Y defined here represents the amount of DPA converted per unit time and per unit light-harvesting volume. The LSC-PM device was placed on non-reflective black paper. The non-reaction zones were covered with black paper to prevent any reaction from occurring.

Results and discussion Flow distribution tests in scaled-up devices Optimization of bifurcation design. An ideal bifurcation distributor should be symmetrical and the channel length (LE) after the elbow should be long enough to result in a fully developed flow profile (Figure 7). The distributor symmetry requirement can be fulfilled by using a 3D printer


12

Page 13 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


with a high level of accuracy. The value of LE to reach a fully developed velocity profile depends on the hydraulic diameter of the channel (D) and the Reynolds number (Re) and can be calculated by47: = 0.05 ∙ #$ ∙ %

(5)

under laminar flow conditions. However, high values of LE would lead in a too large distributor section and thus a larger overall footprint of the device. Consequently, LE needs to be optimized based on the hydraulic diameter of the channels at different Re numbers.

Figure 7. A distributor with the bifurcation structure for 8 parallel reaction channels.

For rectangular channels, the hydraulic diameter D can be calculated using the following equation: %=

2'( ')(

(6)

where ω and h are the channel width and height, respectively.


13


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 29

To optimize LE, three 8-channel devices were made which only differed in their LE values, as shown in Table 1. The reaction channels in these three devices have all the same dimensions, i.e. 0.35 width × 1 height × 150 length mm3. In device I, LE is set to be two times the corresponding hydraulic diameter, i.e. 2·D. Likewise, in device II and III LE is set to be respectively 3·D and 4·D. The exact LE values within these three devices are presented in Table 1, as well as the theoretical entrance length calculated by Equation (5) using the corresponding largest Re in each distributing level investigated in the experiment.

Table 1. An overview for the dimensional parameters of the distributing channels in the three 8channel devices made for the optimization of distributor design (unit: mm).

Distributing level 1 2 3

Entrance length 5.5 2.8 1.5

Device I

Device II

Device III

ω

h

LE

ω

h

LE

ω

h

LE

1.6 0.8 0.35

1 1 1

2.5 1.8

1.6 0.8 0.35

1 1 1

3.7 2.7

1.6 0.8 0.35

1 1 1

5.0 3.6

Next, flow distribution tests were conducted at two extreme flow rate values, namely 0.1 and 0.6 mL/min/channel. The unit mL/min/channel represents the flow rate in an individual reaction channel when the flow is equalized among the parallel reaction channels. In other words, 0.1 mL/min/channel represents a total flow rate of 0.8 mL/min in an 8-channel device. From Table 2, it can be seen that the flow distribution improved with increasing LE values (device I < device II < device III). Furthermore, it can be expected that if we further increase the value of LE (≥ entrance length as presented in Table 1), an improved flow distribution with RSD < 2% can be


14

Page 15 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


achieved. In general, RSD values < 5% can be regarded as excellent and both device II and III meet that criterion. In order to keep the size of the distributor section small, we selected LE =3·D (as in device II) for our further investigations.

Table 2. RSD results for the three 8-channel devices with different values of LE under a low and a high flow rates. Flow rate mL/min/channel 0.1 0.6

Device I 10.9% 4.4%

Device II 3.7% 3.6%

Device III 2.1% 2.3%

Flow distribution of bifurcation design with different reaction channel numbers. Having selected the optimal value of LE, we set out to evaluate the flow distribution efficiency in devices containing respectively 8, 16 and 32 parallel reaction channels. The dimensional parameters of the distributing channels in these three devices are listed in Table 3 and the RSD values for flow distribution are tabulated in Table 4. It can be seen that excellent flow distribution was achieved for the 8-channel device over a wide flow rate range with RSD values all below 4%. When the total number of channels increased, the flow distribution efficiency decreased slightly. However, very good RSD values below 10% can still be obtained even in devices containing 32 parallel channels. However, it should be noted that the occurrence of an increased flow maldistribution is likely caused by the insertion of capillaries at the reaction channel outlets. This was done to estimate the mass flow distribution but in a practical device (vide infra) a flow collector is used. This means that the RSD values will be much smaller in the latter scenario. Another potential


15


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 29

reason for flow non-uniformity is the 3D printing accuracy which is lower for larger devices. This problem could be overcome by using classical soft lithography techniques48 or by using a 3D printer with higher accuracy.

Table 3. An overview of dimensional parameters of the distributing channels in scaled-up devices with 8, 16 and 32 channels (unit: mm).

Distributing level 1 2 3 4 5

8-channel device

16-channel device

32-channel device

ω 1.6 0.8 0.35

ω 3.2 1.6 0.8 0.35

ω 3.2 3.2 1.6 0.8 0.35

h 1 1 1

LE 3.7 2.7

h 1 1 1 1

LE 4.6 3.7 2.7

h 1 1 1 1 1

LE 4.6 4.6 3.7 2.7

Table 4. RSD results for scaled-up devices with 8, 16 and 32 channels under different flow rates. Flow rate mL/min/channel 0.1 0.2 0.3 0.6 1.2

8-channel device 3.6% 3.7% 3.6% 3.7% 2.8%

16-channel device --6.0% 4.9% 4.4% ---

32-channel device --8.5% 7.9% -----

Effect of inter-channel spacing on the device performance One of the crucial design aspects of our LSC-PM is that the reactor material is doped with LR305 to harvest solar light. This energy is subsequently re-emitted and lightguided towards the


16

Page 17 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


reaction channels via total internal reflection. Consequently, it can be simply understood that the size of the harvesting area plays a key role in the photon flux through the reaction channels. The harvesting area can be characterized by the so-called inter-channel spacing which represents the distance between two adjacent channels. The larger the inter-channel spacing, the higher the photon flux will be within the channels. However, there is a limit to this distance due to the reabsorption of the luminescent photons. This reabsorption leads to energy losses caused by the possibility that the light is re-emitted within the escape cone angle and leaves the device, the non-unity fluorescence yield and the background absorption by the polymeric matrix.22 The larger the inter-channel spacing, the higher these losses will become. In order to evaluate the effect of the inter-channel spacing, we commenced our studies by investigating a set of five 8-channel devices, with varying inter-channel spacing of 1.5, 2.5, 5, 10 and 20 mm, respectively. The total reaction channel volume was kept the same and the devices were equipped with a bifurcated distributor and collector (see Figure. 8A). Moreover, to demonstrate the importance of inter-channel spacing for the LSC-PM device, we did the same investigations in five non-doped devices (Figure 8B) with the same reactor design as in the dyedoped ones.


17


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 29

Figure 8. Scaled-up devices with different inter-channel spacings and their reactor performance. (A-B) The five dye-doped devices (A) and the five non-doped devices (B) with inter-channel spacings of 1.5, 2.5, 5, 10 and 20 mm, respectively. (C-D) Plots of conversion (C) and spacetime yield (D) versus the inter-channel spacing under different times in the non-doped and dyedoped devices. (E) Correlation between inter-channel spacing and photon distribution between reaction and non-reaction (distributor, collector, inlet and outlet) zone, calculated by ray-tracing simulations.

As is expected, in non-doped numbered up devices, the conversion was almost kept constant with the increasing of inter-channel spacing when the residence time in the reaction channels was


18

Page 19 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


fixed (Figure 8C). These non-doped devices are able to use only the direct irradiation, which depends on the illumination area of the reaction channels and does not change with the interchannel spacing. In contrast, as shown in Figure 8C, in dye-doped devices the reaction conversion increased with increasing inter-channel spacing due to an augmented photon flux through the reaction channels. As can be observed in Figure 8C, for inter-channel spacing up to 5 mm the observed increase in conversion is quite significant, while a further increase to 10 or 20 mm has a reduced impact on conversion. This effect, due to the lower efficiency in the light guiding over longer distances, is even more significant when the corresponding space-time yields (Figure 8D) are taken into account. Furthermore, due to the device geometry, the percentage of photons reaching the non-reaction zones, as calculated via Monte Carlo ray-tracing modelling49, also increases with the increased inter-channel spacing (Figure 8E). Further in this work, an inter-channel spacing of 5 mm was chosen as more than 90% of the photon flux received by the reaction mixture was localized in the reaction zone. Several aspects influence the optimal inter-channel spacing in LSC-PM based devices including the device material (transparency and refractive index), the luminophore optical properties (fluorescent quantum yield and Stoke shift) the reaction system (reaction kinetics and quantum yield) and the overall process operation (e.g. operational cost for upstream/downstream processes). For solar-based chemical production applications, the inter-channel spacing of scaled-up LSC-PMs should be determined on a case by case scenario by balancing various parameters as mentioned above, based on the needs of the application. The approach we followed coupling experimental investigation and ray-tracing modelling can serve as guidelines to make a judicious choice in this matter.


19


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 29

Reaction performance in numbered-up devices with bifurcation/bifurcation design and bifurcation/chamber design During our previous experiments with the reactors possessing a bifurcation/bifurcation design (Figure 8A), we observed in some situations leakage at the in- or outlet. Especially the 32channel device was prone to failure at higher flow rates. Due to the high pressure drop in such cases, the channels suffered also from expansion or deformation. Feasibility of employment of a collecting chamber. In order to address the above-mentioned undesired phenomena, we also prepared a set of reactors which contained a collecting chamber. This design could effectively decrease the footprint of collecting zone and lower the pressure drop over the entire device (see Table S1 in the Supporting Information). It is easily understood that in such a design the fluid elements from the outer channels have to travel a larger distance to leave the device than those in the inner channels. To ensure flow equalization in the device, the pressure drop in the collecting chamber should be much lower than the total pressure drop so that it will not affect the flow distribution significantly. The flow resistance in the collecting chamber is mainly dependent on its width (Figure 9B). To evaluate the effect of the chamber width on the flow distribution, several devices as shown in Figure 9 were prepared to help us identify qualitatively the flow resistance in the fluid chamber. The flow distribution was evaluated for these devices and the results are presented in Table 5. It can be seen that the RSD values are acceptable in all cases (RSD values < 10%). An improved result was obtained with the 3 mm wide chamber. Furthermore, higher RSD values were observed for the 16-channel device, indicating that this design is less efficient as a distributor than the bifurcated one.


20

Page 21 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 9. A schematic diagram of the device for the investigation on a fluid chamber. (A) Device configuration. The dark blue in the red waveguide represents the microchannels and the light blue outside the red waveguide represents the ETFE tubes inserted at the microchannel outlet. (B) Schema for the chamber width.

Table 5. RSD results for the flow rate distribution among the parallel channels (inter-channel spacing 5 mm) connected with a fluid chamber under different configuration conditions. Flow rate mL/min/channel 0.1 0.3 0.6

Performance

2 mm wide chamber, 8 3 mm wide chamber, 8 3 mm wide chamber, channels channels 16 channels 6.4% 7.6% 4.7%

comparison

7.1% 4.2% 4.2%

between

bifurcation/bifurcation

8.4% 7.1% 8.7%

design

and

the

bifurcation/chamber design. Lastly, devices containing bifurcated collectors (Figure 10A) and chamber collectors (Figure 10B), both with 8, 16 and 32 reaction channels respectively, were prepared all with an inter-channel spacing of 5 mm. Since the pressure drop was lower, no leakages were observed with the bifurcation/chamber design in the experiments.


21


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure

10.

Numbered-up

devices

with

the bifurcation/bifurcation

Page 22 of 29

design

and

the

bifurcation/chamber design and their containing different number of reaction channels and their reaction performance. (A-B) Numbered-up devices with 8, 16 and 32 reaction channels (the channels are filled with MB solution) for the bifurcation/bifurcation design (A) and the bifurcation/chamber design (B), respectively. (C) Conversion-time curves for the devices with different reaction channel numbers for the two designs.


22

Page 23 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 10C displays the conversion in function of the total residence for devices with the bifurcation/bifurcation design (hollow blocks) and the bifurcation/chamber design (solid blocks) containing different number of reaction channels. Gratifyingly, excellent scalability is demonstrated for both designs as shown by the overlapping kinetic curves.

Conclusions We have described for the first time a fast scaling-up strategy of luminescent solar concentratorbased photomicroreactors (LSC-PMs), a reactor concept recently developed in our group which enables energy-efficient solar photochemistry. The use of 3D printed molds was of paramount importance to rapidly prototype different reactors and to investigate the relative importance of several design aspects. A unique parameter for LSC-PM constitutes the inter-channel spacing which determines the photon flux experienced by the reaction medium. We validated experimentally that a symmetrical bifurcated flow distributor resulted in excellent flow partitioning in devices with up to 32 parallel reaction channels (standard deviations below 10%). Furthermore, a photosensitized benchmark reaction proved to be scalable resulting in conversions similar to those obtained in a monochannel device. In order to overcome high pressure drops within the largest devices, we developed a numbered-up LSC-PM design with a bifurcated flow distributor and collection chamber. The use of solar energy to produce pharmaceuticals, agrochemicals and other fine chemicals is a dream expressed by many chemists over the last century.50 We believe that the LSC-PM is one such strategy for meeting this vision. While the productivity of single device is not sufficient to meet the current demand, several numbered-up devices as described in this work might


23


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 29

overcome this particular challenge. Further scaling can be achieved by placing several of these devices in parallel, leading to solar chemical plants similar to the existing solar parks (i.e. photovoltaic power stations).

ASSOCIATED CONTENT Supporting Information The Supporting Information is available free of charge on the ACS Publications website at DOI: Comparison between bifurcation/bifurcation design and bifurcation/chamber design. Light distribution among the reaction channels. Influence of different fluid paths in the chamber design.

AUTHOR INFORMATION Corresponding Author * E-mail: [email protected]; Telephone: 040-2473623; Website: www.NoelResearchGroup.com Notes The authors declare no competing ﬁnancial interest.

ACKNOWLEDGMENT We would like to thank the Dutch Science Foundation (NWO) for a VIDI grant for TN (SensPhotoFlow, grant number 14150). TN and DC acknowledge the European Union for a Marie Curie ITN Grant (Photo4Future, grant number 641861).


24

Page 25 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


REFERENCES (1) Lewis, N. S.; Nocera, D. G. Powering the planet: Chemical challenges in solar energy utilization. Proc. Natl. Acad. Sci. USA 2006, 103, 15729-15735. (2) Obama, B. The irreversible momentum of clean energy. Science 2017, 355, 126-129. (3) Bennett, D. I.; Amarnath, K.; Fleming, G. R. A structure-based model of energy transfer reveals the principles of light harvesting in photosystem II supercomplexes. J. Am. Chem. Soc. 2013, 135, 9164-9173. (4) Mirkovic, T.; Ostroumov, E. E.; Anna, J. M.; van Grondelle, R.; Govindjee; Scholes, G. D. Light absorption and energy transfer in the antenna complexes of photosynthetic organisms. Chem. Rev. 2017, 117, 249-293. (5) Scholes, G. D.; Fleming, G. R.; Olaya-Castro, A.; van Grondelle, R. Lessons from nature about solar light harvesting. Nat. Chem. 2011, 3, 763-774. (6) Bredas, J. L.; Sargent, E. H.; Scholes, G. D. Photovoltaic concepts inspired by coherence effects in photosynthetic systems. Nat. Mater. 2016, 16, 35-44. (7) Mathew, S.; Yella, A.; Gao, P.; Humphry-Baker, R.; Curchod, B. F.; Ashari-Astani, N.; Tavernelli, I.; Rothlisberger, U.; Nazeeruddin, M. K.; Gratzel, M. Dye-sensitized solar cells with 13% efficiency achieved through the molecular engineering of porphyrin sensitizers. Nat. Chem. 2014, 6, 242-247. (8) Nazeeruddin, M. K.; Baranoff, E.; Grätzel, M. Dye-sensitized solar cells: A brief overview. Sol. Energy 2011, 85, 1172-1178. (9) Narayan, M. R. Review: Dye sensitized solar cells based on natural photosensitizers. Renew. Sustainable Energy Rev. 2012, 16, 208-215. (10) Zhou, H.; Ding, L.; Fan, T.; Ding, J.; Zhang, D.; Guo, Q. Leaf-inspired hierarchical porous CdS/Au/N-TiO2 heterostructures for visible light photocatalytic hydrogen evolution. Appl. Catal., B 2014, 147, 221-228. (11) Tong, Z.; Yang, D.; Li, Z.; Nan, Y.; Ding, F.; Shen, Y.; Jiang, Z. Thylakoid-Inspired Multishell g-C3N4 Nanocapsules with Enhanced Visible-Light Harvesting and Electron Transfer Properties for High-Efficiency Photocatalysis. ACS Nano 2017, 11, 1103-1112. (12) Nocera, D. G. The artificial leaf. Acc. Chem. Res. 2012, 45, 767-776. (13) Kubra, K. T.; Rafique, M. S.; Siraj, K.; Sharif, R.; Ghaffar, A. Solar driven hydrogen from photocatalytic water splitting. Sci. Int. (Lahore) 2016, 28, 3821-3824.


25


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 29

(14) Schultz, D. M.; Yoon, T. P. Solar synthesis: prospects in visible light photocatalysis. Science 2014, 343, 1239176. (15) Romero, N. A.; Nicewicz, D. A. Organic Photoredox Catalysis. Chem. Rev. 2016, 116, 10075-10166. (16) Shaw, M. H.; Twilton, J.; MacMillan, D. W. Photoredox Catalysis in Organic Chemistry. J. Org. Chem. 2016, 81, 6898-6926. (17) Tibbetts, J. D.; Carbery, D. R.; Emanuelsson, E. A. C. An in-depth study of the use of Eosin Y for the solar photocatalytic oxidative coupling of benzylic amines. ACS Sustainable Chem. Eng. 2017, DOI: 10.1021/acssuschemeng.7b01754. (18) Oelgemoller, M. Solar Photochemical Synthesis: From the Beginnings of Organic Photochemistry to the Solar Manufacturing of Commodity Chemicals. Chem. Rev. 2016, 116, 9664-9682. (19) Heinemann, C.; Xing, X.-C; Warzecha, K.-D.; Ritterskamp, P.; Görner, H.; Demuth, M. An asymmetric induction principle and biomimetics with photons. Pure Appl. Chem. 1998, 70, 2167-2176. (20) Monnerie, M.; Ortner, J. Economic evaluation of the industrial photosynthesis of rose oxide via lamp or solar operated photooxidation of citronellol. J. Sol. Energy Eng. 2001, 123, 171174. (21) Cambié, D.; Zhao, F.; Hessel, V.; Debije, M. G.; Noël, T. A leaf-Inspired luminescent solar concentrator for energy-efficient continuous-flow photochemistry. Angew. Chem. Int. Ed. 2017, 56, 1050-1054. (22) Debije, M. G.; Verbunt, P. P. C. Thirty years of luminescent solar concentrator research: Solar energy for the built environment. Adv. Energy Mater. 2012, 2, 12-35. (23) Knowles, J. P.; Elliott, L. D.; Booker-Milburn, K. I. Flow photochemistry: Old light through new windows. Beilstein J. Org. Chem. 2012, 8, 2025-2052. (24) Su, Y.-H.; Straathof, N. J. W.; Hessel, V.; Noël, T. Photochemical transformations accelerated in continuous-flow reactors: Basic concepts and applications. Chem. Eur. J. 2014, 20, 10562-10589. (25) Plutschack, M. B.; Correia, C. A.; Seeberger, P. H.; Gilmore, K. Organic photoredox chemistry in flow. Top. Organomet. Chem. 2016, 57, 43-76. (26) Cambié, D.; Bottecchia, C.; Straathof, N. J.; Hessel, V.; Noël, T. Applications of


26

Page 27 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


continuous-flow photochemistry in organic synthesis, material science, and water treatment. Chem. Rev. 2016, 116, 10276-10341. (27) Noël, T.; Su, Y.-H.; Hessel, V. Beyond organometallic flow chemistry: the principles behind the use of continuous-flow reactor for synthesis. Top. Organomet. Chem. 2016, 57, 1-42. (28) Liu, Z.-D; Lu, Y.-C.; Wang, J.-W.; Luo, G.-S. Mixing characterization and scaling-up analysis of asymmetrical T-shaped micromixer: Experiment and CFD simulation. Chem. Eng. J. 2012, 181-182, 597-606. (29) Woitalka, A.; Kuhn, S.; Jensen, K. F. Scalability of mass transfer in liquid–liquid flow. Chem. Eng. Sci. 2014, 126, 1-8. (30) Rebrov, E. V.; Schouten, J. C.; de Croon, M. H. J. M. Single-phase fluid flow distribution and heat transfer in microstructured reactors. Chem. Eng. Sci. 2011, 66, 1374-1393. (31) Guo, X.; Fan, Y.; Luo, L. Mixing performance assessment of a multi-channel mini heat exchanger reactor with arborescent distributor and collector. Chem. Eng. J. 2013, 227, 116-127. (32) Nagaki, A.; Hirose, K.; Tonomura, O.; Taniguchi, S.; Taga, T.; Hasebe, S.; Ishizuka, N.; Yoshida, J.-i. Design of a numbering-up system of monolithic microreactors and its application to synthesis of a key intermediate of valsartan. Org. Process Res. Dev. 2016, 20, 687-691. (33) Su, Y.-H.; Kuijpers, K.; Hessel, V.; Noël, T. A convenient numbering-up strategy for the scale-up of gas–liquid photoredox catalysis in flow. React. Chem. Eng. 2016, 1, 73-81. (34) Kuijpers, K. P. L.; van Dijk, M. A. H.; Rumeur, Q. G.; Hessel, V.; Su, Y.-H.; Noël, T. A sensitivity analysis of a numbered-up photomicroreactor system. React. Chem. Eng. 2017, 2, 109-115. (35) Commenge, J. M.; Falk, L.; Corriou, J. P.; Matlosz, M. Optimal design for flow uniformity in microchannel reactors. AIChE J. 2002, 45, 345-358. (36) Jang, J.-Y.; Cheng, C.-H.; Huang, Y.-X. Optimization of channel width distribution of a gas flow field by integrating computational fluid dynamics code with a simplified conjugate-gradient method. Energy Fuels 2010, 24, 570-580. (37) Mas, N. d.; Gunther, A.; Kraus, T.; Schmidt, M. A.; Jensen, K. F. Scaled-out multilayer gasliquid microreactor with integrated velocimetry sensors. Ind. Eng. Chem. Res. 2005, 44, 89979013. (38) Yue, J.; Boichot, R.; Luo, L.; Gonthier, Y.; Chen, G.; Yuan, Q. Flow distribution and mass transfer in a parallel microchannel contactor integrated with constructal distributors. AIChE J.


27


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 29

2010, 56, 298-317. (39) Amador, C.; Gavriilidis, A.; Angeli, P. Flow distribution in different microreactor scale-out geometries and the effect of manufacturing tolerances and channel blockage. Chem. Eng. J. 2004, 101, 379-390. (40) Min, K. I.; Im, D. J.; Lee, H. J.; Kim, D. P. Three-dimensional flash flow microreactor for scale-up production of monodisperse PEG-PLGA nanoparticles. Lab Chip 2014, 14, 3987-3992. (41) Su, Y.-H.; Lautenschleger, A.; Chen, G.; Kenig, E. Y. A numerical study on liquid mixing in multichannel micromixers. Ind. Eng. Chem. Res. 2014, 53, 390-401. (42) Yap, S. K.; Wong, W. K.; Ng, N. X. Y.; Khan, S. A. Three-phase microfluidic reactor networks

–

Design,

modeling

and

application

to

scaled-out

nanoparticle-catalyzed

hydrogenations with online catalyst recovery and recycle. Chem. Eng. J. 2016, 169, 117-127. (43) Ajmera, S. K.; Delattre, C.; Schmidt, M. A.; Jensen, K. F. Microfabricated Differential Reactor for Heterogeneous Gas Phase Catalyst Testing. J. Catal. 2002, 209, 401-412. (44) Su, Y.-H.; Chen, G.; Kenig, E. Y. An experimental study on the numbering-up of microchannels for liquid mixing. Lab Chip 2015, 15, 179-187. (45) Bhattacharjee, N.; Urrios, A.; Kang, S.; Folch, A. The upcoming 3D-printing revolution in microfluidics. Lab Chip 2016, 16, 1720-1742. (46) Waheed, S.; Cabot, J. M.; Macdonald, N. P.; Lewis, T.; Guijt, R. M.; Paull, B.; Breadmore, M. C. 3D printed microfluidic devices: enablers and barriers. Lab Chip 2016, 16, 1993-2013. (47) Bergman, T. L.; Lavine, A. S.; Incropera, F. P.; DeWitt, D. P. Fundamentals of heat and mass tarnsfer; John Wiley & Sons: Hoboken, 2011. (48) Qin, D.; Xia, Y.; Whitesides, G. M. Soft lithography for micro- and nanoscale patterning. Nat. Protoc. 2010, 5, 491-502. (49) Cambié, D; Zhao, F.; Hessel, V.; Debije, M. G.; Noël, T. Every photon counts: understanding and optimizing photon paths in luminescent solar concentrator-based photomicroreactors (LSCPMs). React. Chem. Eng. 2017, 2, 561-566. (50) Ciamician, G. The photochemistry of the future. Science 1912, 36, 385-394.


28

Page 29 of 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


For Table of Contents Use Only: This paper describes the efficient scale up of solar-driven luminescent solar concentrator based photomicroreactors via a numbering‐up strategy.


29

Scale-up of a Luminescent Solar Concentrator ... - ACS Publications

Recommend Documents