Enhanced Hole Extraction in Perovskite Solar ... - ACS Publications

Subscriber access provided by UNIV OF LETHBRIDGE

Letter

Enhanced Hole Extraction in Perovskite Solar Cells Through Carbon Nanotubes Severin N. Habisreutinger, Tomas Leijtens, Giles E. Eperon, Samuel David Stranks, Robin J. Nicholas, and Henry J. Snaith J. Phys. Chem. Lett., Just Accepted Manuscript • DOI: 10.1021/jz5021795 • Publication Date (Web): 06 Nov 2014 Downloaded from http://pubs.acs.org on November 10, 2014

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.

The Journal of Physical Chemistry Letters 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 20

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

The Journal of Physical Chemistry Letters

1

Enhanced Hole Extraction in Perovskite Solar

2

Cells Through Carbon Nanotubes

3

Severin N. Habisreutinger1, Tomas Leijtens1, Giles E. Eperon1, Samuel D. Stranks1, Robin J.

4

Nicholas1 & Henry J. Snaith1*

5 6

1

University of Oxford, Department of Physics, Clarendon Laboratory, Parks Road, Oxford

7

OX1 3PU, United Kingdom

8

*[email protected]

9

ABSTRACT

10 11

Here, we report on an approach for employing polymer-wrapped carbon nanotubes to

12

enhance charge extraction through an undoped, amorphous hole-conductor matrix enabling

13

good solar cell performance to be achieved without doping of the hole-conductor. We give

14

insights into the mechanisms of how a stratified two-layer approach composed of a network

15

of nanotubes and a layer of undoped spiro-OMeTAD facilitates rapid hole extraction.

16 17

KEYWORDS: Perovskite solar cells, single-walled carbon nanotubes, hole-transporting

18

material, spiro-OMeTAD, photovoltaics

19 20 1 ACS Paragon Plus Environment


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

1

Page 2 of 20

MAIN TEXT

2

Perovskite solar cells have experienced skyrocketing performance gains in the past two

3

years.1,2 The perovskite material CH3NH3PbI3 made from mixed halide precursors has the

4

remarkable ability to function both as a wide-range absorber and as an ambipolar charge

5

transporter.3–5 It can therefore be coated onto an electronically inert alumina scaffold

6

resulting in ‘meso-superstructured solar cells’ (MSSCs) in which charge generation and

7

transport both happen within the perovskite material.4 After photogeneration, both electrons

8

and holes reside in the bulk of the perovskite. To be effectively extracted and to avoid

9

recombination losses the perovskite layer needs to be contacted with charge-selective n-type

10

and p-type materials, which exhibit low resistive losses. The current best practice employs a

11

p-type

12

spirobifluorene (spiro-OMeTAD) as the hole transport layer (HTL). The low intrinsic charge

13

carrier mobility of amorphous organic hole-conductors can be overcome to a certain extent

14

by increasing the carrier density by oxidative doping.6–8 However, we have demonstrated

15

previously, that the stability of a perovskite solar cell depends crucially on the ability of the

16

hole-transporter to withstand the ingress of moisture at elevated temperatures.9 In particular,

17

in the presence of hygroscopic dopants such as Li-TFSI, moisture-induced degradation is

18

significantly accelerated. By using a protective polymer matrix for protection in conjunction

19

with single-walled carbon nanotubes for charge extraction, we demonstrated a strategy to

20

significantly improve the overall stability of perovskite solar cells.9

organic

hole-conductor,

2,2’,7,-7’-tetrakis(N,N-di-p-methoxyphenylamine)-9,9’-

21

In the following study, we elucidate the rationale and basic mechanism for employing a

22

two-layered structure of polymer-wrapped carbon nanotubes embedded within an organic

23

matrix, and show that the nanotubes can fully replace any doping previously required for

24

efficient devices. As model system for the organic matrix, we use undoped spiro-OMeTAD -

25

which can conduct holes but at much slower rates compared to carbon nanotubes. Using the 2 ACS Paragon Plus Environment

Page 3 of 20

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


1

device performance of perovskite solar cells as well as photoinduced absorption spectroscopy

2

and small perturbation photocurrent decays, we propose a mechanism for the enhanced

3

charge extraction in a two-layer approach, which has the potential to serve as a model

4

structure for a range of applications in which carbon nanotubes conduct charges.

5 6

We employ poly(3-hexylthiophene) (P3HT) wrapping to make the SWNTs dispersible in

7

common solvents.10 Such supramolecular nanohybrids (P3HT/SWNT) composed of SWNTs

8

and a polymer-coating have been shown previously to enhance the performance of certain

9

organic solar cells as additives.11,12 Because the energy levels of the wrapping sheath of

10

P3HT make the electron transfer from the perovskite to the SWNTs energetically

11

unfavorable, the polymer coating should lead to selective hole-transfer to the nanotubes

12

(Supporting information).11

13

The solar cell structure employed in this study is shown schematically in Figure 1, where

14

sequential layers of compact n-type TiO2, mesoporous Al2O3, perovskite CH3NH3PbI3-xClx,

15

and SWNT-hole conductor composite are coated on to fluorine doped tin oxide (FTO) glass.

3 ACS Paragon Plus Environment


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 20

1 2

Figure 1: a) Schematic of a device with a blend structure composed of undoped spiro-OMeTAD and

3

P3HT/SWNTs b) Schematic architecture of a device with a P3HT/SWNT layer underneath the hole transport

4

material (spiro-OMeTAD) matrix. c) A cross-sectional scanning electron microscope (SEM) image of a

5

representative device. Individual SWNTs can be seen to protrude from the interface between the alumina and

6

spiro-OMeTAD layers. d) Top view of a bare SWNT layer showing the thick mesh-like structure of the SWNT

7

layer.

8 9

Figure 2 shows the current-voltage characteristics for representative devices in different

10

configurations: without a hole-transport layer, with undoped spiro-OMeTAD and with

11

composites of neat spiro-OMeTAD and P3HT/SWNTs.

12

The device without a p-type collection layer, i.e. direct contact of the hole-collecting

13

electrode to the perovskite exhibits very poor current-voltage characteristics, with especially

14

low short-circuit photocurrent. This suggests that direct metallic contact to the perovskite

15

absorber in the MSSC structure, where both electrons and holes reside within the perovskite,


Page 5 of 20

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


1

is not advantageous for selective charge extraction. It has been shown that if the perovskite is

2

coated onto an electron-accepting mesoporous layer of TiO2, photogenerated holes could be

3

directly collected by a gold electrode,13 because rapid electron transfer to the mesoporous

4

TiO2 competes more favorably than electron transfer to, and recombination at, the gold

5

electrode, however, the efficiency is significantly lower.14 In the MSSC architecture, in

6

contrast, electrons are not extracted so quickly at the electron collection side, so electron

7

transfer from the perovskite to the metal electrode is likely to occur, constituting a

8

recombination pathway. A significant improvement is therefore observed even when undoped

9

spiro-OMeTAD is used as the hole-collection layer. Because this material accepts holes but

10

blocks electrons, it prevents recombination at the electrode. This reduced recombination is

11

particularly evident in the enhanced open-circuit voltage and increased short-circuit

12

photocurrent. Despite the low conductivity in the spiro-OMeTAD film, this configuration can

13

still deliver a device with up to 16.3 mA/cm2 photocurrent, but only 6.8% efficiency due to a

14

poor fill factor (figure 2, table 1). Evidently, charge extraction is limited by the low mobility

15

of the photogenerated charges inside the spiro-OMeTAD layer.

16

Single-walled carbon nanotubes are known for their outstanding charge-carrier transport

17

characteristics both on their own,15–17 and also as additives in polymer matrices, to improve

18

the charge transport within that phase.18,19 We find accordingly, that dispersing

19

P3HT/SWNTs within the spiro-OMeTAD layer results in a significant increase in device

20

efficiency, which we ascribe to improved charge transport throughout the spiro-OMeTAD

21

layer. In terms of device performance this translates into an increase in the extracted

22

photocurrent, as well as an improved fill factor and open-circuit voltage. This enhancement

23

of performance parameters results in an efficiency increase to 10.0% (figure 2, table 1).

24

However, compared to the performance parameters of state-of-the art MSSC perovskite solar

25

cells the photocurrent and fill factor are still low.3,20,21 We suspect that the nanotubes in this



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 20

1

configuration cannot effectively form a closely interconnected percolation network, that can

2

effectively extend throughout the spiro-OMeTAD layer, resulting in the device being limited

3

by series resistance and thus limiting the fill factor. Being randomly dispersed through the

4

spiro-OMeTAD layer, the nanotubes may also not have enough direct interfacial contact to

5

the perovskite absorber to efficiently transport holes away from the interface.

6

7 8

Figure 2: Comparison of current-voltage characteristics of two strategies of P3HT/SWNT incorporation into

9

spiro-OMeTAD and two control structures. The presence of spiro-OMeTAD as hole transport layer (orange

10

diamonds) leads to a significant improvement in efficiency compared to the device without HTL (grey

11

triangles). Dispersing P3HT/SWNTs in the spiro-OMeTAD layer further improve the JV-characteristics (green

12

circles). However, the greatest improvement is achieved by sequential deposition of P3HT/SWNTs and spiro-

13

OMeTAD (red squares), which results in a stratified HTL structure.

14 15 16

Table 1: performance parameters of devices with different hole transport layers. These include two approaches of incorporating P3HT/SWNTs into the spiro-OMeTAD layer and two control structures. 2

2

HTL architecture

Jsc [mA/cm ]

Voc [V]

FF

PCE [%]

Rs [Ω Ω /cm ]

no HTL

2.4

0.68

0.53

0.8

45.0

spiro-OMeTAD

16.3

0.97

0.43

6.8

16.9


Page 7 of 20

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


P3HT/SWNT-spiro-OMeTAD (blend)

19.1

0.97

0.54

10.0

8.8

P3HT/SWNT-spiro-OMeTAD (stratified)

21.4

1.02

0.71

15.4

1.8

1 2 3 4

Based on these considerations, we investigated a second approach in which the nanotubes

5

and spiro-OMeTAD are deposited sequentially. The initial deposition of the P3HT/SWNTs

6

alone ensures that the nanotubes form a densely interconnected network with a direct

7

interface with the perovskite layer allowing efficient direct transfer of photogenerated holes

8

from the absorber to the nanotubes. Subsequent coating and infilling with spiro-OMeTAD

9

prevents the evaporated top electrode from directly contacting the perovskite, and forms the

10

hole-conductor contact to the electrode. The nanotube network thus constitutes a highly-

11

conductive percolation network within the amorphous hole conductor matrix.

12

In figure 1 c) we show a scanning electron microscopy (SEM) image of a representative

13

device cross-section. P3HT/SWNT bundles are apparent protruding from the interface

14

between the Al2O3-perovskite and spiro-OMeTAD layer. We assume that spiro-OMeTAD

15

readily infiltrates the P3HT/SWNT layer filling the openings and gaps of the mesh-like

16

nanotube layer, due to its excellent pore filling properties, and good wettability upon coating

17

this surface. There is clearly stratification, with much higher P3HT/SWNT density near the

18

perovskite surface, which should allow efficient charge transfer away from the interface. In

19

fact, this is evident as a much improved photocurrent in the current-voltage characteristics of

20

a device with such a sequentially coated stratified P3HT/SWNT-spiro-OMeTAD composite

21

layer, shown in figure 2. The other significantly improved performance parameter is the fill

22

factor, which can be attributed to fewer series resistance losses. This device yields a scanned

23

power-conversion efficiency of 15.4% (table 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 8 of 20

1

It is clear that the stratified structure with the formation of a dense, interconnected network

2

of P3HT/SWNT hybrids allows photogenerated holes from the perovskite to rapidly move

3

through the horizontal plane to the most efficient vertical pathway to the top electrode. The

4

fact that the devices work so well indicates that direct contact of the P3HT/SWNTs to the

5

perovskite surface does not introduce an undesired electron recombination pathway but

6

instead facilitates selective hole transport.

7

As previously pointed out, perovskite solar cells often exhibit hysteretic current-voltage

8

characteristics.22,23 Changing the scanning conditions can therefore result in a change of the

9

solar cell performance calculated from such a JV-curve. Determining the stabilized power

10

output by holding a device at constant voltage around its maximum power point is an

11

unambiguous performance metric for such devices.22 We find that well-working devices

12

usually have stabilized power output of around 0.8 of their scanned efficiency (Supporting

13

Information).

14

In order to corroborate the proposed mechanism of charge extraction, we performed

15

photoinduced absorption (PIA) spectroscopy. PIA is a quasi-continuous-wave (CW)

16

technique, which enables probing the optical signature of photogenerated species. Because

17

spiro-OMeTAD and P3HT/SWNTs exhibit different optical signatures when oxidized, it can

18

be determined whether the photogenerated holes are residing on either component or both.

19

The PIA spectrum of a device with undoped spiro-OMeTAD HTL is shown in figure 3 a).

20

The broad feature at ~1300 nm is characteristic of the presence of holes in spiro-OMeTAD

21

(i.e. its oxidized state).4,24,25 For the stratified P3HT/SWNT-spiro-OMeTAD structure, we

22

observe that the broad feature of oxidized spiro-OMeTAD becomes weaker and appears to be

23

overlaid with a second, longer wavelength feature centered around 1450 nm. This suggests a

24

smaller or shorter-lived hole population in the spiro-OMeTAD in the presence of

25

P3HT/SWNTs. To isolate the specific features of P3HT/SWNTs in the PIA spectra, we


Page 9 of 20

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


1

carried out PIA measurements on devices with neat P3HT/SWNT films in place of spiro-

2

OMeTAD. The resulting PIA spectrum is plotted in figure 3 b) together with the absorption

3

spectrum of a P3HT/SWNT film on glass. The absorption spectrum has the characteristic

4

absorption peaks associated with the E11 transitions of s-type P3HT/SWNTs according to

5

their chirality.26 In the PIA spectrum, we observe a series of photobleaching (PB) features,

6

which perfectly coincide with bleaching of the ground state absorption of the P3HT/SWNTs.

7

The PIA absorption feature further into the infrared is presumably absorption of the oxidized

8

P3HT/SWNTs or electrons in the electron-accepting TiO2 layer on the adjacent side of the

9

perovskite film. The photobleaching features were not observed in PIA measurements on

10

films of P3HT/SWNTs on glass without the active perovskite layer. We can thus exclude that

11

the observed PB features are caused by charge transfer between the SWNTs and the

12

encapsulating P3HT sheath.



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 20

1 2

Figure 3: a) shows the photoinduced-absorption (PIA) spectrum of both neat spiro-OMeTAD and the

3

P3HT/SWNT-spiro-OMeTAD stratified HTL. The broad PIA feature at 1300 nm is characteristic of the

4

presence of holes in spiro-OMeTAD. In the presence of the P3HT/SWNT interlayer this feature appears to be

5

much weaker and appears to be overlaid with a second, longer wavelength feature centered around 1450 nm. b)

6

The PIA spectrum of a device with a neat P3HT-wrapped P3HT/SWNT layer as HTL is plotted together with

7

the absorption spectrum of a P3HT/SWNT film on glass. The photobleaching dips correspond well to the

8

exciton absorption peaks of the s-type SWNTs from which hole injection from the perovskite into the

9

P3HT/SWNT valence band can be inferred. c) the absorption spectrum resulting from subtracting the absorption


Page 11 of 20

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


1

spectrum of undoped SWNTs and highly p-doped (oxidized) SWNTs in solution, overlaid with the

2

photoinduced absorption spectrum of a P3HT/SWNT film on top of a perovskite absorber (excited at 515 nm).

3 4

To confirm that the photobleaching features observed in the PIA measurements

5

correspond in fact to the transfer of photogenerated holes to the SWNTs, we chemically p-

6

doped

7

tetracyanoquinodimethane (F4TCNQ).27 This chemical p-doping results in charge depletion

8

of the SWNT valence bands. The change in absorbance between the undoped and the highly

9

p-doped (oxidized) absorption spectrum (figure 3 a)) shows a bleaching of the absorption of

10

the ground state transitions (E11) which resembles closely the photobleaching features

11

detected in the PIA measurements (figure 3 c)). We take this finding as evidence that the

12

photobleaching features can be attributed to the injection of holes photogenerated in the

13

perovskite absorber. These measurements show that the P3HT/SWNT nanohybrids act as an

14

efficient hole-acceptor in their own right.

functionalized

SWNTs

in

solution

using

2,3,5,6-

tetrafluoro-7,7,8,8-

15

To further clarify how direct hole transfer from the perovskite to the nanotubes leads to an

16

increase in efficiency, we performed small perturbation transient photocurrent decay (figure

17

4).28



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 20

1 2

Figure 4: results of the decay measurements of two devices with two different HTLs, one with neat spiro-

3

OMeTAD, one with SWNT-spiro-OMeTAD. The charge extraction lifetime is shown in a). The device with the

4

SWNT interlayer exhibits close to one order of magnitude faster extraction. The recombination lifetime in b)

5

appears to be quite similar in both devices. The faster charge extraction leads to a higher charge extraction

6

efficiency with increasing light intensity, shown in c).


Page 13 of 20

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


1

Figure 4 a) illustrates the charge extraction time measured at open circuit voltage plotted

2

against various light intensities for two devices with a HTL composed of either solely spiro-

3

OMeTAD or of the sequentially coated SWNT-spiro-OMeTAD structure. For all light

4

intensities, it can be seen that charge extraction occurs almost one order of magnitude faster

5

in the presence of the P3HT/SWNT nanohybrids interlayer.

6

The fact that the charge extraction lifetimes decrease with light intensity for both

7

structures indicates that charge transport through undoped spiro-OMeTAD limits hole

8

transport for both structures, because we have previously demonstrated that neat spiro-

9

OMeTAD has a strongly charge density dependent mobility.29 In the case of the nanotube

10

composite structure, however, only a fraction of the photogenerated holes is transported

11

through spiro-OMeTAD whereas others are extracted through the SWNTs, which results in a

12

greatly improved extraction rate. The enhancement of the hole extraction throughout the HTL

13

appears to be the main cause for the overall improved performance, based on the observation

14

that the recombination rate is comparable between the undoped spiro-OMeTAD layer and the

15

nanohybrid structure (figure 4b)). At higher light intensities in particular, the improved

16

transport properties of the stratified HTL structure results in much higher charge collection

17

efficiency values (defined as the ratio of extraction rate to the sum of extraction and

18

recombination rate) (figure 4 c)). This is in agreement with a linear dependence of the short-

19

circuit current on light intensity (Supporting Information).

20

In summary, we have shown that P3HT-wrapped single-walled carbon nanotubes act as

21

highly effective p-type charge transporter in perovskite solar cells and allow us to fully

22

replace Li-TFSI doping of spiro-OMeTAD usually required for comparable efficiencies.

23

Sequentially depositing nanotubes and spiro-OMeTAD is shown to be superior to blend

24

structures, most likely because the nanotubes can form a well-connected percolation network

25

in this configuration. This double-layer structure has the potential to become a model system 13 ACS Paragon Plus Environment


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 20

1

for carbon nanotube based charge-collection layers, which can extract charges as well as

2

protect vulnerable interfaces.

3 4 5

METHODS

6

SWNT Functionalization. Powdered singled-walled carbon nanotubes (SWNTs) produced

7

by the HiPco process were purchased from Carbon Nanotechnologies Incorporated (CNI;

8

now Unidym) with lengths 100–1000 nm and an intermediate diameter distribution of 0.8–1.2

9

nm. The samples used in this work were purchased as ‘purified’ tubes (

Enhanced Hole Extraction in Perovskite Solar ... - ACS Publications

Recommend Documents