Selective Optical Switching of Interface-Coupled Relaxation Dynamics

Subscriber access provided by UNIV OF ALABAMA BIRMINGHAM

Article

Selective Optical Switching of Interface-coupled Relaxation Dynamics in Carbon Nanotube-Si Heterojunctions Stefano Ponzoni, Gianluca Galimbertii, Luigi Sangaletti, Paola Castrucci, Silvano del Gobbo, Maurizio Morbidoni, Manuela Scarselli, and Stefania Pagliara J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/jp506684b • Publication Date (Web): 23 Sep 2014 Downloaded from http://pubs.acs.org on September 30, 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 C 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 22

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

Selective Optical Switching of Interface-Coupled Relaxation Dynamics in Carbon Nanotube-Si Heterojunctions S. Ponzoni1, G. Galimberti1, L. Sangaletti1, P. Castrucci2, S. Del Gobbo4, M. Morbidoni2,3, M. Scarselli2, S. Pagliara1* 1

I-LAMP and Dipartimento di Matematica e Fisica, Università Cattolica, 25121 Brescia, Italy. 2

Dipartimento di Fisica, Università di Roma Tor Vergata, 00133 Roma, Italy. 3

4

Department of Materials, Imperial College London, London, UK.

Solar & Photovoltaics Engineering Research Center, King Abdullah University of Science & Technology, Thuwal, Kingdom of Saudi Arabia.

KEYWORDS: Single wall carbon nanotubes, Time resolved optical spectroscopy, Heterojunction solar cells.

ABSTRACT

By properly tuning the photon energy of a femtosecond laser pump, we disentangle, in carbon nanotube-Si (CNT-Si) heterojunctions, the fast relaxation dynamics occurring in CNT from the slow repopulation dynamics due to hole charge transfer at the junction. In this way

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 2 of 22

we are able to track the transfer of the photo-generated holes from the Si depletion layer to the CNT layer, under the action of the built-in heterojunction potential. This also clarifies that CNT play an active role in the junction and do not act only as channels for charge collection and transport.

INTRODUCTION

One of the key aspects underlying the application of novel materials in photovoltaic devices is the understanding of the steps immediately following the arrival of the light. The overall efficiency of a photovoltaic device is mainly governed by two factors: absorption efficiency and its internal quantum efficiency. The former is influenced by the capability of the active material to photogenerate electron-hole pairs. The latter is mainly determined by the diffusion, dissociation and transfer of the excited charge. These processes occur on ultrafast timescales.

1,2,3,4

Indeed, much of the current research on solar cell devices focuses on the

time averaged properties, even if such steady state measurements give little information about processes actually occurring on fast timescales. In this framework, time resolved optical spectroscopy could be a powerful tool for exploring the sequences following hole-electron pair photo-excitation. 5,6 However, when this spectroscopy is carried out on a junction involving two different materials, as commonly happens for heterojunction cells, the transient signal is quite often an overlap of similar optical contributions coming from the different components of the junction, hindering the possibility to separately track the paths of the excited carriers. In this work, this drawback has been overcome by exploiting the different transient response of the heterojunction materials, namely single walled carbon nanotubes (SWCNT) and n-doped Silicon (n-Si). In the recent years, carbon nanotubes (CNT) have emerged as new building blocks

for

different

technological

applications

including


light-energy

harvesting

Page 3 of 22

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


assemblies7,8. In particular, the CNT/n-Si interfaces are among the most promising heterojunctions for the development of novel hybrid photovoltaic devices, showing efficiencies up to 10-13%9,10,11,12. The solar cells consist of a semitransparent thin film of nanotubes conformally coating a n-type crystalline silicon substrate to create high-density p-n heterojunctions9,10,11,12,13 or Schottky junctions12,14,15. In these junctions, it is not clear whether the CNT behave only as a channel to collect the charges generated on the n-Si side or they can serve also as photogeneration sites16,17. The goal of this work is to track the dynamics of charge transfer processes and ultimately to understand the role of CNT in the SWCNT/n-Si heterojunction. This is achieved by selective optical switching of charge relaxation channels. To single out the dynamics at the junction, two strategies were pursued: i) the pump photon energy has been tuned across the Si absorption edge; ii) the transient optical response from the cell has been directly compared with the ones collected on the same type of SWCNT deposited, in the same experimental conditions, on a glass substrate. This will ensure that differences between the transient behavior at the two interfaces can be ultimately ascribed to the presence of the heterojunction in the CNT layer deposited on silicon. EXPERIMENTAL Fig. 1a shows a scheme of the fabricated solar cell. An n-type Si(100) has been covered by patterned SiO2 layer in order to obtain a 5x5 mm2 bare silicon window in the middle of two SiO2 steps. This process allows to create nanotube/n-Si multijunctions in the n-Si window and, at the same time, to insulate the silver paste deposited on the nanotube film from the n-Si underneath so to avoid short circuits between the metal electrode and the Si. For the present experiments CoMoCat commercial grade (CG200 Southwest Nanotechnology) SWCNT powder with a carbon content ≥90% and SWCNT content ≥70% with various chiralities and diameter ranging between 0.7-1.4 nm has been used. The nanotube dispersion was prepared



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

by sonicating about 100 mg of SWCNT powder in a sodium dodecylsulfate (SDS) (assay > 98.5%, Sigma Aldrich) solution in water, concentration of 3% in weight. After 1 hour of ultrasound treatment, the dispersion was left to settle for one night and the clear supernatant was separated from the precipitate and used to fabricate the film that is than transferred on the patterned Si substrate and on glass, following the procedure reported in ref.13 In the same paper, from the analysis of the Raman radial breathing mode spectrum of the SWCNT films transferred onto glass substrates, the percentage of metallic nanotubes was estimated to be not less than 65%.13 Time resolved optical measurements have been performed with two different laser systems. One-color transient reflectivity (TR) experiments have been carried out with a cavity dumped Ti:Sapphire oscillator (Coherent Mira 900 together with an APE pulse switch), producing 120 fs, 1.55 eV light pulses. The pump beam diameter at the sample is 50 µm, corresponding to fluences (F) in the 0.1 - 1.0 mJ/cm2 range and pulse intensities (Ip) between 0.8 GW/cm2 and 8.3 GW/cm2. The experimental resolution is ∆R/R=10-6. TR experiments in a two-color variable-pump and fixed-probe configuration have been performed by using a 1 kHz amplified Ti:Sapphire laser system capable to deliver 0.5 mJ, 150 fs, 1.55 eV light pulses, together with a Light Conversion traveling wave optical parametric amplifier (TOPAS) capable to generate laser pulses of 150 fs and with selectable photon energies in the 0.75 – 1.07 eV range. The probe energy is kept fixed at 1.55 eV whilst the pump energy is tuned in the 0.75 - 1.01 eV and 1.7 - 2.0 eV energy ranges. For the latter range the TOPAS output is doubled via a nonlinear optical crystal. The experimental resolution is ∆R/R=10-4. External quantum efficiency (EQE) measurements were performed by a dedicated set-up equipped with a 150 W Xenon lamp as light source and a monochromator. The EQE is defined as the fraction of the incident photons, Nph, converted into photocurrent, I(λ). The magnitude of I(λ) was measured recovering the amplified current signal by a lock-in


Page 4 of 22

Page 5 of 22

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


amplifier locked to the light modulation frequency. Simultaneously, Nph is evaluated from light power P(λ) magnitude, measured by a calibrated Si photodiode and acquired by the lock-in amplifier.

Fig.1 (a) Solar cell based on SWCNT/n-Si heterojunction and sketch of its band diagram (b). Here EC and EV represent the minimum energy of the conduction band and the maximum energy of the valence band; Ef the Fermi energy. (c) External quantum efficiency measured by a Xe lamp. (d) Sketch of the transient optical measurements on the reference sample CNT/Glass.

RESULTS AND DISCUSSION



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

Considering that the SWCNT are mainly metallic, the band diagram estimated for the SWCN/n-Si junction interface (Fig.1a) is reported in Fig.1b. The large built-in potential (≈ 0.5 eV) enables the charge separation, where electrons are directed to the n-type Si region and holes are transported through SWCNT. Recently, a thin oxide (SiO2-SiOx) layer has been found between the SWCNT and the n-Si substrate, whose thickness between 1 and 2 nm, slightly alters the band alignment at the junction interface (Fig.1b)18. Whilst it has been reported that a particular thickness of the oxide layer can lead to an increased power conversion efficiency18,19, for sake of simplicity, neglecting it, we model SWCNT/n-Si interface as a simple Schottky junction. The EQE of the solar cell based on SWCNT/n-Si heterojunction measured by using a Xe lamp is shown in Fig.1c. The EQE ranges in the visible and infrared (VIS-IR) region from 10% to 35%, being the latter the maximum values reached above 800 nm. At 800 nm the EQE is 32.5%. In order to shed light on the relaxation dynamics of the carriers excited at the SWCN/n-Si interface, the transient measurements, sketched in Fig.1d, have been carried out by changing the pump photon energy across the Silicon absorption edge. In Fig.2a the transient reflectivity (TR) at pump and probe photon energy of 1.55 eV on SWCNT/n-Si heterojunction and on a bare n-Si substrate is shown. Both signals are collected using the one-color setup. It is worth observing that, to detect a transient response from the bare n-Si substrate, a pump pulse intensity of 6.6 GW/cm2, twice as much as that used to collect the transient signal of SWCNT film (3.3 GW/cm2) is necessary. Considering that the pump pulse intensity at n-Si substrate is reduced by 70% due to the presence of SWCNT film, we can doubtless state that the transient signal detected on SWCNT/n-Si comes from SWCNT film and/or SWCNT/n-Si interface. The small transient signal from the n-Si substrate is highlighted in the insert of Fig.2a. In the two-color experimental setup, being the signal resolution smaller (∆R/R~10-4) than that of the one-color setup, the n-Si signal is


Page 6 of 22

Page 7 of 22

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


undetectable for the all pump pulse intensity values. TR experiments in a two-color variablepump and fixed-probe configuration have been performed on both SWCNT/n-Si and SWCNT/Glass samples by keeping the probe energy at 1.55 eV, whilst the pump energy is tuned in the 0.75-1.01 eV and 1.7-2.0 eV energy ranges. The reference sample (SWCNT/Glass) has been used to discriminate the SWCNT relaxation dynamics from those of SWCNT/n-Si interface. From now on, we will refer to the 0.75-1.01 eV pump energy interval as the IR range whilst to the latter as VIS range. Fig.2b and Fig.2c show TR signals on SWCNT/glass reference sample (Fig.2b) and on SWCNT/n-Si junction (Fig.2c) collected with a pump photon energy in the IR (0.87 eV, dark blue and dark red lines respectively) and in the VIS region (1.97 eV, light blue and orange lines respectively) at the same pump pulse intensity (51.3 GW/cm2). While the relaxation time of SWCNT (Fig.2b) is no longer than one picosecond, the optical response of Si, collected with the one-color setup (Fig.2a), persists for several picoseconds.



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

Fig.2 (a) One color transient reflectivity signals collected on the SWCNT/n-Si (red line) and on the Si substrate (black line) using the low fluence one-color setup with a photon energy of 1.55 eV. The pump pulse intensity is 3.3 GW/cm2 for SWCNT/n-Si and 6.6 GW/cm2 for nSi. In the insert, the n-Si signal has been highlighted to make visible the ∆R/R value. (b, c) Transient reflectivity on the SWCNT/Glass reference sample (b) and on SWCNT/n-Si junction (c) at 0.87 eV (dark blue and dark red lines) and 1.97 eV (light blue and orange lines) pump photon energy. The pump pulse intensity is kept constant at 51.3 GW/cm2. (d) Optical absorption coefficient for the metallic enriched SWCNT/Glass reference sample (blue line), for a free standing CoMoCat commercial grade SWCNT film with a wide distribution of tube diameter (Green line, adapted from24) and for crystalline Silicon (dark grey diamonds, adapted from23). The light blue arrow markers indicate the pump photon energies used in the present experiments.


Page 8 of 22

Page 9 of 22

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 both samples, in the two excitation energy ranges, the sign of the transient signals is always negative, in agreement with a photobleaching (PB) process. In a PB process the absorption of the pump pulse excites electrons in the conduction band, creating holes in the valence band. Until these carriers relax, transient filling effects on the final states decrease the absorption of the probe inducing a positive transient signal in tran smittivity and negative in reflectivity (as well as in the absorption). In the experiment, in fact, we measure the changes of the probe reflectivity (∆R/R) induced by the laser pump at different delay times (Fig.1d). The ∆R/R signal in principle depends on both Δε1/ε1 and Δε2/ε2 where ε1 and ε2 are respectively the real and the imaginary part of the dielectric function. For carbon nanotubes, similarly to graphite, ∆R/R mainly depends on

∆ε2/ε2 and therefore on the absorption of CNT21. In this framework, for CNT and graphitic systems in general, a PB process, as in Fig.2, is usually observed. However, in the case of strong intertube interactions able to induce a free– electron behavior the TR sign becomes positive. This last effect has been observed in pure unaligned SWCNT films directly grown in ultra high vacuum and therefore not affected by other atomic species (e.g. traces of solvents and surfactants)20,21,22. In Fig.2b and Fig.2c, while the TR sign does not change, the shape of transient signal for the heterojunction SWCNT/n-Si changes with the excitation wavelength. It is worth noting that the laser linewidth is slightly longer (about 30 fs) in the VIS than in the IR region thus justifying the symmetric broadening of the transient response around zero delay time. In both samples the transient signal is well fitted by a double exponential decay. The first decay is very fast (about 200 fs) while the second is longer than 800 fs. However, while with the IR pump photon energy the bi-exponential relaxation dynamics for SWCNT/n-Si are similar to those observed on the reference sample for both IR and VIS



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

regime, the transient signal collected in the VIS range (1.97 eV) shows an increase of the relaxation time. It is known in literature that, in the excitation energy range explored in this work, transient response of crystalline silicon strongly depends on pump photon energy. In Fig.2d, the optical absorption coefficient of crystalline silicon23 (dark grey diamonds) is reported together with the absorption coefficient of the metallic enriched SWCNT/Glass reference sample (blue line) estimated by assuming a film thickness of 40 nm18. The absorption coefficient of our SWCNT/Glass reference sample is comparable with the free standing CoMoCat commercial grade SWCNT film24 (green line) reported in literature for a wider photon energy range. While SWCNT absorb in all the explored pump energy range, nSi wafer absorbs mainly in the VIS region being the pump excitation energies in the IR range below the onset of the silicon indirect optical transition (~1.12 eV). Therefore, it is expected that, in the IR range, where the absorption coefficient of SWCNT is orders of magnitude greater than the silicon ones, the transient signal is mainly due to relaxation dynamics of carriers excited in SWCNT layer in contact with n-Si. Conversely, in the VIS region, carriers are excited both in SWCNT layer and in n-Si wafer. Considering that in the two-color setup the transient response on n-Si wafer is undetectable, we can argue that the enhancement of the relaxation dynamics observed in the VIS regime, where the pump photon energy is able to excite carriers in SWCNT as well as in n-Si substrate, has to be ascribed to the response of SWCNT in contact with n-Si. In order to get insights in the role of the SWCNT/n-Si junction in the relaxation dynamics, we have collected the transient response (Fig.3) for different laser pump photon energies ranging from 0.87 eV to 1.97 eV keeping costant the pump pulse intensity. The absolute value of ∆R/R on SWCNT/n-Si junction is displayed in Fig.3, normalized to the ∆R/R maximum and in a logarithmic scale to highlight the differences at longer delay times with the SWCNT/Glass reference. In the IR range, for an excitation energy of 0.78 eV (Fig.3a) the junction TR signal resembles the signal of the reference


Page 10 of 22

Page 11 of 22

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


sample. At 0.87 and 1.01 eV pump photon energies (Fig.3b and Fig.3c) the SWCNT/n-Si TR signal starts to manifest a longer relaxation dynamics deviating from the SWCNT/Glass one. This difference becomes much more evident for the pump excitation above the Si bandgap energy value as shown in Fig.3d, Fig.3e and Fig.3f collected at a pump photon energy of 1.72 eV, 1.82 eV and 1.97 eV, respectively. This behavior unambiguously proves that when the density of the charge carriers excited in the Si substrate is no more negligible, the transient response of SWCNT/n-Si junction deviates from the response of the reference sample. In order to quantify the difference observed in the relaxation dynamics, the transient signals of both samples are interpolated with a two-exponential decay fitting function.



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

Fig.3 Comparison between selected TR signals from the SWCNT/Glass reference sample (blue line) and from the SWCNT/n-Si junction (red line) in logarithmic scale. All TR signals are peak-value normalized and reversed in sign. Pump photon energies are (a) 0.78 eV, (b) 0.87 eV, (c) 1.01 eV, (d) 1.72 eV, (e) 1.82 eV, (f) 1.97 eV respectively. For all the collected signals the pump pulse intensity is kept fixed at 51.3 GW/cm2.


Page 12 of 22

Page 13 of 22

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


Fig.4 Comparison between fast (a) and slow (b) relaxation time constants, and fast (c) and slow (d) relaxation dynamics initial amplitudes of the SWCNT/Glass reference sample (blue squares) and SWCNT/n-Si junction (red circles) upon the variation of pump photon energy as given by a bi-exponential interpolation of the TR signals. Both the numerical values and the error bars magnitude are extracted by the interpolation of the bare TR signals with a double exponential decay fitting function. Considering the statistical error, in the IR region both the fast (about 225 fs) and the slow (about 1000 fs) relaxation times of the SWCNT/n-Si junction are comparable with those of SWCNT/Glass reference sample. Conversely, in the visible region where carriers are excited also in n-Si wafer, while the fast decay of the two samples is comparable (about 300 fs), the slow relaxation time of the SWCNT/n-Si junction is about twice as much as the one from the reference sample (Fig.4a and Fig.4b). The value of the fast decay time is in agreement with



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 data available in literature25,26 on similar systems and is ascribed to the rapid thermalization, via electron-electron scattering, of the laser-excited carrier population in CNT. Subsequently, the hot electron population thermalizes with the lattice via phonon emission processes giving rise to the slow component of the relaxation dynamics of about 1 ps26 in accordance with the value here measured on SWCNT on glass. Therefore, the origin of the slow dynamics enhancement observed for SWCNT/n-Si heterojunction in the VIS region has to be found in the presence of the carriers excited in the Si wafer. For both samples, the amplitude of the fast dynamics increases, going from the IR to the VIS range, (Fig.4c and Fig.4d) because of the excitation photon energy approaching the resonance with the optical transitions in the metallic tubes, localized among 1.7 and 2 eV (Fig.2d). From the data presented so far, the different optical response observed in the VIS region between the SWCNT/n-Si junction and the reference sample has to be mainly ascribed to an increase in the slow relaxation time. In the SWCNT/n-Si junction formation, the balance of the chemical potentials requires a net transfer of electrons from the n-Si towards the tubes via ionization of the silicon donor impurities in a region close to the junction region. This process leads to the creation of a Schottky barrier in which the depleted region extends almost entirely into the semiconductor for a depth of a few microns27. In the metallic tubes the excess electrons form a charge layer that can induce a significant band bending in the semiconductor substrate and the subsequent formation of a potential barrier that gives to the junction its characteristic current-rectifying and light-to-electric current conversion properties. The electron-hole pairs generated in the n-Si depleted region upon absorption of the pump photons are quickly separated by the built-in field. Whilst the electrons are swept away from the junction by the built-in field, the holes are forced towards the metallic SWCNT layer where they can easily cross the junction and become trapped in. This process,


Page 14 of 22

Page 15 of 22

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


in principle, can still be considered valid in the case the junction was formed by semiconducting SWCNT/n-Si. This picture suggests that the increased relaxation time observed in the VIS range (Fig.2 and Fig.3) is primarily due to a net flux of photoexcited holes that drift from the n-Si depleted region into the valence band of the SWCNT layer, under the effect of the built-in field in the device. Since the magnitude of the light absorption coefficient depends on the difference between the density of initial and final states available for the optical transitions27,28, injected holes could lead to a transient bleaching in the SWCNT transient signals. The slow relaxation time of the excited carriers in the n-Si substrate together with the low mobility of the holes in doped crystalline silicon29 make the hole injection, at the junction interface, acting as a slow source for the photobleaching signal, accounting for the enhancement of the slow relaxation dynamics in SWCNT/n-Si heterojunction. When the pump photon energy is below the absorption edge of the n-Si substrate, in the IR range measurements, only a small amount of electron-hole pairs are excited in the silicon giving rise to a negligible hole flux and the measured transient signal from the SWCNT layer resembles the one collected on the SWCNT/Glass reference sample. To provide further support to our claim, we have compared the transient response of two solar cells with a different EQE due to a different thickness of the SWCNT films. In Fig.5a the transient reflectivity collected on the more efficient cell, due to a smaller SWCNT network thickness (Fig.5b green line), and on the less efficient cell (Fig.5b red line), used for the measurements reported before, is shown. Both spectra are collected in the same experimental conditions with a pump and probe energy of 1.55 eV. The pump pulse intensity is 3.7 GW/cm2.



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 22

Fig.5(a) Comparison between the peak-value normalized TR signals collected on the less efficient (red line) and on the more efficient (green line) SWCNT/n-Si cells. For both traces the pump pulse intensity is 3.7 GW/cm2 at 1.55 eV. (b) External quantum efficiency spectra of the two SWCNT/n-Si cells ( red and green lines) and of a conventional p-n silicon solar cell (dark grey points). The blue line is a marker highlighting the EQE values at the pump photon energies (1.55 eV, i.e. 795 nm in term of photon wavelength) used in the one-color experiments. The more efficient cell (EQE at 795nm (1.55 eV) =60%) shows an enhancement of the second dynamics, whose relaxation time is about twice (3700 fs) the value (1640 fs) obtained on the less efficient cell (EQE at 1.55 eV =32.5%). In accordance with our picture, a highest external quantum efficiency means a larger number of electron-hole pairs excited in the depletion layer and thus a larger hole injection rate in the SWCNT layer, that we reveal through an enhancement of the second slow dynamics in the SWCNT transient response. In order to estimate the concentration of electron-hole pairs excited in the depletion layer, we have calculated, starting from the laser pump fluence, the photon number that, after crossing the SWCNT layer, whose average optical trasmittivity is 40%, arrives on the Si depletion layer. The thickness of the depletion layer has been estimated as


/ 27],

where

Page 17 of 22

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


φ is the built-in potential, n is the density of the Si doping, εr and ε0 are the silicon and the vacuum dielectric constants and e the elementary charge. The resulting depletion layer thickness is about 1 µm. With a pump fluence of 0.15 mJ/cm2 and a pump photon energy of 1.55 eV, the average density of the pairs excited in the n-Si depletion layer (calculated by assuming that 1 photon generates 1 e-h pair) is 2.3x1017 pairs/cm3 per pulse 30. This value can be compared with the average density of the photoexcited pairs that contribute to the EQE. From preliminary measurements of the EQE with the pulsed laser source, an average value of 20.5% has been obtained at a pump pulse intensity of 1.2 GW/cm2 and at 1.55 eV photon energy. Therefore, the pair density contributing to the EQE results to be 1.2x1018 pairs/cm3 per pulse, which is about 5 times higher than the average pair density excited in the n-Si depletion layer taking the SWCNT layer transmission into account. This ratio is confirmed as long as we are in a linear regime. With increasing pump fluence, in fact, the photocurrent displays a sublinear increase, due to difficulty to dissipate the created carriers through the device, that hampers the EQE behavior. This result suggests that the pairs that contribute to the EQE of the solar cell based on SWCNT/n-Si heterojuncion mainly come from the n-Si depletion layer. However, a contribution seems to come also from the carriers excited in the carbon nanotubes that, in the working solar cell, behave as photogeneration sites as well as transport transparent coating for charge carriers (holes) 13.

CONCLUSIONS

In conclusion, two relaxation channels have been identified in hybrid SWCNT/n-Si heterojunction: a fast relaxation channel typical of SWCNT and a slow relaxation channel, that appeared to be related to the presence of the heterojunction. Indeed, for energies above the Si absorption threshold, in the SWCNT/n-Si junction the slow relaxation channel



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

increases along with an enhancement of the photobleaching effect. This was recognized as an intrinsic behavior of the junction, not found on the reference SWCNT/glass interface, and was interpreted as a charge transfer of holes from n-Si to SWCNT across the junction, depleting the occupied levels of the SWCNT. Further evidence is provided by the correlation between the EQE and the slow relaxation dynamics, the latter being enhanced in the more efficient solar cells.

AUTHOR INFORMATION Corresponding Author *Stefania Pagliara, Tel: +39(030)2406.711. E-mail: [email protected] Author Contributions The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. ACKNOWLEDGMENT S.P., L.S. and S.P. thank the MIUR for supporting this work under Contract No. PRIN 2010BNZ3F2. S.P. acknowledges partial support from D.2.2 grants of the UCSC. ABBREVIATIONS CNT Carbon Nanotube; SWCNT Single Walled Carbon Nanotube; TR Transient Reflectivity; EQE External Quantum Efficiency; PB Photobleaching. REFERENCES 1. Vithanage, A.D.; Devizis, A.; Abramavicius, V.; Infahsaeng, Y.; Abramavicius, V.; MacKenzie, R.C.L; Keivanidis, P.E.; Yartsev, A.; Hertel, D.; Nelson, J.; et al. Visualizing


Page 18 of 22

Page 19 of 22

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


charge separation in bulk heterojunction organic solar cells. Nature Communications 2013, 4, 2344. 2. Cooke, D.G.; Krebs, F.C.; Jepsen, P.U. Direct Observation of Sub-100 fs Mobile Charge Generation in a Polymer-Fullerene Film. Phys. Rev. Lett. 2012, 108, 056603-5 3. Cowan, S.R.; Banerji, N.; Leong, W.L.; Heeger, A.J. Charge Formation, Recombination, and Sweep-Out Dynamics in Organic Solar Cells. Adv. Funct. Mater. 2012, 22, 1116-1128. 4. Chan, W.L; Ligges, M.; Jailaubekov, A.; Kaake, L.; Miaja-Avila. L.; Zhu, X-Y. Observing the multiexciton state in singlet fission and ensuing ultrafast multielectron transfer. Science 2011, 334,1541-1545. 5. Žídek, K.; Zheng, K.; Ponseca , C. S. Jr.; Messing M. E.; Wallenberg, R. L.; Chábera, P.; Abdellah, M.; Sundström, V.; Pullerits, T. Electron Transfer in Quantum-DotSensitized ZnO Nanowires: Ultrafast Time-Resolved Absorption and Terahertz Study. J. Am. Chem. Soc. 2012, 134, 12110-12117. 6. Zheng, K.; Karel Žídek , K.; Abdellah, M.; Zhang, W.; Chábera, P.; Lenngren, N.; Arkady Yartsev, A.; Pullerits, T. Ultrafast Charge Transfer from CdSe Quantum Dots to pType NiO: Hole Injection vs Hole Trapping. J. Phys. Chem. C 2014, 118, 18462-18471. 7. Eder, D. Carbon Nanotube - Inorganic Hybrids Chem. Rev. 2010, 110, 1348-1385. 8. Kamat, P.V. Meeting the Clean Energy Demand: Nanostructure Architectures for Solar Energy Conversion. J. Phys. Chem. C 2007, 111, 2834-2860. 9. Jia, Y.; Wei, J.; Wang, K.; Cao, A.; Shu, Q.; Gui, X.; Zhu, Y.; Zhuang, D.; Zhang, G.; Ma, B.; et al. Nanotube-Silicon Heterojunction Solar Cells. Adv. Mater. 2008, 20, 45944598.



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

10. Li, Z.; Kunets, V.P,; Saini, Y.; Xu, Y.; Dervishi, E.; Salamo, G.J.; Biris, A.R.; Biris, A.S. Light-Harvesting Using High Density p-type Single Wall Carbon Nanotube/n-type Silicon Heterojunctions. ACS Nano 2009, 3, 1407-1414. 11. Jia, Y.; Cao, A.; Bai, X.; Li, Z.; Zhang, L.; Guo, N.; Wei, J.; Wang, K.; Zhu, H.; Wu, D.; et al. Achieving High Efficiency Silicon-Carbon Nanotube Heterojunction Solar Cells by Acid Doping. Nano Lett. 2011, 11, 1901-1905. 12. Jung, Y.; Li, X.; Rjan, N.; Taylor, A.D.; Reed, M. Record high Efficiency SingleWalled Carbon Nanotube/Silicon p-n Junction Solar Cells. Nano Lett. 2013, 13, 95-99. 13. Del Gobbo, S.; Castrucci, P.; Fedele, S.; Riele, L.; Convertino, A.; Morbidoni, M.; De Nicola, F.; Scarselli, M.; Camilli, L.; De Crescenzi, M. Silicon spectral response extension through single wall carbon nanotubes in hybrid solar cells. J. Mater. Chem. C 2013, 1, 67526758. 14. Wadhwa, P.; Liu, B.; McCarthy, M.; Wu, Z.; Rinzler, A.G. Electronic Junction Control in a Nanotube-Semiconductor Schottky Junction Solar Cell. Nano Lett. 2010, 10, 5001-5005. 15. Del Gobbo, S.; Castrucci, P.; Scarselli, M.; Camilli, L.; De Crescenzi, M.; Mariucci, L.; Valletta, A.; Minotti, A.; Fortunato, G. Carbon Nanotube Semitransparent Electrodes for Amorphous Silicon Based Photovoltaic Devices. App. Phys. Lett. 2011, 98, 183113. 16. Zhu, X-Y.; Yang, Q.; Muntwiler, M. Charge-Transfer Excitons at Organic Semiconductor Surfaces and Interfaces. Acc. Chem. Res. 2009, 42, 1779-1787. 17. Scarselli, M.; Scilletta, C.; Tombolini, F.; Castrucci, P.; Diociaiuti, M.; Casciardi, E.; Gatto, E.; Venanzi, M.; De Crescenzi, M. Multiwall Carbon Nanotubes Decorated with


Page 20 of 22

Page 21 of 22

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


Copper Nanoparticles: Effect on the Photocurrent Response. J. Phys. Chem. C 2009, 113, 5860-5864. 18. Pintossi, C.; Salvinelli, G.; Drera, G.; Pagliara, S.; Sangaletti, L.; Del Gobbo, S.; Morbidoni, M.; Scarselli, M.; De Crescenzi, M.; Castrucci, P. Direct Evidence of Chemically Inhomogeneous, Nanostructured, Si-O Buried Interfaces and Their Effect on The Efficiency of Carbon Nanotube/Si Photovoltaic Heterojunctions. J. Chem. Phys. C 2013, 117, 18688-18696. 19. Jia, Y.; Cao, A.; Kang, F.; Li, P.; Gui, X.; Zhang, L.; Shi, E.; Wei, J.; Wang, K.; Zhu, H.; et al. Strong and Reversible Modulation of Carbon Nanotube-Silicon Heterojunction Solar Cells by an Interfacial Oxide Layer. Chem. Phys. 2012, 14, 8391-8396. 20. Huang, L.; Pedrosa, H.N.; Krauss, T.D. Ultrafast Ground-State Recovery of SingleWalled Carbon Nanotubes. Phys. Rev. Lett. 2004, 93, 017403-1. 21. Galimberti, G.; Pagliara, S.; Ponzoni, S.; Dal Conte, S.; Cilento, F.; Ferrini, G.; Arshad, M.; Cepek, C.; Parmigiani, F. The Photoinduced Charge Transfer Mechanism in Aligned and Unaligned Carbon Nanotubes. Carbon 201, 42, 5246-5252. 22.

Galimberti, G.; Ponzoni, S.; Ferrini, G.; Hofmann, S.; Arshad, M.; Cepek,

C.; Pagliara, S. Transient Reflectivity on Vertically Aligned Single-Wall Carbon Nanotubes. Thin Solid Films 2013, 543, 51-5. 23. Green, M.A.; Keevers, M.J. Optical properties of intrinsic silicon at 300 K. Progress in photovoltaics 1995, 3, 189-192. 24. Pekker, A.; Kamaràs, K. A General Figure of Merit for Thick and Thin Transparent Conductive Carbon Nanotube Coatings. J. Appl. Phys. 2010, 108, 054318.



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

25. Korovyanko, O.J.; Sheng, C.X.; Vardeny, Z.V.; Dalton, A.B.; Baughman, R.H. Ultrafast Spectroscopy of Excitons in Single-Walled Carbon Nanotubes. Phys. Rev. Lett. 2004, 92, 017403-017434. 26. Hertel, T.; Moos, G. Electron-Phonon Interaction in Single-Wall Carbon Nanotubes: A Time-Domain Study. Phys. Rev. Lett. 2000, 84, 5002-5. 27. Grosso, G.; Pastori Parravicini, G. Solid State Physics. 2nd ed. Academic Press: San Diego, USA, 2000. 28. Breusing, M.; Ropers, C.; Elsaesser, T. Ultrafast Carrier Dynamics in Graphite. Phys. Rev. Lett. 2009, 102, 086809. 29. Wang, C.H.; Misiakos, K.; Neugroschel, A. Minority-Carrier Transport Parameters in N-Type Silicon. IEEE Trans. on electron devices 1990, ED-37, 1314-1322. 30. The average density of photoexcited pairs in the n-silicon depleted region is estimated by using the following relation / where , 40 % is the fraction of the pump fluence () that achieves the silicon substrate, α is the silicon absorption coefficient at the pump wavelength, s the sample thickness and is the photon frequency.


Page 22 of 22

Selective Optical Switching of Interface-Coupled Relaxation Dynamics

Recommend Documents