solidi status pss physica Phys. Status Solidi C 9, No. 12, 2415–2419 (2012) / DOI 10.1002/pssc.201200197 c www.pss-c.com current topics in solid state physics Simulation and process flow of radiation sensors based on chalcogenide glasses for in situ measurement capability Mahesh Ailavajhala1, Maria Mitkova*,1, and Darryl P. Butt2 1 2 Department of Electrical and Computer Engineering, Boise State University, 1910 University Dr., Boise, ID 83725-2075, USA Department of Material Science and Engineering, Boise State University, 1910 University Dr., Boise, ID 83725-2090, USA Received 29 May 2012, accepted 20 August 2012 Published online 16 November 2012 Keywords chalcogenide glasses, gamma radiation, radiation sensing, silver diffusion * Corresponding author: e-mail mariamitkova@boisestate.edu In this work we present data about electronic devices based on a planar structure; inert electrode/nanophase chalcogenide glass/inert electrode in close proximity with a source of silver (Ag) oriented laterally over the chalcogenide glass film. The conductivity of the devices changes with radiation and it can be measured by contacting the two inert electrodes. Spacing of the electrodes was chosen after an in-depth investigation into the electric field (E-field) and E-field energy displacement simulations with the aid of COMSOL multi-physics software. Simulations have been used to enlighten a specific device structure to perform in situ measurements. Bias voltages, inert electrode material and thickness of the films were used as standards for all different types of simulations while only varying the spacing and the geometries to affect the E-fields. It has been established from the experimental results that a 1 V bias is the most appropriate for the device performance and this is the voltage used in the simulations. The key motivation for this research is to find appropriate dimensions and geometry, which do not cause a change in conduction as a direct result of the applied E-field, but rather effectively contribute to the establishment of only the radiation induced change of the device conductivity. The process flow for the device fabrication is described and data from the device performance are presented as well. The radiation induced Ag doping is mapped in the device volume using electron dispersion X-ray spectroscopy (EDS). © 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1 Introduction Chalcogenide glasses continue to attract a widespread interest in condensed mater science, largely because they have opened new physics of disordered state and since they have made new applications possible. Specifically, these amorphous semiconductors exhibit a great variety of radiation-induced phenomena, because of the freedom and flexibility associated with their atomic structure. One of the interesting aspects in the chalcogenide glass performance are the radiation induced effects caused by ionizing radiation such as γ-quanta. Electronically these glasses are a type of semiconductors with an abundance of states within the band gap due to lone-pair electrons localized at chalcogen atoms and for that reason they can be excited by a wide range of wavelengths including γ-rays. In addition, the excited carriers are effectively localized in disordered and defective glass structures, and the carriers undergo strong electron-lattice coupling. As a result, the electrical properties of the glasses undergo remarkable change which can be utilized for radiation sensing. These effects have not been explored in depth, because of the expectation that due to the lack of order in the structure of the glasses and the presence of a large number of defects, fast recombination of radiation induced generation of electron-hole pairs will also occur. This is in fact true but never the less, transient effects [1] and more long-term effects [2] on the fundamental absorption, ageing and other properties have been reported [3,4]. Similarly, these effects have been used in the x-ray radiology [5] where the occurring electron-hole pairs are captured by a transistor array contributing to the formation of an image [6]. Structurally the chalcogenide glasses have moderate atomic connectivity. This brings about the formation of © 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim solidi status physica pss c 2416 M. Ailavajhala et al.: Simulation and process flow of radiation sensors structure with different densities allowing for the diffusion of metals. Silver (Ag) is a metal of great interest since it has very high mobility; its diffusion can be accelerated by photons to give rise to the effect of photodiffusion [7]. As a result, a tremendous change in the electrical properties of the hosting glass can occur even after the introduction of few ppm Ag+, as shown for Ge-Se glass [8]. We applied the combination of the two effects - radiation induced generation of electron-hole pairs and radiation induced Ag diffusion in chalcogenide glasses to create elements for radiation sensing [9]. The proof of concept for the sensing devices performance has been demonstrated by the formation of a generic device shown in Fig. 1. It has a lateral structure, consisting of a chalcogenide film deposited on an insulating substrate, two electrodes made from electrochemically inert material, for example Al, W, Ni, or ID + - VD Figure 1 General presentation of the radiation sensor. or Pt and a source of Ag. The device has been characterized through measurement of its I-V characteristics at discrete radiation doses with the application of a DC voltage sweep applied onto the inert electrodes and simultaneously measuring the current. The initial pre irradiated state of the devices can be recovered by the application of an E-field between the Ag electrode being a cathode and the two inert electrodes being anodes. Impedance measurements have been performed to get exact data related to radiation induced Ag diffusion causing changes in the conductivity of the material while excluding the influence of an applied electrical field between the sensing electrodes [9]. One of the most important characteristics of these devices is their ability for real time radiation measurement. During the real time measurements, the applied electrical field is constant throughout the entire measurement; one needs such device structure which can ignore the effects of the electrical field induced Ag diffusion. To find out which device structure will correspond to such requirement, we conducted device simulations with the aid of COMSOL software and after illustrating the appropriate device structure, masks were prepared and a specific process flow for the fabrication of the device structure was developed. We tested the devices and conducted materials characterization. These data are presented below. 2 Experimental details COMSOL software was used to model the devices for E-field and electric potential that is generated due to the application of a constant DC © 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim voltage biased. Two dimensional studies have been performed to study the dispersive effect of the electric field and resultant E-field effects that occur when a 1 V bias is applied between the two inert electrodes while leaving the Ag pads at a floating voltage. Various geometries have been simulated, to study which geometry will provide the least disruptive E-field and therefore will result in measurements corresponding solely to radiation induced effects. Each geometry was also simulated for a large device (1 mm spacing between the pad directly across from one another), small device (10 µm spacing between pad directly across from one another) and most of the geometries were also simulated with the case where the top and bottom pads (simulating the Ag pads) were moved 5 µm away from the previous location to study whether a specific ratio of Al spacing to Ag spacing will result in reducing the E-field effects. Materials data used for simulation were carefully selected to ensure the closest possible similarity to the thin films used in the study. A large square was created to represent the large area for the thin chalcogenide film and the material similar to it was the ternary Ge22As20Se58 since this composition closely resembles the actual chalcogenide glass composition and was preexisting within the materials library. The contacts were simulated using material data for aluminum and silver contacts. Application of a potential bias causes an E-field which has been simulated for each of the geometries. Same method has been applied for the amount of electrical energy that is imparted into the film as a direct relation to the electric potential. The electrical energy density provides evidence of the amount of energy transferred to an ion in certain location, measured in Joules/m3. The sizes of the shapes have been scaled to make them proportional to the distance between the pads. The circle pad sizes used in the simulation were 2 mm in diameter with 1 mm spacing. Same was the length of the rectangles, but the height of the triangles was 1 mm. The dimensions of all shapes in certain simulations were the same to prevent any anomalies due to distorted shapes. The simulations also contained a parameter that accounts for the 100 nm film thicknesses, which is referred in COMSOL as “Out of Plane Thickness” this corresponds to the film thicknesses for the fabricated devices. To verify the Ag ions distribution after the radiation, mapping of the materials composition of the devices with Hitachi S-3400N-II has been performed at the 20 keV beam using energy dispersive X-ray spectroscopy (EDS) operation mode. 3 Results Two representative results of COMSOL simulation are shown in Fig. 1 whereas the results in the Table 1 are the maximum values presented on the graphs. The electrical potential (voltage) experienced by the top and bottom Ag sources is approximately 0.5 V with slight deviations in the areas around these electrodes due to the geometries. From these graphs, the potential voltage exwww.pss-c.com Contributed Article Phys. Status Solidi C 9, No. 12 (2012) 2417 perienced by the Ag pads is going to be half of the applied potential on one Al pad if the other Al pad is grounded. The dark blue color indicates low values approximately 0 and the red – high values. Each topology has similar values for the maximum electric field and energy density but the regions that are affected vary depending on the geometry. Visually, circle geometry provides a dispersive E-field as well as large amounts of E-fields directly affecting the top and bottom Ag sources. To improve this design, the top and bottom pads were moved farther apart from each other which substantially reduce the effective E-field. Further reduction has been achieved by introducing the device design with antenna-type structure pointing to the center point between the electrodes as shown in Fig. 2. Based on the results from the COMSOL simulation, it has been established that devices with the antenna type of structure as shown in Fig. 2 confines the E-field and the electric density effects to very small areas between the two antennas. Another characteristic of these simulations is the ability to create many devices of different dimensions using a specific ratio of 1/1.5 which corresponds to Al electrode spacing to Ag source spacing respectively. Using the aforementioned ratio and with the aid of Visio software, masks for these devices were created of which one such device is shown in Fig. 3(a). Geometry Electrical Energy Density Electric Field Potential Figure 2 Geometry, electrical energy density, E-field and potential of circular and triangle shapes with antenna structure. Al Al Ag (a) Ag (b) Figure 3 (a) Mask arrangement example for devices with antennas. (b) Devices post fabrication process. The process flow for device fabrication starts with a photolithography, on top of an insulator, that creates openings for Al deposition followed by a secondary photolithography step to allow for the deposition of the Ag pads. This step is followed by a third lithography which covers the large pads of the Al and Ag pads with photoresist while allowing for the deposition of a 100 nm chalcogenide glass film to deposit over the entire structure with the exception of the areas covered by the photoresist. This process ensures that the chalcogenide glass film is not exposed to the basic solutions that are customary in the contemporary photolithography processes since these glasses dissolve in a basic solution. Following the Al deposition process, a thin layer of Ag was deposited capping the Al pad, without breaking vacuum to prevent the oxidation of Al. Devices prepared using this process are shown in Fig. 3b. The devices were placed in a 60Co camera and radiated with a 3 Mrad dose of γ-rays after which electrical testing and material characterization were performed. 1 0 0 .0 p 3 5 0 μ m A l sp a cin g 3 0 0 μ m A g sp a cin g Table 1 Maximum data for the devices and fields used in the simulation. Circle Energy (J/m^3) E_field (V/m) Rectangle and Triangle Energy (J/m^3) E_field (V/m) www.pss-c.com 1mm 5.21E06 10mico 0.095 0.0777 0.0761 1084.8 1.46E+05 1.32E+05 1.31E+05 1mm 8.23E06 1348.8 20micro C urrent (A) 5 0 .0 p 0 .0 D e v ic e D e v ic e D e v ic e D e v ic e 30micro P re R a d s w e e p 1 P re R a d s w e e p 2 P o s tR a d (3 M ra d ) s w e e p 2 P o s tR a d (3 M ra d ) s w e e p 1 - 5 0 .0 p 10 micro Adjusted 0.0638 0.0638 1.18E+05 1.19E+05 0 .0 0 .2 0 .4 0 .6 0 .8 1 .0 V o lt a g e ( V ) Figure 4 I-V characteristics obtained by two sweeps before and following radiation. The difference in the current flowing through the devices before and after radiation is relatively low compared to the big dot devices tested in [9] which we relate to the lower supply of Ag in the antenna type devices in which the source in the active device area has very limited dimensions and big spacing (Al at 350 μm and Ag at 300 μm). © 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim solidi status physica pss c 2418 M. Ailavajhala et al.: Simulation and process flow of radiation sensors electric displacement field, and μ magnetic permeability. Then the total energy could be calculated as follows: U= 1 ( E ◊ D + H ◊ B) 2 (2) U= 1 Ê 2 B2 ˆ εE + ˜ μ¯ 2 ËÁ (3) The total energy has been contemplated in the simulations which guarantee that all external field factors that additionally could affect Ag diffusion are considered. When discussing Ag doping in chalcogenide glasses Figure 5 EDS data of Ag distribution after radiation . due to radiation induced effects, it is important to consider Figure 5 shows the Ag distribution (black dots) after radia- that an intrinsic E-field is also generated. It occurs because tion using EDS mapping technique which illustrates a large the thermally generated electrons in the doped region move concentration of black dots that signify Ag atoms, as towards the non-doped part generating a field at the intershown on the Ag pads. From the figure one can see that face, which is limited by the equilibrium conditions, charAg has diffused into the area between the two inert elec- acteristic for diffusion and drift of the charge carriers. It trodes which substantially changes the electrical conduc- contributes to the changes of the electrical conductivity of tivity of the devices. These devices can be reversed using the film as a sub set of the change in conductivity during radiation and is accompanying the Ag diffusion. After cesthe aforementioned procedure. sion of the radiation, the relaxation of the radiation induced electron is associated with ion relaxation of the diffused 4 Discussion The effect of Ag migration in chalco- Ag. This intrinsic field dominates at low external voltage genide glasses and specifically Ge containing glasses under that is used to measure the conductivity in the device. Simthe influence of electrical field is well known [10]. The ilarly, as stated in [12] and accordingly to our simulation electric field also affects the Ag photodoping [11]. It is ob- results, there is an external threshold voltage of 10.8 V, vious, that application of an E-field during the testing of above which the radiation induced field, that is present unthe radiation sensing devices can influence their diffusion der the radiation doping, is strongly affected, i.e. massive parameters and introduce E-field related effects. Based on Ag diffusion is induced through this field. Our tests are the modeling, we realized that for in situ measurements, limited at a lower voltage to avoid this effect. The EDS rethe best results can be achieved with the devices shown in sult showing homogeneous distribution of Ag between the Figs. 2b and 3a, where the Ag diffusion is unaffected by device electrodes is a proof that at the conditions of the the E-field since its value within the diffusion region is measurement, the external electrical field did not affect the very limited or close to zero. It was proved to be the least Ag diffusion process. At the applied voltage and fields in the modeling secinvasive method of applying a bias without ion displacement, while some of the other device models with robust tion (Table 1) it is obvious that in the case shown in Fig. 2a contact areas in close proximity allow better Ag diffusion there would be very strong Ag migration due to the electrical field applied. We experimented with a large number of when used towards measuring at discrete radiation doses. Another important aspect for consideration, included shapes and electrodes orientation but because of space limin the software, is the amount of energy imparted into a itation, in this work, only two cases have been illustrated. given space or medium as a direct result of the electric bias. The devices that have an antenna type of electrodes, disThis is also known as the energy density due to electric and tributed on a distance of 10 mm to 250 µm at applied bias a magnetic field, of which the electric field is proportional of 1 V can be applied in real time tests. to the permeability and the E-field as shown in the follow5 Conclusions Applying COMSOL software the oping equation: erational conditions of radiation sensing devices were sim- ηE = energy 1 2 = εE volume 2 (1) Here ηE is the energy density due to electric field, ε the dielectric constant of the material. It is critical to measure this energy since electron–hole pair (EHP) could be generated as a direct result of the E-field rather than radiation. Complete energy equation also considers the role the magnetic fields plays in the system, with U being the total energy, B the magnetic field, H the magnetizing field, D the © 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ulated. It is shown that the structure of the devices strongly affects the dependence of their performance upon the applied external field. Devices with antenna structure of their electrodes with various spacing ranging from 10 mm to 250 µm and external field of 1 V bias are shown to ensure stable performance of the sensors for real time measurement. At these conditions the internally forming field due to radiation induced Ag diffusion dominates the devices performance. The data about Ag distribution between the source and inert electrodes support this conclusion. www.pss-c.com Contributed Article Phys. Status Solidi C 9, No. 12 (2012) Acknowledgements The authors thank to P. Chen for the high resolution post fabrication devices image and D. Oleski for the electrical test measurement. This work is supported by a grant from Battelle Energy Alliance under Blanket Master Contract No. 41394. References [1] T. Minami, A. Yoshida, and M. Tanaka, J. Non-Cryst. Solids 7, 328 (1972). [2] V. Balitska, A. Kovalsky, O. Shpotyuk, and M. Vakiv, Radiat. Meas. 42, 941 (2007). [3] A. Kovalskiy, Radiat. Eff. Defects Solids 158, 391 (2003). [4] R. Golovchak, A. Kozdras, S. Kozyukhin, and O. Shpotyuk, Nucl. Instrum. Methods Phys. Res. B 267, 2958 (2009). [5] K. Shimakawa, K. Fukami, H. Kishi, G. Belev, and S. Kasap, Jpn. J. Appl. Phys. 46, L192 (2007). www.pss-c.com 2419 [6] K.S. Karim, A. Nathan, J.A. Rowlands, and S.O. Kasap, IEEE Proc. Circuits Devices Syst. 150, 267 (2003). [7] M. Frumar and T. Wagner, Curr. Opin. Solid State Mater. Sci. 7, 117 (2003). [8] M. Ribes, E. Bychkov, and A. Pradel , J. Opt. Adv. Mater. 3, 665 (2001). [9] P. Chen, M. Ailavajhala, M. Mitkova, D. Tenne, I. Esqueda, and H. Barnaby, IEEE WMED, 22 April (2011), p. 1. [10] V. Kh. Kudoyarova, J. Non-Cryst. Solids 90, 593 (1987). [11] M. Mitkova, I. Iliev, V. Boev, and T. Petkova, J. Non-Cryst. Solids 227-230, 748 (1998). [12] T. W. Kang, C. Y. Hong, C. S. Chong, and T. W. Kim, J. Mater. Sci. 27, 5620 (1992). © 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim