IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 59, NO. 6, DECEMBER

IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 59, NO. 6, DECEMBER 2012 3093 Effects of Cobalt-60 Gamma-Rays on Ge-Se Chalcogenide Glasses and Ag/Ge-Se Test Structures Y. Gonzalez-Velo, Member, IEEE, H. J. Barnaby, Senior Member, IEEE, A. Chandran, Student Member, IEEE, D. R. Oleksy, P. Dandamudi, Student Member, IEEE, M. N. Kozicki, Member, IEEE, K. E. Holbert, Senior Member, IEEE, M. Mitkova, Member, IEEE, M. Ailavajhala, and P. Chen Abstract Solid state electrolytes fabricated with chalcogenide glass (ChG) are considered viable candidates for the next generation of non-volatile memory technologies. These glasses, which are composed of group IV and/or group V elements with those of group VI chalcogens (S, Se, and Te), are excellent metal ion conductors. Because of this property, the resistance across structures composed of ChG films sandwiched between active metal (e.g., Ag) and inert metal (e.g., Ni) electrodes can be switched upon the application of sufficient bias, thereby enabling memristive action. In this paper, the effects of gamma-ray irradiations on test structures are investigated. The results show that exposure to high-energy photons can trigger the transport of from an active Ag top layer into an underlying ChG film. Post-irradiation annealing experiments also indicate that this photo-doping process is reversible once the radiation stress is removed. Numerical simulations which model the mechanisms of radiation-induced photo-doping and recovery are shown to agree well with the data. The results and analysis presented in this paper suggest the ChG-based memristors may be more susceptible to transient radiation effects than cumulative radiation damage. Index Terms Chalcogenide glass, memristors, photo-diffusion, photo-doping, radiation effects, total ionizing dose. I. INTRODUCTION ALLOYED melts of group IV and/or group V elements with those of group VI chalcogens (S, Se, and Te) upon water quench usually form bulk glasses, the so called chalcogenide glasses (ChG). They have been recognized as technologically promising materials since their inception several decades ago [1]. Due to their unique ion conduction properties, ChG materials are now used in the fabrication of two terminal devices exhibiting voltage- or current-controlled resistance switching characteristics [2], [3]. These devices are variously referred to as memristors [4] [7], programmable metallization cells [3], or electrochemical metallization memory cells [6]. Novel non- Manuscript received July 14, 2012; revised September 13, 2012; accepted September 25, 2012. Date of current version December 11, 2012. This work supported in part by the Defense Threat Reduction Agency under Grant HDTRA1-11-1-0055 and the Nuclear Energy University Program. Y. Gonzalez-Velo is with the School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85287-5706 USA (e-mail: yago.gonzalezvelo@asu.edu). H. J. Barnaby, A. Chandran, D. R. Oleksy, P. Dandamudi, M. N. Kozicki, and K. E. Holbert are with the School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85287-5706 USA. M. Mitkova, M. Ailavajhala, and P. Chen are with the Department of Electrical and Computer Engineering, Boise State University, Boise, ID 83725 USA. Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TNS.2012.2224137 volatile memories using these ChG-based memristors are currently being commercialized [8]. The chalcogenide glass of interest in this work is as it is a candidate material for manufacturing a ChG-based memory device [5], [6]. The device is composed of a ChG film sandwiched between active metal (e.g., Ag) and inert metal (e.g., Ni, W) electrodes. It operates through a combination of bias dependent ion conduction and reduction/oxidation (RedOx) processes, which are used to grow or conversely remove a conductive Ag bridge within the ChG film [6], thereby enabling a bias-controlled switch of resistance in the cell [5], [6]. As early as the mid-1960s, it was shown that Ag can be photo-diffused into chalcogenide materials, i.e., diffusion of Ag into chalcogenide materials can be activated by ultra-violet (UV) light [9]. This process, also known as photo-doping, is initiated by the ionization of the ChG film by incident light [10] [12]. In this work we study the effects of gamma ray exposures on ChG materials and on specially designed test structures. The experimental results suggest that while gamma rays have a negligible effect on the structure of the ChG film, they can trigger the transport of into the material in ways similar to conventional UV light-induced photo-doping [9], [12]. Post-irradiation annealing experiments also indicate that the photo-doping process is reversible once the radiation stress is removed. In Section II, the process used to manufacture the samples and a description of the test structures used in this work are provided. Section III describes the experimental protocol. The results of as well as UV light exposures performed on the ChG test structures are presented in Section IV. In Section V, the processes of radiation-induced photo-doping and post-irradiation recovery are analyzed and modeled with numerical device simulations. The simulations model the mechanisms of electron-hole pair generation as well as charge transport and reactions within ChG. The experimental results and analysis provide strong evidence of radiation-induced photo-doping, i.e., that exposure to high energy ionizing radiation will, like UV light, induce to be released from an Ag top layer and transport into the film. Moreover, once the radiation stress is removed, the ions will transport back to the top layer, enabling a recovery in the test structure response. The results suggest that ChG-based memristors, particularly those that are not pre-doped during the fabrication process, may be more susceptible to transient ionizing radiation effects (soft errors) than cumulative ionization damage. 0018-9499/$31.00 2012 IEEE

3094 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 59, NO. 6, DECEMBER 2012 Fig. 1. Illustration of the four-step process used to manufacture the test structures examined in this study. II. PROCESSING OF DEVICES AND DESCRIPTION OF TEST STRUCTURES The test structures developed for this study are composed of a ChG thin film deposited on a substrate [13]. A Ag top layer is deposited on the ChG film. Two Ni electrodes are patterned above the Ag layer to enable measurements of the lateral resistance across the stack. An increase in the resistance indicates dissolution of the Ag layer caused by photo-doping [14]. The four processing steps used in the fabrication of the structures are illustrated in Fig. 1. The first step is the deposition of (80 nm) on top of a p-doped silicon wafer using a TorrVac VC-320 electron-beam evaporator. The second step is the blanket deposition of the (60nm)usingaCressington 308 thermal evaporator. During the third step, 40 nm of Ag is deposited on top of the ChG using the Cressington 308 evaporator and patterned by lift-off. Finally, 60 nm of Ni is deposited using the Torrvac and again patterned using lift-off techniques. An illustration of the two different types (A and B) of test structures formed on each die is shown in Fig. 2(a) (d). The figures show layouts and photomicrographs (obtained prior to radiation exposure) of the devices as well as cross-sections. In Fig. 2(a) and (c), the Ag mask is the rectangle in between the two electrodes. The micrographs show that prior to radiation exposure some diffusion of Ag into the ChG film has already occurred (the halo surrounding the original patterned rectangle of Ag) as a result of processing. III. RADIATION TEST AND EXPERIMENTAL PROTOCOL Irradiations of the samples were performed with gamma-rays at room temperature in a Gammacell 220 irradiator at Arizona State University at a dose rate of 10.5 rad(gese)/s. A maximum total ionizing dose (TID) of 3.5 Mrad(GeSe) has been reached on the devices. Samples have been left floating during exposure. Electrical characterizations were performed at several dose levels below the maximum dose in order to retrieve the evolution of the electrical characteristics with increasing dose. UV exposures were also conducted on an additional set of samples in order to observe the photo-doping process already reported in the literature [14]. The UV light intensity was 1.77 at a wavelength of 324 nm. The devices were electrically characterized on a probe-station by performing DC current-voltage sweeps with an Agilent 4156C parameter analyzer. The voltage between the electrodes is swept from 0 to 100 mv. Current-voltage (I V) measurements were performed prior to and shortly after the irradiations as well as after room temperature annealing. Raman spectroscopy and energy dispersive X-ray spectroscopy (EDS) have also been performed. Raman spectroscopy enables measurements of the short range order in the ChG films. Thus any radiation-induced modifications in the types or vibrational states of bonds within the layer may be identified. Raman spectra were obtained with a Horiba Jobin Yvon T64000 triple monochromator. The samples were excited with a 441.6 nm HeCd laser. The power on the sample was 60 mw focused onto a circular spot of in diameter. The sample chamber was pumped down to Torr to avoid oxidation and the samples were cooled to 100 K during Raman measurements to reduce the probability of photo-induced effects due to laser irradiation. EDS was performed on a set of samples in order to quantify the distribution of Ag within the devices prior to and after irradiations. EDS has been conducted with a Hitachi S-3400N-II. A beam of electrons was generated from a tungsten filament, and the electrons are accelerated with 20 kv onto the sample. Interaction between the electrons and the sample generates characteristic X-rays corresponding to the elemental composition across the sample. IV. EXPERIMENTAL RESULTS A. Gamma-Ray Exposure of Structures I V measurements were performed on five devices of each type before and after radiation exposure. All devices were on the same die and showed similar response characteristics. Fig. 3 plots the pre- and post-irradiation (820 krad(gese)) responses of one type A device with 110 spacing between the Ni electrodes The responses (pre-rad and after 290 krad(gese) and 820 krad(gese)) of one type B device with 40 spacing between electrodes is shown in Fig. 4. Figs. 3 and 4 show that for both device types, the current flowing between the Ni terminals decreases after exposure revealing an increase in the resistance in the stack. The data on the type B devices indicates that the reduction in current (increase in resistance) is monotonic with irradiation dose. In Fig. 5, average Raman spectra obtained on two different die from the same processing batch are presented. Raman spectra have been obtained on areas of the dice where the 0.2 laser spot can be focused directly on the ChG films (without the Ag top layer). The black curve is the average spectra for a sample that has not been exposed to gamma-rays, whereas the grey curve is the average spectra obtained on a different sample from the same manufacturing batch and exposed to 3.5 Mrad(GeSe). The peaks observed are signatures of the bond configurations. In the case of glass, the mode assignments are as follows: the corner-sharing (CS) main

GONZALEZ-VELO et al.: EFFECTS OF COBALT-60 GAMMA-RAYS 3095 Fig. 2. (a) Layout and photomicrograph of type A devices. (b) Cross section of type A devices. (c) Layout and photomicrograph of type B devices. (d) Cross section of type B devices. mode is around 200, the edge-sharing (ES) main mode is around 215, the ethane-like structure (ETH) main mode is found close to 175, and the selenium chain (Se-chain) is located around 250 [15], [16]. It can be observed from the spectra that similar peaks are present for both the exposed and non-exposed samples. This means that no clear modification of the film structure occurs as a result of gamma ray exposure, at least for the TID levels investigated in this work. In Fig. 6, the currents extracted at a voltage of 50 mv on a set of five control devices (plus symbols), five devices exposed to 820 krad(gese) (square symbols), and five devices exposed to 3.5 Mrad(GeSe) (triangle symbols) are presented for type A structures. Control devices are those that have not been exposed to any gamma rays and are used to compare against the evolution of current flow in the exposed parts. Two die from the same processing batch have been exposed to gamma-rays, one being exposed in one step up to 820 krad(gese) and the second one up to 3.5 Mrad(GeSe), also in one step. In Fig. 6, we observe that the higher the TID, the lower the current flowing through the devices, which is equivalent to an increase in resistance across the structure with increasing TID. This has also been observed on type B devices (see Fig. 4). These data clearly demonstrate an increase in resistance across the stack after exposure to ionizing radiation, which is an indication of dissolution in the Ag top layer caused by radiation-induced photo-doping. EDS analysis was performed on unexposed devices as well as on parts exposed to a TID of 3.5 Mrad(GeSe) to measure the distribution of Ag between the Ni electrodes. The Ni electrode spacing for the parts used for the EDS is 20.On Fig. 7 the atomic percentage of Ag measured with EDS is presented for unexposed devices (squares) and parts exposed to 3.5 Mrad(GeSe) (circles). The figure shows that prior to exposure there is a large variation in the Ag concentration. Indeed

3096 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 59, NO. 6, DECEMBER 2012 Fig. 3. I V characteristics measured on one representative type A device before and after exposure to gamma-ray. Fig. 6. Current as a function of dose for type A devices. Current is extracted at a voltage of 50 mv. Square symbols represent the current on devices exposed to 820 krad(gese), triangle symbols are for parts exposed to 3.5 Mrad(GeSe), and the plus symbols represent the control part responses. Fig. 4. I V characteristics measured on one representative type B device before and after exposure to gamma-rays. Fig. 7. Atomic percent of Ag measured with EDS along devices with 20 spacing between electrodes: pre-irradiation (squares) and after to 3.5 Mrad(GeSe) (circles). be due to local fields that exist under the contacts which attract. However, after the sample is exposed to 3.5 Mrad(GeSe) the Ag has been transported out from underneath the Ni electrodes and into the ChG film. These EDS data demonstrate further that exposure to gamma rays has likely induced the photo-doping process. Fig. 5. Raman spectrum obtained on 60 nm films. In black the film isnotexposedtogammaraysandingreythefilm is exposed to 3.5 Mrad(GeSe). the pre-rad data shows that Ag tends to build up significantly under the Ni contacts. Device simulations suggest that this may B. UV Exposure of Structures Exposure to UV light has also been performed on an additional set of samples in order to investigate the effects of UV light and support comparisons to irradiation [8], [14]. The UV test was performed with an intensity of 1.77 and a photon wavelength of 324 nm (i.e., energy of 3.82 ev). The evolution of the current obtained after UV exposure on a type A device is presented in Figs. 8 and 9. As can be observed in

GONZALEZ-VELO et al.: EFFECTS OF COBALT-60 GAMMA-RAYS 3097 Fig. 8. I V characteristics measured on a type A device prior to exposure, and after 68 minutes and 120 minutes of exposure. Fig. 10. Current at 50 mv as a function of time for devices exposed to gamma rays and then annealed, unbiased at room temperature. Plus symbols represent the control data and square symbols represent the data for parts exposed to 820 krad(gese) and then annealed. the transport of stress is removed. back to Ag top layer once radiation Fig. 9. Average of five current measurements on type A devices as a function of UV exposure time (error bars are one standard deviation), extracted at a voltage of 50 mv. Fig. 9, the current is mostly constant as a function of time for control devices (squares), while the current decreases in the case of devices exposed to UV light (circles). C. Room Temperature Annealing Behavior of Exposed Structures Type A devices exposed to a TID of 820 krad(gese) have been tested at different time intervals after exposure in order to determine if any post-irradiation room temperature annealing response occurs. The results are plotted in Fig. 10. The control devices (plus symbols) have also been tested during this period and show little variation over time. For the exposed samples (square symbols), the data show a significantamountof post-irradiation annealing. In fact the structures have recovered essentially to their pre-irradiation current levels before 100 hours of unbiased, post-rad anneal. Modeling results discussed in Section V suggest that the annealing mechanism is related to V. DISCUSSION Photo-doping of Ag into chalcogenide films from exposures to visible or UV light has been observed since the 1960s [9]. Many researchers ascribe the interaction between electron-hole pairs generated in the ChG films and the active Ag layer as a principle mechanism for photo-doping [14], [17]. One theory [17] contends that holes generated by ionizing radiation transport to the Ag layer where they react with neutral Ag to produce. The ions subsequently diffuse and drift into the film. The structures fabricated and analyzed for this study were specially designed to characterize this photo-doping process [13]. Sampling current flow (or resistance) between the Ni terminals provides measurements of the continuity of the Ag top layer. A decrease in the sampled current between the electrodes after irradiation indicates a radiation-induced dissolution of the Ag layer caused by the photo-doping process. This process is illustrated in Fig. 11(a) (c). It should be noted that the structure resistance values prior to irradiation are in the range of 10 to 100, which are higher than the resistance of an ideal Ag film. These differences may be attributed to the initial diffusion of Ag observed after manufacturing (see Fig. 2(a) and (c)) as well as an agglomeration of the Ag top layer when it is deposited on the ChG. Nevertheless, the data reported in this paper clearly show much lower currents through the devices after exposures (Figs. 3, 4 and 6), which indicates that high energy radiation exposures facilitate transport into the glass (i.e., photo-doping). This conclusion is supported by the EDS results (Fig. 7). Moreover, similar effects are observed after UV light exposures, although the reductions in current are not as great compared to the high energy photons (Figs. 6, 8 and 9). This difference between the and UV responses may be due to a lower generation rate of electron-hole pairs within the ChG during UV exposure due to reflection or attenuation of the

3098 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 59, NO. 6, DECEMBER 2012 Fig. 11. Illustration of the photo-doping process induced by exposures: (a) initially, the devices exhibit a low resistance that can be attributed to the Ag thin-film; (b) during exposure, are generated that can photo-dope the ChG; (c) when performing I V characteristics on exposed samples, the resistance is increased due to a dissolution of the initial Ag layer; (d) 2D stack modeled in Silvaco s device simulator, Atlas. light by the Ag or Ni layers. Another potential cause for the differences between and UV may be that the high energy gamma rays can also cause trapped charge to build up in the layer underneath the glass. While the impact of these oxide defects are not examined in detail here, it may be speculated that perturbations to local electric fields caused by oxide charge accumulation may influence the transport and reactions within the ChG film. Concerning the post-irradiation annealing response (Fig. 10), we hypothesize that a reversal of the photo-doping process occurs once the radiation stress is removed. That is, during room temperature anneal, the within the film transport back to the interface, leading to a partial reformation of the Ag top layer. Two-dimensional devices simulations were performed on structures representative of the radiation sensitive region of the lateral device in order to model both the post irradiation and annealing behaviors. Simulations were performed with Silvaco s device simulator ATLAS. In order to capture radiation-induced ion transport and reactions we utilized Silvaco s Generic Ion Transport and Reaction module. This feature when combined with the code s intrinsic radiation-induced carrier generation capability makes it possible to numerically model the photo-doping process as well as the post-irradiation recovery. Fig. 11(d) shows the sensitive region of the device that has been modeled. It has been identified as the vertical stack below the Ni contacts, i.e., the region inside the box of Fig. 11(d). In this stack, the intrinsic (un-biased) electric field is greatest prior to irradiation. Work-function differences between the Ni terminals and underlying substrate create a field directed vertically upward. Prior to exposure, this field keeps near the Ag-ChG interface. The key equations used in the model are where is the electron-holes pair generation rate, is the radiation dose rate, is the density, and is the ionization energy for electron-hole pairs, which expresses the hole capture reaction between neutral Ag and mobile holes that releases,and which expresses the reduction reaction between and mobile electrons that produce neutral Ag. Fig. 12 displays the device simulation results as contour plots of the Ag density within the ChG film and Ag top layer metal. Fig. 12(a) plots the Ag distribution in these regions after an arbitrary fixed dose,. Fig. 12(b) plots distribution after 10. The figures show that as the ChG film is ionized by radiation, the Ag moves into the film away from the Ag metal layer. This reduces the integrated concentration of Ag throughout the (1) (2) (3)

GONZALEZ-VELO et al.: EFFECTS OF COBALT-60 GAMMA-RAYS 3099 Fig. 13. Simulated normalized integrated Ag concentration vs. normalized time in stack during irradiation and post-rad anneal (solid line). Also plotted on the figure are the normalized experimental current values (square symbols). Fig. 12. Contour plots of Ag in Ag/GeSe stack after (a) an arbitrary fixed dose,, and after (b) 10. stack, thereby increasing resistivity, and reducing current flow during irradiation. Simulations of post-irradiation annealing were also performed. For these calculations, the carrier generation was suspended and the transport of neutral and ionized Ag was monitored over time. The results showed that upon removal of radiation stress, transports back to the Ag-ChG interface, thereby increasing the integrated Ag concentration in the stack. The simulation (TCAD) results during exposure and post-irradiation anneal are shown in Fig. 13. The simulated response (solid line) plots the normalized integrated Ag density vs. normalized time. This integrated density is defined as where is the Ag concentration (plotted in Fig. 12) and is the thickness of the stack. The simulation results show that the integrated density of Ag in the stack drops dramatically during radiation exposure and then returns to almost pre-irradiation levels fairly quickly during the post-rad anneal. These results are compared to the normalized current measurements (assumed proportional to ), also plotted in Fig. 13. The normalized experimental current values (square symbols) shown in Fig. 13 are taken from the data used to generate Fig. 10. Fig. 13 shows that the trends in the data are in good agreement with the simulation results. (4) Analysis of the simulation results reveals plausible mechanisms for both radiation-induced photo-doping and the post-irradiation recovery. During exposure, radiation-induced holes transport quickly to the interface, where they react with neutral Ag to create. These newly created ions in addition to those ions already present near the interface eventually transport into the ChG layer. This transport is primarily a drift process, caused by an inversion of the electric field in the ChG layer. The model shows the buildup negative charge from the slower moving electrons is the mechanism for the inversion. Once the radiation stress is removed, these electrons eventually transport out of the structure and return the field to its initial state, i.e., directed towards the interface, under the Ni contact. This pushes the un-neutralized back to the Ag top layer during the anneal process, enabling the observed recovery. While more study is necessary to confirm the validity of or refine these suggested mechanisms, if true, they indicate that radiation-induced photo-doping in un-doped ChG films is primarily a transient effect and not a necessarily an indicator of permanent damage to the film. VI. CONCLUSION In this work, we investigated the effect of the photo-doping process on test structures made with a thin-film of Ag deposited on top of a ChG. We demonstrated that an increase of resistance across the structure is observed when photo-doping is induced by UV [14]. The effects of gamma-ray exposures were also investigated. For structures exposed to gammarays we show that photo-doping is similar to that obtained for exposure to UV light. As a consequence, this work shows that a photo-doping process can be triggered in structures by higher energy photons. The experimental results also reveal that the photo-doping process is apparently reversible. Two-dimensional simulations using ion transport capabilities of the Silvaco TCAD tools demonstrate that the behavior observed during exposure and room temperature annealing can be modeled by taking into account the transport of neutral Ag and

3100 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 59, NO. 6, DECEMBER 2012 into the ChG as well as their reactions with electrons and holes generated during exposures. The simulations enable us to identify potential mechanisms for the photo-doping and annealing responses as being causedinpartbythebuildupand subsequent removal of negative charge (i.e., electrons) in the ChG layer. It should be noted that the observed radiation enhanced dissolution of Ag into the ChG in our current study may not be indicative of how an actual memory device will behave under similar exposure. Most ChG-based memory cells are deliberately photo-doped with Ag during fabrication to the point where no more Ag will dissolve, i.e., the film is essentially saturated with Ag [3], [18]. This pre-doping of the material should make it more difficult for any more metal to diffuse into the ChG during subsequent exposure, although we intend to perform studies on actual memory devices to confirm or refute this. In any case, the observed effect may have utility in radiation sensors, in which the change in resistance due to dissolution of the film into the relatively lightly doped material would be calibrated to the amount of exposure. The post-irradiation recovery to the low resistance state may make such a sensor reusable. ACKNOWLEDGMENT The authors would like to thank Dr. D. J. Silversmith and Dr. J. Calkins of DTRA for their support of this work. REFERENCES [1] Z.U.Borisova, Glassy Semiconductors. New York: Plenum Press, 1981. [2] M. N. Kozicki, M. Mitkova, M. Park, M. Balakrishnan, and C. Gopalan, Information storage using nanoscale electrodeposition of metal in solid electrolytes, Superlattices Microstruct., vol. 34, pp. 459 465, Dec. 2003. [3] M. N. Kozicki, M. Park, and M. Mitkova, Nanoscale memory elements based on solid-state electrolytes, IEEE Trans. Nanotechnol., vol. 4, pp. 331 338, May 2005. [4] L. Chua, Memristor-The missing circuit element, IEEE Trans. Circuit Theory, vol. 18, pp. 507 519, Sep. 1971. [5] R. Waser, R. Dittmann, G. Staikov, and K. Szot, Redox-based resistive switching memories Nanoionic mechanisms, prospects, and challenges, Adv. Mater., vol. 21, pp. 2632 2663, Jul. 2009. [6] I. Valov, R. Waser, J. R. Jameson, and M. N. Kozicki, Electrochemical metallization memories Fundamentals, applications, prospects, Nanotechnology, vol. 22, pp. 254003 254003, Jun. 2011. [7] R. Waser and M. Aono, Nanoionics-based resistive switching memories, Nature Mater., vol. 6, pp. 833 840, 2007. [8] [Online]. Available: http://www.adestotech.com [9] M.T.Kostyshin,E.V.Mikhailovskaya,andP.F.Romanenko, Sov. Phys. Solid St., vol. 8, pp. 451 453, 1966. [10] M. Kawasaki, J. Kawamura, Y. Nakamura, and M. Aniya, Ionic conductivity of glasses, Solid State Ionics, vol. 123, pp. 259 269, Aug. 1999. [11] M. A. Ureña, A. A. Piarristeguy, M. Fontana, and B. Arcondo, Ionic conductivity in AgGeSe glasses, Solid State Ionics, vol. 176, pp. 505 512, Feb. 2005. [12]A.Piarristeguy,M.Ramonda,A.Ureña,A.Pradel,andM.Ribes, Phase separation in Ag-Ge-Se glasses, J. Non-Cryst. Sol., vol. 353, pp. 1261 1263, May 2007. [13] M. N. Kozicki, Personal Electronic Dosimeter, United States Patent number 5,500,532, 1996. [14] A. V. Kolobov and S. R. Elliott, Photo-doping of amorphous chalcogenides by metals, Advances in Physics, vol. 40, pp. 625 684, Feb. 1991. [15] X. Feng, W. J. Bresser, and P. Boolchand, Direct evidence of stiffness threshold in chalcogenide glasses, Phys. Rev. Lett., vol. 78, pp. 4422 4425, Jun. 1997. [16] P. Boolchand, J. Grothaus, M. Tenhover, M. A. Hazle, and R. K. Grasselli, Structure of glass: Spectroscopic evidence for broken chemical order, Phys.Rev.B, vol. 33, pp. 5421 5434, Apr. 1986. [17] N. Terakado and K. Tanaka, Electrical response of chalcogenide films in the photo-doping process, Thin Film Solids, vol. 519, pp. 3773 3777, Mar. 2011. [18] M. Mitkova, M. N. Kozicki, H. C. Kim, and T. Alford, Local structure resulting from photo and thermal diffusion of Ag in Ge-Se films, J. Non-Crystalline Solids, vol. 338 340, pp. 552 556, 2004.