Article pubs.acs.org/JPCC
High Quantum Yield Dual Emission from Gas-Phase Grown Crystalline Si Nanoparticles A. M. P. Botas,†,‡ R. A. S. Ferreira,*,‡ R. N. Pereira,*,†,§ R. J. Anthony,∥ T. Moura,†,‡ D. J. Rowe,∥ and U. Kortshagen∥ †
Department of Physics and I3N and ‡Department of Physics and CICECO, University of Aveiro, Aveiro 3810-193, Portugal Walter Schottky Institut and Physik-Department, Technische Universität München, Am Coulombwall 4, Garching 85748, Germany ∥ Department of Mechanical Engineering, University of Minnesota, Minneapolis, Minnesota 55455, United States §
S Supporting Information *
ABSTRACT: We studied the light emission of crystalline Si nanoparticles (SiNPs) with hydrogen and native oxide shell terminations using temperature- and time-dependent photoluminescence. We demonstrate that the broad emission normally observed for SiNPs after natural oxidation is in fact formed of two components, which originate from distinct recombination mechanisms that take place simultaneously in the same SiNPs sample. To identify the two spectral components, we exploited the different time scales associated with different emission mechanisms by carefully choosing the measurement time window at which only one of the emission mechanisms is active. Moreover, our experiments indicate that one of the emissions is due to recombination of photogenerated electrons and holes located in the crystalline core of the SiNPs (excitonic emission) whereas the other component originates from donor−acceptor recombination pairs involving states associated with the native oxide shell. These conclusions are supported from experiments carried out with the same SiNPs but where the surface-oxide shell is replaced by H termination. We conclude that both emission components are excited through electronic states of the SiNPs core, pointing out an effective core-to-shell energy/charge transfer. Furthermore, we show that the light emission quantum yield of SiNP ensembles is strongly affected by inter-NP charge transfer and therefore is not determined solely by the properties of the individual NPs. High quantum yields of up to 43%, observed for our surface-oxidized SiNP samples in solution, result from inhibition of inter-NP charge transfer.
1. INTRODUCTION Crystalline silicon nanoparticles (SiNPs) are currently under intense investigation because they combine the unique features of Si at the nanoscale (for instance, wavelength tunable light emission,1 multiple exciton generation,2 and possibility of doping3−5) with the versatile and inexpensive device fabrication associated with nanoparticle (NP) processing.6 These efforts have readily resulted in the practical demonstration of thermoelectrics, solar energy to electricity conversion,7,8 and light emission9−11 with thin films fabricated from SiNPs. A variety of techniques have been used to produce SiNPs, such as solid−gas reaction,12 liquid-phase synthesis,13,14 laser pyrolysis of silane,15−18 laser ablation,19 laser vaporization controlled condensation,20 and plasma-assisted decomposition of silane,1,6,21,22 silicon tetrachloride,23,24 and silicon tetrabromide25 and grown by microemulsion in inverse micelles.26 Among these, the silane plasma method has proved to be very efficient in producing large quantities of high-quality SiNPs passivated with Si−H bonds with control over diameter in the range 3−50 nm.1,27 The SiNPs display a photoemission characterized by a broad/unstructured band with a spectral distribution that shifts from the near IR to the visible spectral regions as the size of the © 2014 American Chemical Society
crystalline core of the SiNPs (mentioned hereafter simply as SiNPs core) is decreased.1,15,16 This effect, similar to that observed originally for silicon nanocrystals (NCs) in porous Si,28 has also been explained in terms of the energy increase associated with the increased confinement of excitons that results from the reduction of the SiNPs core size, similarly to other nanosilicon materials.29−31 When exposed to air, at room temperature, H-terminated SiNPs undergo surface oxidation, which is one of the most important surface phenomena both from the fundamental and technological points of view.5,16,32−37 The formation of an oxide shell also transforms the as-grown SiNPs (inherently hydrophobic due to the H termination) into hydrophilic NPs conferring water solubility and biocompatibility.34 After surface oxidation, the light-emitting features of SiNPs change, and distinct effects (and in some cases opposite) have been described, for which there is not a consensual interpretation. In previous investigations carried out on SiNPs,32,33,37 it has been found that oxidation may result in a redshift of the emission Received: January 3, 2014 Revised: April 8, 2014 Published: April 8, 2014 10375
dx.doi.org/10.1021/jp5000683 | J. Phys. Chem. C 2014, 118, 10375−10383
The Journal of Physical Chemistry C
Article
spectra,33,37 attributed to a change in the Si−Si bond length38 or to the formation of defects localized at the Si/SiO2 interface.38 Studies indicate that this defect-related emission could originate from the introduction of intragap energy levels by interfacial SiO bonds,32,33,37,39−41 SiOSi surface bridges,37 or oxygen incorporation via passivation with OH groups.42 Other reports mention an opposite effect, suggesting that oxidation induces an emission blueshift in SiNPs.5,16,17,32−34,43−45 Two main arguments have been put forward to explain this blueshift: (i) with oxidation the Sicrystalline core size decreases, resulting in a wider band gap when compared with that prior to oxidation5,16,17,34,40,42,43 and (ii) surface defect species appear upon oxidation.32,44,45 We should note that most of the reports of SiNPs mention that surface oxidation leads to a fade of the photoemission. Importantly, previous studies of SiNPs exclude the possibility of the coexistence of an excitonic-related emission and of an emission associated with surface/interface states. Moreover, little has been established about the effect of inter-NP interactions (e.g., inter-NP energy/charge transfer) on the photoemission properties of SiNP ensembles. In this work, using photoluminescence spectroscopy in steady-state and time-resolved modes, we studied the emission of the same SiNPs with surface-oxide termination and with termination formed by Si−H bonds. Experimental evidence for the simultaneous contribution of two recombination mechanisms to the photoemission of naturally surface-oxidized SiNP ensembles will be given. The role of the inter-NP charge transfer on light emission was also studied. For example, we found that an increase in the inter-NPs separation toward noninteracting SiNPs leads to an enhancement of the emission quantum yield, which is rationalized in terms of competition between radiative recombination and nonradiative inter-NP charge transfer.
the measurements were started. The XRD was measured using a Bruker-AXS microdiffractometer with a 2.2 kW sealed Cu Xray source. The FTIR measurements were done with a nitrogen-purged Nicolet series II magna-IR system 750, equipped with a glowbar light source, a KBr beam splitter, and a mercury−cadmium−telluride detector. All spectra were recorded in diffuse reflection mode at room temperature with a resolution of 2 cm−1 and averaged over 1000 scans. A bare gold-covered silicon wafer was used as a reference. For transmission electron microscopy (TEM), dilute dispersions of SiNPs were pipetted on carbon-coated copper grids. The XRD pattern of the as-grown SiNPs is shown in Figure 1a, from which the mean NP diameter was calculated to be d =
Figure 1. (a) XRD pattern and TEM image (inset) of the as-grown SiNPs. (b) FTIR spectra, in the frequency regions of vibrational modes of Si−H and Si−O−Si bonds, recorded for as-grown (solid line), surface-oxidized (dashed line), and HF-etched (dotted line) SiNPs. The spectra of the as-grown and HF-etched SiNPs are dominated by bands at 866, 908, and ∼2100 cm−1 due to modes from surface Si−Si− H3 (deformation), Si2−Si−H2 (scissor and wag), and Si4−x−Si−Hx (stretching) bonds, respectively.35,36 The surface-oxidized SiNPs spectrum shows a strong band in the region 1050−1150 cm−1 due to Si−O−Si bonds (stretching) and a band at 2255 cm−1 due to O3− Si−H bonds (stretching), both from the surface oxide, plus a much reduced band due to Si4−x−Si−Hx hydrides, as it is typically observed for SiNPs with a fully developed native oxide shell.35,36
2. EXPERIMENTAL SECTION The SiNPs used in this study were synthesized in a nonthermal RF (13.56 MHz) flow-through plasma reactor through the dissociation of injected silane, as detailed elsewhere.27 The system consisted of a borosilicate glass tube equipped with two copper ring electrodes used to apply the RF power. A mixture of 5% silane (SiH4) in He was used as precursor gas at a flux of 13 standard cubic centimeters (sccm), and the carrier gas was Ar at a flux of 35 sccm. The pressure in the reactor was kept constant at 1.4 Torr by using an electronically controlled throttle valve. Hydrogen gas was injected into the plasma afterglow at a flow of 100 sccm to reduce surface dangling bond defects.36 H-terminated SiNPs were collected from the plasma as a powder using a mesh. To avoid surface oxidation, the asgrown SiNPs were transferred under nitrogen from the synthesis reactor to a nitrogen-purged glovebox (oxygen and moisture level below 30 ppm), where further sample preparation for X-ray diffraction (XRD) and Fourier transform infrared (FTIR) spectroscopy was carried out. The samples for XRD and FTIR measurements consisted of a glass and a goldcoated silicon wafer, respectively, covered with SiNPs. These were prepared by firmly pressing the respective substrates onto the SiNPs powder. The transfer of the XRD and FTIR samples from the glovebox to the respective spectrometers was carried out using a gas-tight chamber that was sealed inside the glovebox. Onsite the chamber was opened, the samples were immediately introduced in the respective spectrometers, and
4.0 nm using the Scherrer equation, in good agreement with values estimated by TEM; see inset in Figure 1a. The FTIR spectrum of the as-grown SiNPs, shown in Figure 1b, confirms that these are terminated with Si−H bonds in the form of Si4−x−Si−Hx (x = 1, 2, 3) surface hydride groups, with only a minute contamination with Si−O−Si bonds. The SiNPs were afterward removed from the glovebox and kept under ambient conditions for about one month. As known from our previous investigations,35,36 during this time, a natural oxide shell of 0.3 ± 0.1 nm thickness is formed on the SiNPs surface.35 Thus, after air exposure, we have a powder of SiNPs formed by a crystalline Si core with a diameter of 3.4 ± 0.1 nm and a surface shell of native silicon oxide. The FTIR spectrum recorded with these surface-oxidized SiNPs is shown in Figure 1b, confirming the formation of a fully developed native oxide shell.35,36 10376
dx.doi.org/10.1021/jp5000683 | J. Phys. Chem. C 2014, 118, 10375−10383
The Journal of Physical Chemistry C
Article
Photoluminescence measurements were carried out at 12 K and at room temperature, using a modular double grating excitation spectrofluorimeter with a TRIAX 320 emission monochromator (Fluorolog-3, Horiba Scientific) coupled to a R928 Hamamatsu photomultiplier. Emission spectra (acquired using the front face mode) were corrected for the spectral response of the monochromators and detector, using the correction spectrum provided by the manufacturer, and the excitation spectra were corrected for the spectral distribution of the lamp intensity recorded using a photodiode reference detector. The time-resolved emission spectra and emission decay curves were acquired with the same instrumentation using a pulsed Xe−Hg lamp (6 μs pulse at half width and 20− 30 μs tail). All these photoluminescence measurements were carried out in vacuum (10−6 Pa). The absolute emission quantum yields were measured at room temperature using a quantum yield measurement system C9920-02 from Hamamatsu with a 150 W xenon lamp coupled to a monochromator for wavelength discrimination, an integrating sphere as sample chamber, and a multichannel analyzer for signal detection. Three measurements were made for each sample, and the average values obtained are reported with an accuracy within 10% according to the manufacturer. Photoluminescence measurements of the surface-oxidized SiNPs were carried out both with films of randomly distributed SiNPs deposited on glass substrates and with dispersions of the SiNPs in absolute ethanol (Fisher Chemical, analytical grade) placed inside a quartz tube. The films of the surface-oxidized SiNPs on glass were deposited by spin-coating using a dispersion of the SiNPs in ethanol. We have also carried out photoluminescence measurements of the same SiNPs after replacement of the native oxide shell with a hydrogen termination. As in previous work,46 this was done by etching the NPs in diluted HF (10% in water) for 3 min followed by rinsing with ethanol. The resulting H-terminated SiNPs were afterward dispersed in toluene, and the solution was spin-coated onto a glass substrate, forming a film of randomly distributed SiNPs, with which photoluminescence measurements were carried out. These procedures were done inside a nitrogen-purged glovebox (oxygen below 300 ppm). For measurements, the sample was quickly mounted in the spectrometer cryostat, which was immediately after evacuated. Our procedure allowed reducing the exposure time of the H-terminated SiNPs to air down to 5 min before the first measurements were carried out.
Figure 2. Emission spectra of the surface-oxidized SiNPs excited between 275 and 525 nm and measured at (a) 12 K and (b) 300 K and of the H-terminated SiNPs recorded at (c) 12 K and (d) 300 K, as well as of the (e) surface-oxidized SiNPs suspended in ethanol measured at 300 K.
3. EXPERIMENTAL DATA AND DISCUSSION Figure 2a, b compares the emission spectra of the surfaceoxidized SiNPs recorded at different excitation wavelengths at 12 K and at room temperature, respectively. The roomtemperature emission spectra excited between 300 and 410 nm display a broad emission in the red spectral region, peaking at ∼808 nm (1.54 eV) with-width at half-maximum (fwhm) of ∼250 meV (Figure 2b). Increasing the excitation wavelength to 520 nm, the emission spectrum shifts to the red by ∼8 meV (Figure 2b). The room-temperature excitation spectra were monitored in the low- and high-wavelength side of the emission spectrum (700 and 800 nm, respectively) revealing a similar broad band spanning the 260−590 nm spectral region as shown in Figure 3b. The low-temperature emission spectra excited within 300−420 nm resemble those acquired at room temperature, whereas for lower excitation wavelengths (250− 290 nm), the emission spectra become broader, revealing the raising of the emission intensity on the lower wavelength side
Figure 3. Excitation spectra of surface-oxidized SiNPs at (a) 12 K and (b) 300 K and of the (c) H-terminated SiNPs at 12 K, as well as of the (d) surface-oxidized SiNPs suspended in ethanol recorded at 300 K. The monitoring wavelengths were 700 and 800 nm. The dashed curve in (d) represents the absorption spectrum of the crystalline bulk Si measured at 10 K.55
10377
dx.doi.org/10.1021/jp5000683 | J. Phys. Chem. C 2014, 118, 10375−10383
The Journal of Physical Chemistry C
Article
nm.44 Other studies report quantum yield values in the range 1−30% for films of surface-oxidized SiNPs synthesized from laser pyrolysis of silane, depending on the average diameter (2.5−8 nm) and size distribution.15,16 In these studies, values above 18% were obtained only for molecular beam size-selected SiNPs with average diameters in the range 3−4 nm.15 For asgrown SiNPs without size selection and with an average diameter similar to that of the SiNPs studied here (3.4 nm), these investigations report quantum yield values (1−10%)16 analogous to those we obtained for surface-oxidized SiNPs also in film. We have also carried out time-resolved emission spectroscopy of the surface-oxidized SiNPs acquired at distinct starting delay (SD) values (Figure 5). At both room temperature and
of the spectra. At 12 K, the excitation spectra monitored at 700 nm (solid line in Figure 3a) and at 800 nm (line-dot curve in Figure 3a) are similar to those acquired at room temperature. Nevertheless, in the case of the spectrum monitored at 700 nm, an additional excitation path around 270 nm (4.60 eV) is revealed (curve line in Figure 3a). The emission of the surface-oxidized SiNPs was quantified through the measurement of the absolute emission quantum yield (ϕ) as a function of the excitation wavelength (Figure 4).
Figure 4. Absolute emission quantum yield values obtained as a function of the excitation wavelength for surface-oxidized SiNPs in film (solid squares) and suspended in ethanol (open squares).
The maximum ϕ values (9.5−11.5 ± 1.1%) are attained at excitation wavelengths within 270−350 nm. At higher excitation wavelengths (365−420 nm), ϕ decreases from 9.5 ± 1.0% to 7.0 ± 0.7%. Despite several reports on the absolute emission quantum yield of surface-oxidized SiNPs in solution, involving distinct solvents (see below),17,25,26,37,47 to the best of our knowledge, only few reports mention the emission quantum yield for surface-oxidized SiNPs deposited as films (Table 1).15,16,48,49 In particular, a similar quantum yield value of 9% excited at 266 nm was previously measured for surfaceoxidized SiNPs also grown by gas phase,48 and a smaller value of 5% excited at 350 nm was reported for surface-oxidized SiNPs prepared in a high-temperature aerosol apparatus.49 Higher quantum yield (30%) was reported for SiNPs fabricated by microplasma synthesis with average diameters less than 3
Figure 5. Time-resolved emission spectra (excited at 365 nm) of surface-oxidized SiNPs acquired at (a) 300 K and (b) 12 K for SD between 0.02 and 0.30 ms. The integration window was 20 ms.
12 K (Figure 5a, b, respectively), the emission spectra deviate toward the red and become narrower as the SD increases from 0.02 to 0.30 ms, indicating that the time scale behind the emission at lower wavelengths (700 nm) is faster than that at higher wavelengths (800 nm). The emission decay curves of the lower and higher wavelength sides of the emission band were selectively monitored at 690 and 815 nm, respectively. For both monitoring wavelengths, the emission decay curves display a non single-exponential behavior (Figure 6). The non singleexponential behavior is observed for both 12 and 300 K. We obtained experimental lifetimes (τ) for the emission at 690 nm and at 815 nm (denoted τ1 and τ2, respectively), which correspond to the time at which the experimental emission intensity is reduced to 1/e, yielding to the values gathered in Table 2. For τ1 and τ2, an increase of temperature from 12 to 300 K induces a decrease in τ value. Attending to the fact that the experimental transition probability may be expressed as τ −1 = τnr−1 + τr−1, where τnr and τr represent the nonradiative and radiative lifetimes, respectively, and that at 12 K we have τ −1 ∼ τr −1, the decrease in τ as the temperature increases suggests the presence of competing thermally activated nonradiative mechanisms.50 To gain further insight about the recombination paths, the 12 K time-resolved emission spectra were analyzed according to the following methodology. Figure 7a shows the emission spectrum of the surface-oxidized SiNPs acquired at SD = 0.55 ms. This spectrum is well described by a single-Gaussian band with energy peak position equal to 1.53 ± 0.02 eV (810 nm)
Table 1. Emission Quantum Yield Values of SurfaceOxidized SiNPs in Films and Suspended in Different Solventsa sample form film
suspended in
ethanol
water unspecified toluene a
av diameter (nm)
quantum yield (%)
3.4 3 1−2
