Influence of TID on Lateral Diffusion of Structures made of Novel Metal-Chalcogenide Glass Combinations: A Flexible Radiation Sensor Development Perspective A. Mahmud, Y. Gonzalez-Velo, H. J. Barnaby, M. N. Kozicki, K. E. Holbert, M. Mitkova, T. L. Alford and M. Goryll ABSTRACT We report the results of our study on the influence of total ionization dose (TID) on lateral diffusion of some selected metals into chalcogenide glasses of different composition and variable atomic ratio. Presenting Author: A. Mahmud, is with the School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ, 85287-5706, USA, cell: (631) 682-4333, fax: (480) 965-2811, amahmud1@asu.edu Contributing Authors: Y. Gonzalez-Velo, School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85287-5706, yago.gonzalezvelo@asu.edu H. J. Barnaby, School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85287-5706, hbarnaby@asu.edu M. N. Kozicki, School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85287-5706, michael.kozicki@asu.edu K. E. Holbert, School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85287-5706, keith.holbert@asu.edu M. Mitkova, Department of Electrical and Computer Engineering, Boise State University, Boise, ID 83725-2075, mariamitkova@boisestate.edu T. L. Alford, School for Engineering of Matter, Transport and Energy, Arizona State University, Tempe, AZ 85287-5706, ta@asu.edu M. Goryll, School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85287-5706, mgoryll@asu.edu Presentation Preference: Oral Session Preference: Facilities and Dosimetry ACKNOWLEDGEMENTS This work was funded in part by the Defense Threat Reduction Agency under grant no. HDTRA1-11-1-0055 and Air Force Research Laboratory Det 8/RVKVE under grant no. FA9452-13-1-0288. The authors would like to thank Dr. Jacob Calkins of DTRA and Dr. Art Edwards of AFRL for their support. Also, we gratefully acknowledge the use of facilities within the LeRoy Eyring Center for Solid State Science (LE-CSSS) and Center for Solid State Electronic Research (CSSER) at Arizona State University.
INTRODUCTION Chalcogenide glasses (ChG) are well known for their unique properties and functionalities that made them a desired material for a wide range of applications such as computer memory devices, xerography, image sensors, optical fibers, solar cells, etc. [1-5]. They basically represent the wide range of chemical compositions which are based on elements from column 16 in the periodic table, or chalcogen group elements namely sulfur (S), selenium (Se) and tellurium. Chalcogen elements can form glass-like binary or ternary chemical compounds when they covalently bond with elements from column 14 or 15 (i.e., Ge, As, Sb, Ga, etc.) in a wide range of atomic ratios. In our previous works [6-9], we demonstrated the potential use of thin ChG films as radiation detection sensors using horizontal structures with metal electrodes formed on top of the film. When exposed to ionizing radiation, the deposited energy (i.e. dose) results in the migration of metal ions from the electrodes into the intermediate ChG film region. Ions migrating into the ChG have the ability to dope it and reduce its resistivity. Radiation sensing can be quantitatively performed by measuring the substantial changes in the electrical resistance of the glass, especially when the doping fronts from two nearest electrodes make contact. Fig. 1(a) shows a typical resistance evolution plot of this metal-chg sensor in response to increased dose level. The term limit of detection (LOD) in the plot indicates the dose level where the undoped/high resistance state (HRS) starts to decrease sharply in response to increased dose level and the dynamic range (DR) is the ratio between the undoped/hrs and doped/low resistance state (LRS). Fig. 1(b) shows the progression of the metal doping fronts of the metal-chg radiation sensors in response to increased dose level [6]. The radiation-induced diffusion/migration of metal can depend on several important parameters including but not limited to physical design of the sensor structure, type of metals used as electrode, chemical composition, atomic ratio of the ChG film, etc.. Previously we investigated Ag_Ge 20 based flexible sensors where the radiation-induced metal diffusion can be controlled by tuning physical design parameters (i.e. spacing between electrodes) [6-7]. In this work, we present new experimental results assessing the compatibility of other Ge x systems with different atomic ratio of selenium, in order to investigate if ChG composition can be exploited to improve the control and performance of this metal-chg based radiation sensors. We demonstrate that composition variation can be used to vary parameters such as the LOD or the DR, as well as improve the shelf life of the samples. In the present study, metals from column 11 (i.e. Cu, Ag, Au) are used to investigate the impact of the diffusing atom on the sensor response (Cu and Ag are known to be highly diffusive atoms in dielectric films). The full length version of the manuscript and presentation will incorporate detailed discussions and results of additional experiments on samples made with sulphide based ChG films with different atomic ratio (Ge xs 1-x systems). (a) (b) Fig. 1: (a) Evolution of resistance plot of a metal-chg flexible sensor devices that denotes LOD and DR. (b) shows the progression of the metal doping fronts from the adjacent electrodes into ChG films in response to increasing dose level (increasing dose on the samples from left to right). Figures from [6]. DEVICE FABRICATION AND TEST PROTOCOL Out of many substrate choices, we choose one side pre-treated, heat stabilized flexible polyethylene-naphthalate (PEN) substrates manufactured by DuPont Teijin Films due to their smooth surface, low coefficient of thermal expansion, low shrinkage and rigidity. This lightweight and transparent substrate is ideal for fabricating metal- ChG based flexible radiation sensors due to their compatibility for use on non-flat objects. In order to help the readers to easily follow the experimental process, a simple naming format is used to denote different samples. For instance, samples with Ge 20 thin film and Au electrodes are named as Au_Ge 20. Similar naming formats are used for other Ag_Ge x and Cu_Ge x based systems. The manufacturing process of the sensors begins with the deposition of thin ChG film. To fabricate samples with Ge 40Se 60 film (i.e., Ag_Ge 40Se 60 and Cu_Ge 40Se 60 samples), first a 100 mm by 100 mm square shaped PEN substrate was loaded inside a thermal evaporator (Cressington 308). Then, 15nm of Ge 40Se 60 film was deposited at a rate of 0.1 nm/sec at room temperature. The substrate with ChG film was then diced into two 100 mm by 50 mm pieces. Each of these two pieces were used to manufacture samples with two different metal electrodes (i.e., Ag and Cu). To deposit the metal electrodes, a shadow mask was used that has several array of electrodes. For all the arrays, the diameter of the electrodes was fixed to and spacings between nearest electrodes were varied from to. We opted for this
particular geometry in order to observe how the spacing between electrodes plays role in the photo diffusion process. Ag and Cu electrodes were deposited using an e-beam evaporator (PVD 75, Kurt J. Lesker Company ). For both cases, 75 nm of thick metal films were deposited at a rate of 1 Å/s at room temperature to form the electrodes. Upon completion, the shadow mask was removed. Then, each of the 100 mm by 50 mm substrates were further diced into 100 mm by 25 mm pieces to have both control and exposure samples from the same manufacturing batch in order to reduce the effects of process related variability. Since the devices are sensitive to the UV component of visible light, they were immediately stored inside dark boxes to minimize uncontrolled exposure. Similar processes were performed to manufacture Ag_Ge 33Se 67, Cu_Ge 33Se 67, Ag_Ge 30Se 70, Cu_Ge 30Se 70, Ag_Ge 20, and Cu_Ge 20 control and exposure samples. A pictorial process flow and images of the completed samples with different spacing can be found in [6]. For devices with Au electrodes, we only used Ge 20 for our preliminary study since the worst case of metal diffusion occurs for selenium-rich compositions. If diffusion of Au into Ge 20 is not observed on the exposed samples for TID >3 Mrad(Ge x), it is less likely to see the diffusion with other Ge x and Ge xs 1-x compositions. For Au_Ge 20 devices, a similar technique was used to deposit the 15 nm of Ge 20. A shadow mask with an array of diameter of electrodes and 1 mm spacing was attached on top of the Ge 20 coated substrate and 75 nm of Au was deposited at 0.1 nm/s rate using a thermal evaporator (Edwards Auto 306) at room temperature. After removal of the shadow mask, the substrate was diced into two pieces to create the control and test samples and were stored in a dark box as well. All exposure samples were placed into a Gammacell 220 60 Co irradiator, capable of delivering high total doses. The exposure samples were exposed at a dose-rate of 358 rad(ge x)/min and were periodically unloaded from the Gammacell chamber for a brief amount of time to measure the electrical resistance between adjacent electrodes within a particular array using a semiconductor parameter analyzer (Agilent 4156C). Similar measurements were performed on unexposed control samples within the same time. The 60 Co gamma ray exposure continued up to a maximum dose level of 3.34 Mrad(Ge x) for Au_Ge 20 samples and 3.43 Mrad(Ge x) for rest of the other exposure samples. During exposure, all the test samples were left floating by keeping the electrodes unconnected. To avoid oxidizing the metal electrodes, a very small bias voltage (10 mv) has been used to measure the resistance level. RESULTS AND DISCUSSION In Fig. 2 the evolution of resistance as a function of time for control devices and resistance as a function of dose for exposed devices is presented. Fig. 2(a) presents resistance of Ag_Ge 20 based control devices with different spacing between electrodes, and similarly, Figs. 2(b)-(d) show evolution of resistance for Ag_Ge 30Se 70, Ag_Ge 33Se 67 and Ag_Ge 40Se 60 systems. Fig. 2(e)-(h) shows the resistance level of corresponding exposed samples which were measured within the same time frame. In Fig. 2 it can be observed on one side, that apart from the spacing device on Ag-Ge 20 (Fig. 2(a)) all control samples stayed at high resistance levels (HRS) during the course of the experiment. On the other side, exposed device exhibit varying behaviors with ChG composition: for Ag-Ge 40Se 60, only the spacing device (Fig. 2(h)) exhibit a transition from HRS to LRS. 10 13 Control: Ag_Ge 20 10 13 Control: Ag_Ge 30 Se 70 10 13 Control: Ag_Ge 33 Se 67 10 13 Control: Ag_Ge 40 Se 60 (a) (b) (c) (d) 10 13 Exposed sample: Ag_Ge 20 10 13 Exposed sample: Ag_Ge 30 Se 70 10 13 Exposed sample: Ag_Ge 33 Se 67 (e) (f) (g) (h) Figure 2: Resistance measured on control devices (a,b,c,d) and 60 Co gamma ray exposed (e,f,g,h) Ag_GexSe1-x based systems with different atomic ratio. From left to right, the atomic percent of Selenium of the film increases (Ge20Se80, Ge30Se70, Ge33Se77, Ge40Se60). Results for devices with different spacing between electrodes (1mm, 2mm, 3mm, 4mm) are also presented. Exposed sample: Ag_Ge 40 Se 60
Devices made with the ChG with the poorest selenium composition exhibit the lowest resistance variation trend with total dose. It can be observed in Figs. 2(h)-(e) that the parts with higher selenium content exhibit the greater Exposed sample: 1mm spacing, Ag electrodes 10 13 Ge 40 Se 60 Ge 33 Se 67 Ge 30 Se 70 Ge 20 Figure 3: Comparison among exposed samples with fixed spacing with Ag electrodes and with different atomic ratio of Ge x sensitivity to TID. The impact of the ChG film composition on the response of parts with a fixed spacing of between electrodes is summarized on Fig. 3 showing the variation of response when selenium content is increased. The results presented in Figs. 2 and 3 have two important ramifications. First, they reveal a clear dependence of the observed resistance variation (i.e., radiation-induced diffusion process) on the atomic ratio of the Ge x system. From a point of view of sensor characteristics, for the systems with Se-rich ChG films, the LOD is observed to go down to smaller TID values (see Fig. 3). By changing the atomic ratio of the Ge x film (i.e., selecting a Se poor or Se rich ChG), we can regulate the LOD of the sensor systems to a desired range. Second, both the LOD and DR of the Ag_Ge x based devices displayed dependence on spacing regardless of the atomic ratio of the Ge x. Our previous work demonstrated this effect on Ag_Ge 20 parts (i.e., with increase of spacing, LOD was found moving to higher TIDs and DR decreasing to lower values). This present study demonstrates that the above mentioned trend is valid for other Ag_Ge x systems as well. This happens due to the fact that when spacing is increased, the Ag doping fronts require higher levels of TID to transport additional amounts of Ag into the intermediate ChG film in order to make contact. This increases the LOD. On the other hand, a bigger spacing essentially increases the LRS or post contact resistance level that ultimately brings the DR down to a lower value. Here the DR is demonstrated to vary with the ChG composition. Figs. 4(a)-(d) show the evolution of resistance for Cu_Ge 20, Cu_Ge 30Se 70, Cu_Ge 33Se 67 and Cu_Ge 40Se 60 control samples. Corresponding resistance evolution plots of the exposed samples are shown in Figs. 4(e)-(h). 10 13 Control: Cu_Ge 20 Resistance [ ] 10 13 Control: Cu_Ge 30 Se 70 10 13 Control: Cu_Ge 33 Se 67 (a) (b) (c) (d) 10 13 Exposed sample: Cu_Ge 20 10 13 Exposed sample: Cu_Ge 30 Se 70 (e) (f) (g) (h) Figure 4: Resistance measured on control devices (a,b,c,d) and 60 Co gamma ray exposed (e,f,g,h) Cu_GexSe1-x based systems with different atomic ratio. From left to right, the atomic percent of Selenium of the film increases (Ge20Se80, Ge30Se70, Ge33Se77, Ge40Se60). Results for devices with different spacing between electrodes (1mm, 2mm, 3mm, 4mm) are also presented For Cu_Ge x parts, we observed lateral diffusion of Cu into ChG film for both control and exposed samples. This is shown on control parts (Fig. 4(a)-(d)), where resistance variation is observed to occur faster on selenium rich parts. We assume that exposure to visible light (which occurred during the film deposition processes and periodical resistance level testing) could not be responsible for Cu ionization at the electrode ChG film interface, as that contribution was minimized by keeping the samples in the dark. Room temperature might have provided adequate energy to Cu atoms to migrate into the Ge x film. On exposed samples, variation of resistance is observed but cannot be used effectively (or easily) for radiation sensing due to the diffusion happening on the control samples. The migration of the copper was certainly impacted by its smaller atomic radius (compared to 10 13 Exposed sample: Cu_Ge 33 Se 67 10 13 Control: Cu_Ge 40 Se 60 10 13 Exposed sample: Cu_Ge 40 Se 60
Ag and Au, from column 11) which might be an important reason for the control devices to exhibit such a notable diffusion. Finally, Fig. 5 shows the evolution of resistance of Au_Ge 20 control and exposed sample devices with increase dose level (i.e., 325.8 krad (Ge x), 787.6 krad (Ge x), 1.75 Mrad (Ge x), and 3.34 Mrad (Ge x)). As can be observed, no radiation induced change of resistance and diffusion of Au has been observed even after the maximum TID dose level pointing their incompatibility for the radiation detection application. It is shown here that by using atoms from column 11 in the periodic table, it is possible to modify the radiation induced diffusion within the ChG film. The size of the atom plays a key role in both the thermal diffusion and photodiffusion processes, showing that the smaller the atoms the easier it is for it to diffuse into the amorphous structure of the chalcogenide films. Control: Au_Ge 20 Sample: Au_Ge 20 Control and Sample: Au_Ge 20 Fig. 5: Evolution of resistance of the control and exposed samples of Au_Ge20Se80 system with increased dose level. The electrode diameter and spacing of the devices were respectively and. CONCLUSION In this work, we have reported test results on several metal-ge x devices in order to understand their suitability for radiation detection application. ChG composition variation and metal electrode type impacts are investigated and will be discussed in more detail in the final paper. Nevertheless, in this abstract it is already presented that photo-diffusion is affected by the metal atom size as well as the chalcogenide atom concentration within the ChG films. Both Cu and Au metal electrodes on Ge x films are shown to be incompatible for radiation sensing since no clear radiation-induced diffusion can be determined. In the full length version of this manuscript, we will extend our study to report experimental results performed on sulfide based systems (i.e., Ag_Ge xs 1-x, Cu_Ge xs 1-x), since the bandgap of sulfide ChG is higher. This might allow generation of fewer electron-hole pairs for a given TID level, enabling a better controlled photo-diffusion process and also permitting improved regulation of LOD and DR values. Eventually, a slower diffusion process can be expected from these sulfide based systems which might be helpful to reduce the unstable behavior of Cu based systems. Moreover, in the final paper, results obtained on samples with electrodes separated by several tens of micrometers will be presented, in order to illustrate and investigate the LOD and DR capability of such parts for lower total doses as the ones used until now. REFERENCES [1] M. A. Popescu, Non-Crystalline Chalcogenicides. vol. 8. Springer Science & Business Media, 2001. [2] K. Tanaka, and S. Koichi, Amorphous chalcogenide semiconductors and related materials. Springer Science & Business Media, 2011. [3] D. Mahalanabis, R. Liu, H. J. Barnaby, M. N. Kozicki, A. Mahmud, E. Deionno, "Single Event Susceptibility Analysis in CBRAM Resistive Memory Arrays," IEEE Trans. Nucl. Sci., vol. 62, no. 6, pp. 2606-2612, Dec. 2015. [4] S. Rajabi, M. Saremi, H. J. Barnaby, A. Edwards, M. N. Kozicki, Y. Gonzalez-Velo, M. Mitkova, D. Mahalanabis, A. Mahmud, Static impedance behavior of programmable metallization cells, Solid State Electronics, vol. 106, pp. 27-33, 2015 [5] J. L. Taggart, Y. G. Velo, D. Mahalanabis, A. Mahmud, H. J. Barnaby, M. N. Kozicki, "Ionizing radiation effects on non-volatile memory properties of Programmable Metallization Cells", IEEE Trans. Nucl. Sci., vol. 61, no. 6, p: 2985-2990, Dec. 2014. [6] A. Mahmud, Y. Gonzalez-Velo, M. Saremi, H. J. Barnaby, M. N. Kozicki, K. E. Holbert, M. Mitkova, T. L. Alford, M. Goryll, W. Yu, D. Mahalanabis, W. Chen, J. Taggart Flexible Ag-ChG Radiation Sensors: Limit of Detection and Dynamic Range optimization through physical design tuning IEEE Trans. Nucl. Sci. (in press). [7] A. Mahmud, Y. Gonzalez-Velo, H. J. Barnaby, M. N. Kozicki, K. E. Holbert, M. Mitkova, T. L. Alford, M. Goryll, M. Saremi, W. Yu, D. Mahalanabis, W. Chen, J. Taggart, "Optimization of Flexible Ag-Chalcogenide Glass Sensors for Radiation Detection," presented at the IEEE Radiation Effects Components and Systems Conf., Moscow, Russia, Sep. 14-18, 2015. [8] P. Dandamudi, A. Mahmud, Y. Gonzalez-Velo, M. N. Kozicki, H. J. Barnaby, K. E. Holbert, M. Mitkova, Flexible Sensors Based on Radiation-Induced Diffusion of Ag in Chalcogenide Glass, IEEE Trans. Nucl. Sci., vol. 61, no. 6, pp. 3432-3437, Dec. 2014. [9] A. Mahmud, Y. Gonzalez-Velo, H. J. Barnaby, M. N. Kozicki, K. E. Holbert, M. Mitkova, T. L. Alford, M. Goryll, M. Saremi, W. Yu, D. Mahalanabis, W. Chen, J. Taggart, "Optimization of Flexible Ag-Chalcogenide Glass Sensors for Radiation Detection," presented at the IEEE Radiation Effects Components and Systems Conf., Moscow, Russia, Sep. 14-18, 2015.