Open Access
Add to BibSonomyAbstract
It is demonstrated that as-grown e-beam deposited SiOx thin films on fused silica substrates show a second-order nonlinear response that is dependent on film thickness. Using a Maker fringes method the effective nonlinear coefficient for a SiO thin film is estimated to be comparable to that of crystalline quartz. Variation of process parameters has been used to investigate the origin of the nonlinear response. The second-harmonic signal is very sensitive to annealing of the film and can be totally removed by annealing at a few hundred degrees. It is also demonstrated that a retarding grid that traps charged particles between the crucible and the sample reduces the nonlinear signal from a SiO thin film significantly. It is suggested that oriented dipoles arise during deposition due to a negatively charged film from oxygen ions, thus, resulting in a non-centrosymmetric film. Finally, using e-beam lithography, well-defined nonlinear 2D structures can be synthesized, thus opening the door to a new and practical way to create nonlinear structures for planar waveguide technology.
© 2012 Optical Society of America
1. Introduction
The development of new nonlinear optical materials compatible with Si process technology has been an active research field over the past two decades. Poling of glasses is considered a promising process that can lead to large second-order nonlinearities over a broad spectral range for a wide range of glass compositions. In the thermal poling process electric field induced rearrangement of ions at elevated temperatures lead to a built-in electric field after cooling to room temperature [1]. In combination with bulk allowed third-order nonlinearities this results in an effective second-order effect allowing second-harmonic generation (SHG) as well as electro optic effects. The two processes, SHG and linear electro optic effect, have the same selection rules and SHG is often used to map out the spatial distribution of induced nonlinearities [2]. Typically the built-in field exists in a few micrometers thick layer at the surface of the material. Effective nonlinear coefficients up to a few pm/V have been obtained depending on the glass composition [3, 4]. More than two orders of magnitude enhancement of SHG compared to a bulk poled silica glass has been obtained from a poled stack of alternating doped and un-doped layers of silica [5]. Very recently, it has been demonstrated that silicon nitride films prepared on fused silica substrates by Plasma Enhanced Chemical Vapor Deposition (PECVD) has a strong SHG response even without poling of the deposited film [6].
SiOx films are amorphous with a range of oxidation stages of Si atoms (Sin+ with 0<n<4) depending of deposition conditions and external excitations. Such layers have a large number of applications, for instance as protective coatings or as part of dielectric stacks for mirrors and antireflective coatings. High temperature annealing of SiOx films leads to light-emitting layers through formation of Si nanocrystals embedded in a silica matrix [7]. The relatively open structure of the oxide allows for inclusion of ionic species, where hydrogen is expected to be the most prominent impurity.
Electron-beam irradiation of doped SiO2 glass have shown to induce a second harmonic response due to an electrostatic space-charge field formed by the electrons [8–10]. This SHG mechanism resembles that of thermally poled glass. However, also high-energy electron beam deposition techniques are known to affect structural, electrical and optical properties of thin films [11,12]. In this paper we show that e-beam deposition of SiO leads to a film with oriented dipoles and thus an effective second order nonlinearity giving rise to efficient SHG. The focus of this work is on description of the second-order nonlinearity of e-beam grown SiO layers and the deposition conditions leading to SHG rather than a detailed microscopic analysis of the materials aspects behind the nonlinear effect. SHG signals as functions of incidence angle of the light, thickness of the SiO layers, and the wavelength of the pump light are presented. It is suggested that the nonlinearity is formed directly through inclusion of ionized species during the deposition process and a negatively charged film due to oxygen ions. It is thus not necessary to apply electrical contacts and a thermal process as in conventional poling. Furthermore, the process allows straight forward use of conventional lithography for in-plane definition of nonlinear optical elements. The symmetry of the material with a preferred direction perpendicular to the surface is similar to that of poled materials. One can thus consider the same type of active optical components produced from SiO films as from poled glass films.
2. Experimental
2.1. Thin film deposition
Several SiO thin films with different thickness have been deposited at room temperature onto fused silica and Si(100) substrates using an −7.5 kV electron beam evaporation technique with a deposition rate of 0.5–1Å/s. The electron beam heated 3 mm-6 mm SiO grains with a 99.99 % purity placed in a graphite crucible. The advantage of using fused silica substrates is its lack of SHG response and that it is transparent in the UV to ∼250 nm. The substrates were cleaned in an ultrasonic bath of acetone followed by ethanol and then blow dried by N2 before they were loaded into the deposition system through a load-lock. A Si wafer was placed next to the silica substrates and received the same deposition of SiOx. Prior to deposition the system was pumped down to a base pressure of 1·10−6 mbar (to prevent oxidation of SiO during deposition). By crunching SiO2 and SiO grains to a finer powder a mixture for deposition of SiOx∼1.5 films was obtained. Additionally, an e-beam deposited SiO2 film on fused silica was prepared. Furthermore, the same deposition chamber included a rf-magnetron source that was used for sputter deposition of a SiO thin film using a 99.9 % pure SiO target in an argon plasma.
2.2. Material composition
The stoichiometry was investigated with a Rutherford backscattering spectrometry (RBS) technique. He+ ions of 2 MeV and a backscattering angle between 30° and 70° were used. Measurements on three different SiO samples with the thickness of 344 nm, 550 nm, and 902 nm revealed that the films deposited from the SiO grains are SiO∼0.95 thin films with a homogeneous distribution of the Si and O atoms throughout the films (data not shown). The SiOx stoichiometry was calculated from the powder mixture. Additionally, secondary ion mass spectrometry (SIMS) was used to determine the hydrogen content in the SiO films.
2.3. Linear optical characterization methods
A Sentech SE 850 ellipsometer was used to estimate the refractive index and the absorption coefficient utilizing the Jellison and Modine model [13] to represent the SiO films on the silicon substrates. This includes an estimation of the film thickness and was verified with RBS. A UV-3600 Shimadzu UV-VIS-NIR spectrometer was used to investigate the change in linear optical properties of the SiO thin films on fused silica after annealing in N2 or air atmospheres at various temperatures.
2.4. Nonlinear optical characterization methods
The SHG spectroscopy experiments were performed using a Nd:YAG-laser pumped optical parametric oscillator (OPO) system delivering 5-ns pulses at 10 Hz repetition rate. Pulse energies of the order of 1 mJ focused to a 1 mm spot diameter on the sample were used. No signs of sample damage were observed, even after illumination for several hours. The spectral composition of the up-converted transmitted light was investigated with a monochromator. A photomultiplier tube coupled to gated electronics was used to detect the nonlinear signals.
After analyzing the up-converted light at a few selected pump wavelengths the monochromator was removed from the setup during spectroscopic measurements. Extraction of the SHG light in the transmitted beam was done with a few combinations of colored glass filters to cover the investigated spectral region. A normalizing reference signal without Maker fringes and with a magnitude comparable to that of the thicker SiO samples was obtained by placing a wedge-shaped quartz crystal at the position of the samples.
Measurements as a function of angle of incidence were performed at pump wavelengths of 760 nm and 1350 nm, and again the measurements were normalized to the reference in order to compare the signal magnitudes of the different samples.
Maker fringes appearing in the angle of incidence scans of SHG from a 1.15-mm thick quartz disc was used to estimate absolute values of the nonlinearity of a SiO film.
3. Theoretical considerations
Herman and Hayden [14] showed that the transmitted SHG signal from an isotropic and absorbing slab with the thickness L on a substrate pumped with λp is given by:
The effective deff coefficient for C∞υ space symmetry are Here we have assumed d15=d31 (Kleinman symmetry). The γ denotes if the fundamental wavelength is s or p polarized. The Fresnel transmission coefficients at the fundamental and SHG wavelengths are denoted as t1γ and t2p, respectively, and subscripts af, fs and sa refer to air-film, film-substrate and substrate-air interfaces, respectively. Furthermore, Here θ is the angle of incidence of the fundamental wavelength. The subscripts m =1 and m =2 denote the fundamental and SHG wavelength, respectively. The refractive index and absorption coefficient are nm and κm, respectively. Thereby using the work by Herman and Hayden the ratio between the two nonlinear coefficients d33/d31 can be estimated.In order to estimate the effective nonlinear coefficient, deff, the measured SHG signal from a 1.15-mm thick quartz disc was used. By combining Eq. (1) with the work by Jerphagnon and Kurtz [15] to describe the quartz crystal the signal radio between the SiO film and quartz signals can be expressed, thus, a estimation of deff can be made.
4. Experimental results and discussions
Figure 1(a) shows a monochromator scan of the transmitted SHG signal from a 500-nm thick SiO∼1.5 film pumped at 760 nm. The relatively narrow spectral peak centered at approximately at 380 nm indicates that the measured signal has to be SHG arising from the SiO∼1.5 film. The same was realized for several SiO films. On the other hand, neither a 500-nm thick e-beam deposited SiO2 film nor a 160-nm thick rf-magnetron sputtered SiO film showed detectable SHG signals. Already from these observations it is clear that charging effects from the e-beam and the composition of the deposited film are important in the creation of the second order nonlinearities.
Fig. 1 Spectral and angular dependence of SHG from SiO films. a) The second harmonic signal between 300 nm and 500 nm from a 500-nm thick SiO∼1.5 film excited by a pump beam at 760 nm. b) Angular dependences of the SHG signal from the SiO thin films with the thickness between 104 nm to 902 nm. On the left hand side and the right hand side, respectively, 1350 nm and 760 nm excitation wavelength were used. Symbols are experimental results and the solid lines are the best fits using Eq. (1). Notice that a factor of 15 has been multiplied to the signal strength on left hand side.
4.1. SHG signal dependence on angle of incidence and film thickness
The transmitted p-p polarized SHG as a function of angle of incidence from different SiO samples with various thicknesses is presented in Fig. 1(b). On the left and right hand sides of the figure, pump wavelengths of 1350 nm and 760 nm were used, respectively. Clearly, there is a strong angular dependence and the SHG intensity peaks around 55–60 degrees angle of incidence, which has also been observed for poled silica [1]. The solid lines are the best fits achieved using Eq. (1) together with the refractive index and absorption coefficients estimated from ellipsometric measurements on SiO films on silicon.
According to Kazansky and Russel [16] the ratio d33/d31 should be 3 for an electric-field-induced mechanism. In the current experiments this ratio has been extracted from the measured angle-of-incidence-dependence for p-polarized SHG. The ratio varied in the range from 2 to 4 for the different films with no particular tendency with film thickness. It is suspected that observed variations in SHG among films may be caused by differences in the traces of the e-beam in the crucibles during deposition. With the variations in the ratio seen in the current experiments it is not possible to use the ratio d33/d31 to determine the origin of the nonlinearity.
In Fig. 1(b) it is noticed that with 760 nm pump light the growth in intensity seems to saturate for thick films as the signal levels are the same for the 550- and 902-nm thick films. To understand this we need to take a look at the linear optical properties.
The transmittance and reflectance curves alongside with the calculated absorption coefficient, α, are presented in Fig. 2(a) and Fig. 2(b) for a 550-nm thick SiO film on fused silica. By using a Tauc plot [17] the estimated energy band gap for the as-deposited film is approximately Eopt ∼2.5 eV. Thus, an absorption, α ∼ 4 · 10−4cm−1, of our SHG signal within the SiO film occurs when using 760 nm as the pump wavelength as in this present experiments.
Fig. 2 The linear optical properties of a 550-nm thick SiO film. a) The measured transmittance (T) and reflection (R) curves of the as-deposited film and after annealing at 440°C for 30 min in a N2 atmosphere. The oscillations are due to interference effects which are included in the model of the absorption coefficient (α). b) The calculated absorption spectra from the data in Fig. 2(a).
The left hand side of Fig. 1(b) shows the behavior of the SHG signal when using a pump wavelength of 1350 nm. Now the 902-nm thick sample has an average signal strength about ∼4.8 times that of the 550-nm thick sample between 50–60 degrees angle of incidence. According to Eq. (1) the signal strength grows with the thickness squared, meaning we should expect the 902-nm thick film to have a signal strength ∼2.7 times larger than the 550-nm thick film without absorption. The higher experimental value may be explained by Fig. 3, which shows the variation of the SHG signal at 57 degrees angle of incidence taken from the right hand side of Fig. 1(b). The curves are the calculated SHG signal using Eq. (1) with different dependencies on film thickness and normalized to fit the data. The form of the curves can be understood from a simplified version of Eq. (1):
with the first factor representing the signal growth with increasing film volume (β=2 in Eq. (1)) while the exponential factor represent the absorption of the SHG light. The almost linear initial growth of SHG with thickness is best represented by β=2.3 when absorption is taken into account (the red curve). As the thickness grows β appears to be closer to 2 as expected. It is suggested that variations in β may reflect the complicated charge distribution in the film. A value of β larger than 2 is consistent with the fast growth of the signal observed at 1350 nm as mentioned above.Fig. 3 The SHG signal as a function of film thickness. The stars are the SHG intensities at 57 degrees angle of incidence taken from the right side of Fig. 1(b). The curves are the calculated SHG signal using Eq. (1) with different dependencies on film thickness.
Figure 4(a) shows the SHG signal, λp=760 nm, as a function of angle of incidence from a 1.15-mm thick quartz disc and from a 550-nm thick SiO film, respectively, presented with blue circles and red squares. The green curve is the envelope curve that represents the maximum SHG signal from the quartz disc. The SHG signal in Fig. 4(a) from the SiO film is multiplied by 176 for proper presentation. This large difference in signal strength is due to the large difference between the thickness of the SiO thin film and the optical coherence length of quartz. The calculated coherence length of quartz at these wavelengths is ∼7.6μm. Notice that the squared ratio between this coherence length and the SiO film thickness is 190. Thus, the SHG response of SiO is compatible to that of quartz.
Fig. 4 The angular and pump energy dependence of the SHG signal. a) The SHG signal measured from an 1.15-mm thick quartz disc presented by blue circles and fitted with the work from [15] represented by at black curve. The green curve is the expected SHG signal from the quartz disc without any Maker fringes. The red squares are the SHG measurements from a 550-nm thick SiO film fitted with Eq. (1). The SiO SHG signal is multiplied by 176. Here λp=760 nm. b) The transmitted SHG from a 550-nm thick SiO film on fused silica as a function of SHG photon energy. The OPO is not tunable in the region around degeneracy at 710 nm corresponding to SHG photon energies around 3.5 eV. This leads to the “hole” in the spectrum at this energy. The solid curve is a fit of a Lorentzian that help to determine the resonance position to ∼3.3 eV.
By using Eq. (1) to describe the transmitted SHG signal from the absorbing 550-nm thick SiO film on a substrate and the work by Jerphagnon and Kurtz [15] to describe the quartz crystal the radio between the SiO film and quartz signals can be expressed. Then, by calculating the ratio between the expected SHG signal at normal incidence from quartz, without Maker fringes, and the SiO film at signal peak, in this case ∼57 degrees angle of incidence, see Fig. 4(a), the ratio can be evaluated.
By using the value d11= 0.31 pm/V reported by Hagimoto and Mito [18] for quartz at an excitation wavelength of 633 nm, and considering Miller’s rule [19], the is calculated to 0.17 pm/V. Thus, the e-beam deposited SiO thin film has an effective nonlinear coefficient that is comparable to that of crystalline quartz.
4.2. SHG signal dependence on wavelength
Figure 4(b) shows the measured SHG signal as a function of the SHG photon energy normalized to the reference SHG signal from a wedge shaped quartz crystal. The data were fitted with a Lorentzian to determine the resonance peak position to be at ∼3.3 eV. The nonlinear response (deff in Eq. (1)) is thus dominated by a two-photon resonance near ∼3.3 eV [20]. The peak at 3.3 eV corresponds very well to the linear calculated absorption spectra shown in Fig. 2(b).
4.3. SHG signal dependence on annealing temperature
The nonlinearity of the e-beam deposited SiOx films is sensitive to the sample temperature and can be removed completely through annealing. For instance, annealing a 500-nm thick SiO1.5 film in N2 at 100°C for 30 minutes reduced the SHG signal by a factor of 1.5. Figure 5(a) shows the loss of SHG signal when 11 pieces of an as-deposited 500-nm thick SiO1.5 sample are annealed for 30 min in a N2 atmosphere at different temperatures. Every point is the average SHG signal measured by using pump wavelengths between 720 nm and 1000 nm. Notice that the SHG signal can be significantly reduced by annealing at 500°C. Following the work of Bucci and Fieschi [21] on annealing of induced polarizations in insulators an activation energy is estimated as shown in the inserted Arrhenius plot in Fig. 5(a). Here the square root of the SHG signal is used as the SHG signal is proportional to . As indicated in the insert two processes may be at play, one with an activation energy of 0.11 eV and another at 0.36 eV.
Fig. 5 Temperature dependence of SHG SiO films. a) 11 pieces of a 500-nm thick SiO1.5 thin film annealed at different temperatures in a N2 atmosphere. The activation energy was estimated by a linear fit as illustrated in the insert. b) The SHG signal from a 902-nm thick SiO film while annealed to 440°C in an air atmosphere.
In another experiment the SHG signal was measured from a 902-nm thick SiO film while heating in free air up to 440°C. Fig. 5(b) shows the decay of the SHG signal recorded during constant heating of the sample. The three vertical lines starting from the left to the right denotes when the temperature is turn on, when it reaches 400°C and finally stabilizes at 440°C. When the sample subsequently was cooled down to room temperature the SHG signal remained unchanged and the nonlinearity was thus permanently reduced.
The linear transmittance curve (not shown here) behaved the same way as seen in Fig. 2(a), where the 550-nm thick SiO film annealed in N2 atmosphere became more transparent. This suggests that the reduction in SHG signal is not due to oxidation of the SiO films.
4.4. Origin of the nonlinear response
It is well known that the energetic electron beams used in e-beam deposition systems will affect the properties of the deposited film. Often a grid is used in front of the evaporator in order to prevent charged particles from reaching the substrate. Hoffman and Leibowitz [12] demonstrated that Al2O3 films deposited on glass slides from an e-beam source became brown in color if a buildup of negative charge on the insulating substrate was present. Furthermore, they argued that the electron beam created dissociated positive Al ions and negative O ions, thus, the negatively charged substrate repels the oxygen ions thus altering the stoichiometry, resulting in a brown Al2O3 film. In analogy with these experiments we placed a mesh grid (∼50% transmittion) consisting of stainless steel 15 cm above the crucible containing the SiO grains. Applying the same −7.5 kV on the mesh grid as for the e-gun, works as an ion trap that repels both O ions and electrons. A blueish plasma about 3–4 cm high rising from the crucible towards the grid was visible during deposition of a 230-nm thick SiO sample. The blueish color is consistent with an oxygen plasma. Figure 6 shows the SHG signal, λpump = 760 nm, divided by the squared thickness for two different samples with and without the −7.5 kV mesh grid. Compared to the film deposited without the retarding grid the SHG signal was reduced by a factor ∼47. Thus, electrons and/or O− ions play a significant role in the creation of the optical nonlinearity.
Fig. 6 The SHG signal from two different SiO films prepared with and without a retarding grid. Using a retarding grid reduces the SHG signal by a factor of ∼47 in this case.
It is expected that room temperature deposited SiO films have a high density of defects. Furthermore, O− ion may be accelerated towards the sample and create defects on collisions with atoms within the film [22]. Thus, as the film grows one must expect that a high density of vacancies is generated throughout the film. One could imagine that the generation of defects by ions impinging from a direction perpendicular to the surface could lead to a second-order nonlinearity. However, these defects are expected to have a considerably higher activation energy than the 0.36 eV (Fig. 5(a)) observed here. This energy is more consistent with desorption of foreign species attached to the defects. Such adsorbed species would form dipoles that could be oriented by charges on the film surface. SIMS measurements (not showed here) performed on a 200-nm thick SiO film showed the presence of hydrogen atoms in a concentration of 2 hydrogen atoms per 13 silicon atoms. Thus, these vacancies may be passivated by hydrogen.
The necessity for defect generation in order to obtain second-order nonlinearities in glasses has been demonstrated by Kameyama and Yokotani [23]. They showed that introduction of defects into highly pure silica glasses (using a KrF excimer laser) was necessary in order to create a nonlinear optical response in the subsequent thermal poling process. Moreover, the presence of hydrogen in the glass films was important for the magnitude of the nonlinearity.
The effect of foreign ions on defects in oxide films has been studied by Argalland and Jonscher [24] through electrical measurements on thermally deposited silicon monoxide. They found a polarization of the films which they ascribed to hydrogen atoms bound to silicon. Moreover, they found activation energies around 0.5 eV for the induced polarization. The activation energy found in the present work (Fig. 5(a)) is thus comparable to those obtained from electrical measurements. In addition to Si-H bonds also Si-OH and Si-O bonds are expected in the films, but due to the relatively low observed activation energy (0.1–0.36 eV) it is expected that Si-H dipoles dominate the nonlinearity. However, since two activation energies are found more than one contribution may be present, e.g. desorption of hydrogen from Si-H and Si-OH species.
A second-order nonlinearity requires orientation of the bonds generated at defect sites. This orientation may be driven by a field generated in a way similar to that described for corona poling. In this process ions generated in the gas above the thin film are accelerated towards the film where they are deposited and create an internal field across the film that orients bonds to impurity atoms [25]. Following this analogy, the O− ions in the e-beam deposition process renders the film surface negatively charged, resulting in process similar to that of a corona poling. Thus, the O− ions both create defects and deliver the charges leading to orientation of the dipoles at defect sites.
In order to further investigate the origin of SHG a 160-nm thick SiO film was rf-magnetron sputtered onto a fused silica substrate. Here the rf-field traps the electrons and the argon ion plasma is close to the target. Thus no charging of the substrate takes place. The sample appeared almost transparent with an energy band gap of Eopt ≤ 4.2 eV and no measurable nonlinear response. The same was the case for a 250-nm thick SiO2 film e-beam deposited without the retarding grid. Moreover, it appeared as transparent as fused silica. Furthermore, while deposition of a new SiO2 film with the retarding grid no blueish plasma was visible even with a fiber spectrometer. It is suggested that the lack of a nonlinear response from the e-beam deposited SiO2 film is due to saturation of Si bonds with oxygen atoms in a dense structure that does not allow formation of oriented Si-H dipoles.
Ning et al. [6] found a strong second order nonlinear susceptibility of 2.5 pm/V in silicon nitride films grown at 300°C by PECVD onto fused silica. They suggested that very small Si nanocrystals with Si-Si dimers showing an anharmonic response to the applied field are responsible for the strong SHG. If Si nanocrystals are formed in the SiO films in the present work the orientational mechanism leading to a macroscopic nonlinear coefficient could be the oxygen ions as described above. However, a signature of nanocrystal formation in glass is photoluminescence which was only observed after annealing of the room temperature grown films at temperatures were the SHG signal had disappeared. This does not support formation of nanocrystals as the origin of the SHG response in the SiO films.
The observations described above lead us to suggest that the nonlinearity of the SiO film is related to the formation of oriented dipoles. The dipoles are formed at defects created during film growth. The dipoles are oriented perpendicular to the film surface by the negative charging of the sample from the O− ions during deposition, as sketched in Fig. (7) Thus, as the film grows the oxygen ions at the surface maintain the preferred orientation of the dipoles resulting in a non-centrosymmetric film, i.e. the orientation of the dipoles takes place continuously during the deposition due to the dc- field across the growing surface. This would be consistent with the quadratic dependence of SHG with film thickness.
Fig. 7 A sketch illustrating oriented dipoles arising due to a charging of the film during deposition caused by oxygen ions. Drawing inspired from [24]. The sketch is based on a SiO2 structure since is the difficult to represent an amorphous SiO structure.
The dc-field generated in the SiO film will also give rise to an effective second order nonlinearity through the third order susceptibility as describe for thermal poling [26] and electron beam irradiation of glass [8–10]. Assuming constant third-order susceptibility this contribution can give a quadratic dependence on film thickness only if the created dc-field is independent of thickness. The separated opposite charge distributions giving rise to this field are expected to be concentrated at the interfaces. To compensate for the growing separation between interfaces the charge densities should grow proportional to the film thickness. It is not clear how such interface charge layers can build up. It is thus suggested that the dipole orientation described above dominates the nonlinear response of the film.
5. 2D structures
One of the advantages of direct deposition of a nonlinear film could be the possibility of defining structures on substrates by lithography. In order to demonstrate this possibility a structure was defined by UV lithography in a 2-μm thick photoresist. After development a 200-nm thick SiO film was deposited and a lift-off process left 200-μm wide SiO structures on the fused silica substrate. Figure 8(a) shows a SHG scan with 130x100 points recorded with a focused beam from a mode-locked Ti:Sapphire laser delivering 100-fs p-polarized pulses at a wavelength of 786 nm. Again colored glass filters were used to separate the excitation line from the SHG signal. The SHG signal was measured with a photomultiplier coupled to a photon counter. The SHG scan shows a structure corresponding to that defined with lithography. A number of defects appearing in the scan are most likely due to poor lift-off.
Fig. 8 SHG from SiO structures defined by lithography. a) SHG scan from a well-defined structure with 130x100 points recorded with a focused beam from a Ti:Sapphire laser operating at a wavelength of 786 nm. b) The SHG diffracted signal measured from a ∼90-nm thick SiO grating consisting of ∼1.4 μm wide lines that are spaced d ∼3-μm apart, see the inserted SEM image, scale bar is 10 μm wide.
Finally, we demonstrate by using e-beam lithography, that it is possible to define well-ordered nano-sized nonlinear structures. A 350-nm thick 3.5 % PMMA film was spin-coated onto a cleaned fused silica substrate and sputtering a thin aluminum layer on top made it possible to ground the sample and thus perform e-beam lithography. After the lithography process the aluminum was removed in a 30 % KOH solution and the structrue was developed. Then a 90-nm thick SiO film was e-beam deposited on the sample and after the deposition the PMMA was removed with acetone, thus resuling in the structure shown in the SEM image in the insert in Fig. 8(b). The lines are 1.4 μm wide and with a period d ∼ 3 μm resulting in a linear- and a nonlinear diffracting SiO grating.
Figure 8(b) shows the measured SHG diffracted signal from the SiO grating. Again we used the mode-locked Ti:Sapphire laser and a photomultiplier coupled to a photon counter. The SHG photons were measured during rotation of the photomultiplier behind the SiO diffracting grating. The green stars indicate the predicted maxima for the m = −2, −1, 0, 1, 2 and 3 order modes by using the grating equation
Here λp/2 is the wavelength of the SHG signal. In the experiment the incident angle is θi = 30° and θm are the angles of the m’th order of diffraction. It is clear that the nonlinear response of the SiO film is intact after formation of nanoscale structures.6. Conclusion
It has been demonstrated that e-beam deposition of SiO is an simple way to synthesize nonlinear thin films with a built-in second order nonlinearity that is comparable in magnitude to that of crystalline quartz. The nonlinearity is stable at room temperature, but decays at elevated temperatures. Spectroscopic measurements show that the nonlinear response has a resonance corresponding to the band gap at 3.3 eV found in linear absorption spectra. The resonance is however broad and leads to appreciable SHG for pump wavelengths from the near-infrared through the visible region. It is suggested that the nonlinearity is caused by Sin+– H− dipole pairs oriented in the direction perpendicular to the surface by charging of the film during the e-beam deposition. Direct formation of nonlinear structures with in-plane dimensions of a few micrometers by a lift-off process has been demonstrated. It has thus been shown that nonlinear optical components with sizes in the region of interest for planar integrated optics technology may be fabricated by e-beam deposition of SiO.
Acknowledgments
We thank John Lundsgaard Hansen, Department of Physics and Astronomy at Aarhus University (AU), for the assistance with the RBS and SIMS measurements. A thanks to Peter Kjær Kristensen at the Department of Physics and Nanotechnology in Aalborg for assistance with EBL. Finally, a special thanks to Arne Nylandsted Larsen (AU) for the fruitful discussions concerning the microscopic structure of the films.
References and links
1. R. A. Myers, N. Mukherjee, and S. R. J. Brueck, “Large second-order nonlinearity in poled fused silica,” Opt. Lett. 16, 1732–1734 (1991). [CrossRef] [PubMed]
2. K. Pedersen, S. I. Bozhevolnyi, J. Arentoft, M. Kristensen, and C. Laurent-Lund, “Second-harmonic scanning optical microscopy of poled silica waveguides,” J. Appl. Phys. 88, 3872–3878 (2000). [CrossRef]
3. M. Guignard, V. Nazabal, J. Troles, F. Smektala, H. Zeghlache, Y. Quiquempois, A. Kudlinski, and G. Martinelli, “Second-harmonic generation of thermally poled chalcogenide glass,” Opt. Express 13, 789–795 (2005). [CrossRef] [PubMed]
4. M. Dussauze, E. Fargin, M. Lahaye, V. Rodriguez, and F. Adamietz, “Large second-harmonic generation of thermally poled sodium borophosphate glasses,” Opt. Express 13, 4064–4069 (2005). [CrossRef] [PubMed]
5. K. Yadav, C.L Callender, C.W. Smelser, C. Ledderhof, C. Blanchetiere, S. Jacob, and J. Albert, “Giant enhancement of the second harmonic generation efficiency in poled multilayer silica glass structures,” Opt. Express 26975–26983 (2011). [CrossRef]
6. T. Ning, H. Pietarinen, O. Hyvrinen, J. Simonen, G. Genty, and M. Kauranen, “Strong second-harmonic generation in silicon nitride films,” Appl. Phys. Lett. 100, 161902 (2012). [CrossRef]
7. F. Iacona, G. Franzo, and C. Spinella, “Correlation between luminescence and structural properties of Si nanocrystals,” J. Appl. Phys. 87, 1295–1303 (2000). [CrossRef]
8. P. G. Kazansky, A. Kamal, and P. St. J. Russell, “High second-order nonlinearities induced in lead silicate glass by electron-beam irradiation,” Opt. Lett. 18, 693–695 (1993). [CrossRef] [PubMed]
9. Q. Liu, X. Zhao, K. Tanaka, A. Narazaki, K. Hirao, and F. Gan, “Second-harmonic generation in GeAsS glasses by electron beam irradiation and analysis of the poling mechanism,” Opt. Commun. 198, 187 (2001). [CrossRef]
10. Q. Liu, B. Poumellec, R. Blum, G. Girard, J.-E. Boure, A. Kudlinski, and G. Martinelli, “Stability of electron-beam poling in N or Ge-doped H:SiO2 films,” Appl. Phys A 81, 1213 (2005). [CrossRef]
11. G. Myburg and F. D. Auret, “Influence of the electron-beam evaporation rate of PT and the semiconductor carrier density on the characteristics of PT/normal-GAAS schottky contacts.,” J. Appl. Phys. 71, 6172–6176 (1992). [CrossRef]
12. D. Hoffman and D. Leibowitz, “Effect of Substrate Potential on Al2O3 Films Prepared by Electron Beam Evaporation,” J. Vac. Sci. Technol. 9, 326–329 (1972). [CrossRef]
13. G. E. Jellison and F. A. Modine, “Parameterization of the optical functions of amorphous materials in the inter-band region,” Appl. Phys. Lett. 69, 371–373 (1996). [CrossRef]
14. W. N. Herman and L. M. Hayden, “Maker fringes revisited: second-harmonic generation from birefringent or absorbing materials,” J. Opt. Soc. Am. B 12, 416–427 (1995). [CrossRef]
15. J. Jerphagnon and S. K. Kurtz, “Maker Fringes: A Detailed Comparison of Theory and Experiment for Isotropic and Uniaxial Crystals,” J. Appl. Phys. 41, 1667–1681 (1970). [CrossRef]
16. P.G. Kazansky and P.St.J. Russel, “Thermally poled glass: frozen-in electric field or oriented dipoles?,” Opt. Commun. 110, 611–614 (1994). [CrossRef]
17. J. Tauc, “Optical properties of non-crystalline solids,” F. Abeles (Ed.), Optical Properties of Solids (North-Holland, Amsterdam, 1972), p. 277.
18. K. Hagimoto and A. Mito, “Determination of the second-order susceptibility of ammonium dihydrogen phosphate and α-quartz at 633 and 1064 nm,” Appl. Opt. 34, 8276–8282 (1995). [CrossRef] [PubMed]
19. R. C. Miller, “Optical second harmonic generation in piezoelectric crystals.,” Appl. Phys. Lett. 5, 17–19 (1964). [CrossRef]
20. R. W. Boyd, Nonlinear Optics, 3rd Ed. (Elsevier, 2008).
21. C. Bucci and R. Fieschi, “Ionic Thermoconductivity. Method for the Investigation of Polarization in Insulators,” Phys. Rev. Lett. 12, 16–19 (1964). [CrossRef]
22. SRIM simulation software based on: J. F. Ziegler and J. M. Manoyan, “The Stopping of Ions in Compounds,” Nucl. Instr. Meth. B35, 215–228 (1989).
23. A. Kameyama, A. Yokotani, and K. Kurosawa, “Identification of defects associated with second-order optical nonlinearity in thermally poled high-purity silica glasses,” J. Appl. Phys. 89, 4707–4713 (2001). [CrossRef]
24. F. Argalland and A. K. Jonscher, “Dielectric properties of thin films of aluminium oxide and silicon oxide,” Thin Solid Films 2, 185–210 (1968). [CrossRef]
25. H. L. Hampsch, J M. Torkelson, S J. Bethke, and S G. GrubbSecond harmonic generation in corona poled, doped polymer films as a function of corona processing,” J. Appl. Phys. 67, 1037–1042 (1990). [CrossRef]
26. T. G. Alley, S. R. J. Brueck, and R. A. Myers, “Space charge dynamics in thermally poled fused silica,” J. Non-Cryst. Solids 242, 165–176 (1998). [CrossRef]
