Telluride glasses possess the widest infrared window of all amorphous materials and are key to a number of long-wavelength applications such as bio-sensing. However they are not intrinsically good glass formers and require significant materials engineering for device fabrication. Strategies for stable glass engineering are presented and the fabrication of far infrared optical fibers is described. A new type of optical sensor based on electrophoretic capture of protein is also presented. This sensor is based on a conducting telluride glass which can act as both a capture electrode and an infrared optical element for collecting vibrational signatures of target molecules such as proteins.
©2013 Optical Society of America
Telluride glasses have recently been the subject or renewed interest for applications in far-infrared optics and non-volatile memories [1–5]. Non-volatile memories are based on materials that can switch between a glassy and crystalline phase within nanosecond time scale to take advantage of the difference in reflectivity (for optical memories), or resistivity (for electrical memories) . The amorphous tellurides selected for these applications are therefore chosen to be intrinsically unstable toward crystallization . Conversely, telluride glasses selected for optical application must exhibit a very strong resistance towards crystallization in order to prevent the formation of nuclei which results in unacceptable scattering losses . However, tellurium is intrinsically not a glass-forming element and such amorphous materials are hard to come by. Hence, the design of Te-based optical elements requires significant materials engineering.
Chalcogenide glasses (composed of S, Se or Te) constitute the only class of amorphous materials that offer wide transparency windows over the whole infrared domain up to 20 μm . The width of the transmission window is determined by the phonon energy spectrum of the amorphous network and is therefore controlled by the mass of the component elements. Hence, telluride glasses have wider transparency windows than selenide and sulphide . This is particularly relevant for optical elements involving long optical paths such as fibers where even the smallest absorption mechanism rapidly leads to opacity . Indeed, sulfide-based fibers start cutting off near 6 μm [12,13] while selenide-based fibers tend to cut off near 9 μm . However, many applications require operation in the long wave infrared region beyond 9 μm. This includes thermal imaging around 10 μm, defense applications in the second atmospheric window at 8-12 μm (such as missile counter measure or optical communication), but also the detection of bio-molecules which can be identified with high selectivity from their infrared signature in the 6-12 μm region. Such applications require telluride-based glasses which can transmit up to 12 μm in the fiber form and above 20 μm through thin windows . But while selenium is an excellent glass former, tellurium has a strong tendency for crystallization and it must be combined with an appropriate selection of modifier elements in order to form stable glasses.
In this paper we describe materials strategies for obtaining such stable telluride glasses and show that they can be effectively applied to the design of optical fibers as well as optical electrophoretic sensors for the capture and detection of bio-molecules. We describe the fabrication of single index and core/clad fibers as well as the capture and selective detection of proteins using a tellurium based bio-sensors acting as both an electrode and an infrared optical element.
2. Structural considerations for telluride glass fabrication
Elemental tellurium readily crystalizes at 450°C into a trigonal phase composed of hexagonally packed helicoid chains. Indeed, significant p-orbital overlap result in a metallic bonding character which is not conducive to glass formation . It is therefore necessary to break the chain symmetry and the metallic character through addition of neighboring elements to produce a random covalent network that promotes the formation of glass. Germanium is a tetravalent element with an electronegativity almost identical to Te (2.01 for Ge, 2.1 for Te) which effectively serves that purpose and leads to glass formation in the range near 15-20% Ge . For example the compound Ge15Te85 has a ΔT = Tx - Tg ~70°C (where Tg is the glass transition and Tx the crystallization temperature) and it has been used in the film form to produce waveguides for infrared interferometry [4,17]. Structural analysis show that the amorphous network is composed of GeTe4 tertrahedra connected by short Te chains as depicted in Fig. 1 [18,19].
The structure of Ge-Te glasses is therefore composed of a flexible three-dimensional covalent network that has sufficient degrees of freedom for glass formation. Nevertheless, a ΔT ~70°C is still too low for sensitive glass fabrication methods such as fiber drawing which inevitably leads to crystallite formation during manipulation above Tg. Hence, further reduction of the network connectivity is desirable to reduce structural rigidity and improve glass formation. This can be achieved through addition of elements with lower coordination that can further open up the structure. Iodine is an ideal candidate since it is monovalent but also because it has the same mass as Te and consequently will preserve the low phonon energy of the network. Structural analysis of a Ge20Te73I7 glass has revealed that iodine atoms bind exclusively to Ge atoms as depicted in Fig. 2.
Interestingly, iodine does not form terminal atoms at the end of Te chains and therefore serves its purpose of opening up the network by generating larger rings without breaking the continuity of the structure. This approach considerably improves glass formation and ΔT = 124°C adequate for glass molding has been achieved . But while iodine-based glasses provide outstanding optical and thermal performances, they are difficult to synthesize due to the high vapor pressure of iodine. Indeed, purification methods involving glass distillation are required for optical applications which result in significant losses of iodine and complicate a precise control of the glass stoichiometry. Another strategy for reducing the network connectivity therefore involves substituting tetravalent Ge by trivalent As. This is a viable approach and the Ge-As-Te system shows a broad glass formation domain . Structural analysis of these glasses indicate a strong violation of chemical order with large fraction of Ge/As-As bonds (Fig. 3) even in Te-rich compositions where each Ge and As could be surrounded entirely by Te [22,23].
Nevertheless, Ge-Te-As glasses show extremely good thermal stability and ΔT = 145°C has been achieved . With this level of stability, Ge-As-Te glasses can be safely molded to produce complex optical element such as aspheric and diffractive lenses. However fiber drawing requires higher fluidity (107 poise for drawing versus 109 poise for molding) which further raises the risk of crystallization and consequently low-loss optical fibers are very challenging to obtain with pure telluride glass compositions. Hence, additional strategies are required to further stabilize the glass against crystallization in order to produce fibers. To that avail, the most effective strategy is to add small fractions of Se to the glass composition. Indeed, such modified Te-based glasses have been successfully used to produce fibers [3,11,24]. But adding Se comes at a cost to the phonon energy and the infrared transparency as will be described below.
3. Long wave infrared fibers
Ge-As-Te-Se glasses for fiber fabrication are produced from 6N purity elements introduced in a silica tube under 10−6 Torr vacuum and purified in situ by distillation. After 12h of homogenization at 800°C, the melt is quenched in water and the resulting glass is annealed ~5°C below Tg for 3h. Typical glass rods for fiber preforms are 8mm in diameter and drawn under protective He atmosphere into fibers of diameter 150-300μm. The transmission of a Te-rich Ge10As15Te70Se5 single index fiber is shown in Fig. 4. The losses are quite considerable (>10db/m) despite the addition of Se into the structure. The inset of Fig. 4 shows a differential scanning calorimetry (DSC) curve of that glass which exhibit a sizeable crystallization peak near 250°C. This indeed suggests that crystallite nucleation may have occurred during fiber drawing, thereby resulting in scattering losses. But it is also important to note that glasses with large fractions of Te (~70%) tend to have small band gap energies (Eg<1 eV)  which tend to increase the population of charge carriers. Indeed, the conductivity of the Ge10As15Te70Se5 is 3.3 × 10−5 (Ω∙cm)−1 which indicate a significant population of free carriers that are known to contribute to background absorption in amorphous semiconductor fibers [11,24]. It is therefore necessary to reduce the fraction of Te in order to minimize this loss mechanism.
Accordingly, a Te-poor Ge15As40Te40Se5 single index fiber was fabricated and its transmission is shown in Fig. 5. The attenuation is much lower and reach < 3dB in the 8-11 μm range. The DSC inset shows that this glass not only has a higher Tg but also shows no crystallization peak. In addition the conductivity is now several orders of magnitude lower at 9.6 × 10−8 (Ω∙cm)−1. Hence, a lower fraction of Te improves the attenuation in two ways, first by reducing free carriers losses and second by averting nucleation and scattering losses. Nevertheless, it should be emphasized that the reduced fraction of Te comes at the cost of a reduction in long wave transmission. Indeed, it can be seen that while the Ge10As15Te70Se5 fiber had a minimum in transmission near 11.5 μm, the Ge15As40Te40Se5 fiber has a minimum in transmission near 9.5 μm. This is the result of higher phonon energies in the Te-poor glass which shift the cut off to lower wavelengths. But regardless, the overall losses are still lower in the Ge15As40Te40Se5 fiber even at longer wavelengths.
Many applications such as modal filters or interferometers require fibers with single mode propagation. In terms of fiber design this implies a core/clad structure with a core size smaller than tens of microns. Conventional fibers preforms have a core/clad size ratio around 1/2 which would imply drawing a fiber with a total diameter much smaller than 100 μm in order to obtain the proper core size for single mode propagation. Considering the relatively poor mechanical resistance of telluride glass, this is not a practically viable approach and instead a double draw method must be adopted as depicted in the inset of Fig. 6. In this method a first core/clad preform is drawn once to reduce the core size and it is subsequently introduced in a second cladding to reduce the core/clad size ratio. This method results in low-loss single mode fibers with low attenuation of ~3dB/m in the 6-10 μm range as shown in Fig. 6. The single mode propagation is evidenced by imaging the output of a fiber coupled with a CO2 laser at 10.6 μm. The output image shown in the inset of Fig. 6 reveals a single circular mode as expected from the refractive index ration of 3.025/3014 between the core and the cladding.
Due to the two subsequent drawings required for this approach, the risk of crystallization is greatly enhanced and the glass must be further stabilized with an additional fraction of Se (~16%). It should also be noted that significant absorption peak appears near 4.5 and 5 μm which can be assigned to Se-H impurity arising from water interfacial contamination during the rod-in-tube fabrication of the core/clad structure.
In summary, it is found that the fabrication of long wave transmission fibers is faced with an intrinsic limitation due to the correlation between band gap and optical window. Fibers compositions exhibiting long wave transmission suffer from free carrier losses due to low energy band gap, while low loss fiber with wider band gap inherently have lower wavelength cut-off due to higher phonon energy. In term of fiber applications, the electrical conductivity is therefore very clearly detrimental; however we show in the next section that it can be turned into an advantage for the development of bio-sensors.
4. Conducting telluride glasses for opto-electrophoretic sensing of proteins
In the previous section, it is shown that glass compositions containing very large fractions of Te (>70%) are inadequate for fiber applications due to their high electronic conductivity and lower resistance against crystallization which induces significant optical losses. However, the stringent requirements applied to fiber drawing are not necessary for the production of smaller optical elements such as Attenuated Total Reflection (ATR) crystals. These ATR elements can be produced by quenching a single batch of glass, thereby avoiding the remelting process that leads to nucleation and scattering losses during fiber drawing. And while a significant population of charge carrier creates a measurable attenuation in a fiber with a long optical path, the corresponding attenuation in a small optical element such as an ATR crystal is comparatively negligible. This effect is illustrated in Fig. 7 for a Ge10As15Te75 glass with a conductivity of 7.8 × 10−5 (Ω∙cm)−1 similar to that of the fiber depicted in Fig. 4. The transmission curve of the pure Ge10As15Te75 glass (x = 0) plotted in Fig. 7(b) does not show any significant losses (here it should be noted that the low baseline transmittance of this glass is due to reflection losses resulting from its high refractive index). However, if the conductivity is significantly increased, carrier losses become significant. Indeed, the electrical conductivity of the Ge10As15Te75 glass can be raised by doping with Cu and it is shown that the conductivity increases exponentially with Cu% (Fig. 7(a)). The resulting attenuation shown in Fig. 7(b) is now quite significant and become detrimental for sensing applications.
While electrical conductivity is not a useful attribute for an optical fiber, it could be exploited for the design of an optical element that serves simultaneously as an electrode and an infrared sensor. The goal of such a device is to induce an electrophoretic migration of target molecules towards the surface of the sensor where the evanescent optical wave is most intense. This is of particular interest for bio-sensing because most bio-molecules such as bacteria, virus or proteins carry a net surface charge that depends on the local pH and that is defined by their isoelectric point pI [25,26]. For local pH larger than their pI, bio-molecules carry a negative charge while they carry a positive charge for local pH lower than their pI. These charged molecules should therefore respond to an electric field and undergo electrophoretic migration under an applied voltage. A sensing device for capture and detection of bio-molecules based on that principle is depicted in Fig. 8(a). The sensor is composed of an ATR plate made of Ge10As15Te75 glass as the positive electrode and an Indium Tin Oxide plate as the negative counter-electrode. An applied voltage between these two electrodes drives negatively charged bio-molecule onto the ATR surface for optical detection using a conventional Fourier Transform Infrared spectrometer in the ATR mode.
Most bio-molecules have a pI lower than 7 and consequently carry a negative charge in deionized water. For example E. Coli bacteria have a pI = 4.5  and an example of successful electrophoretic migration in neutral water is depicted in Fig. 9(a) using a Ge15As10Te65Se10 ATR plate and an applied voltage of 12 V. The signal at time zero when only few bacteria diffuse through the evanescent wave is barely resolvable, but as soon as the voltage is applied, the signal intensity rapidly increases as the bacteria accumulate on the sensor surface.
After a deposition time of 35 min, the E Coli signal is well resolved and comparable to that of a dry E Coli culture as shown in Fig. 9(b). Indeed, the rich signature region of the bio-molecule including Amide I and II, phospholipids, polysaccharides and amino acids  is clearly observable and should therefore enables selective identification of a given target. The selectivity of the sensor is demonstrated in Fig. 10 using a pair of protein solutions: Lysozyme and Bovine Serum Albumin (BSA) deposited on a Ge10As15Te75 ATR with a voltage of 2.5 V. While Fig. 10(a) suggests that both spectra are almost identical, a Principal Component Analysis (PCA) of the two systems show that each protein solution can be selectively resolved using statistical spectral analysis as shown in Fig. 10(b). More details on the PCA method can be found in ref . It should be emphasized that each data point on Fig. 10(b) corresponds to a different electrodeposition experiment with a new proteins solution. It is also noticeable that the BSA spectrum in Fig. 10(a) is noisier than the Lysozyme spectra which results in larger data scatter in the PCA map of Fig. 10 (b).
It is expected that higher conductivity glasses containing significant Cu concentration would lead to a stronger electrical driving force for electrophoretic migration. This in turn would permit faster collection and detection rates for sensing applications. However, Fig. 7(b) shows that higher electrical conduction is associated with a significant drop in transmission which would be detrimental for optical detection. Hence, a trade-off must be found between collection speed and optical resolution.
Finally it should be mentioned that the electrodeposition process is reversible and bio-molecules can be lifted off the surface by reversing the applied voltage . This opens the potential for rejuvenating the sensor after a measurement, however at the moment a significant number of bio-molecules remain on the surface either due to chemisorption or due to loss of surface charge after denaturation. This issue may be solved in the future through functionalization of the sensor surface with a protective layer.
Telluride based glasses can be engineered to gain sufficient stability against crystallization for the production of complex optical elements such as single mode fibers. Stable glasses can be formed by optimizing the network rigidity through addition of lower coordination elements. Ultimately, small amounts of selenium must be added to help inhibit crystallization without significantly altering the infrared transparency window. It is found that a strong correlation exist between the band gap energy, the telluride content and the width of the infrared window. This correlation implies that glasses with far infrared transparency tend to suffer from optical losses due to free charge carriers. While this is detrimental for optical fiber engineering, the conducting nature of low band gap telluride glasses can be exploited for the development of sensors with short optical path. Indeed, an opto-electrophoretic sensor acting as a trapping electrode for charge bio-molecules and an optical sensor for collection of infrared signatures is shown to be effective for selective detection of protein and bacteria. Further materials engineering is required to optimize these sensors or to minimize losses in Te-based fibers.
This work was supported by the National Science Foundation under Grant Number ECCS-1201865, the CNRS International Associated Laboratory for Materials & Optics and the Partner University Fund.
References and links
4. C. Vigreux, E. Barthélémy, L. Bastard, J.-E. Broquin, M. Barillot, S. Ménard, G. Parent, and A. Pradel, “Realization of single-mode telluride rib waveguides for mid-IR applications between 10 and 20 μm,” Opt. Lett. 36(15), 2922–2924 (2011). [CrossRef] [PubMed]
5. Z. Yang, M. K. Fah, K. A. Reynolds, J. D. Sexton, M. R. Riley, M.-L. Anne, B. Bureau, and P. Lucas, “Opto-electrophoretic detection of bio-molecules using conducting chalcogenide glass sensors,” Opt. Express 18(25), 26754–26759 (2010). [CrossRef] [PubMed]
7. J. Kalb, F. Spaepen, and M. Wuttig, “Atomic force microscopy measurements of crystal nucleation and growth rates in thin films of amorphous Te alloys,” Appl. Phys. Lett. 84(25), 5240–5242 (2004). [CrossRef]
10. J. Lucas, “Infrared Glasses,” Curr. Opin. Solid State Mater. Sci. 4(2), 181–187 (1999). [CrossRef]
11. Z. Yang, O. Gulbiten, P. Lucas, T. Luo, and S. Jiang, “Long-wave infrared-transmitting optical fibers,” J. Am. Ceram. Soc. 94(6), 1761–1765 (2011). [CrossRef]
12. M. F. Churbanov, G. E. Snopatin, V. S. Shiryaev, V. G. Plotnichenko, and E. M. Dianov, “Recent advances in preparation of high-purity glasses based on arsenic chalcogenides for fiber optics,” J. Non-Cryst. Solids 357(11-13), 2352–2357 (2011). [CrossRef]
13. V. S. Shiryaev, J. Troles, P. Houizot, L. A. Ketkova, M. F. Churbanov, J. L. Adam, and A. A. Sibirkin, “Preparation of optical fibers based on Ge-Sb-S glass system,” Opt. Mater. 32(2), 362–367 (2009). [CrossRef]
14. J. Troles, V. Shiryaev, M. Churbanov, P. Houizot, L. Brilland, F. Desevedavy, F. Charpentier, T. Pain, G. Snopatin, and J. L. Adam, “GeSe4 glass fibres with low optical losses in the mid-IR,” Opt. Mater. 32(1), 212–215 (2009). [CrossRef]
15. R. M. Martin, G. Lucovsky, and K. Helliwell, “Intermolecular bonding and lattice dynamics of selenium and tellurium,” Phys. Rev. B 13(4), 1383–1395 (1976). [CrossRef]
16. A. A. Piarristeguy, E. Barthelemy, M. Krbal, J. Frayret, C. Vigreux, and A. Pradel, “Glass formation in the GexTe100-x binary system: Synthesis by twin roller quenching and co-thermal evaporation techniques,” J. Non-Cryst. Solids 355(37-42), 2088–2091 (2009). [CrossRef]
17. C. Vigreux, E. Barthelemy, L. Bastard, J. E. Broquin, S. Menard, M. Barillot, G. Parent, and A. Pradel, “Fabrication and testing of all-telluride rib waveguides for nulling interferometry,” Opt. Mater. Express 1(3), 357–364 (2011). [CrossRef]
18. P. Jóvári, A. Piarristeguy, R. Escalier, I. Kaban, J. Bednarčik, and A. Pradel, “Short range order and stability of amorphous Ge(x)Te(100-x) alloys (12 ≤ x ≤ 44.6),” J. Phys. Condens. Matter 25(19), 195401 (2013). [CrossRef] [PubMed]
19. P. Jóvári, I. Kaban, B. Bureau, A. Wilhelm, P. Lucas, B. Beuneu, and D. A. Zajac, “Structure of Te-rich Te-Ge-X (X = I, Se, Ga) glasses,” J. Phys. Condens. Matter 22(40), 404207 (2010). [CrossRef] [PubMed]
20. A. A. Wilhelm, C. Boussard-Pledel, Q. Coulombier, J. Lucas, B. Bureau, and P. Lucas, “Development of far-infrared-transmitting Te based glasses suitable for carbon dioxide detection and space optics,” Adv. Mater. 19(22), 3796–3800 (2007). [CrossRef]
21. Z. Yang and P. Lucas, “Tellurium-based far-infrared transmitting glasses,” J. Am. Ceram. Soc. 92(12), 2920–2923 (2009). [CrossRef]
22. P. Jóvári, P. Lucas, Z. Yang, B. Bureau, I. Kaban, B. Beuneu, and J. Bednarcik, “The structure of Ge-As-Te glasses,” J. Phys. Condens. Matter . in press.
23. S. Sen, S. Soyer Uzun, C. J. Benmore, and B. G. Aitken, “Structure, topology and chemical order in Ge-As-Te glasses: a high-energy x-ray diffraction study,” J. Phys. Condens. Matter 22(40), 405401 (2010). [CrossRef] [PubMed]
24. I. Inagawa, S. Morimoto, T. Yamashita, and I. Shirotani, ““Temperature dependence of transmission loss of chalcogenide glass fibers,” Jpn. J. Appl,” Phys., Part 1 36(4A), 2229–2235 (1997).
26. H. Miörner, P. A. Albertsson, and G. Kronvall, “Isoelectric points and surface hydrophobicity of Gram-positive cocci as determined by cross-partition and hydrophobic affinity partition in aqueous two-phase systems,” Infect. Immun. 36(1), 227–234 (1982). [PubMed]
27. P. Lucas, D. Le Coq, C. Juncker, J. Collier, D. E. Boesewetter, C. Boussard-Plédel, B. Bureau, and M. R. Riley, “Evaluation of toxic agent effects on lung cells by fiber evanescent wave spectroscopy,” Appl. Spectrosc. 59(1), 1–9 (2005). [CrossRef] [PubMed]
28. P. Lucas and B. Bureau, “Advanced infrared glasses for biochemical sensing,” in Biointerface Characterization By Advanced IR Spectroscopy, (Elsevier 2011), 217–243.