Open Access
Add to BibSonomy
Abstract
We demonstrate for the first time functional arrayed waveguide gratings (AWGs) fabricated using the femtosecond laser direct-write technique. This fabrication technique is a mask-less alternative to lithography enabling design flexibility and rapid prototyping. It is ideal for customized small scale production for new applications. The devices were demonstrated in the visible region at 632.8 nm with a measured free spectral range (FSR) of 22.2 nm, and 1.35 nm resolution. To highlight the advantages of using a 3-dimensional fabrication technique, a 3-port photonic lantern was integrated with an AWG in a single monolithic chip. Integration of this type is not feasible with lithography-based AWG fabrication and can increase the functionality of AWGs for sensing applications.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Arrayed waveguide gratings (AWGs) were first proposed in 1988 by M. K. Smit as compact (de)multiplexers for wavelength division multiplexing [1]. Since inception, driven by the telecommunication industry, AWGs fabricated using lithography have been extensively engineered, to become a highly reliable, compact, and low loss device. These attributes have enabled AWGs to be successfully applied in other integrated sensing applications such as: spectral domain optical coherence tomography [2], miniaturized Raman spectroscopy [3], compact biomedical sensors [4], and as astronomical spectrographs [5].
Although AWGs have proven themselves in various applications, uptake is restricted by development costs and time; or restricted to spectral regions covered by commercial lithographic telecommunication AWGs. Therefore, a rapid and mask-less AWG prototyping platform is highly desirable. Previously we have demonstrated the fabrication of key AWG components using the femtosecond laser direct-write technique [6]. Now with suitable component integration we have developed fully functional AWGs in the visible spectral region; a region attractive for material and analyte sensing [7], as well as sensing in astronomy. However due to the flexible nature of the fabrication process, the operational wavelength of the AWG can be tailored within the spectral transmission of the substrate material for other applications.
AWGs are inherently single-mode devices. However, many applications require multimode fibers for efficient light collection, which leads to large coupling losses. A solution is to utilize a device known as a photonic lantern [8–10]. This device efficiently converts N-modes of a multimode fiber into N individual single-mode cores. Each core can then be individually injected into an AWG. While each single-mode waveguide from a photonic lantern could be injected into an individual AWG, the number of AWGs required for large multimode inputs would reduce the inherent benefits in terms of cost and volume. An alternative option is to launch multiple single-mode fibers simultaneously into a single AWG [11]. This was demonstrated when 12 single-mode outputs from a photonic lantern were fiber coupled to a lithographic AWG [5]. While this device was successful it was fragile, suffered transmission losses from multiple fiber splices, and large preparation time. More recently N. Cvetojevic et al. (2017) directly coupled an integrated photonic lantern and AWG, minimising losses and preparation time [12]. Here we take this further by demonstrating for the first time the integration of a photonic lantern and AWG into a single robust monolithic chip.
2. AWG operation
An AWG works as follows (see Fig. 1): light is injected via a single-mode waveguide into a large free propagation zone (FPZ). The FPZs are large index modified regions enabling light to freely diffract in the horizontal plane while being confined in the vertical. This slab mode is then captured in phase by an array of tapers. These tapers adiabatically guide the slab mode into a waveguide array, reducing the mode mismatch and thus losses. Each waveguide of the array is incrementally longer than the adjacent, creating a phase tilt across the waveguide array similar to a bulk diffraction grating. These waveguides are then injected into a second FPZ where each wavelength of light constructively interferes, creating a horizontally dispersed spectrum at the output. An AWG is analogous to a diffraction grating, and depending on the design can have multiple diffraction orders across the output. For a more detailed explanation of AWG operation see [13].
Fig. 1 Schematic of an arrayed waveguide grating with its individual components. This schematic demonstrates how the AWG’s input and outputs have been aligned perpendicular with the glass samples edges for simple post-processing and characterisation.
3. Fabrication
The AWGs were fabricated using an ultrafast Ti:sapphire oscillator (FEMTOSOURCE XL 500, Femtolasers GmbH), which emits 50 fs pulses with a repetition rate of 5.1 MHz centred at 800 nm. Using a 40 ×, 0.65 NA microscope objective, light was focused at a depth of 170 μm into an alkaline earth boro-aluminosilicate glass sample (Corning Eagle 2000). Due to non-linear absorption at the focal point, the refractive index of the glass can be locally modified [14]. By moving the sample on a set of Aerotech 3-axis air-bearing translation stages, the position of the focal point within the sample is moved leaving a refractive index modified region.
Modifications are written with a pulse energy of 55 nJ and a translation speed of 2000 mm/min. These modifications create single-mode waveguides at 633 nm with a width of 4.8 ± 0.2 μm and a mode-field diameter of 7.3 × 8.1 μm (1/e2). The waveguides exhibit propagation losses of 0.82 dB/cm. A peak refractive index contrast of 1.5 × 10−3 was determined using the inverse Helmholtz technique [15]. These modifications can be multi-scanned to form slab waveguides that can act as FPZs and taper regions. Using a multi-scan pitch of 0.4 μm, uniform slabs with a refractive index uniformity of 1.97 % (standard deviation) were created. The high translation speed enables the inscription of a single AWG in approximately 100 minutes. This corresponds to 2,300 scans across the sample to create the FPZs. The total fabrication time could be further reduced by using a spatial light modulator to focus multiple beams into the material simultaneously [16]. The solitary AWG device demonstrated has been shown to work 1.5 years after initial fabrication suggesting the modification is permanent. A more in-depth discussion on the fabrication details can be found in our previous paper [6].
4. Design
The AWGs demonstrated in this paper utilize a simple horse-shoe design (see Fig. 1). The total chip footprint is 35.5 × 4.3 mm, the input and output FPZs are 7.7 mm long and 0.9 mm wide. Due to the low refractive index contrast, the waveguides have a weakly confined mode. This low mode confinement leads to coupling between waveguides which becomes even more severe in the bends of the array. Such cross talk is detrimental to device operation, thus the minimum waveguide spacing in the waveguide array has to be kept larger than 35 μm to ensure that no cross coupling occurs. The waveguide array has 19 waveguides spaced at 40 μm with a physical path length difference between adjacent waveguides of 11.76 μm. The design specifications are as follows: FSR of 22.6 nm, resolving power of 532, central diffraction order 28th and a central wavelength of 632.8 nm. The angle between the two FPZs is 12.4 degrees. This small angle maintains laser writing in one direction, reducing modification variations caused by laser pulse front tilt [17]. It also reduces the length of the waveguide array when using large bend radii and enables input and output surfaces to be parallel to simplify testing.
To minimize bend loss in the waveguide array, the losses of free-standing curved waveguides were measured. Two opposing arcs each subtending 30 degrees with radii between 10 and 40 mm were inscribed into a Corning Eagle 2000 sample. Straight waveguides were extended from the arcs to the ends of the sample. A Thorlabs fiber-coupled 635 nm diode laser was launched via a single-mode SM600 fiber into the waveguides. The output was imaged onto a camera and captured by the Spiricon LBA-PC software. The propagation loss and the mode-field mismatch were removed to calculate the bend loss, as shown in Fig. 2. There is good agreement between the experimental measured bend losses and the bend losses calculated using RSoft BeamProp for a 4.8 μm wide step-index waveguide with 1.5 × 10−3 index contrast. From these results a minimum bend radius of 26 mm was chosen for the waveguide array. This bend radius has a simulated loss of 0.050 dB/cm, equating to a total bend loss of 0.024 dB. The total length of the minimum radius arrayed waveguide is 12.52 mm, resulting in a total loss of 1.05 dB.
5. Results and discussion
The fabricated devices were tested by launching a 632.8 nm HeNe laser via a single-mode fiber into the AWG input waveguide. The near-field output was imaged by a microscope objective onto a Pulnix TM-745E camera to capture the output profile. To obtain a single profile the 2-dimensional output was traced, summed and normalized as a single profile. The experimental AWG output near-field profiles were then compared to a full beam propagation model (RSoft BeamPROP) of the device, as shown in Fig. 3. At the designed central wavelength of 632.8 nm the simulated output consists of a central diffraction order, surrounded by symmetrically decreasing diffraction orders confined by a diffraction envelope. However, for the fabricated device the central wavelength is shifted by 7.7 nm or physically 28 μm at the FPZ output surface. This central wavelength shift has two plausible causes: a constant path length error in the waveguide array, or a lateral beam drift in the free propagation zone. To account for a 7.7 nm central wavelength shift the physical path length error that needs to be applied to each waveguide of the array is 145.8 nm, totalling 2.77 μm across 19 waveguides. A length error on this scale is challenging to measure and a plausible manufacturing error. Furthermore, if the optical path length error was due to a miscalculation in the waveguide index contrast, an unrealistic index contrast of 2.12 × 10−2 is required.
Fig. 3 Comparison between the experimental AWG output for 632.8 nm and theoretical 640.5 nm. The theoretical diffraction envelope for a 40 and 42 μm wide taper have been added to guide the eye.
To test for lateral propagation five identical FPZs with a single-mode waveguide for light injection were fabricated. Each FPZ has a second single-mode waveguide written 30 μm above the centre of the FPZ. 632.8 nm light was injected into the FPZs and the near-field outputs were fitted with a Gaussian. The displacement between the centre of the diffraction envelope and the centre of the FPZ, indicated by the single-mode waveguide above, is the measured lateral drift. Four FPZs had a lateral displacement of 12–14 μm (FPZs were written left to right causing a lateral propagation drift to the right-hand side), while one FPZ showed no lateral drift. While this lateral displacement was in the correct direction, the overall lateral drift was not large enough to fully account for the central wavelength shift. Therefore the shift of the central wavelength is most likely a combination of lateral propagation and path length error in the array. It is also noted that this lateral drift is not consistent, conceivably due to pulse energy variations during the fabrication process.
The shape of an AWG diffraction envelope is determined by the far-field of the output taper [13]. The Simulated diffraction envelope of a 42 μm wide step-index taper was found to fit the experimental output profile, seen in Fig. 3. While the physical width of the tapers are 40.3 ± 0.9 μm. The variation is due to the fabricated taper not having a perfect step-index profile, thus exhibiting a slightly different mode-profile and far-field profile.
The experimental spacing between the diffraction orders match simulations well, it is noted that the outer diffraction orders are offset to the left; the cause of this misalignment is unknown. The experimental output shown in Fig. 3 has noise between the diffraction orders. This is due to random phase errors in the waveguide array and/or refractive index irregularities in the FPZs. To estimate the phase errors in the fabricated device the fraction of total power contained as noise between the diffraction orders was compared to a full beam propagation model with a random uniformly distributed phase error across the waveguide array. It was found that the phase error is equivalent to random length errors uniformly distributed between ±95 nm.
To measure the AWGs output spectral response a HeNe laser and two tunable single-mode laser diodes were injected (Mitsubishi 638 nm/150 mW and Oclaro 642 nm/150 mW). For reference, the spectra of the laser diodes where recorded using a commercial spectrograph (Acton 2500i, Camera Pixis100 model 7515-0001; Princeton Instruments). The output spectra for wavelengths ranging from 632.8–645.6 nm were normalized and plotted in Fig. 4. The spectra were fitted with a Gaussian and the peak position and wavelength was recorded. Using this data, the FSR was extrapolated to be 22.2 ± 0.5 nm, compared to the designed FSR of 22.6 nm. At 632.8 nm the physical FWHM is 4.9 ± 0.1 μm and the associated spectral FWHM is 1.35 ± 0.03 nm yielding a resolving power R of 468.7, in reasonable agreement with the design R of 532. The discrepancy between the measured and theoretical R is attributed to manufacturing imperfections and the fact that the waveguide array is illuminated by an approximately Gaussian field rather than an ideal flat top which reduces the resolving power [18]. Birefringence in AWG components are known to cause a small polarization dependent wavelength shift [19], while laser written waveguides have also been shown to exhibit birefringence [20]. With a waveguide birefringence of 2 × 10−5 a polarization dependent wavelength shift of 0.01 nm is expected [21]. However, due to the low resolving power of the AWG, polarization dependency can be neglected.
Fig. 4 Emission spectra of two tunable laser diodes (638 and 642 nm) recorded with the fabricated AWG. The spectra correspond to the central diffraction order of the AWG. The 642 nm laser diode exhibits a broader emission spectrum.
To measure the AWG throughput, a Thorlabs fiber coupled 635 nm diode laser source was launched into the AWG input waveguide via a single-mode fiber. This source was used due to its stability. The output was again imaged onto a camera. The power of each of the 5 main diffraction orders was calibrated against the input power. By numerically evaluating the mode overlap integral a 14.48% coupling loss was removed. The AWG features a throughput of 11.5 ± 0.2 % summed across 5 orders, and a throughput of 3.97 ± 0.04 % for the central 28th diffraction order. The two main sources of losses are propagation losses in the waveguides and non-optimized mode conversion in the adiabatic taper. Future work will look to minimize propagation losses in the waveguide array. While taper losses can be reduced by decreasing the taper width. Narrower tapers can more efficiently convert the diffracted light from the first FPZ into a single-mode. However, the minimum spacing of the waveguides is limited by cross coupling in the waveguide array. This could be improved by using a 3-dimensional waveguide array, where the waveguides are alternatively displaced upward and downward avoiding cross coupling.
6. Photonic lantern AWG integration
The laser direct-write technique has the intrinsic benefit of facilitating 3-dimensional fabrication. This enables devices such as the 3-dimensional photonic lantern to be integrated on-chip with an AWG. To demonstrate this capability a simple 3-moded photonic lantern, as shown in Fig. 5(a), was integrated with an AWG. This integration enables AWGs to be compatible with multi-mode fibres in a single robust chip, reducing processing time and improving device stability compared to fabricating these devices separately. Before discussing photonic lantern design, the implications of multiple AWG inputs will be discussed. AWGs are re-imaging devices, hence shifting the input position from the center of the FPZ results in a physical shift of the output spectrum [18, 22]. Therefore, if two inputs were injected simultaneously the output would contain the spectrum of both inputs offset by an amount dependent on the injection spacing. The maximum spacing between inputs is limited by the physical FSR of the device. If the FSR is exceeded neighbouring diffraction orders overlap and become spectrally indistinguishable. There are two options for separating each input: cross dispersion which requires free space optics, or wavelength filtering which reduces the FSR depending on the number of inputs.
Fig. 5 a) 3 port photonic lantern schematic. The photonic lantern is designed to inscribe the waveguides in the following order −35, 0, +35 μm offset. This ensures that the lower waveguides are inscribed first to avoid focusing through previous modifications. b) End-on bright field image of the multimode region consisting of 3 waveguides radially separated by 2.5 μm. c) Top-down differential interference contrast microscope images of the photonic lantern AWG interface. The inputs are separated by 35 μm, and the central input is aligned with the FPZ central axis. d) End-on image of the photonic lanterns output, the 3 single-mode waveguides are spaced by 35 μm. e) Near-field output from the photonic lantern characterized at 632.8 nm.
The photonic lantern is 15 mm long and consists of 3 regions: multimode waveguide, multimode to single-mode transition, and re-mapper. The multimode region is 1 mm long and consists of 3 waveguides in a triangular arrangement spaced 2.5 μm radially from the centre (Fig. 5(b)). This multimode region has a similar structure and mode-profile to the 3 × 1 conventional fibre lantern demonstrated by Leon-Saval et al. [23]. The transition region is 7 mm long, over which the radial spacing of the waveguides increase to 30 μm. Over the next 7 mm the 3 waveguides are remapped to the horizontal plane with a spacing of 35 μm (70 μm outer waveguide spacing), for injection into the AWG’s input FPZ (Fig. 5(c–e)). This spacing was chosen as it avoids coupling between inputs and utilizes a majority of the AWGs 80.7 μm physical FSR. To increase the number of photonic lantern inputs, AWGs can be designed with a larger physical FSR. Due to spherical aberrations laser inscribed waveguide properties change as a function of depth. To calibrate for these changes waveguides of various pulse energies were fabricated at the minimum and maximum photonic lantern depths. A linear fit was then used to control the laser power as a function of laser inscription depth.
The integrated photonic lanterns and AWGs were tested by injecting 632.8 nm light into the multimode input of the photonic lantern. The near-field output was recorded using the same method described above. Experimental AWG near-field profiles were then compared to a full beam propagation model of an AWG device with 3 inputs spaced at 35 μm, as shown in Fig. 6. The position of the diffraction orders align with the simulated positions, while the intensity of the diffraction orders only loosely follows the simulation. This can be partly explained as the simulation assumes that the power of each input from the photonic lantern is equal. However, in this example the three waveguides of the photonic lantern were not evenly illuminated.
Fig. 6 Beam propagation modelling showing the designed optical output at 632.8 nm for a 3-port photonic lantern injection into an AWG, compared to the experimental optical response. The two spectral lines close together are from the input waveguides spaced at −35 and +35 μm, respectively. While the lone spectral line was injected at the centre of the AWG.
7. Conclusion
We successfully demonstrated for the first time the fabrication of a functional laser written arrayed waveguide grating. The fabricated AWG has a resolving power of 468.7 and FSR of 22.2 ± 0.5 nm, which agrees with the design values of 532 and 22.6 nm. Currently the designed central wavelength 632.8 nm and measured are offset by 7.7 nm, which we attribute to a lateral propagation drift and a manufacturing path length error. The device throughput is 3.97% for the central order and 11.5% across 5 orders. Future work will look to improve this value. The mask-less fabrication technique is also ideal for prototyping as it enables design flexibility and rapid device production, with devices taking 100 minutes to fabricate. Next we integrated a 3-moded photonic lantern and AWG into a single monolithic chip. This demonstrates the ability to make AWGs compatible with multimode fibers. Whilst also removing coupling losses and improving device stability, robustness and processing time compared to fabricating the devices separately. Integration of 3-dimensional devices of this type is not feasible with lithography-based AWG fabrication.
Funding
Australian Research Council Centre of Excellence for Ultrahigh bandwidth Devices for Optical Systems (CUDOS)(CE110001018); Australian Research Council Discovery Early Career Researcher Award (DECRA)(DE160100714).
Acknowledgments
This research was supported by the Australian Research Council Centre of Excellence for Ultrahigh bandwidth Devices for Optical Systems and was performed in part at the OptoFab node of the Australian national Fabrication Facility utilizing Commonwealth as well as NSW and SA state government funding. G. Douglass acknowledges the support of the MQRES scholarship. S. Gross was supported by an ARC Discovery Early Career Researcher Award.
References and links
1. M. K. Smit, “New focussing and dispersive planar component based on an optical phased array,” Electron. Lett 24(7), 385–386 (1988). [CrossRef]
2. B. I. Akca, V. D. Nguyen, J. Kalkman, N. Ismail, G. Sengo, F. Sun, A. Driessen, T. G. van Leeuwen, M. Pollnau, K. Wörhoff, and R. M. de Ridder, “Toward Spectral-Domain Optical Coherence Tomography on a Chip,” IEEE J. Sel. Top. Quantum Electron. 18(3), 1223 (2012). [CrossRef]
3. N. Ismail, L. P. Choo-Smith, K. Wörhoff, A. Driessen, A. C. Baclig, P. J. Caspers, G. J. Puppels, R. M. de Ridder, and M. Pollnau, “Raman spectroscopy with an integrated arrayed–waveguide grating,” Opt. Lett. 36(23), 4629 (2011). [CrossRef] [PubMed]
4. Y. Komai, H. Nagano, K. Kodate, K. Okamoto, and T. Kamiya, “Application of arrayed-waveguide grating to compact spectroscopic sensors,” Jpn. J. Appl. Phys. 43 (8B), 5795–5799 (2004). [CrossRef]
5. N. Cvetojevic, N. Jovanovic, C. Betters, J. S. Lawrence, S. C. Ellis, G. Robertson, and J. Bland-Hawthorn, “First starlight spectrum captured using an integrated photonic micro-spectrograph,” Astron. Astrophys. 544, L1 (2012). [CrossRef]
6. G. Douglass, F. Dreisow, S. Gross, S. Nolte, and M. J. Withford, “Towards femtosecond laser written arrayed waveguide gratings,” Opt. Express 23(16), 21392 (2015). [CrossRef] [PubMed]
7. K. Suzuki, Y. Hida, T. Shibata, Y. Inoue, H. Takahashi, and K. Okamoto, “Silica-based arrayed-waveguide gratings for the visible wavelength range,” NTT Tech. Rev. 4(6), 48–51 (2006).
8. S. G. Leon-Saval, T. A. Birks, J. Bland-Hawthorn, and M. Englund, “Multimode fiber devices with single-mode performance,” Opt. Lett. 30(19), 2545–2547 (2005). [CrossRef] [PubMed]
9. R. R. Thomson, T. A. Birks, S. G. Leon-Saval, A. K. Kar, and J. Bland-Hawthorn, “Ultrafast laser inscription of an integrated photonic lantern,” Opt. Express 19(6), 5698–5705 (2011). [CrossRef] [PubMed]
10. N. Jovanovic, I. Spaleniak, S. Gross, M. Ireland, J. S. Lawrence, C. Miese, A. Fuerbach, and M. J. Withford, “Integrated photonic building blocks for next-generation astronomical instrumentation I: the multimode waveguide,” Opt. Express 20(15),17029 (2012). [CrossRef]
11. J. Bland-Hawthorn, J. Lawrence, G. Robertson, S. Campbell, B. Pope, C. Betters, S. Leon-Saval, T. Birks, R. Haynes, N. Cvetojevic, and N. Jovanovic, “PIMMS: photonic integrated multimode microspectrograph,” Proc. SPIE 7735, 77350N (2010). [CrossRef]
12. N. Cvetojevic, N. Jovanovic, S. Gross, B. Norris, I. Spaleniak, C. Schwab, M. J. Withford, M. Ireland, P. Tuthill, O. Guyon, F. Martinache, and J. S. Lawrence, “Modal noise in an integrated photonic lantern fed diffraction-limited spectrograph,” Opt. Express 25(21), 25546–25565 (2017). [CrossRef] [PubMed]
13. M. K. Smit and C. Van Dam, “PHASAR-Based WDM-Devices Principles, Design and Applications,” IEEE J. Sel. Top. Quantum Electron. 2(2), 236–250 (1996). [CrossRef]
14. K. M. Davis, K. Miura, N. Sugimoto, and K. Hirao, “Writing waveguides in glass with a femtosecond laser,” Opt. Lett. 21(21), 1729 (1996). [CrossRef] [PubMed]
15. I. Mansour and F. Caccavale, “An improved procedure to calculate the refractive index profile from the measured near-field intensity,” J. Light. Technol. 14(3), 423–428 (1996). [CrossRef]
16. M. Sakakura, T. Sawano, Y. Shimotsuma, K. Miura, and K. Hirao, “Fabrication of three-dimensional 1 × 4 splitter waveguides inside a glass substrate with spatially phase modulated laser beam,” Opt. Express 18(12), 12136 (2010). [CrossRef] [PubMed]
17. P. G. Kazansky, W. Yang, E. Bricchi, J. Bovatsek, A. Arai, Y. Shimotsuma, K. Miura, and K. Hirao, “‘Quill’ writing with ultrashort light pulses in transparent optical materials,” Appl. Phys. Lett. 90, 151120 (2007). [CrossRef]
18. N. Cvetojevic, N. Jovanovic, J. Lawrence, M. Withford, and J. Bland-Hawthorn, “Developing arrayed waveguide grating spectrographs for multi-object astronomical spectroscopy,” Opt. Express 20(3), 2062 (2012). [CrossRef] [PubMed]
19. X. Xia, J. Zou, T. Lang, and J. J. He, “Experimental demonstration of birefringence compensation using angled star couplers in silica-based arrayed waveguide grating,” IEEE Photonics J. 4(6), 2236–2242 (2012). [CrossRef]
20. T. Meany, M. Gräfe, R. Heilmann, A. Perez-Leija, S. Gross, M. J. Steel, M. J. Withford, and A. Szameit, “Laser written circuits for quantum photonics,” Laser Photonics Rev. 9(4),363–384 (2015). [CrossRef]
21. G. Corrielli, A. Crespi, R. Geremia, R. Ramponi, L. Sansoni, A. Santinelli, P. Mataloni, F. Sciarrino, and R. Osellame, “Rotated waveplates in integrated waveguide optics,” Nat. Commun. 5, 1–6 (2014). [CrossRef]
22. J. Lawrence, J. Bland-Hawthorn, N. Cvetojevic, R. Haynes, and N. Jovanovic, “Miniature astronomical spectrographs using arrayed-waveguide gratings: capabilities and limitations,” Proc. SPIE 7739, 77394I (2010). [CrossRef]
23. S. G. Leon-Saval, N. K. Fontaine, J. R. Salazar-Gil, B. Ercan, R. Ryf, and J. Bland-Hawthorn, “Mode-selective photonic lanterns for space-division multiplexing,” Opt. Express 22(1), 1036 (2014). [CrossRef] [PubMed]
