+ All Categories
Home > Documents > Morphology, structure, chemical composition, and light emitting properties of very thin anodic...

Morphology, structure, chemical composition, and light emitting properties of very thin anodic...

Date post: 27-Apr-2023
Category:
Upload: auth
View: 0 times
Download: 0 times
Share this document with a friend
8
Morphology, structure, chemical composition, and light emitting properties of very thin anodic silicon films fabricated using short single pulses of current S. Gardelis, 1,a A. G. Nassiopoulou, 1 F. Petraki, 2,3 S. Kennou, 2,3 I. Tsiaoussis, 4 and N. Frangis 4 1 IMEL/NCSR Demokritos, P.O. Box 60228, Aghia Paraskevi, 15310 Athens, Greece 2 Department of Chemical Engineering, University of Patras, 26500 Patras, Greece 3 FORTH/ICE-HT, 26504 Patras, Greece 4 Solid State Physics Section, Department of Physics, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece Received 7 February 2008; accepted 24 March 2008; published online 29 May 2008 In this work, the morphology, structure, surface chemical composition, and optical properties of very thin 10–70 nm anodic silicon films grown on a silicon substrate by electrochemical dissolution of bulk crystalline silicon in the transition regime between the porous formation and electropolishing were investigated in detail. Anodization was performed by using short single pulses of anodization current in low and high hydrofluoric acid HF concentration electrolytes. A systematic comparison was made between films grown at low and high HF concentration electrolytes. The morphology and structure of the films were investigated by combining atomic force microscopy and transmission electron microscopy TEM, while x-ray and ultraviolet photoelectron spectroscopies were used to investigate the chemical composition of the films. Photoluminescence was used to investigate the optical properties. It was found that films that formed at low HF concentrations were much thinner than films that formed at high HF concentrations due to surface dissolution of the films during anodization. High resolution TEM images revealed an amorphouslike structure porous in all of the films in which discrete Si nanocrystals NCs were identified. NC size was, on the average, larger in films fabricated in low HF concentration electrolytes and these films were not luminescent. On the other hand, films fabricated in high HF concentration electrolytes were thicker and contained smaller NCs. A silicon oxide layer covered the internal surface of all films, this oxide being much thinner in films grown at high HF concentrations. This last effect was attributed to self-limiting oxidation of the very small NCs constituting these films. © 2008 American Institute of Physics. DOI: 10.1063/1.2936317 I. INTRODUCTION Silicon nanocrystals NCs embedded in very thin SiO 2 layers are interesting for different applications in nanoelec- tronics, photonics, and sensor devices. They are fabricated by different techniques, including low pressure chemical va- por deposition of Si, followed by high temperature thermal oxidation, low energy ion beam synthesis, or deposition of SiO 2 / SiO x / SiO 2 layers, followed by annealing. 15 Porous silicon fabricated by electrochemistry is another well known form of nanostructured Si composed of NCs and voids, the size of which depends on the concentration of the electrolyte, the anodization current density, and the type and resistivity of the starting bulk crystalline silicon used for the anodic reaction. 6 In oxidized porous silicon, the Si nanostructures of the porous skeleton are covered by SiO 2 . In most of the existing literature, thick porous silicon layers are investi- gated and used in different applications. However, for some applications, as in Si NC nonvolatile memories or in NC electroluminescent devices, there is a need for fabrication of ultrathin, almost two-dimensional 2D NC layers embedded in SiO 2 . In this work, we investigate the growth of thin porous silicon layers by anodization using short single pulses of current, with the overall objective of fabricating, after oxida- tion, very thin SiO 2 layers with discrete Si NCs embedded therein. We compare two different cases of silicon anodiza- tion in the transition regime between pore formation and electropolishing 6 by using short single pulses of current, one using low hydrofluoric acid HF concentration electrolyte and the other using high HF concentration electrolyte. The resulting material is very different in these two different cases. It has been reported that by using anodization in low HF concentration electrolytes and a current density in the transition regime between pore formation and electropolish- ing, using short single pulses of anodization current, the re- sulting films are composed of laterally separated Si NCs. 7,8 This conclusion was based solely on atomic force micros- copy AFM investigation of the films. In a previous inves- tigation by the authors of the present article, 9,10 by using similar anodization conditions but higher HF concentration electrolytes, light-emitting thin porous films with embedded silicon NCs were fabricated. Here, we compare the above two cases of low and high HF concentration electrolytes in the transition regime between pore formation and electropol- ishing by using AFM, transmission electron microscopy TEM, x-ray photoelectron spectroscopy XPS, and ultra- violet photoelectron spectroscopy UPS, which give addi- a Electronic mail: [email protected]. JOURNAL OF APPLIED PHYSICS 103, 103536 2008 0021-8979/2008/10310/103536/8/$23.00 © 2008 American Institute of Physics 103, 103536-1 Downloaded 11 Jun 2008 to 150.140.191.81. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
Transcript

Morphology, structure, chemical composition, and light emitting propertiesof very thin anodic silicon films fabricated using short single pulsesof current

S. Gardelis,1,a� A. G. Nassiopoulou,1 F. Petraki,2,3 S. Kennou,2,3 I. Tsiaoussis,4 andN. Frangis4

1IMEL/NCSR Demokritos, P.O. Box 60228, Aghia Paraskevi, 15310 Athens, Greece2Department of Chemical Engineering, University of Patras, 26500 Patras, Greece3FORTH/ICE-HT, 26504 Patras, Greece4Solid State Physics Section, Department of Physics, Aristotle University of Thessaloniki,54124 Thessaloniki, Greece

�Received 7 February 2008; accepted 24 March 2008; published online 29 May 2008�

In this work, the morphology, structure, surface chemical composition, and optical properties ofvery thin �10–70 nm� anodic silicon films grown on a silicon substrate by electrochemicaldissolution of bulk crystalline silicon in the transition regime between the porous formation andelectropolishing were investigated in detail. Anodization was performed by using short single pulsesof anodization current in low and high hydrofluoric acid �HF� concentration electrolytes. Asystematic comparison was made between films grown at low and high HF concentrationelectrolytes. The morphology and structure of the films were investigated by combining atomicforce microscopy and transmission electron microscopy �TEM�, while x-ray and ultravioletphotoelectron spectroscopies were used to investigate the chemical composition of the films.Photoluminescence was used to investigate the optical properties. It was found that films thatformed at low HF concentrations were much thinner than films that formed at high HFconcentrations due to surface dissolution of the films during anodization. High resolution TEMimages revealed an amorphouslike structure �porous� in all of the films in which discrete Sinanocrystals �NCs� were identified. NC size was, on the average, larger in films fabricated in lowHF concentration electrolytes and these films were not luminescent. On the other hand, filmsfabricated in high HF concentration electrolytes were thicker and contained smaller NCs. A siliconoxide layer covered the internal surface of all films, this oxide being much thinner in films grownat high HF concentrations. This last effect was attributed to self-limiting oxidation of the very smallNCs constituting these films. © 2008 American Institute of Physics. �DOI: 10.1063/1.2936317�

I. INTRODUCTION

Silicon nanocrystals �NCs� embedded in very thin SiO2

layers are interesting for different applications in nanoelec-tronics, photonics, and sensor devices. They are fabricatedby different techniques, including low pressure chemical va-por deposition of Si, followed by high temperature thermaloxidation, low energy ion beam synthesis, or deposition ofSiO2 /SiOx /SiO2 layers, followed by annealing.1–5 Poroussilicon fabricated by electrochemistry is another well knownform of nanostructured Si composed of NCs and voids, thesize of which depends on the concentration of the electrolyte,the anodization current density, and the type and resistivityof the starting bulk crystalline silicon used for the anodicreaction.6 In oxidized porous silicon, the Si nanostructures ofthe porous skeleton are covered by SiO2. In most of theexisting literature, thick porous silicon layers are investi-gated and used in different applications. However, for someapplications, as in Si NC nonvolatile memories or in NCelectroluminescent devices, there is a need for fabrication ofultrathin, almost two-dimensional �2D� NC layers embeddedin SiO2.

In this work, we investigate the growth of thin porous

silicon layers by anodization using short single pulses ofcurrent, with the overall objective of fabricating, after oxida-tion, very thin SiO2 layers with discrete Si NCs embeddedtherein. We compare two different cases of silicon anodiza-tion in the transition regime between pore formation andelectropolishing6 by using short single pulses of current, oneusing low hydrofluoric acid �HF� concentration electrolyteand the other using high HF concentration electrolyte. Theresulting material is very different in these two differentcases. It has been reported that by using anodization in lowHF concentration electrolytes and a current density in thetransition regime between pore formation and electropolish-ing, using short single pulses of anodization current, the re-sulting films are composed of laterally separated Si NCs.7,8

This conclusion was based solely on atomic force micros-copy �AFM� investigation of the films. In a previous inves-tigation by the authors of the present article,9,10 by usingsimilar anodization conditions but higher HF concentrationelectrolytes, light-emitting thin porous films with embeddedsilicon NCs were fabricated. Here, we compare the abovetwo cases of low and high HF concentration electrolytes inthe transition regime between pore formation and electropol-ishing by using AFM, transmission electron microscopy�TEM�, x-ray photoelectron spectroscopy �XPS�, and ultra-violet photoelectron spectroscopy �UPS�, which give addi-a�Electronic mail: [email protected].

JOURNAL OF APPLIED PHYSICS 103, 103536 �2008�

0021-8979/2008/103�10�/103536/8/$23.00 © 2008 American Institute of Physics103, 103536-1

Downloaded 11 Jun 2008 to 150.140.191.81. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

tional information on the chemical composition of the sur-face and of the inner structure of the porous layers. By usingthe combined information obtained from all of the tech-niques mentioned above, we were able to get an importantinsight into the structure, morphology, and chemical compo-sition of the films. A thorough investigation of the opticalproperties of the films was also carried out by photolumines-cence �PL� and time-resolved PL measurements.

II. EXPERIMENTAL DETAILS

Three films, A–C, which were grown by using differentelectrolyte concentrations and anodization current densities,were investigated. The silicon substrates were �100�-orientedboron doped �resistivity of 1–2 � cm� silicon wafers. An-odization was carried out in an electrochemical cell in day-light at room temperature. The electrolyte solutions wereprepared by mixing hydrofluoric acid �HF� �50 wt %� withabsolute ethanol in various volume ratios. All films weregrown by using a single pulse of anodic current with 400 msduration. Specifically, for film A, the anodic current density�single pulse height� was 130 mA /cm2 and the electrolytesolution consisted of one part hydrofluoric acid �HF� andthree parts ethanol. Film B was formed by using the samecurrent density but a much higher HF concentration in theelectrolytic solution, composed of three parts of HF and onepart of ethanol. Finally, film C was formed by using an an-odization current density three times higher than that in filmB �390 mA /cm2� and the same electrolytic solution as thatin film B. In all of the cases, the anodization current densityused was situated in the transition regime of the anodizationcurve, between porous silicon formation andelectropolishing.7,9 The anodization conditions for thegrowth of films A–C are summarized in Table I.

AFM operating in the tapping mode was used to inves-tigate the morphology of the surface of the films by using aDigital Instruments Nanoscope III multimode scanningprobe microscope. High resolution TEM �HRTEM� was car-ried out in a cross-sectional configuration by using a Jeol2011 microscope with a point resolution of 0.194 nm. Detailsof the preparation of cross-sectional TEM samples are givenelsewhere.9 The photoelectron spectroscopy �XPS, UPS�measurements were carried out in a commercial UHV sys-tem, at a base pressure of 5�10−10 mbar. For the XPS mea-surements, the Mg Ka line at 1253.6 eV and a constant ana-lyzer pass energy of 36 eV were used, while for UPS, theHe I �21.22 eV� radiation was used. The spectrometer wascalibrated by the Au 4f7/2 core level �84.00�0.05 eV forclean Au foil�. The XPS resolution measured by the fullwidth at half maximum �FWHM� of the Au 4f7/2 was 1.1 eVfor a constant energy of 36 eV. The analyzer resolution for

the UPS measurements was 0.16 eV. XPS and UPS measure-ments were carried out on the as-grown films and on thefilms after sputtering with Ar ions at 2 keV applied for 30min in order to remove the adventitious carbon and to moni-tor the chemical composition of the inner structure of thefilms. PL measurements were performed at temperatures be-tween 70 K and room temperature by using the 457.9 nmline of an Ar+-ion laser for excitation. Time-resolved PLmeasurements were performed by the gated photon countingmethod, which is based on pumping to steady state, switch-ing off the pump beam by using a mechanical light chopper,and finally detecting the PL intensity as a function of time ata fixed wavelength.

III. RESULTS AND DISCUSSION

A. Morphological and structural investigation by AFMand TEM

The surface topography of all films was investigated byAFM. Figures 1�a� and 1�b� show the AFM images of thesurface of films A and B, respectively. Specifically, film Ashows a grainlike surface morphology with an averageroughness height of about 10 nm and a lateral size of about30–40 nm. Film B shows a much smoother surface with anaverage roughness height of 1.5–2 nm and with a lateral sizeof about 10–20 nm. Film C demonstrated a featureless to-pography implying an even smoother surface. The differencein surface morphology of films A and B is consistent with thefact that anodization in electrolytes containing low HF con-centration �case of film A� generally results in rougher sur-faces as a consequence of chemical dissolution of the surfaceof the film during anodization.11,12

From the cross-sectional TEM images, it was found thatfilm A had a thickness of about 10 nm, which is equal to themeasured film roughness, while films B and C were 20 and70 nm thick, respectively �Table I�. Since the anodizationtime and current density were the same in films A and B, theobserved difference in film thickness in these two cases is inagreement with the above mentioned result that at low HFconcentrations, part of the grown porous film is dissolvedduring anodization, where this dissolution proceeds from thefilm surface.11,12 Figure 2�a� shows an example of a brightfield, low magnification cross-sectional TEM image obtainedfrom film A. In this image, it can be observed that the inter-face of the film with the substrate is relatively rough, whichshows nanocrystalline protrusions of 2–3 nm extending fromthe Si substrate into the film. TEM images of sample B pub-lished elsewhere9 showed similar protrusions at the interfacebut they were denser and more pronounced than those infilm A.

HRTEM images have revealed, in all cases �films A–C�,an amorphouslike structure with embedded Si NCs that wereonly slightly disoriented relative to the substrate. In film A,the NC sizes were, on the average, larger than those in filmsB and C; this latter film showed the smallest NC sizes. Anexample of a HRTEM image from film A is given in Fig.2�b�.

The rough surface of films fabricated at low HF concen-trations was also observed by Nychyporuk et al.,8 who con-

TABLE I. Growth conditions and thickness of the investigated films.

Film Electrolytic solution�HF:ethanol�

Current density�mA /cm2�

Anodization time�ms�

Thickness�nm�

A 1:3 130 400 10B 3:1 130 400 20C 3:1 390 400 70

103536-2 Gardelis et al. J. Appl. Phys. 103, 103536 �2008�

Downloaded 11 Jun 2008 to 150.140.191.81. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

cluded, based on AFM images, that such films consisted oflaterally separated silicon nanoparticles at the surface of thesilicon wafer. We have to mention here that AFM alone can-not provide a full picture of the film structure. A more thor-ough examination of film A in this work by HRTEM showedthat this film was also porous, with larger NCs than in thosein films B and C.

B. Chemical analysis by XPS and UPS

The XPS wide scan spectra of the as-grown films beforesputtering showed the presence of Si, oxygen, and carbon onthe surface. An example of a typical wide scan XPS spec-trum is shown in Fig. 3. We note that all of the films studiedhere have been subjected to identical aging conditions before

FIG. 1. AFM image obtained from �a� film A and �b� film B. The imagesshow the surface topography of the films. FIG. 2. �a� Cross-sectional bright field TEM image of film A. �b� HRTEM

image of film A. �c� HRTEM image of film B. Si NCs are marked in circles.

103536-3 Gardelis et al. J. Appl. Phys. 103, 103536 �2008�

Downloaded 11 Jun 2008 to 150.140.191.81. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

their insertion into the analysis chamber. This allows us tocompare the spectra obtained from the films. Oxygen in thefilms was due to oxidation during aging in air. Carbon wasadsorbed on the surface during the exposure of the films inthe atmosphere.

By comparing Si 2p core level XPS spectra obtainedfrom the as-grown films before and after sputtering in UHVconditions with spectra obtained from the Si substrate, wecan get significant insight into the structure of the films ifthis information is combined with the AFM and HRTEMimages. Figure 4 shows the normalized Si 2p core level XPSspectra of the surface of films A–C and that of the Si sub-strate. Two peaks were present in the Si 2p XPS spectra ob-tained from films A and B and from the Si substrate. In filmC, only one peak could be detected. The Si 2p spectra weremodified significantly when we removed the surface layer ofthe films by sputtering and investigated their internal surface,as discussed below. In order to determine the exact positionof the peaks in Fig. 4, a fitting procedure has been performedby using the XPS peak program. According to this proce-dure, the Si 2p peak is fitted by mixed Lorentzian–Gaussianfunctions after the subtraction of a Shirley-type background.

Figure 5 shows an example of the fitting procedure for filmA. The peak corresponding to a binding energy of 99.6 eV isattributed to Si0 and the peak located at 104.0 eV is attrib-uted to SiO2. The difference in Si 2p binding energy betweenSi0 and SiO2 is constant at 4.2 eV. From a previous investi-gation, we know that in the case of a uniform SiO2 film onSi, this difference depends on the thickness of the siliconoxide and it increases gradually from 3 to 4 eV when theSiO2 thickness increases from 0.5 to 3.6 nm.13 In the filmsstudied here, this difference was measured to be 4.4 eV forfilm A and 3.6 eV for film B, whereas for the reference Sisubstrate �covered with a native oxide�, it was 4.1 eV. Theabove result is the first indication that the surface oxide isthicker in film A than in the flat Si substrate and also thickerthan that in film B. The absence of the oxide peak in film Csuggests at first sight that its surface is not oxidized. How-ever, the presence of the O 1s peak in the wide scan XPSspectrum suggests that some surface oxide layer is present,although it is much thinner than in films A and B since theratio of the O 1s peak intensity to the Si0 peak intensity infilm C �0.35� is much smaller than in those in films A �2.0�and B �0.95�. In addition, the fact that the Si 2p peak fromSiOx was not detected suggests that the oxide thickness at theoutermost surface of the film was thinner than that deeper inthe film since the probing depth is larger in O 1s comparedto that in Si 2p.

Quantitatively, the average oxide thickness d can be cal-culated in all of the cases from the intensity ratio of the XPSSi 2p peaks attributed to SiO2 and Si0, assuming that theattenuation of the substrate �Si0� and the oxide overlayer�SiO� signals for a uniform oxide film follow the exponentialrelations

ISi0

ISi00 = exp�−

d

��Si0�SiO2

� , �1�

ISi0SiOx

ISiO2

0 = 1 − exp�−d

��SiOx�SiO2� , �2�

where ISio0 and ISiO2

0 are the peak intensities measured inde-pendently in our system for the clean silicon substrate and

FIG. 3. Typical wide scan x-ray photoelectron spectrum obtained from theas-grown films �aged in atmosphere�.

FIG. 4. Si 2p core level XPS peaks of the as-grown films, normalized to theintensity of the Si0 peak for each film. The dotted lines are fits to theexperimental data.

FIG. 5. Fitting to the experimental Si 2p core level XPS peaks of the as-grown film A.

103536-4 Gardelis et al. J. Appl. Phys. 103, 103536 �2008�

Downloaded 11 Jun 2008 to 150.140.191.81. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

the bulk oxide surface, which gives a ratio of ISi0 / ISiO2

0

=0.164 and ��Si�o ���SiOx�

SiO2 =3.26 nm. Based on the above for-mulas, we have calculated that the average surface oxidethicknesses on the Si substrate on films A and B were 0.8,1.8, and 0.4 nm, respectively. From the above qualitativeestimation of film thickness and from the difference in Si 2pbinding energy between Si0 and SiO2, the SiO2 thickness waslarger in film A than that in the Si substrate. This differenceis due to the fact that all of these calculations correspond touniform, planar layers, while both films A and B are notplanar. As a consequence, by comparing film A to the Sisubstrate, we see that the calculated surface oxide thicknessfrom the ratio of intensities of the SiO2 peak relative to theSi0 peak gives a thicker oxide on the substrate compared tofilm A. We can understand this difference if we take intoaccount the larger internal surface of film A compared to thatof the flat Si substrate, which results in an overestimation ofthe surface oxide thickness in film A. We assume that theincident beam probed similar volume sizes in all of thesamples.

If we compare film A to film B, the resulting oxide thick-ness from the Si 2p peak position difference in Si0 and SiO2

is consistent with what is obtained from the intensity ratio ofthe corresponding peaks. In the case of sample B, the surfaceoxide is estimated to be much thinner than the native oxideof the reference Si bulk sample. If we consider that the oxidethickness on film B is overestimated, due to its large internalsurface, this thickness is even smaller than the calculatedone. This is indeed confirmed by the considerable reductionin the difference between the binding energies of Si0 andSiO2 peaks observed in film B.

The thinner oxide that covers the internal nanostructuredsurface of film B compared to film A is attributed to theself-limiting oxidation process of the smaller nanostructuresthat compose this film and also film C compared to film A.This is a well known effect reported for oxidation of silicondots at high temperatures14 and more recently for silicon dotsaged in atmosphere.15 It has been observed that due to per-pendicular stresses developed at the silicon-silicon oxide in-terface during oxidation of silicon dots, the oxidation processslows down and after a critical stress is reached, the oxida-tion essentially stops. We argue that this effect is importantin films B and C. Indeed, in these films, the observed Si NCswere smaller in size and denser than those in film A. Inparticular, in film C, the surface structure was the finest of allof the films and its internal structure looked more amor-phous.

UPS measurements, which give information about va-lence energy states, support our description of the surfacestructure and morphology of the films. Specifically, Fig. 6shows the normalized UPS spectra of the as-grown films andthat of the Si substrate covered with native oxide. In all ofthe films, including film C in this case, we observe a broadpeak at around 7 eV, which corresponds to the O 2p state�observed in H2O-adsorbed Si surfaces, where the oxidationproceeds by means of dissociative adsorption of H2O �Ref.16��, while only in films B and C �more evidently in film C�can one observe the onset of a second peak at around 4 eV,which appears as an extended tail toward the Fermi level �at

0 eV on the graph�. The existence of this second peak infilms B and C, which corresponds to Si0 states in the valenceband, shows that the surface oxide in these two samples ismuch thinner than the native oxide of the reference Sisample and that of sample A. This confirms once more thatthe surface oxide is much thinner in films B and C than thatin film A. It is worth noting that the oxidation of the outer-most surface of film C is more evident with UPS than withXPS since UPS has a smaller probing depth; therefore, it ismore sensitive to the oxide of the outermost surface.

Interesting remarks can be made about the inner struc-ture of the films by investigating the Si 2p core level spectraafter sputtering. Figure 7 shows the spectra of the films aftersputtering. It is evident that after sputtering, all of the filmsdemonstrate the presence of silicon suboxide at their newsurface, which was revealed after sputtering. In particular, infilms B and C, the relative integrated intensity which corre-sponds to the silicon suboxide region of the spectra to thatcorresponding to Si0 increases considerably after sputtering.This, together with a significant increase in the O 1s peak

FIG. 6. Ultraviolet photoelectron spectra of the as-grown films, normalizedto the intensity of the main peak for each film.

FIG. 7. Si 2p core level XPS peaks of the films after sputtering, normalizedto the intensity of the Si0 peak for each film.

103536-5 Gardelis et al. J. Appl. Phys. 103, 103536 �2008�

Downloaded 11 Jun 2008 to 150.140.191.81. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

intensity/Si0 peak intensity ratio �from 0.97 to 1.48 in film Band from 0.35 to 1.32 in film C� observed after sputtering, isevidence that films B and C have porous texture �i.e., therewas an inner surface area, partially oxidized, under the filmsurface which was revealed after sputtering�. This is sup-ported by the HRTEM images of the films which show SiNCs embedded in an amorphous matrix with the amorphousmatrix being most probably the oxidized surface area at thepore walls. On the contrary, in film A, sputtering consider-ably reduced the O 1s /Si0 ratio from 2.0 to 1.30. This is alsoevident in the XPS spectrum, where the relative intensity inthe oxide region of the spectrum was reduced after sputter-ing. Also after sputtering, the surface stoichiometric oxidewas reduced to suboxides. This is similar to the reduction inthe planar oxide in the case of the Si substrate after sputter-ing �in this case, the O 1s /Si0 ratio was reduced from 1.48 to0.81�. However, the HRTEM image of film A �Fig. 2�b��,which shows silicon NCs embedded in an amorphous matrix,suggests that in film A, there is also a surface area surround-ing the silicon NCs which is exposed to oxidation. The factthat the oxide was reduced after sputtering, as in the case ofthe silicon substrate, although the structure still containedvoids/pores, agrees with the AFM image of the film well,which showed larger grains with larger separation �i.e., moreopen structure�, arranged in almost 2D arrays of NCs. Thisimplies that the ions, during sputtering, could reach the re-gion between the NCs more easily, which results in a moreuniform removal of surface oxide than in the case of films Band C, which have a fine porous structure buried under thesurface.

Taking into account both the XPS and UPS measure-ments for the three films, one can construct the followingpicture for the structure of the films. Film B and more evi-dently film C �both grown in high HF concentration electro-lyte� are composed of very small Si NCs at their outermostsurface, which are smaller than those in the internal surface.These films showed little oxidation at their outermost surfaceand much larger oxidation at their internal surface, whichwas revealed after sputtering. The smaller NCs at the outer-most surface of the films were not easily oxidized due to theself-limiting oxidation effect and were mostly responsible forthe very bright luminescence obtained particularly from filmC, as discussed below. The case of film A is different. TheNCs at its outermost surface are larger than those in films Band C, as the more extensive oxidation of the outermost sur-face suggests. In addition, the high degree of oxidation of theinternal surface of this film, revealed after sputtering, sug-gests that the NCs are homogeneous in size both in the out-ermost and in the internal film surface. Figure 8 shows sche-matically these differences in the structure of the two typesof films.

C. Optical properties

In order to fully characterize the films regarding the sizeof the silicon NCs, we investigated their optical properties byPL and time-resolved PL measurements. Film A did not emitany detectable light, which is consistent with the observedlarger NC size in this sample; films B and C were light

emitting, which have similar spectral characteristics, but filmC emitted more efficiently. The details of the room tempera-ture PL characteristics of these two films were presentedelsewhere.9,10 Here, we studied in detail the PL and time-resolved PL from sample C performed at temperatures be-tween 70 K and room temperature.

Figure 9 shows PL obtained at different temperatures.The PL peak position shifted to shorter wavelengths by de-creasing the temperature �from 652 nm �1.90 eV� at roomtemperature to 632 nm �1.96 eV� at 70 K, which correspondsto a shift of 60 meV�. This is consistent with electron-hole

FIG. 8. �a� Schematic of films B and C. The Si NCs at the outermost surfaceare smaller than those in the internal surface revealed after sputtering. �b�Schematic of film A. The Si NCs at the outermost surface are generallylarger than those at the outermost surface of films B and C and comparableto those in the internal surface of the film.

FIG. 9. PL spectra obtained from film C at different temperatures.

103536-6 Gardelis et al. J. Appl. Phys. 103, 103536 �2008�

Downloaded 11 Jun 2008 to 150.140.191.81. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

recombination in the silicon NCs that compose the film, asthis shift can be mainly attributed to the decrease in the en-ergy band gap of the NCs with increasing temperature. Notethat the band gap variation with temperature for bulk siliconis about 40 meV over the temperature range from 70 to 300K.17 This shift was accompanied by a considerable decreasein PL intensity with temperature.

Figure 10 shows PL temporal decay graphs measured ata wavelength that corresponds to the PL peak position atthree different temperatures. The curves can be fitted to astretched exponential function of the following form:

I = I0 exp�− � t

��� , �3�

where � is the lifetime and � is a dispersion factor which cantake values between 0 and 1.

This decay law has been reported for various nanocrys-talline silicon systems18–24 and is expected in materials withsignificant disorder, where there are various pathways of de-excitation with different decay times. In the case of porousSi, the dispersion in the size and shape of the NCs dictatesthe above law, as the silicon NCs keep their indirect bandgap25 behavior, which results in phonon-assisted recombina-tion processes with a size- and shape-dependent decay time.

We have measured the PL decay time for different tem-peratures and at three different wavelengths on the PL spec-tra, specifically at 630 �around the peak of the spectra�, 590,and 710 nm. Because � deviates from 1, an average decaytime �̄ has been calculated by using the followingexpression:23

�̄ = �1

��� 1

�� , �4�

where � is the lifetime extracted from Eq. �3�. �̄ decreasedmonotonically with increasing temperature from 73 �s at 70K to 13 �s at room temperature. This was accompanied by amonotonic decrease in PL intensity, which indicates that non-radiative recombination dominates over the radiative one athigher temperatures.26 �̄ also decreased with decreasing de-tection wavelength. At 70 K, �̄ was reduced from 95 �s at710 nm to 78 �s at 590 nm. At room temperature, �̄ was

reduced from 25 �s at 710 nm to 10 �s at 590 nm. Thisreduction in the detection wavelength is a well documentedeffect on silicon nanocrystalline systems19,23 and is attributedto the decrease in the size of the silicon NC.25 The dispersionfactor � was independent of the detection wavelength andtemperature and ranged between 0.7 and 0.8, which suggeststhat this parameter was rather dependent on the macroscopiccharacter of the medium, particularly on its disorder and noton the microscopic properties related to the size of the indi-vidual NCs.27

The energy range of the PL spectra of film C centered at630 nm generally corresponds to silicon NC sizes of lessthan 3 nm.15,26 This agrees with the HRTEM images of thefilms. Specifically, films A and B show the presence of NCswith a mean size larger than 3 nm and with a broad sizedistribution, whereas film C showed NCs, on the average,smaller than 3 nm with a narrower size distribution.

IV. CONCLUSION

In conclusion, we have investigated the structure, sur-face chemical composition, and optical properties of thin po-rous anodic silicon films, grown by the electrochemical dis-solution of bulk crystalline silicon by using short singlepulses of anodization currents, in the transition regime be-tween pore formation and electropolishing in low and highHF concentration electrolytes. AFM images showed that thesurface roughness was much larger in films grown in a lowHF concentration electrolyte. This roughness is attributed tothe dissolution of the surface of the films during anodization,as verified by the reduced film thickness, measured by TEM,compared to that of a film prepared under the same condi-tions in high HF concentration electrolytic solutions. Themean NC size was larger in films fabricated at low HF con-centration electrolytes and, consequently, these films werenot luminescent. The much thinner oxide covering the outer-most surface of films grown in high HF concentration elec-trolytes is attributed to the fact that these films were com-posed of smaller NCs, in which self-limiting oxidation wasmore pronounced. These very small silicon NCs at the out-ermost surface of the films are expected to emit light moreefficiently than the larger NCs in the bulk of the film. De-tailed temperature dependent PL and temperature dependenttime-resolved PL measurements suggested that light emis-sion was due to exciton recombination in the silicon NCs ofsizes less than 3 nm of the porous films.

1A. G. Nassiopoulou, in Encyclopedia of Nanoscience and Nanotechnol-ogy, edited by H. S. Nalwa �American Scientific, Valencia, CA, 2004�,Vol. 9, pp. 793–813.

2A. Salonidou, A. G. Nassiopoulou, A. Travlos, V. Ioannou-Sougleridis,and E. Tsoi, Nanotechnology 15, 1233 �2004�.

3A. G. Nassiopoulou and A. Salonidou, J. Nanosci. Nanotechnol. 7, 368�2007�.

4T. Müller, K. H. Heinig, and W. Möller, Appl. Phys. Lett. 81, 3049 �2002�.5M. Zacharias, J. Heitmann, R. Scholz, U. Kahler, M. Schmidt, and J.Bläsing, Appl. Phys. Lett. 80, 661 �2002�.

6R. L. Smith and S. D. Collins, J. Appl. Phys. 71, R1 �1992�.7T. Nychyporuk, V. Lysenko, B. Gautier, and D. Barbier, Appl. Phys. Lett.86, 213107 �2005�.

8T. Nychyporuk, V. Lysenko, B. Gautier, and D. Barbier, J. Appl. Phys.100, 104307 �2006�.

9S. Gardelis, I. Tsiaoussis, N. Frangis, and A. G. Nassiopoulou, Nanotech-

FIG. 10. PL decay curves obtained from film C at different temperatures.

103536-7 Gardelis et al. J. Appl. Phys. 103, 103536 �2008�

Downloaded 11 Jun 2008 to 150.140.191.81. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

nology 18, 115705 �2007�.10S. Gardelis and A. G. Nassiopoulou, Phys. Status Solidi C 4, 2165 �2007�.11A. Halimaoui, in Properties of Porous Silicon, Datareviews Series No. 18,

edited by L. T. Canham �Inspec, London, 1997�, p. 12.12A. G. Nassiopoulos, S. Grigoropoulos, L. T. Canham, A. Halimaoui, I.

Berbezier, E. Gogolides, and D. Papadimitriou, Thin Solid Films 255, 329�1995�.

13V. Papaefthimiou, A. Siokou, and S. Kennou, Surf. Sci. 569, 207 �2004�.14H. Heidemeyer, C. Single, F. Zhou, F. E. Prins, D. P. Kern, and E. Plies, J.

Appl. Phys. 87, 4580 �2000�.15G. Ledoux, O. Guillois, D. Porterat, C. Reynaud, F. Huisken, B. Kohn, and

V. Paillard, Phys. Rev. B 62, 15942 �2000�.16D. Schmeisser, F. J. Himpsel, and G. Hollinger, Phys. Rev. B 27, 7813

�1983�.17M. Fujii, S. Hayashi, and K. Yamamoto, J. Appl. Phys. 83, 7953 �1998�.18X. Chen, B. Henderson, and K. P. O’ Donnel, Appl. Phys. Lett. 60, 2672

�1992�.

19L. Pavesi and M. Ceschini, Phys. Rev. B 48, 17625 �1993�.20I. Mihalcescu, J. C. Vial, and R. Romestain, J. Appl. Phys. 80, 2404

�1996�.21J. Linnros, N. Lalic, A. Galeckas, and V. Grivickas, J. Appl. Phys. 86,

6128 �1999�.22F. Priolo, G. Franzo, D. Pacifici, V. Vinciguerra, F. Iacona, and A. Irrera, J.

Appl. Phys. 89, 264 �2001�.23O. Guillois, N. Herlin-Boime, C. Reynaud, G. Ledoux, and F. Huisken, J.

Appl. Phys. 95, 3677 �2004�.24I. Mihalcescu, J. C. Vial, and R. Romestain, Phys. Rev. Lett. 80, 3392

�1998�.25C. Delerue, G. Allan, C. Reynaud, O. Guillois, G. Ledoux, and F. Huisken,

Phys. Rev. B 73, 235318 �2006�.26O. Bisi, S. Ossicini, and L. Pavesi, Surf. Sci. Rep. 38, 1 �2000�.27M. Dovrat, Y. Goshen, J. Jedrzejewski, I. Balberg, and A. Sa’ar, Phys.

Rev. B 69, 155311 �2004�.

103536-8 Gardelis et al. J. Appl. Phys. 103, 103536 �2008�

Downloaded 11 Jun 2008 to 150.140.191.81. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp


Recommended