Gamma radiation effects in amorphous siliconand silicon nitride photonic devicesQINGYANG DU,1 YIZHONG HUANG,1 OKECHUKWU OGBUU,1 WEI ZHANG,1,2 JUNYING LI,1 VIVEK SINGH,1

ANURADHA M. AGARWAL,1 AND JUEJUN HU1,*1Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA2Key Laboratory of Photoelectric Materials and Devices of Zhejiang Province, Ningbo University, Ningbo, Zhejiang 315211, China*Corresponding author: hujuejun@mit.edu

Received 14 December 2016; revised 6 January 2017; accepted 10 January 2017; posted 10 January 2017 (Doc. ID 282830);published 30 January 2017

Understanding radiation damage is of significant importancefor devices operating in radiation-harsh environments. Inthis Letter, we present a systematic study on gamma radia-tion effects in amorphous silicon and silicon nitride guidedwave devices. It is found that gamma radiation increases thewaveguide modal effective indices by as much as 4 × 10−3 inamorphous silicon and 5 × 10−4 in silicon nitride at 10 Mraddose. This Letter further reveals that surface oxidation andradiation-induced densification account for the observedindex change. © 2017 Optical Society of America

OCIS codes: (350.5610) Radiation; (130.0130) Integrated optics;

(350.1820) Damage; (070.5753) Resonators.

https://doi.org/10.1364/OL.42.000587

High-energy radiation tends to create defects in materialsand, thereby, modifies their electronic and optical properties.Mitigating radiation-induced damage associated with such de-fect generation is an important design consideration for devicesused in outer space, near nuclear reactors, and close to particlecolliders. As satellite communications and data links fornext-generation particle accelerators start turning to integratedphotonics solutions [1], the demand to understand radiation-induced degradation mechanisms in photonic devices becomescritical. In crystalline optical materials such as silicon oninsulator (SOI), optical property changes occur through bothionization damage and displacement damage [2]. Ionizationleads to free carriers that modify refractive index and opticalabsorption via the plasma dispersion effect. Ionization damagealso creates trapped charges at the Si∕SiO2 interface whichlikely accounts for the phase drift observed in SOI modulators[3]. Displacement damage results in lattice defects whichimpacts optical properties through local polarizability changeand carrier density modulation due to trap states. Gammaradiation-induced effects in SOI have been characterized byboth in situ [4] and ex situ [5] measurements using opticalresonator devices. Besides the aforementioned bulk materialstructural transformations, surface oxidation was identifiedas a contributor to observed resonance spectral shift [5].

In this Letter, we focus on gamma radiation-induced effectsin amorphous silicon (a-Si) and silicon nitride (SiNx) photonicdevices. Both a-Si and SiNx have a low optical loss near1550 nm telecommunication wavelength [6]. For instance,SiNx waveguides with a remarkably low optical loss down to0.045 dB/m have been demonstrated [7]. Their amorphousnature and compatibility with standard Si CMOS processingalso qualify them as promising candidates for the back endof line photonic integration [8]. Additionally, both materialsexhibit large Kerr coefficients useful for nonlinear opticalapplications [9,10].

Radiation damage in a-Si has been investigated in electronicdevices, including field effect transistors [11], photodetectors[12], and solar cells [13]. In electronic devices, it was concludedthat ionization damage is the primary degradation mechanism[14]. For passive photonic applications, however, bulk opticalproperty changes in a-Si are likely dictated by atomic structuralperturbations resulting from displacement defects. This isdifferent from the case of crystalline silicon, because free carriereffects (plasma dispersion and free carrier absorption) areminimal in a-Si, given its small carrier mobility. Likewise, weanticipate that displacement is the main damage mechanism inSiNx materials.

On the experimental side, radiation effects on a-Si and SiNxphotonic devices have only been reported in a few recentstudies. Grillanda et al. studied gamma radiation damage ina-Si resonators covered with a silicon dioxide cladding grownby chemical vapor deposition (CVD) and found minimalresonance shift up to a 15 Mrad absorbed dose [15]. Thisobservation is likely a consequence of cladding compensationsince CVD silicon dioxide exhibits a negative refractive indexchange upon gamma ray exposure. In nitride devices, protonirradiation up to a total fluence of ∼1010 protons∕cm2 pro-duced a negligible loss change [16]. In silicon oxynitride wave-guides, a large index change in the order of 0.01 was measuredupon exposure to alpha radiation [17].

This Letter aims to provide a precise quantitative measure-ment of total dose effect of gamma irradiation on a-Si and SiNxphotonic device. Radiation-induced optical property modifica-tions are evaluated using micro-ring resonators. The modal

Letter Vol. 42, No. 3 / February 1 2017 / Optics Letters 587

0146-9592/17/030587-04 Journal © 2017 Optical Society of America

effective index change of a-Si and SiNx waveguides can becalculated from the resonant peak shift, whereas the opticalabsorption increase, if any, is inferred from the quality factor(Q-factor) and the extinction ratio change [18]. The irradiateda-Si and SiNx materials were also characterized by high-resolution transmission electron microscopy (HRTEM), graz-ing incidence x-ray diffraction (XRD), and x-ray photoelectronspectroscopy (XPS) to elucidate the underlying structural trans-formation mechanisms responsible for the observed opticalproperty modifications.

The fabrication of a-Si and SiNx resonators followed thestandard Si microfabrication processes. Thin film a-Si andsilicon nitride were first deposited onto silicon wafers with3 μm of thermal oxide using plasma-enhanced chemical vapordeposition and low-pressure chemical vapor deposition, respec-tively. The SiNx film has a refractive index of 2.1 at a 1550 nmwavelength (compared to an index of 2.0 for stoichiometricSi3N4) and, thus, is silicon-rich. The a-Si devices were pat-terned using electron beam lithography, whereas photolithog-raphy performed on an i-line stepper was used to define theSiNx device structures. Subsequently, the patterns were trans-ferred to the film by a reactive ion etch with Cl2 for a-Si and amixture of CF4 and CHF3 gases for nitride. The as-fabricateda-Si waveguide has dimensions of 450 (width) by 250 nm(height), and the SiNx waveguide has a cross section of800 × 400 nm. The Fig. 1(b) inset shows the top view of ana-Si resonator used in our experiment.

Optical transmission spectrum of the devices was character-ized on a Newport Auto Align workstation equipped withan optical vector analyzer (LUNA Technologies OVA-5000)and a built-in tunable laser. A near-infrared light was coupledin and out of the resonators through bus waveguides using ta-pered lens-tip fibers. In our measurements, we used the trans-verse electric polarization. The loadedQ-factors of as-fabricateda-Si and SiNx resonators are both around 105.

Gamma irradiation of the devices was conducted usinga Co-60 source with photon energies of 1.17 MeV and1.33 MeV. For the experiments described herein, every as-fabricated device chip was diced into two nominally identicalpieces. One of them (the “sample”) was exposed to gammairradiation, while the other (the “reference”) was kept in anotherwise identical environment (i.e., at the same temperature,humidity, and lighting conditions). Post-irradiation measure-ments were performed with the devices mounted on athermostat stage maintained at a constant temperature of

�19.7� 0.2�°C. The radiation-induced resonance shift Δλ iscalculated via

Δλ � �λ0 − λ0;r� − �λ1 − λ1;r�; (1)

where λ0 and λ1 denote the resonant wavelengths of the samplebefore and after gamma irradiation, and λ0;r and λ1;r are theresonant wavelengths of the reference measured under identicalconditions. The measured spectra from an a-Si resonatorsample and its reference in their as-fabricated state and afterexposure to gamma irradiation (dose calibrated to silicon) arepresented in Fig. 1 as an example. The extinction ratio andQ-factor remain unchanged, signaling a minimal loss increaseafter gamma irradiation. The use of a reference chip allows usto eliminate environmental noise interfering with precise indexquantification. In the experiments, we can reliably measure aneffective index change down to 2 × 10−6 in SiNx and 5 × 10−5in a-Si, with the measurement accuracy ultimately limited bythe temperature stability of the thermostat stage.

To ascertain that the thermal oxide under cladding does notcontribute significantly to the observed peak shift, we quanti-fied the refractive indices of thermal oxide films before and aftergamma irradiation using variable angle spectroscopic ellipsom-etry. The ellipsometry measurement consistently yielded iden-tical indices for thermal oxide before and after gammairradiation doses up to 20 Mrad. The standard deviation ofthe index measurement is less than 2 × 10−4. Considering thatthe modal confinement factor in the under cladding is onlyabout 6% and 23% in a-Si and SiNx devices, respectively,we conclude that the contribution from thermal oxide tothe post-irradiation resonance shift is insignificant.

HRTEM images (shown in Fig. 2) were taken on a-Si andSiNx films before and after irradiation. The HRTEM sampleswere prepared by ion milling to obtain a ∼20 nm thick lamellafor imaging. Images before [Figs. 2(a) and 2(c)] and after[Figs. 2(b) and 2(d)] receiving 20 Mrad radiation dose do notexhibit discernable structural differences. No signs of crystalli-zation were found after examining multiple (>10) HRTEMimages. The absence of radiation-induced crystallization isfurther confirmed by the XRD spectra plotted in Figs. 2(e)and 2(f ), where no sharp diffraction peaks were identified afterirradiation. The results exclude radiation-induced crystalliza-tion found in several amorphous material systems, includinga-Si [19–21] as a possible mechanism for the observed refractiveindex increase.

While HRTEM and XRD do not reveal clear signatures ofradiation effects, XPS analysis shows surface chemistry modi-fication as a result of gamma irradiation. Figure 3(a) plots theXPS spectrum of the silicon 2 p peak in as-deposited a-Si film.The spectra can be deconvolved into two peaks, correspondingto Si–Si bonds at a binding energy of ∼99 eV and Si–O bondsat ∼103 eV. Relative concentrations of the two types of bondscan be determined by comparing the areas under the decon-volved peaks. The black curve in Fig. 3(b) plots the calculatedSi–O bond fractions in as-deposited and gamma irradiated(irradiated at 1 Mrad and 10 Mrad doses in air) a-Si films.The increased Si–O bond fraction indicates native oxide layergrowth catalyzed by gamma irradiation. This is not surprisingsince highly energetic photon beams such as gamma and x-raysare known to decompose oxygen molecules in air into reactivespecies such as O− and ozone, and it has been shown that thesehighly reactive species contribute to accelerated surface oxide

Fig. 1. (a) Transmission spectra of an a-Si resonator (a) beforeradiation (black) and after receiving 4 Mrad (red), 8 Mrad (green),and 10 Mrad (blue) radiation dose in air. (b) Transmittance of thereference sample under identical conditions; (inset) top view of ana-Si resonator.

588 Vol. 42, No. 3 / February 1 2017 / Optics Letters Letter

growth in crystalline silicon [5,22]. To unequivocally identifythe surface oxidation effect, we performed a second set ofirradiation experiments using samples sealed inside ultrahigh-purity (oxygen impurity <5 ppm) argon gas-filled glass tubes.Since gamma rays are highly penetrating in glass, the sealingtube has a minimal impact on the irradiation dose received bythe sample. The Si–O bond fraction hardly increases in thesample irradiated in argon [Fig. 3(b)], illustrating the suppres-sion of oxide growth. A similar trend was found in SiNxfilms, as shown in Figs. 3(c) and 3(d), which confirms thataccelerated surface oxidation is also present in SiNx devicesexposed to gamma radiation in air.

The impact of surface oxidation on optical resonance drift ischaracterized by sequentially irradiating resonator samplesmultiple times with 2 Mrad dose increments up to a totalcumulative dose of 10 Mrad. Optical transmittance of the de-vices was collected before and after every irradiation session.The radiation-induced resonant wavelength shift calculatedfrom Eq. (1) was measured in two sets of samples, each con-taining five devices, and the average spectral shift Δλ was usedto infer the waveguide effective index change via

Δneff �ngΔλλr

; (2)

where λr denotes the resonant wavelength, ng represents themode group index, and Δneff gives the effective refractive index

change of the waveguide mode. The group index ng is relatedto the free spectrum range in the wavelength domain (FSR,defined as the spacing between two adjacent optical resonantpeaks) by

ng �λ2r

FSR · L; (3)

where L is the round-trip length of the optical resonator. Theradiation-induced index changes in a-Si and SiNx measured us-ing this method are plotted in Fig. 4. The error bars in Fig. 4take into account the experimental uncertainty due to bothtemperature fluctuations and sample-to-sample variations. InFig. 4(b), the error bars are small (<10−5) and, thus, hardlyvisible from the plot. The refractive indices for both a-Si andSiNx increase monotonically with the increasing radiation dosewhen irradiated in argon. The index increase follows approxi-mate linear dependences on the cumulative total radiation

Fig. 2. HRTEM images of a-Si sample (a) before and (b) after20 Mrad radiation. (c) and (d) are the corresponding images of a SiNxfilm before and after radiation under the same dose. (e) and (f ) are theXRD spectra of a-Si and SiNx thin film before and after 20 Mradgamma irradiation, respectively.

Fig. 3. (a), (c) Si 2 p peak from a high-resolution XPS scan ofas-deposited (a) a-Si and (c) SiNx films. The measured spectra (black)are deconvoluted into different Si bonding states, where the red curvescorrespond to the Si–O bond, the blue curves are assigned to Si–Sibond, and the pink curves are associated with the Si–N bond. Theinsets show the sums of the deconvoluted peaks (green), indicatinggood fitting quality. (b) and (d) Calculated surface Si–O bond fractionfor (b) a-Si and (d) SiNx samples irradiated in argon (red) and ambientair (black). The dotted lines are added as visual aids.

Fig. 4. Dependences of effective index changes on cumulativegamma radiation dose in (a) a-Si resonators and (b) SiNx devices,inferred from optical resonator measurements.

Letter Vol. 42, No. 3 / February 1 2017 / Optics Letters 589

doses in both materials. Since the possibility of radiation-induced crystallization is excluded, the observation suggeststhat densification or compaction of the amorphous atomicnetwork is the likely cause of the index increase [23]. In com-parison, the index increase is retarded in devices irradiated inambient air and plateaus beyond a 6 Mrad dose in a-Si. Therefractive index in SiNx drops slightly after an initial rise at a2 Mrad dose. Since silicon dioxide has a lower refractive index(nSiO2

� 1.45) than those of a-Si (na−Si � 3.6) and SiNx(nSiNx

� 2.1), the trend is attributed to surface oxide formationwhich counteracts the index increase due to densification.

In summary, we quantified for the first time, to the best ofour knowledge, refractive index increases in a-Si and SiNx ma-terials induced by gamma irradiation through on-chip opticalresonator characterization. No measurable changes in opticalloss were observed in both materials. When the devices wereirradiated in an inert (argon gas) environment, an index in-crease in a-Si and SiNx follow approximate linear dependenceon the total radiation dose attributed to radiation-inducedamorphous network densification in the materials. The indexincrease is retarded in devices irradiated in ambient air dueto the added effect of surface oxidation confirmed by our XPSanalysis. The quantitative information on gamma radiationeffects in a-Si and SiNx devices provides useful designguidelines for photonic systems operating in radiation-harshenvironments.

Funding. Defense Threat Reduction Agency (DTRA)(HDTRA1-13-1-0001, HDTRA1-15-1-0060).

Acknowledgment. The authors acknowledge thematerial characterization and fabrication facility support bythe MIT Center of Materials Science & Engineering (CMSE)and the Microsystems Technology Laboratories (MTL).

REFERENCES

1. M. Zerler, “CERN’s LHC explores silicon photonics for data links,”SPIE Professional, October 2015, https://spie.org/membership/spie‑professional‑magazine/spie‑professional‑archives‑and‑special‑content/2015_oct_archive/silicon‑data‑links‑at‑cern‑lhc.

2. F. De Leonardis, B. Troia, C. E. Campanella, F. Prudenzano, andV. M. Passaro, IEEE Trans. Nucl. Sci. 62, 2155 (2015).

3. S. S. El Nasr-Storey, F. Boeuf, C. Baudot, S. Detraz, J. M. Fedeli, D.Marris-Morini, L. Olantera, G. Pezzullo, C. Sigaud, C. Soos, and J.Troska, IEEE Trans. Nucl. Sci. 62, 329 (2015).

4. P. Dumon, R. Kappeler, D. Barros, I. McKenzie, D. Doyle, and R.Baets, Optics and Photonics (International Society for Optics andPhotonics, 2005), p. 58970D.

5. S. Bhandaru, S. Hu, D. M. Fleetwood, and S. M. Weiss, IEEE Trans.Nucl. Sci. 62, 323 (2015).

6. R. Sun, J. Cheng, J. Michel, and L. C. Kimerling, Opt. Lett. 34, 2378(2009).

7. J. F. Bauters, M. J. Heck, D. D. John, J. S. Barton, C. M. Bruinink, A.Leinse, R. G. Heideman, D. J. Blumenthal, and J. E. Bowers, Opt.Express 19, 24090 (2011).

8. Y. H. D. Lee and M. Lipson, IEEE J. Sel. Top. Quantum Electron. 19,409 (2013).

9. J. J. Wathen, V. R. Pagán, R. J. Suess, K. Y. Wang, A. C. Foster, andT. E. Murphy, Opt. Express 22, 22730 (2014).

10. Y. Okawachi, K. Saha, J. S. Levy, Y. H. Wen, M. Lipson, and A. L.Gaeta, Opt. Lett. 36, 3398 (2011).

11. J. M. Boudry and L. E. Antonuk, Med. Phys. 23, 743 (1996).12. L. E. Antonuk, J. Boudry, J. Yorkston, C. F. Wild, M. J. Longo, and

R. A. Street, Nucl. Instrum. Methods Phys. Res. Sect. A 299, 143(1990).

13. D. Nikolić, K. Stanković, L. Timotijević, Z. Rajović, and M. Vujisić, Int.J. Photoenergy 2013, 843174 (2013).

14. J. R. Srour, G. J. Vendura, D. H. Lo, C. M. C. Toporow, M. Dooley,R. P. Nakano, and E. E. King, IEEE Trans. Nucl. Sci. 45, 2624(1998).

15. S. Grillanda, V. Singh, V. Raghunathan, F. Morichetti, A. Melloni, L. C.Kimerling, and A. M. Agarwal, Opt. Lett. 41, 3053 (2016).

16. V. Brasch, Q. F. Chen, S. Schiller, and T. J. Kippenberg, Opt. Express22, 30786 (2014).

17. F. Morichetti, S. Grillanda, S. Manandhar, V. Shutthanandan, L. C.Kimerling, A. Melloni, and A. M. Agarwal, ACS Photon. 3, 1569 (2016).

18. J. Hu, N. Carlie, L. Petit, A. Agarwal, K. Richardson, and L. C.Kimerling, J. Lightwave Technol. 27, 5240 (2009).

19. R. G. Elliman, J. S. Williams, W. L. Brown, A. Leiberich, D. M. Maher,and R. V. Knoell, Nucl. Instrum. Methods Phys. Res. Sect. B 19, 435(1987).

20. M. P. Krulikovskaya, A. A. Kuzub, L. I. Chirko, and A. M. Shalaev, Bull.Acad. Sci. USSR Phys. Ser. 6, 37 (1985).

21. H. Cui, E. T. Pashuck, Y. S. Velichko, S. J. Weigand, A. G. Cheetham,C. J. Newcomb, and S. I. Stupp, Science 327, 555 (2010).

22. S. Bhandaru, E. X. Zhang, D. M. Fleetwood, R. A. Reed, R. A. Weller,R. R. Harl, B. R. Rogers, and S. M. Weiss, IEEE Trans. Nucl. Sci. 59,781 (2012).

23. F. Piao, W. G. Oldham, and E. E. Haller, J. Non-Cryst. Solids 276, 61(2000).

590 Vol. 42, No. 3 / February 1 2017 / Optics Letters Letter