Optical characterization of polysilazane based silica thin films on silicon substrates

Applied Surface Science 265 (2013) 470–474

Contents lists available at SciVerse ScienceDirect

Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc

Optical characterization of polysilazane based silica thin films on silicon substrates Pier Carlo Ricci a,∗ , Gianluca Gulleri b , Francesco Fumagalli b , Carlo Maria Carbonaro a , Riccardo Corpino a a b

Dipartimento di Fisica, Università di Cagliari, s.p. n 8 Km 0.700, 09042 Monserrato, Cagliari, Italy Micron Semiconductor Italia, s.r.l. via Camillo Olivetti, 2 20864 Agrate Brianza (MB), Italy

a r t i c l e

i n f o

Article history: Received 26 July 2012 Received in revised form 6 November 2012 Accepted 8 November 2012 Available online 16 November 2012 PACS: 71.55.−I 71.55.Jv 68.60.−p 68.65.−k

a b s t r a c t In this work polysilazane based silica thin films grown on multilayer structures of different ultra-thin barriers (UTBs) on silicon substrates were studied. The silica thin films were obtained by polysilazane spin coating deposition (also called SOD, spin-on dielectrics) upon different UTB liners (silicon nitride or silicon dioxide). By curing the SOD with thermal treatments the polysilazane is converted into silica thin films. The degree of conversion to SiO2 was analyzed and the oxide local structure was studied in terms of Si O Si bridges by FTIR spectroscopy. Steady state and time resolved luminescence were applied to further characterize the oxide structure, the substrate–silica interfaces and the presence of defects. The analysis revealed the presence of dioxasilirane, Si(O2 ), and silylene, Si:, defect centers in the samples grown on silicon nitride UTB, while these defects are not observed in samples grown on silicon oxide UTB. © 2012 Elsevier B.V. All rights reserved.

Keywords: Silica thin film Polysilazane Ultra thin barriers Spin-on dielectrics SOD FTIR Photoluminescence Structural properties Surface defects Dioxasilirane Silylene

1. Introduction In the spite of Moore’s Law, ultra-large-scale integrated (ULSI) circuits require an increasing density of devices every new generation of circuits. Due to the continuous decrease in the dimensions and spacing of devices on integrated circuits (ICs), insulating layers (typically SiO2 ) have to be deposited to electrically isolate the different active components of the circuit (transistors, resistors, capacitors). Indeed the architecture of the circuits plays a key role and insulating structures such as shallow trench isolation (STI) regions are formed in trenches within the substrate between components. Such trenches can have a width of about 50 nm or even smaller, and filling such narrow gaps can be challenging [1,2]. In

∗ Corresponding author. E-mail address: [email protected] (P.C. Ricci). 0169-4332/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.apsusc.2012.11.030

addition, the dielectric material must be able to withstand subsequent processing steps such as etching and cleaning steps. Spin-on dielectric (SOD), that is the deposition of specific silica precursors by spin coating, represents a valid alternative to high density plasma or other techniques based on chemical vapor deposition (CVD), in particular regarding the demand of gap-filling in STI with aspect ratio greater than 5:1 [3–7]. Indeed flowable materials with high gap filling properties such as SOD and spin-on polymers like silicates, siloxanes, silazanes or silisesquioxanes, have been recently developed [5,6,8,9]. Dealing with SOD materials, the key point is the chemical physical quality of the obtained insulating layer: the silica conversion degree by thermal treatment with furnace has to guarantee a reliable oxide in terms of density, robustness and resistivity to breakdown [10]. Increasing furnace annealing temperature gives good results on flat wafers in terms of material conversion degree, but it seems to have not the same impact concerning the filling of narrow trenches [11].

P.C. Ricci et al. / Applied Surface Science 265 (2013) 470–474 Table 1 List of analyzed samples and parameters growth of the UTBs.

471

Furnace atmosphere

Pressure (Pa)

Temperature (◦ C)

Treatment time (min)

O1 O2 N1

N2 + O2 N2 O + SiH4 NH3 + SiH2 Cl2

Atm. 66.6 40

1000 790 775

300 180 110

The effect of the substrate composition on the properties of thin layer was recently studied [12], similarly, in this work we focus our attention on the ultrathin barrier (UTB) deposited between the active areas and the insulating dielectrics and how different UTBs can affect the silica conversion of selected polysilazane based SOD. We present a FTIR analysis of the silica converted SOD as a function of different UTBs (silicon nitride and silicon oxide), after furnace curing, focusing on the effective silica conversion in terms of SOD bulk properties. The optical properties were studied by steady state and time resolved photoluminescence (PL) in order to assign the nature of defects at the SOD layer and to correlate the results of the different experimental techniques. 2. Material and methods Polysilazane based SOD thin films were spinned on large high purity circular p-silicon mono-crystal wafers (diameter 200 mm), with (1 0 0) orientation. Different ultra-thin barriers (UTB) of silicon nitride or silicon dioxide, each one with thickness less than 80 nm, were deposited as substrate liners upon silicon wafer by thermal treatment inside a furnace in different chamber atmospheres (N2 + O2 , NH3 + SiH2 Cl2 and N2 O + SiH4 ), see Table 1 for details. Polysilazane was deposited on top of UTBs of the silicon wafer by spin coating, then a first bake treatment was performed in an inert atmosphere at 150 ◦ C for 180 s in order to remove the solvent. A three step curing procedure was applied for all the samples: a first step at 400 ◦ C (30 min) followed by 700 ◦ C (30 min), both in a steam atmosphere and finally a 900 ◦ C (30 min) in a dry atmosphere. Samples with SOD thickness of about 290 nm were obtained (the thickness was estimated by interferometric analysis). Time resolved photoluminescence (TR-PL) measurements were performed with excitation provided by an optical parametric oscillator with a frequency doubler device (Spectra Physics MOPO), excited by the third harmonic of a pulsed Nd-YAG laser (Spectra Physics QuantaRay PRO-270), with pulse width at half maximum of 8 ns and 10 Hz of repetition rate. The PL signal was dispersed by a spectrograph (ARC-SpectraPro 300i) with a spectral bandpass <2.5 nm and detected by a gateable intensified CCD (PI MAX Princeton Inst.). Spectra were corrected for the optical transfer function. FTIR measurements were performed in the 2500–400 cm−1 spectral range by using a FTIR spectrometer Nicolet Eco1000 model with a spectral resolution of 4 cm−1 . The list of samples and the UTBs growth parameters are reported in Table 1. 3. Results Fig. 1a reports the FTIR spectra of the different ultra-thin layers before SOD spin coating process. Both O1 and O2 samples display the FTIR spectrum of pure silica, with typical absorption peaks of Si O bonds occurring at about 1080 cm−1 (stretching mode region), at 456 cm−1 (rocking mode region) and at 810 cm−1 (bending mode region) [13,14]. The FTIR spectrum on sample N1 presents two main bands at 830 cm−1 and 490 cm−1 which can be ascribed to Si N vibration and to Si breathing vibrations, respectively [10,15]. Fig. 1b reports the spectra of the samples after SOD spin coating and curing by thermal annealing in different atmosphere: the main

N1 O2 O1

500

1000

1500

2000

B

2500

N1 O2 O1

500

1000

1500

2000

2500

Wavelength (cm-1) Fig. 1. (a) FTIR spectra of the ultra-thin layer before SOD spinning process. (b) FTIR spectra after SOD spinning process and thermal annealing.

vibrations of pure SiO2 oxide are clearly observable for the whole set of samples. However, small differences in the FTIR spectra can be evidenced mainly in the region of the Si O stretching mode; an enlarged view (900–1300 cm−1 ) is reported in Fig. 2a. In this region the spectrum is dominated by the vibration of the Si O Si bridges and can be deconvoluted by assuming four Gaussian contributions: two main components, peaked at about 1050 cm−1 and 1090 cm−1 and two larger and less intense bands at 1150 and 1230 cm−1 . The four bands pertain to Si O Si vibrations in different local environments, as reported in the discussion section [14,16,17]. Fig. 2b reports the experimental data, the four Gaussian bands and the fit result (square correlation factor R2 > 0.98). The fitting results are summarized in Table 2. The fitting procedure assumes as initial parameters the values reported in the literature for each band (see Table 1 for the assignment) [18–21] and it was performed, constraining the parameters of the fit of each curves to vary less than 10 cm−1 and 5 cm−1 for the center wavenumber and the FWHM, respectively. In Fig. 3 the emission spectra recorded by exciting the different samples at 250 nm and gathered in the same experimental conditions are reported. Both the PL spectra from O1 and O2 samples present two maxima with relative low intensity, peaked at 295 (310) and 375 (392) nm in the O1 (O2) case (inset of Fig. 3). In the O2 case the overall PL intensity is larger than in the O1 case. Concerning

N1

Absorbance

Sample name

FTIR intensity (arb. units)

A

O2

O1

900

950

1000

1050

1100

1150

1200

1250

1300

Wavelength (cm-1) Fig. 2. FTIR spectra in the region of the SiO stretching mode (900–1300 cm−1 ) and related fitting results: in red the deconvoluted contribution form bulk Si O Si bridges, in blue the deconvoluted contribution from boundary Si–O–Si bridges.

472

P.C. Ricci et al. / Applied Surface Science 265 (2013) 470–474

Table 2 Fitting results of the FTIR spectra in the region of the Si O stretching mode. O1

Peak A Bulk – in-phase mode Peak B Surface in-phase Peak C Bulk – out of phase mode Peak D Bulk – out of phase mode

O2

FWHM (cm−1 )

Area %

Xc (cm−1 )

FWHM (cm−1 )

Area %

Xc (cm−1 )

FWHM (cm−1 )

Area %

1051.0

62.3

38

1050

60

35

1048.4

54.6

32

1091.9

46.6

27

1091.0

49.4

30

1089.6

48.4

33

1147.5

93.1

20

1150.0

94.2

19

1149.5

94.2

17

1230

96.5

15

1239.9

94.2

16

1239.6

94.1

18

Fig. 3. Photoluminescence spectra with excitation at 250 nm. Inset, enlarged view of samples O1 and O2 PL spectra.

the N1 sample a single very large PL band peaked at about 417 nm is evidenced. This emission was characterized by TR-PL measurements (Fig. 4). The observed decay profile was successfully fitted by a double exponential function, as reported in the figure, yielding two decay time constants in the nanosecond region:  1 = 2.9 ns and  2 = 20.0 ns. 4. Discussion Spin-on dielectrics can be an easy, economic and versatile solution to the request of nanometric dielectric layers in the down

Adj. R-Square

PL Intensity (arb. units)

0.1

y0

0.998 Value 0

Standard 0.00219

A1 t1 (ns) A2 t2 (ns)

0.212 2.99 0.0620 20.0

0.008 0.15 0.006 3.5

20

40

Time (ns) Fig. 4. Time decay at 417 nm, excitation 250 nm.

scaling of the integrated circuits. The aim of the present work is to study how the conversion to silica of the selected SOD (polysilazane) is affected by different ultra thin barriers (liners) grown on Si wafer by analyzing the structure of the silica network and the presence of defects at the liners-SOD interface. The liner is twice useful: from one side it prevents the formation of uncontrolled silicon oxide layers at the Si wafer because, i.e., exposure to ambient atmosphere, from the other side it allows the bonding of SOD to be cured and converted in silica. The FTIR spectra of the different liners (Fig. 1a) indicate the formation of the requested ultra thin barriers, silicon dioxide and silicon nitride in the O1/O2 and N1 samples. It can be observed that beside the fingerprints of the liners, no other bands were detected, for examples the one due to Si H (2150 cm−1 ), Si C, Si N H (1550 cm−1 ), Si NH (1180 cm−1 ), or OH (3470 cm−1 ), indicating the absence of strong environmental absorption at the surface [20]. FTIR spectra also confirmed the formation of a pure silica layer after SOD coating and curing (Fig. 1b) in the wall set of samples, in particular no absorptions at 610 cm−1 due to unsaturated Si Si bonds were detected, indicating the absence of structural imperfections at the UTB-SOD interface. Concerning the dependence on different liners, the detailed analysis of the FTIR spectra in the 900–1300 cm−1 spectral range (Fig. 2) reveals small but significant differences to the Si O stretching vibrations. In order to analyze the differences, the spectra were fitted with four Gaussian bands related to different vibrational contributions: two bands at 1050 cm−1 and 1090 cm−1 (A and B respectively) and two bands at 1150 and 1230 cm−1 (B and C). These bands were previously assigned to the asymmetric stretching of Si O Si bridges where two adjacent O atoms move in phase (A and B) or out of phase (C and D) [19,21]. In addition the A and C bands correspond to Si O Si bridges located in the bulk of the SiO2 film and far away from defects or grain boundaries, while the B and D bands correspond to Si O Si bridges located at the interfaces of the film, at the grain boundaries and at defects [18,19,22]. It should be noted that N1 sample, grown on Silicon Nitride UTB, presents the largest contribution of the B band, suggesting a larger contents of defect and inclusions. In order to estimate the fraction of bulk-like Si O Si bridges (NBulk ) with respect to the boundary ones (Nboundary ) we can

ω

0.01

0

N1

Xc (cm−1 )

60

evaluate the integral 0 ωε2 dω which is proportional to the total number of oscillators (ω being the wavenumber and ε2 the imaginary part of the dielectric constant) [19,23]. The NBulk /Nboundary ratio NBulk /Nboundary is 0.96 for N1 sample, while increases up to 1.38 and 1.17 in O1 and O2 samples respectively. A higher number of boundary bridges with respect to bulk ones indicate a larger concentration of voids and defects, which can be related to a lower degree of SOD conversion into silica layer. However the FTIR spectra are mainly connected to the study of bulk species and can be less influenced by the presence of surface or interfaces defects. The deconvolution procedure that separates the contribution of the “bulk” Si O Si bridges (at 1050 cm−1 ) from the “boundary” Si O Si ones (at about 1090 cm−1 ), can be a tricky process when

P.C. Ricci et al. / Applied Surface Science 265 (2013) 470–474

473

Fig. 5. In the upper panel the formation of a perfect SiO2 structure starting from oxide surface and thermal annealing. Bottom part, the proposed model of defects generation in the silicon nitride based SOD. The dioxasilirane and silylene centers are evidenced.

the surface effects are negligible and the final FTIR spectrum is broad; therefore the results need complementary experimental investigation techniques. In this respect photoluminescence measurements can be very useful and can be focused on the detection and assignment of bulk and/or surface defects [24–28]. In respect of this discussion, PL measurements confirmed the presence of defects, with emission properties typically ascribed to surface defects in nanosized silica, and differences in the defectiveness of the samples related to a different contribution to the overall PL in different samples. In addition also the recorded spectroscopic features depend on the different UTB-SOD interface. By considering that the integrated area of the PL response is proportional to the total amount of the luminescent defects, we can give a rough estimate of the defect concentration in different samples as referred to sample O1, giving a defect concentration 20 times larger in N1 than in O2. It is worth to note that FTIR and PL analysis can be considered as complementary: the luminescence is strictly connected to the kind of defects and vacancies while the FTIR analysis identify the presence of Si O Si boundary bridges around to voids and defects in the samples or at the interfaces, being the siloxane bridge lacking of any visible PL emission. Focusing on the attribution of the large emission observed in N1 sample, the reported spectral features strongly resemble previous results on fumed and porous silica samples and assigned to a surface defects pair, Si(O2 ) (dioxasilirane) and Si: (silylene), formed by a dehydroxylation reaction between two surface geminal silanols (interacting Si OH groups) [26–30]. The presence of this defects pair can be understood by means of the following descriptive model. The polysilazane based SOD (Fig. 5) is a flowable thermally processed curable material: during the thermal heating process, the material is engineered to be converted to a dense Si O network. As the SOD is cured, the nitrogen and hydrogen atoms in the SOD leave the material and are replaced by oxygen to form Si O bonds; the Si H, Si N, and N H bonds at the UTB-SOD interface are replaced by a Si O Si network, as shown in Fig. 1. It should be considered that the UTB surface, being exposed to ambient atmosphere,

is expected to react with H2 O molecules. In the O1 and O2 cases the reaction causes the surface coverage by OH groups which allow the reaction with Si N groups when the surface is coated with polysilazane leading to the formation of Si O metal bonds and promoting a good adhesion of the coating to the substrate. In the N1 case, the Si3 N4 surface reacts with H2 O from the environmental, yielding silanol species (Si OH) and amine groups (Si N H2 and Si2 N H) [31,32]. Even though the silanol groups are the dominant species and the most reactive with the SOD (as in the case of pure oxide), the presence of the amine limits the bonding ability of silicon nitride, generating voids at the silicon nitride-polylazanes interface. The nanopores can favor the dehydroxylation reaction between two surface geminal silanols and the formation of the defects pair responsible of the observed emission at 417 nm [33]. On the contrary, in O1 and O2 samples FTIR analysis suggests a more regular distribution of Si O Si bridges with respect to sample N1: we can hypothesize that nanopores are not formed and the low intensity luminescence bands at 375 and 295 nm are connected to rarely distributed hydrogen related surface groups. 5. Conclusions Polysilazane based spin-on dielectrics (SOD) film with thickness of 290 nm were grown on different ultra-thin barrier (UTB) and analyzed by FTIR and photoluminescence spectroscopy. We found that the UTB layers strongly influence the SOD bulk properties, in particular, sample grown on silicon nitride presents a larger amount of defects with respect to samples grown on silicon oxide thin barrier obtained in different furnace atmosphere. The analysis of the FTIR spectra was conducted on the basis of Si O Si bridges in order to estimate the fraction of bulk-like over boundary bridges. We found a larger concentration of boundary bridges in the sample grown on silicon nitride, the fingerprint of a more defective structure. The presence of defects is confirmed by photoluminescence spectra a very large band at 417 nm is observed only in the sample grown on silicon nitride and is ascribed to the dioxasilirane ( Si(O2 )) and silylene ( Si:) defects pair. The less intense bands at

474

P.C. Ricci et al. / Applied Surface Science 265 (2013) 470–474

375 and 295 nm in samples grown on silicon oxide are connected to rarely distributed hydrogen related surface groups. We proposed that dioxasilirane and silylene centers are located at the surface between the ultra-thin barriers and the SOD and connected to the formation of SiNH2 , and Si2 NH groups on the silicon nitride surface, limiting the bonding ability of the surface to the polysilazane solution. Acknowledgements We are grateful to P. Comite, R. Quarenghi and L. Castelletti Micron Italy for his support in sample preparation and characterization. References [1] A. Tavernier, L. Favennec, T. Chevolleau, V. Jousseaume, Innovative gap-fill strategy For 28 nm shallow trench isolation, ECS Transactions 45 (2012) 225–232. [2] Y.T. Huang, S.L. Wu, S.J. Chang, C.K. Hung, T.J. Wang, C.W. Kuo, C.T. Huang, O. Cheng, Enhancement of CMOSFETs performance by utilizing SACVD-based shallow trench isolation for the 40-nm node and beyond, IEEE Transactions on Nanotechnology 10 (2011) 433–438. [3] B. Lee, Y.-H. Park, Y.-T. Hwang, Weontae, O.H. Yoon, M. Ree, Ultralow-k nanoporous organosilicate dielectric films imprinted with dendritic spheres, Nature Materials 4 (2005) 147–151. [4] A. Kshirsagara, P. Nyaupane, D. Bodas, S.P. Duttagupta, S.A. Gangal, Deposition and characterization of low temperature silicon nitride films deposited by inductively coupled plasma CVD, Applied Surface Science 257 (2011) 5052–5058. [5] C.W. Holzwarth, T. Barwicz, H.I. Smith, Optimization of hydrogen silsesquioxane for photonic applications, Journal of Vacuum Science and Technology B 25 (2007) 2658–2662. [6] W.G. Lee, The effects of C on the low-temperature formation and the properties of the spin-on dielectric films used for sub-50 nm technology and beyond, Thin Solid Films 520 (2012) 3003–3008. [7] G. Gulleri, C. Carpanese, C. Cascarano, D. Lodi, R. Ninni, G. Ottaviani, Deposition temperature determination of HDPCVD silicon dioxide films, Microelectronic Engineering 82 (2005) 236–241. [8] L. Toniutti, S. Mariazzi, N. Patel, R. Checchetto, A. Miotello, R.S. Brusa, Porosity depth profiling of spin-coated silica thin films produced by different precursors sols, Applied Surface Science 255 (2008) 170–173. [9] S. Hemmilä, J.V. Cauich-Rodríguez, J. Kreutzer, P. Kallio, Applied Surface Science 258 (2012) 9864–9875. [10] K. Trivedi, C. Floresca, S. Kim, H. Kim, D. Kim, J. Kim, M.J. Kim, W. Hu, Void-free filling of spin-on dielectric in 22 nm wide ultrahigh aspect ratio Si trenches, Journal of Vacuum Science and Technology B 27 (2009) 3145–3151. [11] C.Y. Ho, K.-Y. Shih, J.H. He, Gate oxide wear out using novel polysilazane-base inorganic as nano-scaling shallow trench filling, Microelectronic Engineering 87 (2010) 580–583. [12] M. Nicolescua, M. Anastasescu, S. Preda, H. Stroescu, M. Stoica, V.S. Teodorescu, E. Aperathitis, V. Kampylafk, M. Modreanu, M. Zaharescu, M. Gartner, Influence of the substrate and nitrogen amount on the microstructural and optical properties of thin r.f.-sputtered ZnO films treated by rapid thermal annealing, Applied Surface Science 261 (2012) 815–823.

[13] R. Qu, M. Wang, C. Sun, Y. Zhang, C. Ji, H. Chen, Y. Meng, P. Yin, Chemical modification of silica-gel with hydroxyl- or amino-terminated polyamine for adsorption of Au(III), Applied Surface Science 255 (2008) 3361–3370. [14] M. Kopani, M. Mikula, M. Takahashi, J. Rusnák, E. Pincík, FT IR spectroscopy of silicon oxide layers prepared with perchloric acid, Applied Surface Science (2012), http://dx.doi.org/10.1016/j.apsusc.2012.09.081. [15] V. Godinho, V.N. Denisov, B.N. Mavrin, N.N. Novikova, E.A. Vinogradov, V.A. Yakovlev, C. Fernandez-Ramos, M.C. Jimenez de Haro, A. Fernandez, Vibrational spectroscopy characterization of magnetron sputtered silicon oxide and silicon oxynitride films, Applied Surface Science 256 (2009) 156–164. [16] I.W. Boyd, Deconvolution of the infrared absorption peak of the vibrational stretching mode of silicon dioxide: evidence for structural order? Applied Physics Letters 51 (1987) 418–421. [17] P.G. Pai, S.S. Chao, Y. Takagi, G. Lucovsky, Infrared spectroscopic study of SiOx films produced by plasma enhanced chemical vapor deposition, Journal of Vacuum Science and Technology A 4 (1986) 689–694. [18] V.Em. Vamvakas, M. Theodoropoulou, S.N. Georga, C.A. Krontiras, M.N. Krontiras, Correlation between infrared transmission spectra and the interface trap density of SiO2 films, Microelectronics Reliability 47 (2007) 834–837. [19] V.E.M. Vamvakas, D. Davazoglou, Influence of the growth temperature on the atomic distribution of TEOS deposited SiO2 films, Journal of Vacuum Science and Technology B 23 (2005) 1956–1964. [20] G. Scardera, E. Bellet-Amalric, D. Bellet, T. Puzzer, E. Pink, G. Conibeer, Formation of a nanocomposite from plasma enhanced chemical vapour deposition multilayer structures, Journal of Crystal Growth 310 (2008) 3685–3689. [21] C.T. Kirk, Quantitative analysis of the effect of disorder-induced mode coupling on infrared absorption in silica, Physical Review B 38 (1988) 1255–1273. [22] P.N. Sen, M.F. Thorpe, Phonons in AX2 glasses: from molecular to band-like modes, Physical Review B 15 (1977) 4030–4038. [23] D. Davazoglou, V.Em. Vamvakas, Arrangement of Si and O atoms in thermally grown SiO2 films, Journal of the Electrochemical Society 150 (2003) F90–F96. [24] C.M. Carbonaro, P.C. Ricci, R. Corpino, M. Marceddu, A. Anedda, Photoluminescence characterization of aged and regenerated mesoporous silica, Journal of Non-Crystalline Solids 357 (2011) 1904–1907. [25] C.M. Carbonaro, P.C. Ricci, A. Anedda, Thermal quenching properties of ultraviolet emitting centers in mesoporous silica, Physical Review B 76 (2007) 125431–125436. [26] T. Uchino, N. Kurumoto, N. Sagawa, Structure and formation mechanism of blue-light-emitting centers in silicon and silica-based nanostructured materials, Physical Review B 73 (2006) 233203–233209. [27] A. Nishimura, N. Sagawa, T. Uchino, Effect of cross-linking and organic groups on the visible photoluminescence characteristics of n-octadecylsiloxanes, Journal of Physical Chemistry C 114 (2010) 8568–8574. [28] A. Nishimura, N. Sagawa, T. Uchino, Structural origin of visible luminescence from silica based organic–inorganic hybrid materials, Journal of Physical Chemistry Letters 113 (2009) 4260–4262. [29] A. Anedda, C.M. Carbonaro, F. Clemente, R. Corpino, P.C. Ricci, Time resolved ultraviolet photoluminescence of mesoporous silica, Journal of Physical Chemistry B 109 (2005) 1239–1242. [30] L. Vaccaro, A. Morana, V. Radzig, M. Cannas, Bright visible luminescence in silica nanoparticles, Journal of Physical Chemistry C 115 (2011) 19476–19481. [31] S. Mezzasalma, D. Baldovino, Characterization of silicon nitride surface in water and acid environment: a general approach to the colloidal suspensions, Journal of Colloid and Interface Science 180 (1996) 413–420. [32] M. Grattarola, G. Massobrio, S. Martinoia, Modeling H+-sensitive FETs with SPICE, IEEE Transactions Electron Devices 39 (1992) 813–819. [33] A. Anedda, C.M. Carbonaro, F. Clemente, R. Corpino, P.C. Ricci, Low temperature investigation of the blue emission in mesoporous silica, Materials Science and Engineering C 25 (2005) 631–634.