s0218863510005042 journal of nonlinear optical physics ...web.mit.edu/hujuejun/www/my papers/journal...

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042

Journal of Nonlinear Optical Physics & MaterialsVol. 19, No. 1 (2010) 75–99c© World Scientific Publishing CompanyDOI: 10.1142/S0218863510005042

PROGRESS ON THE FABRICATION OF ON-CHIP,INTEGRATED CHALCOGENIDE GLASS

(CHG)-BASED SENSORS

K. RICHARDSON∗,†, L. PETIT†, N. CARLIE†, B. ZDYRKO†,I. LUZINOV†, J. HU‡, A. AGARWAL‡, L. KIMERLING‡,

T. ANDERSON§ and M. RICHARDSON§†School of Material Science and Engineering, COMSET,

Clemson University, Clemson, SC 29634, USA

‡Microphotonics Center, Massachusetts Institute of Technology,Cambridge, MA 02139, USA

§College of Optics and Photonics CREOL/FPCE,University of Central Florida, Orlando,

4000 Central Florida Boulevard, FL32816, USA∗[email protected]

Received 4 February 2010

In this paper, we review ongoing progress in the development of novel on-chip, low lossplanar molecular sensors that address the emerging need in the field of biochemicalsensing. Chalcogenide glasses were identified as the material of choice for sensing dueto their wide infrared transparency window. We report the details of manufacturingprocesses used to realize novel high-index-contrast, compact micro-disk resonators. Ourfindings demonstrate that our device can operate in dual modalities, for detection of theinfrared optical absorption of a binding event using cavity enhanced spectroscopy, orsensing refractive index change due to surface molecular binding and extracting micro-structural evolution information via cavity enhanced refractometry.

Keywords: Chalcogenide glass; thin film; waveguide; sensor; resonator; photonic integra-tion; refractometry; absorption spectroscopy; micro-fluidics; lab-on-a-chip; femtosecondlaser irradiation.

1. Introduction

Recent advances in micro-fabrication technology have enabled a paradigm shift inthe field of chemical and biological detection. Miniaturized sensing devices can nowbe created that outperform and possibly replace their conventional bulky counter-parts. These devices build on the knowledge base and technologies developed inmicro-photonics, micro-fluidics, and micro-electronics and have given rise to sens-ing/characterization platforms commonly referred to as a “sensor-on-a-chip” or“lab-on-a-chip”.1 The competitive advantages of such miniaturized sensing devices

∗Corresponding author.

75

http://dx.doi.org/10.1142/S0218863510005042

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042

76 K. Richardson et al.

are three-fold. Firstly, the Moore’s law paradigm driving the rapid strides ofmicro-electronics can be applied to revolutionize bio-/chemical detection, leveragingmature silicon CMOS manufacturing technology to achieve high volume productionand thus very low cost, as well as scalable performance improvement.2 Secondly, asmall device footprint opens up avenues of new device and system applications suchas remote sensor network deployment. These networks, for example, are impossible(cost prohibitive) with a conventional, large footprint design. Lastly, integration ofdifferent functional components onto a planar platform enables multi-modal detec-tion and significantly enhanced intelligence capabilities.

Recent efforts in our group have focused on the development of a novel, inte-grated sensor system with low loss, enhanced sensitivity and specificity suitablefor use in advanced detection in chemical–biological warfare and other intelligentsensing applications.3–9 These efforts have resulted in the development of Si-CMOScompatible processing procedures and demonstration of viability of manufacturinguniformity to large areas (> 6′′ diameter wafers). Such progress provides a means toevaluate scalability and economics in fabricating and integrating optical structuresand devices for use in chip-based chemical and/or biological sensors. The noveltyof the resulting sensor design is three-fold:

(1) Novel high refractive index chalcogenide glass (ChG) materials suitable formulti-spectral chemical and biological sensing have been developed which allowextension of the operating wavelength range for devices from the UV to visi-ble to far infrared. ChGs enable use of a low temperature process to fabricatean ultra-high-Q optical resonant cavity with an atomically smooth surface.Although being widely used as phase change materials for optical disks andnon-volatile random access memories, ChG films are also good candidates forplanar integrated nonlinear optical devices due to their high nonlinearitiesand low linear and nonlinear loss, their fabrication and structural flexibility,a diverse range of optical and electrical properties, large capacity for doping,and tailorable photosensitivity.10–12 Several micro-photonics devices such asgratings, optical storage units, holographic recordings, optical amplifiers andlasers that utilize these unique properties of chalcogenides, have already beendemonstrated.13 ChGs exhibit a broad optical transparency window stretchingfrom the visible to the far-infrared, allowing the realization of multi-modal opti-cal detection in the same material platform, a significant advantage for deviceprocessing and integration.14 In addition, the high refractive index of chalco-genide glasses enables compact photonic integration essential for array-formathigh-throughput analysis. Most importantly, their almost infinite capability ofcompositional alloying enables application-specific tailoring of glass propertiessuch as thermo-optic coefficient, optical response (damage resistance, photo-induced nonlinear response) and chemical compatibility with sensing moleculesin different environments.

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042

Fabrication of On-Chip, Integrated Chalcogenide Glass Sensors 77

(2) In conventional bulk optical sensor designs, high sensitivity is achieved viaincreased optical path length not amenable to device miniaturization andscaling.15 In contrast, our approach relies on planar optical resonators toenhance the device sensitivity utilizing strong photon–molecule interactions in aresonant cavity while dramatically reducing the sensor footprint. Furthermore,optical resonators offer a versatile device platform suitable for cavity-enhancedabsorption and fluorescence spectroscopy as well as label-free refractometricsensing. Finally, compared to fiber-based counterparts, planar ChG-based reso-nator devices are more mechanically robust and amenable to lower cost, larger-scale integration with other on-chip photonic and electronic devices, enablinga full spectrum of optical signal read-out and processing functions.

(3) The sensor device fabrication strategies developed and optimized in this effortemploys CMOS-compatible processing technology suitable for mass productionand further scalable performance improvement. Such scalability reduces persensor costs, allowing multi-sensor arrays tailorable to single or multi-speciesdetection to be deployed for large sampling area evaluation. No other currenttechnology offers such a possibility.

In this paper, we review the choice of the materials used for the device fabrica-tion. We demonstrate the ability to fabricate (i) single-mode waveguides with coresizes down to the sub-micron range possessing reduced sidewall roughness using alift-off technique that has broad applicability to a diverse range of non-silica glasscompositions and also (ii) chalcogenide glass racetrack and micro-disk resonators,the key elements for biochemical sensing. We discuss the use of (i) high powerlaser to fabricate micro-fluidic vias for sensor devices and (ii) polymer coatings toenhance the sensitivity and selectivity of the devices. Lastly, we describe how thesedevices can be used for detection of the infrared optical absorption using cavityenhanced spectroscopy, sensing refractive index change due to surface molecularbinding and extracting micro-structural evolution information via cavity enhancedrefractometry.

2. Glass Material Down-Selection

To design and fabricate the planar molecular sensor, the candidate glass for theoptical resonator structure must possess the following attributes:

(i) The glass must be stable against devitrification as suggested by a differ-ence between its crystallization (Tx) and glass transition (Tg) temperatures.∆T s larger than 90◦C suggest a glass possesses good crystallization resis-tance, important for film deposition. Tg must be engineered to be high enoughto remain stable during normal back end lithographic processing, prepara-tion/deposition of surface coatings, yet suitably stable to minimize aging forrefractometry sensing. It is also important that Tg be low enough to allow

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


thermal reflow at a suitably low temperature to be used to reduce the surfaceroughness of the resonators and thus reduce the resonator optical loss.

(ii) As most biological and chemical agents have their identification signatures inthe infrared (IR) spectral region, the candidate glass must have low absorption(including electronic and bond vibrational absorption) at the operation wave-lengths. For instance, this translates to low absorption and scattering loss in:

(a) the mid- and far-infrared [MIR and FIR] wavebands for absorption spec-troscopic sensing,

(b) the near-infrared [NIR] water transparency window for refractometric sens-ing and fluorophore excitation wavelength for fluorescence spectroscopy)for sensitivity improvement, which in turns dictates the application-specificglass phonon energy and optical band gap.

Additionally, the candidate glass must not exhibit any fluorescence nearthe target absorption/emission wavelength of the source (in its use as athin film resonator and/or when used as a spin on glass).

(iii) The glass also needs to possess a high linear refractive index to allow compactdevice design and to minimize substrate leakage loss.

(iv) Candidate glass must exhibit process compatibility with large-area film deposi-tion techniques (e.g. evaporation or sputtering) critical for device fabrication.Also of importance is the thermal/chemical stability during all steps in thelithographic process.

(v) Post-processing stability is required, specifically chemical stability and mate-rial durability in the sensing environment, as well as biological compatibility forbiosensor applications (which could also be possibly achieved through appro-priate surface coatings using specific polymers) also define further materialrequirements.

(vi) It would be desirable if the candidate glass was suitable for solution-based pro-cessing and possessed the requisite solubility and stability for such application.This would include post-deposition moisture stability and compatibility withany over-coating of a functional polymer.

Silica-based glasses exhibit high absorption past 3 µm (and low optical non-linearities for use in the above cited applications) which precludes their use in the2–5µm MIR window. Lead-germanate, fluoride, chalcogenide, and bismuth oxideglasses possess attributes that make them suitable for MIR transmission applica-tions but they do not meet all criteria requisite to the proposed application. Ofthese glass systems, only fluoride waveguides have demonstrated lower optical lossthan tellurite waveguides. Unfortunately, fluorides have lower glass stability, limitedchemical durability, and lower optical nonlinear coefficients, than other non-oxidematerials.16 They also cut off around 4.5µm in wavelength, shorter than the MIRwindow goals, but have shown laser power handling capabilities (in short lengths)suitable for consideration in compact MIR source applications.17 The best materialchoice for the 2–5µm MIR spectral region which meets the goals for the proposed

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


application is chalcogenide glass (ChG) materials (the amorphous compound con-taining S, Se and/or Te) due to their very wide optical transparency in the mid-waveand long-wave infrared, covering the characteristic absorption wavebands for mostmolecules of interest. Chalcogenide glasses exhibit unique optical properties attrac-tive for both refractometry biological sensing and infrared absorption spectroscopicchemical sensing. The high refractive index of the ChG facilitates planar integrationand low cost mass production. The amorphous nature of chalcogenide glasses allowstheir monolithic integration on a silicon platform via low-cost deposition techniquessuch as evaporation and sputtering. Their amorphous nature also enables a thermalreflow process for optical loss reduction and hence sensitivity improvement. Spe-cific to the refractometry biological sensing application, intrinsically these glassesexhibit low optical loss in near infrared where water is also optically transparent.This is important since most biological reactions occur in an aqueous environment.The thermo-optic coefficient of chalcogenide glasses lies in the range between 10−5

and 10−4 K−1 which is appropriate for creating refractometry sensors minimallyaffected by temperature fluctuations. Furthermore, their almost unlimited capabil-ity for compositional alloying and property tuning provide immense flexibility fordevice design and fabrication.

Several years ago, researchers within our team demonstrated that As-based ChGglasses (i) are very stable against crystallization, (ii) can be deposited into filmsusing thermal evaporation and (iii) when exposed to near-IR (800 nm) femtosecondlaser, exhibit an increase of refractive index18, 19 lending them suitable to laser-induced structuring (both for waveguide and ablative applications). While versatilefrom the standpoint of photosensitivity and high linear and nonlinear indices, manybinary and ternary ChG glasses in the As-Se, As-S and As-S-Se systems have lowglass transition temperatures (∼ 200◦C), and also exhibit moderate long-term sta-bility in ambient conditions. Limitations in post-fabrication stability are usuallydriven by the glass’ constituent elements, which while suitable for long wave opti-cal transmission within the two color IR bands, possess thermal-mechanical robust-ness far less than oxide counterparts. Hence, versatility in the IR (bandwidth toallow transmission and detection of IR resonant species) is at the expense of goodprocessability and long-term performance.

To offset this limitation, we have extended the stability of glasses by enhancingthe coordination of the glass network, making glasses with higher thermal stabil-ity amenable for use in CMOS processing and high stability sensor devices. Forthis reason, we have included in our study Ge-based glasses which not only bringthe rigidity of a four-coordinated network former, but also serves to eliminate theuse of As which can be undesirable in some CMOS manufacturing environments.Ge-containing glass is known to exhibit higher melting and glass transition temper-atures than the As-based glasses.20 We have demonstrated that Ge23Sb7S70 glasscan be reproducibly deposited into thin films using thermal evaporation.21 Whilesuitable for CMOS lithographic processing, this glass cannot be used for directwrite waveguides as shown in Ref. 22. In our current study, we have primarily

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


utilized As- and Ge-based glasses for fabrication of waveguides, ring and racetrackresonators fabricated using lithographic processing. For the 3D laser direct writing,we have only focused our effort on the As-based glass system.23

3. Device Processing and Characterization

The novel sensor device discussed above is comprised of three parts: a micro-fluidicflow system for analyte transport, an optical sensing unit for generating signalsmodulated by the presence of targeted chemical or biological species, and periph-eral electronics for signal read-out and processing. At the core of the device, theoptical sensing unit consists of several sequential components: a light source toemit the probing light, planar waveguides to couple the light into the resonator,a polymer coated on-chip chalcogenide glass optical resonator where strong light-chemical interactions take place, and finally photo-detectors to receive the signalthrough waveguides that couple the light out of the sensor module. In this section,we describe the device fabrication steps from the film deposition through the planarwaveguide fabrication. We also discuss the use of (i) femtosecond laser direct writ-ing (FLDW) as an alternative technique to fabricate microfluidic sensor devices and(ii) polymer coatings to enhance the sensitivity and the selectivity of the evanescentwave sensors.

3.1. Device processing

3.1.1. Bulk processing and film deposition

The synthesis of the As- and Ge-based glasses used in this study is explained indetail, elsewhere.18, 23 All glasses are prepared from high purity elements (As, Alfa99.999%, Ge, Aldrich 99.999%, Sb, Alpha 99.9% and S, Cerac 99.999%) with nofurther purification. Thus, starting bulk glasses used for target materials for filmdeposition have measurable loss in the near and mid-infrared region of the spectrumdue to quantities of adsorbed oxide and hydride. Starting materials are weighed andbatched into quartz ampoules inside a nitrogen-purged glove box and sealed undervacuum using a gas–oxygen torch. Prior to sealing and melting, the ampoule andbatch are pre-heated at 100◦C for 4 h to remove surface moisture from the quartzampoule and the batch raw materials. The ampoule is then sealed and heatedfor 24 hours at between 800 and 975◦C, depending on the glass composition. Arocking furnace is used to rock the ampoule during melting to increase the melthomogeneity. Once homogenized, the melt-containing ampoule is air-quenched toroom temperature. To avoid fracture of the tube and glass ingot, the ampoules aresubsequently returned to the furnace for annealing for 15hours at 40◦C below therespective glass transition temperature, Tg, of the glass. Glass samples are then cut,optically polished and inspected visually. Glass composition has been verified usingelemental dispersive spectroscopy (EDS) and found to be identical to the initialbatch composition. No loss of sulfur has been observed within the accuracy of the

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


measurement (±2 at %). X-ray diffraction (XRD) has been carried out to confirmthe amorphous nature of our samples.

In the present effort, the thermal evaporation (TE) technique has been used forthe deposition of the films with the composition Ge17Sb12S71, As42S58 and As2S3.TE was chosen as the method of choice given its (1) low-cost, large-area uniformityand simplicity (versus pulsed laser deposition, PLD); (2) high-throughput (com-pared to sputtering) and (3) minimized step coverage which proves excellent forsubsequent lift-off (versus chemical vapor deposition, CVD). Thermal evaporationis often a non-congruent vaporization process which can lead to off-stoichiometryfilms. In this case, it is crucial to modify the starting bulk composition to com-pensate for this change to yield films and structures with desired composition andtarget refractive index. As- and Ge-based thin films on oxide-coated Si wafers (6′′

Si wafers with 3 µm thermal oxide, Silicon Quest International) and glass micro-scope slides were prepared via thermal evaporation. A complete description of thetechniques can be found elsewhere.6, 24 Using a SEM, the films did not exhibit anyobvious signs of local compositional variation or phase separation, as evident fromuniform elemental maps made with energy dispersive spectroscopy (EDS) acrossmultiple 1mm2 areas of sample surface. We demonstrated that the thermal evap-oration process results in films with similar composition and thermal propertiesas compared to those of the parent bulk glasses.21 However, Raman spectroscopyhas shown that the structure of the As-based films contain a significantly highernumber of As4S4 molecular units as well as As4S3 units, not seen in the parentbulk material. Ge-based films have been shown to contain homopolar Ge-Ge bondsand a lower number of homopolar S-S bonds compared to bulk material.18, 19, 21

We believe that these species, found to be only present in films, result from thedifferent thermal history (quench rate) induced during the film deposition process.The study of the structure of the as-deposited As-based and Ge-based films con-firms that the deposition of glasses into thin films can result in variation of thematerial properties which may affect device performance. Knowledge of the effectsof such processing conditions on relevant glass properties, such as refractive index,structure and glass transition temperature, and definition of means to correct orcompensate for them, is crucial for the choice of the glasses used for the devicedesign and fabrication.

3.1.2. Device processing

The waveguide, micro-ring and racetrack resonator structures have been fabricatedusing the lift-off process. As in a standard lift-off process, a photoresist pattern isfirst formed on a substrate. ChG film is then thermally evaporated onto the waferpatterned with photoresist, and sonicated in solvent (usually acetone) to dissolvethe photoresist layer beneath the undesired parts of the film, thus lifting it off. Onlyglass deposited onto areas not covered by photoresist is retained, and thus a chalco-genide pattern reversed to that of the photoresist is defined. The patterned wafer

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


is then rinsed in methanol and isopropanol to clean the surface. To fabricate ribwaveguides, a second film deposition is made sequentially on a lift-off patterned film.In our process, the starting substrates are coated (with a 3µm-thick thermal oxide)6′′ Si wafers. Commercially available negative resist NR9-1000PY (Futurrex Inc.) isused due to its negative-sloping sidewall profile and superior pattern resolution. Theresist is spin-coated onto substrates on a manual photoresist coater (Model 5110,Solitec Inc.). UV exposure is carried out using a Nikon NSR-2005i9 i-line waferstepper (minimum linewidth 500nm). Resist pattern development and subsequentbaking are both completed on an SSI 150 automatic photoresist coater/developertrack. The entire photolithography process is performed in a class-10 CMOS cleanroom. To prevent surface oxidation, a 3 µm thick layer of SU8 polymer is spin-coated to serve as a top cladding after patterning and the devices are subsequentlyannealed at 140◦C for 3 hours to stabilize the glass structure, whereas no SU8 coat-ings are applied on the waveguides and resonators.

The morphology of as-fabricated waveguides has been characterized using aJEOL 6320FV field-emission high-resolution SEM. Figure 1(a) shows a cross-sectional SEM micrograph of a strip Ge23Sb7S70 waveguide before photoresist lift-off and Fig. 1(b) show an AFM scan of a 2 µm by 2 µm square area showing thesurface morphology of a waveguide with a width of 750 nm, respectively. A DigitalInstruments Nanoscope IIIa Atomic Force Micro-scope (AFM) has been used tomeasure the roughness of the as-patterned waveguides. Measurement scans havebeen performed parallel to the propagation direction of the waveguides using thetapping mode. The AFM measurements yield an average sidewall line RMS rough-ness value of (11 ± 2) nm for as-fabricated waveguides and (1.6 ± 0.3) nm for thestrip waveguide’s top surface. In comparison, plasma etched Ge23Sb7S70 waveguidesidewalls exhibit RMS roughness values typically ranging from 20–150nm depend-ing on the etching parameters.7 The relatively low sidewall roughness in waveguides

(a) (b)

Fig. 1. (a) Cross-sectional SEM micro-graph of a sub-micron Ge23Sb7S70 strip waveguide beforephotoresist lift-off. (b) AFM morphological scan of a Ge23Sb7S70 strip waveguide fabricated viathe lift-off method.

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


Fig. 2. (a) Top view optical micro-graph of an As2S3 micro-disk resonator with a pulley couplerconfiguration and (b) a racetrack As2S3 resonator.

fabricated through lift-off can be attributed to the fact that the sidewall is definedduring a deposition process rather than etching. In the lift-off process, we believethe major source of sidewall roughness originates only from edge roughness of pho-toresist patterns.

Figure 2 shows (a) the top view of a fabricated As2S3 racetrack resonator and(b) a 20µm-radius As2S3 micro-disk resonator with a bus waveguide in a “pulley-type” coupling configuration. Compared to a conventional micro-disk/micro-ringcoupler, the pulley coupler design increases the coupling length leading to strongercoupling. In order to achieve the same coupling strength as a conventional couplerdesign, a wider gap between a resonator and a bus waveguide is possible when apulley coupler is employed. With such performance constraints reduced, lower costlithography techniques can be used for fabrication.

Femtosecond Laser Direct Writing (FLDW) has been investigated as an alter-native technique for the fabrication of microfluidic sensor devices, specifically inthe formation of flow vias. FLDW is well-established as a flexible and powerfultechnique for the fabrication of optical elements such as waveguides both on thesurface and within the volume of transparent materials and is thus particularlywell-suited for the fabrication of complex 3D device geometries.26–29 We previouslyshowed that through nonlinear multiphoton absorption of the laser light, femtosec-ond pulses can be used to modify the molecular bond structure of chalcogenideglass, leading to changes in the density and the refractive index.22, 30–32 This local-ized photo-modification also affects the resistance of the glass to chemical etching.Thus, a chalcogenide glass can be patterned using femtosecond lasers and thenimmersed in an etchant to preferentially etch the photo-modified regions. This pro-cess can be likened to a mask-less lithography and could potentially be used toeither supplement traditional lithographic fabrication or in some cases serve as astand-alone fabrication technique.

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


We describe here the fabrication of a 200µm× 200µm surface relief reservoirmeasuring in As42S58 glass using the combination of FLDW and wet chemicaletching in a KOH solution. The laser used for the photo-modification step is afiber-based Yb:YAG glass laser with a wavelength of 1048nm, a repetition rateof 1.5MHz, and a pulse duration of ∼400 fs (µJewel, IMRA Corp). The beam isfocused to a ∼4µm spot size using a 0.3N A (10×) microscope objective and thebulk glass is mounted on a computer-controlled 3D translation stage. A 200µmwide square area of photo-modified material has been formed on the surface ofthe glass by scanning the surface sample at a speed of 500µm/s through the focalvolume of the laser in a raster pattern with a line spacing of 2µm with a 7.5 nJ pulseenergy of the laser, which corresponds to 95% of the ablation threshold under theseconditions. The surface profile (roughness) of the films has been examined usinga Zygo corp. NewView 6300 white light interferometer microscope. In agreementwith Ref. 30, the FLDW process induces a surface photo-expansion of 120nm in thephoto-modified area of the As42S58 glass as illustrated in Figs. 3(a) and 3(b) whichshow a color-mapped contour plot and a cross-sectional surface profile of the photo-modified region after laser irradiation, respectively. After the laser irradiation step,the sample is immersed in a 0.1M KOH solution for 12minutes. The KOH acts asa positive developer and preferentially etches the laser-modified region at a rate of270 nm/min, resulting in a reservoir depth of 3.2µm as seen in Fig. 4. The depth ofthe reservoir has been determined by the depth of the photo-modified material. It isinteresting to mention that further etching does not deepen the reservoir but etchesboth the top surface and the bottom of the reservoir at the same rate. This mask-less technique for the fabrication of fluidic and optical elements is highly versatileand can be used to augment or enhance the fabrication of elements created usingother lithographic techniques towards the development of on-chip biosensors.

3.2. Characterization of sensor/resonator

Waveguide loss measurements have been performed on a Newport AutoAlign work-station. Lens-tip fibers have been used to couple laser light into and out of the

Fig. 3. (a) Color-mapped contour plot and (b) cross-sectional surface profile of a 200 µm× 200 µmsquare photo-expanded region fabricated at the surface of the As42S58 glass by FLDW measuredusing a Zygo NewView 6300 3D Optical Profilometer.

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


Fig. 4. (a) Color-mapped contour plot and (b) cross-sectional surface profile of a 200 µm× 200 µmsquare surface relief reservoir fabricated at the surface of the As42S58 glass by FLDW and wetchemical etching.

Table 1. Measured optical transmission losses and calculated modal parameters of Ge23Sb7S70

waveguides at 1550 nm wavelength.

0.75 µm (strip) 1.2 µm (strip)1.6 µm (strip) 1.2 µm (rib)

Waveguide width, w TM TE TM TE TM TE

Transmission loss (dB/cm) 3.9 ± 0.4 6.4 ± 0.8 3.5 ± 0.3 5.2 ± 0.5 2.3 ± 0.4 <0.5Number of modes supported 1 1 1 2 1 1

waveguides. Highly reproducible coupling between waveguides and fibers can beachieved via an automatic alignment system. Waveguide loss has been determinedvia a standard cutback method using paper-clip waveguide patterns. Transmissionloss measurements made at 1550nm for strip and rib Ge23Sb7S70 waveguides ofdifferent widths and cross-sectional geometry are reported in Table 1. It can clearlybe seen that the Transverse-Magnetic (TM) mode exhibits lower transmission lossthan the Transverse-Electric (TE) mode in strip waveguides with the same width.Moreover, the transmission losses for both modes decrease as the strip waveguidesbecome wider. The loss dependence on width is more significant for TE modethan for TM mode. The rib waveguides show very low loss for both TE and TMmodes due to less mode interaction with sidewall roughness in the rib waveguidegeometry. Statistical analysis reveals that these waveguides have an average loss of(2.3 ± 0.4) dB/cm at 1550nm wavelength, larger than those reported by Maddenet al.33, 34 who measured 0.05 and 0.3 dB/cm optical losses at 1550nm in As2S3

planar rib waveguides and in Ge33As12Se55 rib waveguides of 3, 4 and 5 µm wide,respectively. The standard deviation of 0.4 dB/cm is low enough to confirm theexcellent processing uniformity of our lift-off technique, indicating that the lift-offis intrinsically a wafer-scale processing technique suitable for scale up for massproduction.

Micro-disks with the composition As2S3 and Ge17Sb12S71 have also been tested.Table 2 summarizes the optical properties of the resonators. The higher Q factorof TM polarization compared to that of TE polarization suggests that bending loss

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


Table 2. Coupled cavity Q and free spectral range (FSR)of the micro-disk resonators at 1550 nm.

Cavity Q (±10%) FSR (nm)

Bulk composition TM TE TM TE

As2S3 210,000 150,000 7.62 7.82Ge17Sb12S71 (in air) 110,000 100,000 8.35 8.92Ge17Sb12S71 (in water) 20,000 8.38

(a) (b)

Fig. 5. (Color on line) (a) Measured transmission spectra of a 20 µm-radius As2S3 microdiskresonator. (b) A TM-polarization transmission spectrum averaged over 32 wavelength sweepingscans near a resonant peak. The black dots are experimental data points and the red curve is theLorentzian peak fitted in linear scale.

is insignificant in the micro-disk, and thus it is possible to achieve an even smallercavity mode volume without suffering excess radiative loss. The measured trans-mission spectra of the As2S3 micro-disk with a gap separation of 800 nm betweenbus waveguide and micro-disk are shown in Fig. 5(a). The transmission spectrafeature a set of resonant peaks evenly spaced by a well-defined free spectral range,indicative of single-mode resonator operation as seen in Fig. 5(b). The micro-diskoperates near critical coupling regime for both TE and TM polarizations around1550nm wavelength, an important advantage for applications in the telecommuni-cation bands.

3.3. Sensor enhancement with polymer coatings

Thin polymer films anchored to a substrate are widely used for modulation ofthe surface properties and for fabrication of versatile adaptive surfaces capable ofresponding to changes in the environment. Polymer brushes are frequently used forthis purpose, where the “brush” term is referred to the dense ensemble of polymerchains attached to the substrate.35 The main idea behind better sensitivity andselectivity of an evanescent wave sensor is the deposition of a polymer layer on topof the waveguide to bind the analyte of the interest from the surrounding media.

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


Fig. 6. Schematics of enrichment layer for evanescence sensor.

In the presence of the analyte, the polymer coating is expected to swell, leadingto a variation of the refractive index in the vicinity of the optical waveguide asillustrated in Fig. 6. In this study, poly(glycidyl methacrylate) (PGMA) has beenused as a molecular glue to assemble the layered structure and covalently bondpolymeric layers one to another.36–39 The glycidyl methacrylate units located inthe “loops” and “tails” sections of the attached PGMA chains are not connectedto the substrate. These free groups could serve as reactive sites for the subsequentattachment of chemical moieties/polymers to the PGMA layer as shown in Fig. 7(a).Figure 7(b) shows the PGMA layer uniformly deposited on top of ChG substrate.The coated films are smooth and homogeneous. The thickness of the PGMA primarylayer can be altered to match evanescent wave penetration, thus enhancing sensorresponse.

In order to define the most suitable polymer coating for our study, ATR crystalswere coated with PGMA and with PGMA-polyacrylic acid (PAA) and then exposedto vapor of triethylamine, Et3N. The FT-IR spectra of the uncoated and coatedcrystals are presented in Figs. 8(a)–8(c). The spectra of the uncoated and PGMAcoated crystals do not show any significant signatures of triethylamine, whereas anew peak appears at 1563 cm−1 in the FT-IT spectrum of the PGMA-PAA coatedsample as seen in Fig. 8(c). This peak corresponds to carboxylate ion. The peak

(a) (b)

Fig. 7. (a) Schematic representation of reactive polymer attached to substrate. (b) AFM imageshowing topography of PGMA layer deposited by dip-coating on ChG surface. Size is 2 × 2 µm.

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


Si substrateSi substrate

Si substrate

PGMAPAA

Si substrate

PGMAPAA

Si substrate

PGMA

Si substrate

PGMA

Fig. 8. (a) FT-IR spectra of the triethylamine vapors over bare Si surface; (b) PGMA; (c)PGMA-PAA coatings.

related to the carbonyl at 1700–1730cm−1 decreases in amplitude comparatively tothe baseline, indicating consumption of free carboxyl group in the coating duringinteraction with amine. These variations in the FT-IR spectrum clearly demonstratethe analyte presence with the use of the PGMA-PAA coating which is capableto reversibly bind amines. Based on this result, the ChG thin film was coatedwith PGMA-PAA and exposed to saturated ammonia vapor above 0.01% aqueoussolution. Figure 9 shows the absorption spectrum of the PGMA-PAA coated ChGafter exposure. Also shown is the spectrum of uncoated ChG thin film after exposureto saturated ammonia vapor. The selective binding between NH3 and carboxylicgroups in the coating leads to negative adsorption of carboxyl groups in the presenceof ammonia. This allows distinguishing between ammonia and water present in themixture being analyzed. As seen in Fig. 9, the PGMA-PAA coated ChG has ∼4times higher sensitivity towards ammonia vapor than uncoated ChG, confirmingthat polymer coating can enhance the sensitivity of the evanescence sensor.

To achieve multi-analyte sensing capability, the polymer coating should becomprised of specific layers possessing affinity to different analytes. Thus, ChGwaveguides (O-ring resonators) deposited on Si wafer were modified with three dif-ferent polymer systems: (i) coating comprised of epoxidized polybutadiene (EPB),(ii) layered coatings consisting of 6 alternating layers of PGMA and PAA and(iii) multilayer coatings comprised of four polymers of different nature. To this end,

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


4000 3500 3000 2500 2000 1500 1000 500

-0.05

0.00

0.05

0.10

0.15

0.20

Ab

sorb

ance

Wave numbers [cm-1]

Bare ChG PGMA-PAA layered system

Fig. 9. FT-IR spectra of the ammonia vapor over bare ChG and polymer-coated ChG surfaces.

Si substrate

PGMAPAA

PVP

EPB

Si substrate

PGMAPAA

PVP

EPB

Fig. 10. Schematics of the 5-layered coating consisting from polymers of different chemicalnatures.

we designed a 5-layer system consisting of the following components as shown inFig. 10:

• EPB, and PGMA which can be swollen with hydrocarbons;• PAA which can react with basic nature analytes;• PVP, poly(vinylpyridine) which exhibits change when it interacts with acidic

analytes, and with aromatic compounds, possibly due to the presence of aromaticpyridine fragments.

The modified ChG resonators were exposed to vapor of water, methanol, ethanol,isopropanol, acetone, hexane and aqueous ammonia solution (0.027%). To estimatethe polymer coating influence on the O-ring resonator performance, the resonantpeak shift of the resonator in the presence of various chemical vapors was measuredafter exposure. The maximum shifts of the various modified ChG resonators arereported in Fig. 11. It is clearly shown that, depending on the coating, the modifiedresonators exhibit different patterns in resonance peak shift when exposed to water,methanol, ethanol, isopropanol, acetone and hexane. This type of response can be

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


0.00

0.05

0.10

0.15

0.20

NH3 0.027% in water

hexaneAcetone

IPAEtOH

MeOHwater

Pea

k sh

ift,

nm

PGMA-PAA5 layered coating EPB60

Fig. 11. Resonant peak shift upon exposure polymer coated resonators to various chemicals.

exploited in “electronic nose” type sensor to distinguish between these substances.From data reported in Fig. 11, we can conclude the following:

(i) PGMA-PAA layered system decreases the magnitude of the response whenthe polarity of the chemical is decreased (water-methanol-ethanol-isopropanol-acetone-hexane row).

(ii) Both PGMA-PAA and the 5-layered coating are saturated with ammonia atthe concentration used in the experiments, thus making them suitable for earlydetection.

(iii) The 5-layered system shows the strongest response to alcohols among all testedchemicals.

(iv) EPB coating exhibits a negative peak shift in the presence of hexane vapor,indicating a specific reorientation in the film structure due to contact with thesolvent.

4. Applications

4.1. Detection of the infrared optical absorption using cavity

enhanced spectroscopy

We have previously reported the demonstration of a microfluidic-integrated waveg-uide evanescent sensor for detection of infrared absorption arising from chemicalmolecular species (small molecules),8 which draws a direct analogy to conventionalfree-space infrared spectrophotometer and is physically simple. However, in orderto resolve the trade-off between improving sensitivity and reducing device physi-cal size (optical path length), we introduce another form of sensor configurationbased on cavity-enhanced absorption. The sensing mechanism is based on the high

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


sensitivity of the extinction ratio of resonant peaks to optical absorption varia-tions in the sensor’s surroundings.40 This design leverages the optical resonanceenhancement effect to achieve multiple sampling of the analyte and hence improvessensitivity while still maintaining a small device footprint, and therefore it is highlyappropriate for future generations of on-chip spectrophotometers.

Additionally, it is worth-noting that even though resonators are often consideredas intrinsically narrow-band devices, spectroscopic scans covering a wide wavelengthrange can be performed by simultaneous monitoring of multiple resonant peaks incavity-enhanced spectroscopy. In this regard, the spectroscopic wavelength reso-lution is only limited by the free spectral range (FSR) of the resonator. There-fore, resonators with large radius are better suited for high-resolution spectroscopicapplications, given their small FSR and hence closely-packed resonant peaks.

Cavity-enhanced absorption sensor performance has been tested by monitor-ing the resonant peak extinction ratio variation while injecting a solution ofN-methylaniline in carbon tetrachloride into the micro-fluidic channel. The N–Hbond in N-methylaniline is known to exhibit an absorption peak near 1500nm,which is used as the characteristic fingerprint for chemical identification in ourtest.41 The absorption (in dB), αL, induced by N-methylaniline in our micro-fluidicchannel is derived using the generalized coupling matrix formalism.42 Figure 12shows the transmission spectra around a resonant peak, both before and after injec-tion of 10% vol. N-methylaniline solution in CCl4 into the microfluidic channel. Theresonator works in an under-coupling regime. Therefore, the additional optical loss

Fig. 12. Microdisk transmission spectra around a resonant peak before and after N-methylanilinesolution injection. The dots are experimentally measured data points and the lines are theoreticalfitting results based on the generalized coupling matrix formalism. The extinction ratio changeand resonant peak broadening are due to the optical absorption of N-methylaniline near 1500 nm.Note that the resonant wavelength red shifts are due to the refractive index difference betweenpure CCl4 and N-methylaniline solution.

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


Fig. 13. Optical absorption of N-methylaniline solution in carbon tetrachloride near 1497 nmwavelength measured as a function of N-methylaniline volume concentration.

due to N-methylaniline absorption results in a decrease of the extinction ratio alongwith a broadening of the resonant peak. Since carbon tetrachloride has no absorp-tion band and is transparent in the investigated spectral range, this extinction ratiochange can be unambiguously assigned to N–H bond vibrational absorption. Thepeak absorption in cm−1 at 1496nm has been measured for different concentra-tions of N-methylaniline solution in carbon tetrachloride and the result is shownin Fig. 13. The excellent linear fit suggests that the sensor exhibits linear responsewhen varying analyte concentrations in the range investigated.

4.2. Biological molecular detection: Sensing refractive index

change due to surface molecular binding

Sensitive identification and analysis of biological macro-molecules (proteins andnucleic acids) constitute the basis for applications ranging from clinical diag-nostics, macro-molecular drug discovery and anti-bioterrorism. The conventionallabel-based detection techniques rely on pre-labeling of the target species usingradioactive, chromogenic or fluorescent molecular markers. Those techniques canbe highly sensitive. Nevertheless, the extra labeling step incurs complicated,time-consuming processing as well as additional reagent cost. Furthermore, theintroduction of a labeling marker can interfere with accurate analysis of molecularproperties such as binding affinity. Label-free detection techniques have thus beenproposed as alternatives for biological molecular testing. In addition to overcom-ing the aforementioned drawbacks of label-based tests, several label-free techniquessuch as surface plamson resonance (SPR) are capable of measuring the molecu-lar binding kinetics in real-time, a capability conventional label-based tests can-not accomplish. Despite these advantages, label-free techniques are often criticizedfor their lower sensitivity compared to label-based assays. In addition, some of

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


the label-free techniques require costly machines to operate (e.g. the commercialBiacore SPR automatic protein analysis systems cost up to ∼ $300,000, and theprice of each sensor chip ranges from $200–$600), largely due to the need for costlyreagents, discrete optical components as well as complicated peripheral electronicsfor signal readout and interpretation.

We foresee that resonator refractometry biosensors offer a promising solutionto these challenging issues. In a resonator biosensor, the resonator surface is func-tionalized with antibodies conjugate to the target molecules to be detected. Whenthe resonator is exposed to liquid samples containing the target species, specificsurface molecular binding occurs. The refractive index change resulting from sur-face molecular binding leads to a resonant frequency change of the optical resonatorthat can be detected by a photodiode. The high-Q optical resonance leads to stronginteraction between photons and the molecules to be detected and hence very highsensitivity. Two orders of magnitude sensitivity improvement over SPR sensors ispossible using dielectric optical resonator sensors.43 Additionally, the small foot-print and CMOS-backend compatible processing of the sensor chips also suggestsmuch reduced fabrication and reagent costs.

To test the refractometry sensing capability of our chalcogenide glass resonators,bulk refractive index sensitivity has been measured via injecting liquids with dif-ferent refractive indices into the micro-fluidic channel and measuring the resonantwavelength shift as a function of liquid index change. In our testing procedure,deionized (DI) water solutions of isopropanol (IPA) of varying concentrations havebeen injected into the channels through a syringe pump, and the resonant peak shiftdue to ambient refractive index change has been monitored in situ. The measure-ments with different concentrations of solutions have been repeated twice to con-firm reproducibility. The TM polarization transmission spectra of a Ge17Sb12S71

resonator in IPA solutions of various concentrations are shown in Fig. 14(a).

(a) (b)

Fig. 14. (a) TM polarization transmission spectra of a Ge17Sb12S71 microdisk in IPA solutionswith four different concentrations. (b) Measured resonant peak wavelength shift as a function ofIPA solution mole ratio concentration and corresponding solution refractive index.

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


The resonant wavelength shift as a function of IPA concentration and correspondingsolution refractive index is plotted in Fig. 14(b), and a refractive index (RI) sensi-tivity of (182±5)nm/RIU is inferred from the fitted curve slope. The RI sensitivitycan also be predicted using the equation below:

SensitivityRI =λ

ng× Γiv , (1)

where λ is the operating wavelength, and Γiv denotes the TM mode power confine-ment factor in the interrogation volume (water solution in this case) estimated tobe 0.204 using finite difference simulations.

Based on simulation results, a RI sensitivity of 194 nm/RIU is predicted. Thedifference between the experimentally measured RI sensitivity and the theoreticalprediction is probably due to a slight deviation in waveguide dimension from thetarget design values.

By applying a Lorentzian fit to the resonant peaks, the resonant wavelengthcan be determined to an accuracy of ∼0.1 pm limited by noise. Besides resonantwavelength shift, the cavity Q-factor decreases from 110,000 to 20,000 after solutioninjection, which can be directly attributed to the optical absorption of water. Usingthe absorption coefficient of water at 1550nm wavelength (α = 9.6 cm−1),44 we esti-mate the absorption-limited Q-factor to be ∼19,000, which is in excellent agreementwith our measurement result. By shifting the operating wavelength to near infraredwater transparency windows (e.g. 1.06µm wavelength), we can expect minimizedQ deterioration due to water absorption, which should lead to much higher sensorsensitivity.

4.3. Precision glass metrology: Extracting micro-structural

evolution information via resonant cavity refractometry

While we have employed the resonators’ extreme sensitivity to refractive indexchanges for miniaturized biochemical sensors in the preceding sections, this propertyof resonators can also be utilized to probe micro-structural evolution in chalco-genide glasses, since structural modifications in these glasses are often accompa-nied by a refractive index change.45 Some important examples include index ofrefraction change due to density variation (free volume change), structural relax-ations (e.g. those induced by annealing or light illumination), absorption coefficientchange in the event of multi-photon-excitation, defect state activation or free carrierabsorption. Notably in the aforementioned examples, the index change often con-stitutes an accurate and sensitive indicator for the direction, rate and magnitude ofmicro-structural evolutions in glasses, a mechanism highlighting the relevance andimportance of our cavity-enhanced refractometry technique. In addition, accuratedetermination of the index change is also technically important for applications thatrely on photo-induced index modifications such as direct laser writing. For example,Anderson et al. measured an index increase up to 0.09 at 550 nm in femtosecondlaser written As36Sb6S58 waveguides.23

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


(a) (b)

Fig. 15. (a) Photosensitivity measurement set-up. (b) Transmission spectra of a resonator in theproximity of its resonant peak. Black dots represent experimental data points, and the red dotsare data measured in the same device after exposure to ∼550 nm wavelength light. The curves arefitted Lorentzian peaks. A wavelength shift of (3 ± 0.3) pm is determined by peak fit, correspondingto an index change of (4.5 × 10−6 ± 10%) RIU in As2S3 at 1597 nm wavelength.

To study the photosensitivity of an As2S3, the resonator device has been exposedto near-band gap light (∼550nm wavelength with an irradiance of 6.2mW/cm2

from a band pass filtered halogen lamp) to induce a controlled refractive indexchange using a set-up schematically shown in Fig. 15(a), and the resulting peakshift has been monitored in situ. A Lorentzian fit has been used to extract theaccurate peak position. Refractive index change can be calculated via:

∆n =∆λ

λ× ng/Γ, (2)

where ∆λ stands for the resonant peak shift, neff is the waveguide effective index,and Γ is the confinement factor in the As2S3 core.

Finite-difference simulations show neff and Γ of TM polarization to be 1.88 and0.77, respectively. The group index ng is calculated from the resonator circumferenceand FSR to be 2.31. Figure 15(b) gives an example of two spectra with a peak shift of(3±0.3)pm after light exposure, corresponding to an index increase of 4.5×10−6±10% in As2S3 at 1597nm wavelength. The wavelength resolution of 0.3 pm hasbeen statistically determined by repeating the experiment and comparing multiplemeasurement results. Such a high sensitivity is essential especially for accuratephotosensitivity measurement in annealed glass, which exhibits much smaller photo-induced index change compared to its as-deposited counterpart.30

Figure 16 shows the refractive index increase at 1550nm wavelength as a func-tion of exposure dose for both as-deposited (without annealing) and annealedAs2S3 devices. Excellent data reproducibility has been confirmed by repeatingthe measurement on several resonators. In both cases, the index increase can be

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


Fig. 16. Photo-induced refractive index change as a function of exposure dose (λ = 550 nm) inan as-deposited, unannealed resonator and in a resonator annealed at 130◦C for 3 hours.

well-described with a single-exponential function:

∆n = ∆nsat × [1 − exp(−D/D0)], (3)

where ∆nsat is the maximum index change when the photo-refractive effect hasbeen fully saturated, D represents the exposure dose in J/cm2 and D0 is a materialconstant which depends on processing history.

Our results confirm the validity of exponential empirical fit in previousreports.18, 46 For unannealed As2S3, ∆nsat = 0.0505 ± 0.0002, D0 = (0.036 ±0.001) J/cm2, whereas for annealed As2S3, ∆nsat = 0.0125± 0.0001, D0 = (0.024±0.001) J/cm2. The smaller photo-induced index change in annealed glass was alsoobserved by C. Lopez in As2S3 films prepared using the same deposition technique,after irradiation with an 800nm femtosecond laser.18 It is also important to notethat the annealed and unannealed samples have distinctive D0 values. This obser-vation suggests that even though both annealing and photosensitivity lead to arefractive index increase in As2S3 glass, these two processes actually result fromdistinctive micro-structural changes and thus follow very different kinetic paths.Therefore, even though photosensitivity in chalcogenide glasses is a process accom-panied by an entropy increase,47 it is dissimilar from thermal annealing from a glassmicro-structure point of view. This conclusion also confirms the previous findingsby other authors.18, 48, 49

5. Conclusion

We have demonstrated that ChG materials can be highly flexible, property-tailorable solutions for integrated highly specific and sensitive, integrated sensors.We explained that functionalized polymers add to the resonator specificity as theycan functionalize the surface for protein/antigen size, chemistry specific to a sensor

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


need. Planar micro-disc and ring and racetrack micro-resonators have been devel-oped with low loss (∼1 dB/cm) from unpurified multi-component ChG using scal-able and CMOS-compatible fabrication routes. Further enhancement to lower theloss of these devices is expected with further purification of the starting materialsforming the bulk glass target materials.

We have described how multi-modality detection (absorption+ index change)possible in these components facilitates greater device accuracy and sensitivity.We demonstrated that compact, robust, high sensitivity devices can be utilized inliquid/vapor phase sensing devices for infrared spectroscopy, chemical and biologicalsensing and as a tool for quantifying aging/relaxation/stability mechanisms andtime-scales, in optical materials/structures and devices.

Acknowledgments

The work presented in this article is conducted at Clemson University in collabora-tion with the Laser Plasma Laboratory group at CREOL/UCF and the Micropho-tonics Center at Massachusetts Institute of Technology. Funding support has beenprovided by the Department of Energy under award number DE-SC52-06NA27341.The authors also acknowledge the Center of Materials Science and Engineeringand the Micro-systems Technology Laboratories at MIT for characterization andfabrication facilities.

Disclaimer

This paper is prepared as an account of work supported by an agency of the USGovernment. Neither the US Government nor any agency thereof, nor any of theiremployees, makes any warranty or assumes any legal liability or responsibility forthe accuracy, completeness or usefulness of any information, apparatus or processdisclosed, or represents that its use would not infringe privately owned rights. Ref-erence herein to any specific commercial product, process, or service by trade name,trademark, manufacturer, or otherwise does not necessarily constitute or imply itsendorsement or favoring by the US Government. The opinions of authors expressedherein do not necessarily reflect those of the US Government or any agency thereof.

References

1. P. Yager, T. Edwards, E. Fu, K. Helton, K. Nelson, M. Tam and B. Weigl, Nature442 (2006) 412–418.

2. G. Moore, Electronics 38 (1965) 114.3. J. Hu, V. Tarasov, N. Carlie, L. Petit, A. Agarwal, K. Richardson and L. Kimerling,

Optical Materials 30 (2008) 1560–1566.4. J. Hu, N. Carlie, L. Petit, A. Agarwal, K. Richardson and L. Kimerling, Optics Letters

33 (2008) 761–763.5. J. Hu, N. Carlie, N. Feng, L. Petit, A. Agarwal, K. Richardson and L. Kimerling,

Optics Letters 33 (2008) 2500–2502.

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


6. J. Hu, N. Feng, N. Carlie, L. Petit, J. Wang, A. Agarwal, K. Richardson and L. Kimer-ling, Optics Express 15 (2007) 14566.

7. J. Hu, V. Tarasov, N. Carlie, N. Feng, L. Petit, A. Agarwal, K. Richardson andL. Kimerling, Optics Express 15 (2007) 11798.

8. J. Hu, V. Tarasov, N. Carlie, N. Feng, L. Petit, A. Agarwal, K. Richardson andL. Kimerling, Opt. Express 15 (2007) 2307.

9. J. Hu, N. Carlie, L. Petit, A. Agarwal, K. Richardson and L. C. Kimerling, J. Light-wave Technol. 27 (2009) 5240–5245.

10. Z. Ling, H. Ling and Z. C. Shan, J. Non Cryst. Solids 184 (1995) 1–4.11. H. Nasu, Y. Ibara and K. Kubodera, J. Non Cryst. Solids 110 (1989) 229–234.12. J. A. Savage, J. Non Cryst. Solids 47 (1982) 101–115.13. J. Xu and R. M. Almeida, J. Sol-Gel Sci. Tech. 19 (2000) 243–248.14. G. Delaizir, P. Lucas, X. Zhang, H. Ma, B. Bureau and J. Lucas, J. Am. Ceram. Soc.

90 (2007) 2073–2077.15. K. Nakamur, T. Okada and S. Ueha, Journal of Optics A: Pure and Applied Optics

3 (2001) L17–L19.16. J. S. Wang, E. M. Vogel and E. Snitzer, Optical Materials 3 (1994) 187–203.17. C. Xia, M. Kumar, O. Kulkarni, M. Islam, F. Terry Jr., M. Freeman, M. Poulain and

G. Maze, Opt. Lett. 31 (2006) 2553.18. C. Lopez, Evaluation of the photo-induced structural mechanisms in chalcogenide

glass, PhD Thesis (College of Optics and Photonics at the University of CentralFlorida, 2004).

19. A. Zoubir, Towards direct writing of 3D photonic circuits using ultrafast lasers, PhDThesis (College of Optics and Photonics at the University of Central Florida, 2004).

20. B. G. Aitken and C. W. Ponader, J. Non-Cryst. Solids 256&257 (1999) 143–148.21. N. Carlie, J. Hu, L. Petit, A. Agarwal, L. Kimerling and K. Richardson, submitted

to Journal Acta Materialia (2010).22. T. Anderson, L. Petit, N. Carlie, J. Choi, J. Hu, A. Agarwal, L. Kimerling, K. Richard-

son and M. Richardson, Optics Express 16(24) (2008) 20081–20098.23. T. Anderson, Fabrication of integrated optofluidic circuits in chalcogenide glass using

femtosecond laser direct writing, PhD Thesis (University of Central Florida, 2009).24. L. Petit, N. Carlie, K. Richardson, Y. Guo, A. Schulte, B. Campbell, B. Ferreira and

S. Martin, J. Phys. Chem. Solids 66 (2005) 1788–1794.25. J.-F. Viens, C. Meneghini, A. Villeneuve, T. V. Galstian, E. J. Knystaunas, M. A.

Duguay, K. A. Richardson and T. Cardinal, J. Lightwave Technology 17 (1999) 1184–1190.

26. K. Davis, K. Miura, N. Sugimoto and K. Hirao, Opt. Lett. 21 (1996) 1729–1731.27. A. Szameit, F. Dreisow, T. Pertsch, S. Nolte and A. Tuennermann, Optics Express

15(4) (2007) 1579–1587.28. A. Zoubir, M. Richardson, L. Canioni, A. Brocas and L. Sarger, J. Opt. Soc. Am. B

22 (2005) 2138–2143.29. R. Osellame, G. Della Valle, N. Chiodo, S. Taccheo, P. Laporta, O. Svelto and G.

Cerullo, Appl. Phys. A 93(1) (2008) 17–26.30. A. Zoubir, M. Richardson, C. Rivero, A. Schulte, C. Lopez, K. Richardson, N. Ho and

R. Vallee, Opt. Lett. 29(7) (2004) 748–750.31. L. Petit, J. Choi, T. Anderson, R. Villeneuve, J. Massera, N. Carlie, M. Couzi, M.

Richardson and K. C. Richardson, Opt. Mat. 31 (2009) 965–969.32. C. Meneghini and R. Villeneuve. J. Opt. Soc. Am. B 15(12) (1998) 2946–2950.33. S. Madden, D. Y. Choi, D. A. Bulla, A. V. Rode, B. Luther-Davies, V. G. Ta’eed, M.

D. Pelusi and B. J. Eggleton, Optics Express 15 (2007) 14414–14421.

May 15, 2010 9:11 WSPC/S0218-8635 145-JNOPMS0218863510005042


34. D. Y. Choi, S. Madden, A. Rode, R. Wang and B. Luther-Davies, Appl. Phys. Lett.91 (2007) 011115.1–011115.3.

35. I. Szleifer and M. A Carignano, Advances in Chemical Physics 94 (1996) 165–260.36. I. A. Luzinov, K. L. Swaminatha Iyer, V. Z. Klep and B. V Zdyrko, Surface graft

modification of substrates, US patent 7.026.014 B2 (2006).37. B. Zdyrko, O. Hoy, M. Kinnan, G. Chumanov and I. Luzinov, Soft Matter 4(11) (2008)

2213–2219.38. B. Zdyrko, K. Swaminatha Iyer and I. Luzinov, Polymer 47(1) (2006) 272–279.39. M. Kothe, M. Muller, F. Simon, H. Komber, H. J. Jacobasch and H. J. Adler, “In

Examination of poly(butadiene epoxide)-coatings on inorganic surfaces”, Colloids andSurfaces A-Physicochemical and Engineering Aspects 154 (1999) 75–85.

40. R. Boyd and J. Heebner, Appl. Opt. 40 (2001) 5742.41. S. Shaji, S. Eappen, T. Rasheed and K. Nair, Spectrochim. Acta Part A 60 (2004)

351–355.42. A. Yariv, Electron. Lett. 36 (2000) 321–322.43. J. Hu, X. Sun, A. Agarwal and L. C. Kimerling, J. Opt. Soc. Am. B 26 (2009) 1032–

1041.44. W. M. Irvine and J. B. Pollack, Icarus 8 (1968) 324.45. J. Laniel, J. Menarda, K. Turcotte, A. Villeneuve, R. Vallee, C. Lopez and K. Richard-

son, J. Non-Cryst. Solids 328 (2003) 183.46. A. van Popta, R. DeCorby, C. Haugen, T. Robinson, J. McMullin, D. Tonchev and

S. Kasap, Opt. Express 10 (2002) 639.47. P. Lucas, J. Phys.: Condens. Matter 18 (2006) 5629–5638.48. P. Lucas, E. King, A. Doraiswamy and P. Jivaganont, Phys. Rev. B 71 (2005) 104207.49. H. Fritzsche, Semiconductors 32 (1998) 850.

s0218863510005042 journal of nonlinear optical physics ...web.mit.edu/hujuejun/www/my papers/journal...

Documents