An innovative way to produce chalcogenide glasses and glass-ceramics for infrared devices is reported. This new method of synthesis at low temperature combining ball-milling and sintering by SPS (Spark Plasma Sintering) is a technological breakthrough to produce efficient infrared chalcogenide glasses and glass-ceramics. This technique will offer the possibility to strongly decrease the cost of infrared devices and to produce new chalcogenide glasses. It will also permit to increase the potential of some glass compositions by allowing their shaping at desired dimensions.
© 2011 OSA
Thermal imaging presents a large field of applications for defense and commercial uses thanks to great achievement in uncooled focal plan array detectors. These infrared systems work generally in the wavelength region of 8-12 µm. They take profit of the maximum emission of all objects near room temperature and of the high transmission of the third atmospheric window. Very few materials are transparent in this window and the very expensive single crystalline germanium is still largely used for making the optics. Chalcogenide glasses are substituting partially germanium for this application because of its lower cost and of the possibility of fabricating optics by molding. Theses glasses are always produced in sealed silica ampoules, batch by batch, which are very expensive and not reusable.
In this work, we report an innovative approach for making transparent chalcogenide glasses and glass-ceramics by combining mechano-chemical synthesis and Spark Plasma Sintering (SPS). The advantages of this approach are associated with the synthesis of amorphous chalcogenide powders near room temperature by thinly grinded raw metallic elements and then with the possibility of fast precision molding of infrared optics by Spark Plasma Sintering (SPS) from the obtained amorphous powder. This technique allows the production of thermodynamically unstable glasses which are indispensable for making transparent glass ceramics with greatly improved thermo-mechanical properties.
The ability to synthesize a bulk material from amorphous powders obtained by mechanical-alloying has recently been demonstrated for many systems such as metallic glasses [1–3]. Also, amorphous powders in the Li2S-P2S5 or AgI-As2Se3 systems have been synthesized using the same method for solid electrolyte applications [4–6].
Few papers reported the amorphization of metallic elements used in the manufacture of IR glasses; they are limited to the study of Ge-Se powder [7,8]. It is important to notice that none of these papers are relevant for the production of optical devices.
In this paper, we demonstrate for the first time the possibility to combine the amorphization process by mechanical milling of raw metallic materials at low temperature and the Spark Plasma (SPS) technique to make bulk glasses and glass-ceramics transparent in the infrared range. The SPS technique has recently proven its efficiency to fasten the crystallization step of bulk chalcogenide glasses compared to ventilated furnaces .
2. Experimental section
2.1 Glass powder synthesis
The main goal of this paper takes place in the amorphization process of raw metallic elements without heating them to their melting point under vacuum. Chalcogenide amorphous powders of composition 80GeSe2–20Ga2Se3 were prepared by mechanical alloying and by introducing stochiometric amounts of pure raw metallic germanium (5N, Umicore), gallium (6N, Cerac) and selenium (5N, Umicore) in a tungsten carbide (WC) grinding jar containing 6 WC milling balls, with a ball-to-powder weight ratio of 8:1. The jar was introduced into a planetary ball mill (Retsch PM100). Rotation cycles of 3min at 400rpm were scheduled with direction reversal and a pause of 3min between each cycle during 80h.
2.2 Pulsed current sintering
The bulk samples were prepared by introducing a proper amount of glass powder into a carbon die (diameters from 8mm to 36mm) with a tantalum foil in the inner part of the die as carbon diffusion barrier. The powder obtained after 80h of mechanical milling was sintered under vacuum at a temperature slightly higher than the glass transition temperature (Tg) by Spark Plasma Sintering (Dr. Sinter 505 Syntex). A maximum pressure of 50 MPa was applied. A temperature of 390°C was obtained over 4 minutes and kept constant for different dwell durations ranging from 2 minutes to 60 minutes before cooling.
2.3 Materials characterizations
Samples of powder were taken out from the jar at different milling durations. Thermal properties of the powder were analyzed by differential scanning calorimetry using a ramp of 10°C/min (DSC Q20, TA Instruments). The pure raw material amorphization was analyzed by carrying out X Ray Diffraction (XRD) spectra using a Philips PW3710 diffractometer (Cu Kα 1,5418Ǻ). Microdiffraction experiments were performed on a D8 Bruker equipment in the range 10-70° (2θ) every 1mm from the center of the 36mm SPS glass bulk to the periphery.
The as-prepared bulk glass slices were then polished on both sides and their optical properties were measured in the visible/near infrared (Perkin Elmer Lambda 850) and infrared (Bruker Vector 22 FTIR) regions.
The nanoparticles generated during the Spark Plasma Sintering treatment have been observed using SEM (FEG-SEM JEOL 7400F) and TEM (JEOL 2100F). The crystalline phases were analyzed by XRD analysis of a polished disc.
Mechanical properties such as Vickers hardness and toughness were measured and calculated respectively using a Vickers micro indenter with a charge of 100g during 5s, and Young’s modulus was obtained by measuring the ultrasound propagation speed in the glass. The thermal expansion coefficient (TEC) was measured with a TMA 2940 Calorimeter (TA Instruments) using a heating rate of 2°C/min from room temperature to 200°C.
3. Results and discussion
3.1. Amorphization process
We focused our work on the 80GeSe2-20Ga2Se3 (mol %) glass composition which demonstrated an excellent ability to make reproducible and controllable glass-ceramics. However, this glass has a tendency to crystallize during cooling. Only glass rods with a maximum diameter of 9 mm can be obtained when using the conventional melt-quenching technique with silica tubes. While these glass-ceramics present exceptional mechanical and optical properties, their small diameter, prevents them from any commercial applications.
The first step of this innovative technique consists in making an amorphous powder using mechanical energy. During the mechanical milling of initial metallic elements (Ge, Ga, Se), the resulting powder taken at constant time interval shows a change in color according to the milling duration (Fig. 1 ), indicating a lowering of particle size and a progressive reaction between the raw elements. The initial grey powder, corresponding to a mix of raw selenium, gallium and germanium, progressively turns into a red powder with increasing milling time.
The powder amorphization was checked by both X-Ray Diffraction (XRD) and Differential Scanning Calorimetry (DSC) analysis (Figs. 2 , 3 ). The crystalline raw materials turn into an amorphous material after 40h of milling (Fig. 2). The three main peaks observed at 27.3°, 45.4° and 53.9° (2θ) correspond to the crystalline Ge, while no crystallization peak due to Ga or Se is observed. For longer time, the decreasing shape of the baseline observed on XRD patterns can be attributed to a micro-structural change of amorphous powder.
The first peak observed in DSC analysis (Fig. 3) at about 200°C after 3 and 5 hours of milling is due to the melting of residual crystalline Se which has not reacted. A glass transition temperature (Tg) is clearly identified and confirms the production of an amorphous powder after 3h of milling. With increasing milling time, the melting peak of Se vanishes and a shift of both glass transition temperature and crystallization temperature is observed. What is surprising is the third crystallization peak observed on the DSC curves from 8h to 20h of milling (Fig. 3). The XRD pattern made on the powder after the DSC experiment performed at 10°C/min up to 470°C shows the crystallization of GeSe, GeSe2 and Ga2Se3 or Ga4GeSe8, as both of these two phases present extremely close patterns [10,11]. As shown in Fig. 3, the glass transition temperature, Tg, evolves according to the milling time. In order to make a comparison, the glass transition temperature observed with glasses of composition 80GeSe2-20Ga2Se3 made in sealed silica tube under vacuum is about 350°C . As shown, this Tg is reached after 20 hours of ball milling, assuming that an equivalent glass microstructure is obtained using both techniques.
Moreover, as already demonstrated the competition between two phases to crystallize (two close crystallization peaks observed in the DSC curve) permits to obtain controllable and reproducible glass-ceramics with the 80GeSe2-20Ga2Se3 base glass. The first peak corresponds to the crystallization of a Ga2Se3 or GaGe4Se8 crystalline phase while the second peak is due to GeSe2 . Another interesting point is the increase of the difference between the crystallization temperature and the glass transition temperature after 80h of ball milling leading to an amorphous powder with a higher thermal stability.
The amorphization process could be explained by two phenomena. The first one consists in a fine chemical reaction between each element, while the second one could be attributed to the localized melting/quenching process due to the high energy of the impact characteristic of ball milling process. V. Caron has demonstrated that the first process occurs when the milling temperature is lower than the glass transition temperature, Tg . Considering the high Tg of the 80GeS2-20Ga2Se3 glass composition, we could assume that a fine chemical reaction occurs between the raw materials leading to an amorphous powder. Finally, XPS measurements were performed on the 80h milling powder and confirm the composition 80GeSe2-20Ga2Se3.
3.2. Sintering of infrared bulk glasses
The challenge was then to sinter the powder obtained from a 80h-ball milling treatment by SPS. The main obstacle was to prevent the glass bulk from carbon contamination as usually observed in processed ceramics with such a technique and which is detrimental for optical properties . The process consists of treating powders in conducting dies (graphite die with inner papyex foil or tantalum foil, diameters from 8mm to 36mm) under uniaxial pressure, where direct current pulses allow temperature to increase over a short time period (a few minutes) which limits the uncontrolled crystallization of the glass in our case.
Compared to alternative processes, such an approach offers an advantage and an improvement where energy and time savings are concerned. The density, the optical properties as well as the synthesis of perfect glass and glass-ceramics in situ by SPS was optimized by control of the different parameters (heating rates, temperature, pressure protocols, and holding time at sintering temperature). A sintering treatment at 390 °C under 50MPa (2 minutes dwell time) results in a self supported glass bulk with a density (4.39) equal to the glass 80GeSe2-20Ga2Se3 synthesized by melt-quenching. The study of the shrinkage dz/dt (z: displacement of the punches during SPS experiment) and the load applied (50MPa) as a function of temperature indicates that the densification proceeds in two steps and is over at 370°C (Fig. 4 ). The mechanisms at the origin of densification will be described in a forthcoming paper. The as-produced glasses with diameters ranging from 8mm to 36mm were optically polished on both faces and their optical properties were characterized.
The transmission of the obtained bulk glassy samples whatever their dimension is good in the mid infrared range as observable in the picture taken by a thermal camera working in the third atmospheric window from 8 to 12µm (Fig. 5 ). The absorption observed in this picture is firstly due to reflection on both faces because of the high refractive index of the glass (n~2,41). Secondly, the presence of oxygen leading to the formation of Ge-O bonds inside the bulk glass induces phonon absorption and finally a slight diffusion of C graphite (papyex) or tantalum can also interfere. These parameters can be improved by working under controlled atmosphere from the beginning to the end of the experiments and by pre-compacting the powder at room temperature. The sintering of large diameter is challenging in SPS experiment due to thermal inhomogenities between the center and the periphery of the pellet.
A microdiffraction experiment has been carried out to investigate the amorphous character of the glass bulk along the diameter. Results show that no crystallization peaks are observed in the 36mm diameter glass and the glass is still amorphous and homogeneous (Fig. 6 ).
It is important to point out that the usual way of synthesis using the melt-quenching technique and using a sealed silica tube limits the quenching rate because of the low thermal conductivity of silica. Unstable glasses (Tc-Tg<100°C), such as the composition 80GeSe2-20Ga2Se3 (Tc-Tg = 82°C) present a weak resistance to crystallization. Increasing the silica tube diameter irremediably leads to the melt bath crystallization during quenching. In the case of the 80GeSe2-20Ga2Se3 glass composition, several tries have been made with different silica tube diameters. Homogeneous glasses without crystallization can be synthesized using a maximum silica tube inner diameter of 9mm. As shown in Fig. 5, the innovative way of synthesis we developed ensures the preparation of higher diameter samples needed for optical lenses in the case of infrared cameras for example. In the case of the 80GeSe2-20Ga2Se3 glass composition, the glass surface has been increased 16 times allowing its manufacture for many applications (thermal imaging, new laser source if doped with rare earth ions, non linear applications such as optical switching, etc). One of the limiting factors in the glass or glass-ceramic dimension is the mold size (36mm in our case).
First analysis on bulk samples by XRD reveals the presence of carbon particles inside the glassy matrix (SPS parameters 390°C, 50MPa, 2 minutes dwell time) when using papyex as inner foil (Fig. 7 ). To prevent the glass bulk from such a contamination in the next experiments, tantalum foil has been added in the inner part of the graphite die as a carbon diffusion barrier (Fig. 7). The efficiency of tantalum foil is conclusive as no crystallization peak of carbon or tantalum particles is then observed.
3.3. Generation of nanoparticles
The second challenge was to generate nanoparticles into the glassy matrix to improve its mechanical properties known to be the weak point of chalcogenide glasses. Glass-ceramics are obtained with similar thermo-mechanical cycles, with longer dwell time ranging from 2 to 60 minutes (Fig. 7). The results obtained for a 30 minutes SPS treatment time clearly show that the Ga2Se3 (or Ga4GeSe8) crystalline phase has nucleated and grows homogeneously inside the glassy matrix. The preference to form such crystals has been discussed already . For longer treatment time (60 minutes) a second phase, GeSe2 is appearing (Fig. 7).
Good optical properties are confirmed with transmission spectra in the infrared region, from 2 µm up to 11 µm (Fig. 8 ). A strong absorption band at 12.5 µm corresponding to the vibration of the Ge-O bonds within the glass is observed. In order to understand when the oxidation phenomena occurs, 10g of pure 80GeSe2-20Ga2Se3 glass made by the conventional way using sealed silica tube has been manually and finely grinded. The bulk glass-ceramic obtained by sintering this amorphous powder in a similar way as the previous experiments (390°C, 50MPa, 2 minutes dwell time) shows an excellent infrared transmission up to 16µm without any absorption bands. This last result clearly shows the low airtightness of mechanical milling jars. So to optimize the infrared transmission the planetary grinder will have to be placed in a glove box under controlled atmosphere.
Rayleigh scatterings observed at shorter wavelengths for the samples with longer dwell time (glass-ceramics) are due to the presence of Ga2Se3 nanoparticles in the glassy matrix [10, 14]. Then, the decrease of the maximum of transmission is mainly due to the rapid crystallization of GeSe2 microparticles (from 0.5 to 5 µm) inducing MIE scatterings.
SEM analysis was used to determine the size of the crystals generated in situ by SPS from longer thermal treatments. The particles size has been analyzed according to the sintering time and compared with the mechanical properties of the as-made glass-ceramics. A crystallization of about 0, 15%, 40%, 70% of nanoparticles having a size ranging from 50 to 200 nm is observed in Figs. 9 a, b, c and d respectively. The TEM picture (Figs. 9–e) clearly shows the mechanism of growth of Ga2Se3 nanocrystals ranging from 50 to 200 nm which are progressively constituted of aggregates of particles of 5 to 10 nm. For longer treatment time (60 minutes), GeSe2 crystals appear (Figs. 9–f).
Mechanical properties of glass-ceramics are compared to the base glass made by the conventional melt quenching method in Table 1 . All the samples, including the 36mm diameter, present good mechanical handling and no cracks even if the cooling rate was not controlled during the SPS experiment. The three glasses present similar physical and mechanical properties: a density of 4.40 ± 0.2, hardness of 202 ± 2 Hv, toughness Kc of 0.186 ± 0.005 MPa.m1/2 and a low thermal expansion coefficient of 12.1 ± 0.3.10−6 K−1. These results are in good agreement with the mechanical properties recently observed with the glasses synthesized using the conventional silica tubes .
An important increase of toughness is observed, linked to the growing percentage of generated nanoparticles inside the glassy matrix and then to the growing size of nanoparticules from 50nm to 200nm (Fig. 9, Table 1). The hardness remains constant with increasing number of Ga2Se3 particles but drastically decreases when GeSe2 particles start growing inducing a decrease in the glass-ceramic density. In order to make a comparison, it is important to notice that the GASIR glass (Ge22 As20 Se58) which is one of the most used chalcogenide glasses for IR devices present a hardness of 170 Hv, a Young Modulus of 17.89 GPa and a thermal expansion coefficient of 17.10−6 K−1. Furthermore, glasses and glass-ceramics containing gallium present a lower thermal expansion coefficient compared to antimony or arsenic based glasses usually commercialized. This low thermal expansion coefficient which is constant and equal to 12.10−6 K−1 whatever the crystallization time, leads to an increase in the resistance to thermal chocks.
As demonstrated, the nucleation and growth steps can be easily controlled by optimizing the temperature and time of sintering. Thus, a glass can be obtained after 2 minutes at 390°C under 50MPa while glass-ceramics with adjusted number and size of particles can be synthesized with increasing sintering time from 15 to 60 minutes (tens of hours with a ventilated furnace).
All these results highlight the feasibility to make bulk glasses and glass-ceramics transparent in the infrared range using a new process of synthesis combining ball milling and Spark Plasma Sintering techniques. This new manufacturing process leads to an important breakthrough in the way to make materials transparent in the infrared range. It would first avoid the use of complex and expensive silica set-up, and second, it permits to eliminate the melting reaction at high temperature. Furthermore, many chalcogenide glasses presenting real potential applications were set aside since the synthesis in sealed silica tubes under vacuum did not permit to produce samples big enough for applications. This technique offers the possibility to make glasses or glass-ceramics of desired shape and size sintering an amorphous powder slightly higher than the glass transition temperature in few minutes. Thus, this new process would lead to lower manufacturing costs of new chalcogenide glasses and glass-ceramics which could be adapted to numerous set-ups such as thermal imaging, infrared laser sources, IR waveguides, IR sensors, etc... Ultimately, this new synthetic route will lead to larger scale production of infrared optics, broadening the target market to a more general public.
This work has been supported by the French DGA (Délégation générale pour l’Armement). Authors would like also to thank the MRCT-CNRS (Mission Ressources et Compétences Technologiques) for its financial support, Dr. M. Dollé (CEMES-CNRS, France) for microdiffraction experiments and B. Soulestin (SPCTS-CNRS) for TEM observations.
References and links
1. X. Yan, X. Song, N. Lu, E. Li, and J. Zhang, “A novel route for preparing binary Sm–Co bulk amorphous alloys,” Mater. Lett. 62(17-18), 2862–2864 (2008). [CrossRef]
2. U. Patil, S. J. Hong, and C. Suryanarayana, “An unusual phase transformation during mechanical alloying of an Fe-based bulk metallic glass composition,” J. Alloy. Comp. 389(1-2), 121–126 (2005). [CrossRef]
3. T. S. Kim, J. K. Lee, H. J. Kim, and J. C. Bae, “Consolidation of Cu54Ni6Zr22Ti18 bulk amorphous alloy powders,” Mater. Sci. Eng. A 402(1-2), 228–233 (2005). [CrossRef]
4. A. Hayashi, S. Hama, H. Morimoto, M. Tatsumisago, and T. Minami, “Preparation of Li2S–P2S5 amorphous solid electrolytes by mechanical milling,” J. Am. Ceram. Soc. 84(2), 477–479 (2001). [CrossRef]
5. M. Sekine, Y. Suzuki, H. Ueno, Y. Onodera, T. Usuki, T. Nasu, and S. Wei, “Appearance of fast ionic conduction in AgI-doped chalcogenide glass powders prepared by mechanical milling,” J. Non-Cryst. Sol. 353, 2069 (2007).
6. J. Trevey, J. S. Jang, Y. S. Jung, C. R. Stoldt, and S. H. Lee, “Glass–ceramic Li2S–P2S5 electrolytes prepared by a single step ball billing process and their application for all-solid-state lithium–ion batteries,” Electrochem. Commun. 11(9), 1830–1833 (2009). [CrossRef]
7. Y. Shirakawa, T. Matsuda, Y. Tani, A. Shimosaka, and J. Hidaka, “Amorphization of Ge–GeSe mixtures in mechanical alloying process,” J. Non-Cryst, Sol. 293–295, 764 (2001).
8. K. D. Machado, J. C. De Lima, C. E. M. Campos, A. A. M. Gasperini, S. M. De Souza, C. E. Maurmann, and T. A. Grand andP. S. Pizani, “Reverse Monte Carlo simulations and Raman scattering of an amorphous GeSe4 alloy produced by mechanical alloying,” Solid State Commun. 133, 411 (2005). [CrossRef]
9. G. Delaizir, M. Dollé, P. Rozier, and X. H. Zhang, “Spark plasma sintering: an easy way to make infrared transparent glass–ceramics,” J. Am. Ceram. Soc. 93(9), 2495–2498 (2010). [CrossRef]
10. L. Calvez, H. L. Ma, J. Lucas, and X. H. Zhang, “Selenium-based glasses and glass ceramics transmitting light from the visible to the far-IR,” Adv. Mater. (Deerfield Beach Fla.) 19(1), 129–132 (2007). [CrossRef]
11. M. Rozé, L. Calvez, Y. Ledemi, M. Allix, G. Matzen, and X. H. Zhang, “Optical and mechanical properties of glasses and glass-ceramics based on the Ge-Ga-Se system,” J. Am. Ceram. Soc. 91(11), 3566–3570 (2008). [CrossRef]
12. V. Caron, J. F. Willart, F. Danede, and M. Descamps, “The implication of the glass transition in the formation of trehalose/mannitol molecular alloys by ball milling,” Solid State Commun. 144(7-8), 288–292 (2007). [CrossRef]
13. G. Bernard-Granger, N. Benameur, C. Guizard, and M. Nygren, “Inversion defects in MgAl2O4 elaborated by pressureless sintering, pressureless sintering plus hot isostatic pressing, and spark plasma sintering,” Scr. Mater. 60, 164 (2009).
14. H. Ma, L. Calvez, B. Bureau, M. Le Floch, X. Zhang, and L. Jacques, “Crystallization study of infrared transmitting glass ceramics based on GeS2–Sb2S3–CsCl,” J. Phys. Chem. Solids 68(5-6), 968–971 (2007). [CrossRef]