Alpha particle radiation effects in RF MEMS ... - Semantic Scholar
Microelectronics Reliability 48 (2008) 1241–1244
Contents lists available at ScienceDirect
Microelectronics Reliability journal homepage: www.elsevier.com/locate/microrel
Alpha particle radiation effects in RF MEMS capacitive switches J. Ruan a, E. Papandreou b, M. Lamhamdi a, M. Koutsoureli b, F. Coccetti a, P. Pons a, G. Papaioannou a,b,*, R. Plana a a b
LAAS CNRS, 7, av du Colonel Roche, 31077 Toulouse Cedex 4, France Solid State Physics Section, University of Athens, Panepistimiopolis Zografos, Athens 15784, Greece
a r t i c l e
i n f o
Article history: Received 25 June 2008 Available online 10 August 2008
a b s t r a c t The paper investigates the effect of 5 MeV alpha particle irradiation in RF MEMS capacitive switches with silicon nitride dielectric film. The investigation included MIM capacitors in order to obtain a better insight on the irradiation introduced defects in the dielectric film. The assessment employed the thermally stimulated depolarization currents method for MIM capacitors and the capacitance–voltage characteristic for MEMS switches. Asymmetric charging was monitored in MIM capacitors due different contact electrodes and injected charge interactions. Ó 2008 Elsevier Ltd. All rights reserved.
1. Introduction Microelectromechanical systems (MEMS) are receiving increasing interest for use in space systems. One particular area of interest is in Picosats [1,2] where very little shielding is afforded to the space radiation environment. To date, however, few radiation tests have been performed on MEMS devices [3–7]. Tests performed with gamma rays on MEMS accelerometers have shown the technology prone to radiation effects at moderate dose levels [3,4] and a quantitative model for the persisting electrostatic force due to radiation induced charging has been proposed by Edmonds [6]. In this model it was proposed that charging arises from carrier emission from the adjacent electrodes and injection into MEMS dielectric. This process has been further modeled for structures with different dielectrics materials and dielectric thicknesses [8]. Presently, it is well known that ionizing radiation introduces defects in insulating materials [9]. These defects can be detected and their properties can be identified with the aid of Metal–Insulator–Metal capacitors by applying the discharge current transient (DCT) or the thermally stimulated depolarization (TSDC) method. In the case of silicon nitride it has been shown that alpha particle radiation introduces deep defects [10] that increase the time constant of trapped charge, hence the dielectric charging. The aim of the present work is to provide information on the 5 MeV alpha particle radiation inducing degradation in RF MEMS capacitive switches with SiNx dielectric film. The investigation has been performed by employing the TSDC method in MIM capacitors in order to identify the charging processes. Finally, the shift of
* Corresponding author. Address: LAAS CNRS, 7, av du Colonel Roche, 31077 Toulouse Cedex 4, France. Tel.: +33 561337900; fax: +33 561553577. E-mail address: [email protected] (G. Papaioannou). 0026-2714/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.microrel.2008.06.047
the capacitance–voltage characteristic was used to investigate the degradation of MEMS basic characteristics. 2. Experiment The dielectric film of the MIM capacitors and MEMS switches was SiNx with a thickness of 300 nm. The choice to employ MIM capacitors was based on the fact that these structures can mimic a MEMS switch in the ON state and they provide valuable information on the material properties. The top and bottom electrode of MIM structures is composed of a Ti/Au layer (0.1 lm/1 lm). The top electrode, obtained by lift off, features a diameter of 500 lm. The silicon nitride layer was deposited by PECVD at 200 °C with high frequency (HF, 13.56 MHz). The composition of the dielectric film was determined with the aid of Rutherford backscattering spectroscopy (RBS) method that revealed an average value of x ffi 1.04. The TSDC current was measured through a Keithely 6487 voltage source – picoampere meter in the temperature range of 200 K to 450 K. The bias was applied to the top electrode with respect to the bottom one; hence the polarity is in reference to the one of the top electrode. The sacrificial layer of the MEMS capacitive switches features 2.5 lm thickness. The dielectric charging before and after each successive irradiation was assessed through capacitance–voltage characteristics. The region below pull-in threshold turns out to the determination of the magnitude of charge that is trapped in the dielectric after each positive or negative down state. The shift of pull-in and pull-out voltage was also monitored in order to determine the effect of charging under high electric field. Finally the irradiation experiments have been carried out at room temperature with 5 MeV alpha particles provided by an Am source. The alpha particle fluence was common for MIM and MEMS
1242
J. Ruan et al. / Microelectronics Reliability 48 (2008) 1241–1244
Due to higher sensitivity of used experimental methods, fluence up to 1011 cm2 were used for the MIM capacitors and up to 8 1011 cm2 for MEMS switches. 3. Results and discussion 3.1. Simulation of radiation introduced vacancies Low temperature amorphous silicon nitride constitutes a disordered material which homogeneity and stoechiometry may vary significantly across the material. Due to excess silicon, the material may consist of silicon clusters, stoechiometric silicon nitride as well as silicon nitride of varying composition that gives rise to large potential fluctuations that correspond to energy gap variations in the range of 1.6 eV, for amorphous silicon, to 4.6 eV, for Si3N4 [11]. In our case, where x ffi 1.04, reported assessment based on Raman spectra indicated that the silicon clusters size or density or both are rather small [11]. Nevertheless, the presence of potential fluctuations and silicon nanoclusters affects significantly the trapping of charges. High energy particle radiation introduces defects due to lattice atom displacement. The displaced atoms generate vacancies and interstitials. In the case of an amorphous and disordered material like SiNx, it has been shown that particle (Ar) irradiation contributes to nitrogen and silicon atoms mixing due to which the material exhibits a tendency to form a substitution of solid solution [11]. Since the atom mixing is caused by the displaced atoms it becomes obvious that the concentration of generated vacancies can be considered as a measure of the atoms mixing. In order to obtain a better insight, we calculated the introduction rate of Si (VSi) and N (VN) vacancies through the use of TRIM code [12]. The calculation was performed assuming a homogeneous SiN1.04 material since the ion projectile trajectories were not expected to be affected by the material inhomogeneity due to the small thickness of the SiN film and the high energy of alpha particles. The vacancies distribution was calculated assuming a simple MIM capacitor structure which consisted of 1000 nm Au top electrode, 300 nm SiN1.04 film and 500 nm Au bottom electrode. Calculations were also performed by taking into account the presence of the small air gap (2.5 lm) between the MEMS suspended electrode and the dielectric film. The distribution of Si and N vacancies are presented in Fig. 1. A peak observed at the front interface and the increase of vacancy introduction rate with the film thickness caused by the recoiled heavy Au atoms and the cumulative process of recoiled Si and N atoms, respectively. The effect of backscattered Au atoms at the rear interface is negligible. Finally, the extension of Si and N vacancies beyond the SiN film has to be attributed to displaced atoms that are further displaced by knock-on process (see Fig. 2).
Fig. 1. Structure of MIM capacitor.
Fig. 2. Distribution of Si and N vacancy introduction rates (V) for a MIM capacitor.
3.2. The MIM capacitors In insulators, the time and temperature dependence of polarization and depolarization processes are, in the case of dipolar polarization, driven by the competition, between the orienting action of the electric field and the randomizing action of thermal motion. The current density produced by the progressive decrease in polarization during the TSDC experiment, where time and temperature are simultaneously varied, is approximated by:
J D ðTÞ
PS ðT p Þ
s0
" # 2 EA 1 kT EA exp exp exp bs0 EA kT kT
ð1Þ
where b is the heating rate (K/s), EA the depolarization mechanism activation energy, PS (Tp) is the equilibrium polarization at the polarizing temperature Tp and s0 the corresponding infinite temperature relaxation time. In the case of space charge polarization the processes are much more complex because several mechanisms can be involved simultaneously and [13] and the TSDC spectra may show the characteristic properties of distributed processes, such as extension over a wide temperature range, etc. Since the heating rate b is constant, the TSDC spectrum allows the calculation of the stored charge:
Q¼
1 b
Z
T2
ITSDC ðTÞdT
ð2Þ
T1
The TSDC spectra for radiation fluencies up to 1011 cm2 as well as for positive and negative bias polarity applied to top electrode are plotted in Fig. 3. The TSDC spectra are asymmetric and depend strongly on the sign of polarizing bias. This behavior may be attributed to both the asymmetrical electrodes, since the work functions for Au and Ti are about 5.2 eV and 4.2 eV, respectively [14], and the dielectric film properties. The amorphous silicon-rich alloys behave like a-Si:H with expected differences that arise from nitrogen doping/incorporation and the rapid increase in the strength of the valence band tail as the degree of disorder increases with nitrogen content [15]. The current transport in silicon rich a-SiNx:H MSM structures is strongly controlled by the barriers at the metal–semiconductor interfaces. The injected electrons from cathode electrode may recombine with holes, which are ‘‘injected” from anode electrode and are either trapped at Si dangling bond midgap defects or defects that are located in valence band tail. The later gives rise to defects generation through Staebler–Wronski effect in a manner similar to a-Si:H [15], since 1.8 eV is needed to break the weakest silicon-hydrogen bonds. The probability of defect formation depends on both the number of weak Si–Si bonds and the number of Si–H bonds. These in turn
J. Ruan et al. / Microelectronics Reliability 48 (2008) 1241–1244
1243
Fig. 3. TSDC spectra and stored charge of MIM capacitor obtained before and after each successive irradiation experiment. Continuous lines correspond to bias of +20 V and dashed ones to 20 V.
depend on the amount of nitrogen and hydrogen in the material and the temperature at which it is deposited which concentration is high in HF PECVD SiN [16]. Finally, experiments performed in ni-n a-Si:H structures revealed a current-stress induced asymmetry in the device current–voltage characteristic [16–19]. The asymmetry was found to be generated when electrons are injected from the bottom electrode and attributed to the effects of impurities or the intrinsic differences in quality between first and last-grown material. Finally, the temperature was found to play a major role on the speed of trap generation [17–20]. The analysis of the TSDC spectra indicates the presence of different defects which characteristics depend on the applied bias polarity (Fig. 4a and b). This has to be attributed to effects arising from the contact asymmetry, the defects introduced during irradiation and the interaction of injected holes, which interact with as already mentioned Si midgap and band tail defects. The importance of the contribution of Si defects is also supported by the TRIM code calculations, which clearly show a larger introduction rate for Si vacancies. Finally it must be pointed out that as the presence of potential fluctuations and silicon nanoclusters are expected to further contribute to the generation of complex trapping of ionizing radiation generated charges, hence the radiation induced dielectric charging. These conclusions reveal that upon particle radiation the contribution to dielectric charging arises from space charge polarization that leads to quite important variation in the formation and release of excess carriers due to the differences in work function, blocking factor and the resulting defect interactions. The metal contacts obviously do not affect the dipolar polarization. Moreover, the effect of different metal contact work function is expected to affect in a similar way the dielectric charging hence the degradation in the MEMS switches, although for the latter the metal-dielectric contact will be not uniform. The stored charge for radiation fluencies up to 1011 cm2 is presented in Fig. 5. The dependence of stored charge on the radiation fluence seems to be linear. SRIM code [12] calculations have shown that the vacancy introduction rate is in the range of 8000 cm1 in SiN, which leads to a density of defects in the range of 1015 cm3 for the used fluence. This value is very low compared to the reported density of defects of 1018 cm2 [21]. Therefore, for a small perturbation caused by irradiation up to 1011 cm2, we can assume a linear introduction or change in concentration of stored charge, a fact that is confirmed in Fig. 3. Furthermore, since the injected charge is affected by the metal contact work function and the distribution of charged centers is not known we can calculate the degradation parameter defined from:
Fig. 4. Arrhenius plots of TSDC currents obtained before and after each successive irradiation experiment for (a) 20 V and (b) +20 V.
Fig. 5. Dependence of stored charge on radiation fluence for () positive and ( ) negative bias on top electrode.
rD ðFÞ ¼ rD;0 þ a F
ð3Þ
where rD,0 is the residual charge, a a factor related to the charging introduction rate and F the radiation fluence. The obtained values for the charging introduction rate are about 0.28 1020 C cm2 and 2.3 1020 C cm2 for positive and negative bias, respectively.
1244
J. Ruan et al. / Microelectronics Reliability 48 (2008) 1241–1244
The device degradation arises from the primarily radiation introduced vacancies and the injected charge interation/trapping during the switch down state. These effects enhance significantly the contribution of space charge polarization to the dielectric charging. In the case of capacitive switches where the dielectric and contact roughness affect the charging processes alpha particle radiation shifts the bias for minimum capacitance and varies the magnitude OFF-state capacitance. References
Fig. 6. Dependence of () change capacitance at zero bias and ( ) bias for capacitance minimum vs. radiation fluence.
3.3. The MEMS capacitive switches In MEMS capacitive switches the bias that corresponds to capacitance voltage characteristic minimum is determined by the dielectric film charge density rD [22].
V min ¼
z1 rD e0
ð4Þ
where z1 is the dielectric film thickness. If the capacitance voltage characteristic is restricted in below pull-in region, Vmin allows the calculation of residual charging while if the characteristic is extended beyond pull-in, Vmin provides information on the dielectric charging during down state, which occurs under high electric field. Finally, the capacitance for zero bias gives information on the presence of opposite polarity charges [22,23]. So, in the up-state and zero bias the switch bridge displacement, hence capacitance will be proportional to the electrostatic force. Now, if we assume a linear dependence of the additional charging caused by traps that have been introduced by radiation, the bias for minimum capacitance, see Eq. (4), is expected to vary linearly with radiation fluence. Furthermore, if we assume that the bridge displacement at zero bias is much smaller than the bridge-dielectric gap we expect the capacitance to vary as:
DCð0; FÞ / ðrD;0 þ a FÞ2
ð5Þ
The above hypothesis is supported by the plots of DC(F) and DVmin(F) vs. radiation fluence (Fig. 6) where the experimental data are fitted with a square and linear law, respectively. The degradation parameter extracted from the variation of Vmin on radiation fluence was found to range from 7 1020 C cm2 to 17 1020 C cm2, which is in reasonable agreement with the TSDC ones. 4. Conclusions In conclusion the effect of 5 MeV alpha particle irradiation effects has been investigated in both MIM capacitors and RF–MEMS capacitive switches. The experimental data show an asymmetrical charging due to both different metal work functions and asymmetric trap generation during down state charge injection.
[1] Cass S. MEMS in space. IEEE Spectrum 2001;38:56–61. [2] Yao JJ, Chien C, Mihailovich R, Panov R, DeNatale J, Studer J, et al. Microelectromechanical system radio frequency switches in a picosatellite mission. Smart Mater Struct 2001;10:1196–203. [3] Knudson AR, Buchner S, McDonald P, Stapor WJ, Campbell AB, Grabowski KS, et al. The effects of radiation on MEMS accelerometers. IEEE Trans Nucl Sci 1996;43:3122–6. [4] Lee CI, Johnston AH, Tang WC, Barnes CE, Lyke J. Total dose effects on micromechanical systems (MEMS): accelerometers. IEEE Trans Nucl Sci 1996;43:3127–32. [5] McClure SS, Edmonds LD, Mihailovich R, Jonhston AH, Alonzo P, DeNatale J, et al. Radiation effects in micro-electro-mechanical systems (MEMS): RF relays. IEEE Trans Nucl Sci 2002;49:3197–202. [6] Edmonds LD, Swift GM, Lee CI. Radiation response of a MEMS Accelerometer: AN electrostatic force. IEEE Trans Nucl Sci 1998;45:2779–88. [7] Crunteanu A, Pothier A, Blondy P, Dumas-Bouchiat F, Champeaux C, Catherinot A, et al. Gamma radiation effects on RF MEMS capacitive switches. Microelectron Reliab 2006;46:1741–6. [8] Theonas V, Exarchos M, Konstantinidis G, Papaioannou G. RF MEMS sensitivity to electromagnetic radiation. J Phys: Conf Ser 2005;10:313–7. [9] Cazaux J. A physical approach to the radiation damage mechanisms induced by X-rays in X-ray microscopy and related techniques. J Microscopy 1997;188:106–24. [10] Exarchos M, Papandreou E, Pons P, Lamhamdi M, Papaioannou GJ, Plana R. Charging of radiation induced defects in RF MEMS dielectric films. Microelectron Reliab 2006;46:1695–9. [11] Gritsenko VA, Gritsenko DV, Novikov YuN, Kwok RWM, Bello I. Short-range order, large potential fluctuations and photoluminescence in amorphous SiNx. J Exp Theoret Phys 2004;98:760–9. [12] SRIM code: http://www.srim.org/. [13] Vandershueren J, Casiot J. In: Braunlich P, editor. Topics in applied physics: thermally stimulated relaxation in solids, vol. 37. Berlin: Springer-Verlag; 1979. p. 135. [chapter 4]. [14] http://www.kayelaby.npl.co.uk/atomic_and_nuclear_physics/4_3/4_3.html. [15] Shannon JM, Deane SC, McGarvey B, Sandoe JN. Current induced drift mechanism in amorphous SiNx:H thin film diodes. Appl Phys Lett 1994;65:2978–80. [16] Papandreou E, Lamhamdi M, Skoulikidou CM, Pons P, Papaioannou GJ, Plana R. Structure dependent charging process in RF MEMS capacitive switches. Microelect Reliab 2007;47:1822–7. [17] Shannon JM, Morgan BA. Hole transport via dangling-bond states in amorphous hydrogenated silicon nitride. Appl Phys Lett 1999;86:1548–50. [18] Lau SP, Shannon JM, Sealy BJ. Changes in the Poole–Frenkel coefficient with current induced defect band conductivity of hydrogenated amorphous silicon nitride. J Non-Crystall Solids 1998;277–230:533–7. [19] Lau SP, Shannon JM. Generation and annealing kinetics of current induced metastable defects in amorphous silicon alloys. J Non-Crystall Solids 2000;266–269:432–6. [20] Nieuwesteeg KJBM, Boogaard J, Oversluizen G, Powell MJ. Current-stress induced asymmetry in hydrogenated amorphous silicon n-i-n devices. J Appl Phys 1992;71:1290–7. [21] Krick DT, Lenahan PM, Kanicki J. Electrically active point defects in amorphous silicon nitride: an illumination and charge injection study. J Appl Phys 1988;64:3558–63. [22] Papaioannou G, Papapolymerou J, Pons P, Plana R. Dielectric charging in radio frequency microelectromechanical system capacitive switches: a study of material properties and device performance. Appl Phys Lett 2007;90:233507. [23] Rottenberg X, Nauwelaers B, De Raedt W, Tilmans HAC. Distributed dielectric charging and its impact on RF MEMS devices. In: 12th GAAS symposium EuMW; 2004. p. 475–8.
Contents lists available at ScienceDirect
Microelectronics Reliability journal homepage: www.elsevier.com/locate/microrel
Alpha particle radiation effects in RF MEMS capacitive switches J. Ruan a, E. Papandreou b, M. Lamhamdi a, M. Koutsoureli b, F. Coccetti a, P. Pons a, G. Papaioannou a,b,*, R. Plana a a b
LAAS CNRS, 7, av du Colonel Roche, 31077 Toulouse Cedex 4, France Solid State Physics Section, University of Athens, Panepistimiopolis Zografos, Athens 15784, Greece
a r t i c l e
i n f o
Article history: Received 25 June 2008 Available online 10 August 2008
a b s t r a c t The paper investigates the effect of 5 MeV alpha particle irradiation in RF MEMS capacitive switches with silicon nitride dielectric film. The investigation included MIM capacitors in order to obtain a better insight on the irradiation introduced defects in the dielectric film. The assessment employed the thermally stimulated depolarization currents method for MIM capacitors and the capacitance–voltage characteristic for MEMS switches. Asymmetric charging was monitored in MIM capacitors due different contact electrodes and injected charge interactions. Ó 2008 Elsevier Ltd. All rights reserved.
1. Introduction Microelectromechanical systems (MEMS) are receiving increasing interest for use in space systems. One particular area of interest is in Picosats [1,2] where very little shielding is afforded to the space radiation environment. To date, however, few radiation tests have been performed on MEMS devices [3–7]. Tests performed with gamma rays on MEMS accelerometers have shown the technology prone to radiation effects at moderate dose levels [3,4] and a quantitative model for the persisting electrostatic force due to radiation induced charging has been proposed by Edmonds [6]. In this model it was proposed that charging arises from carrier emission from the adjacent electrodes and injection into MEMS dielectric. This process has been further modeled for structures with different dielectrics materials and dielectric thicknesses [8]. Presently, it is well known that ionizing radiation introduces defects in insulating materials [9]. These defects can be detected and their properties can be identified with the aid of Metal–Insulator–Metal capacitors by applying the discharge current transient (DCT) or the thermally stimulated depolarization (TSDC) method. In the case of silicon nitride it has been shown that alpha particle radiation introduces deep defects [10] that increase the time constant of trapped charge, hence the dielectric charging. The aim of the present work is to provide information on the 5 MeV alpha particle radiation inducing degradation in RF MEMS capacitive switches with SiNx dielectric film. The investigation has been performed by employing the TSDC method in MIM capacitors in order to identify the charging processes. Finally, the shift of
* Corresponding author. Address: LAAS CNRS, 7, av du Colonel Roche, 31077 Toulouse Cedex 4, France. Tel.: +33 561337900; fax: +33 561553577. E-mail address: [email protected] (G. Papaioannou). 0026-2714/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.microrel.2008.06.047
the capacitance–voltage characteristic was used to investigate the degradation of MEMS basic characteristics. 2. Experiment The dielectric film of the MIM capacitors and MEMS switches was SiNx with a thickness of 300 nm. The choice to employ MIM capacitors was based on the fact that these structures can mimic a MEMS switch in the ON state and they provide valuable information on the material properties. The top and bottom electrode of MIM structures is composed of a Ti/Au layer (0.1 lm/1 lm). The top electrode, obtained by lift off, features a diameter of 500 lm. The silicon nitride layer was deposited by PECVD at 200 °C with high frequency (HF, 13.56 MHz). The composition of the dielectric film was determined with the aid of Rutherford backscattering spectroscopy (RBS) method that revealed an average value of x ffi 1.04. The TSDC current was measured through a Keithely 6487 voltage source – picoampere meter in the temperature range of 200 K to 450 K. The bias was applied to the top electrode with respect to the bottom one; hence the polarity is in reference to the one of the top electrode. The sacrificial layer of the MEMS capacitive switches features 2.5 lm thickness. The dielectric charging before and after each successive irradiation was assessed through capacitance–voltage characteristics. The region below pull-in threshold turns out to the determination of the magnitude of charge that is trapped in the dielectric after each positive or negative down state. The shift of pull-in and pull-out voltage was also monitored in order to determine the effect of charging under high electric field. Finally the irradiation experiments have been carried out at room temperature with 5 MeV alpha particles provided by an Am source. The alpha particle fluence was common for MIM and MEMS
1242
J. Ruan et al. / Microelectronics Reliability 48 (2008) 1241–1244
Due to higher sensitivity of used experimental methods, fluence up to 1011 cm2 were used for the MIM capacitors and up to 8 1011 cm2 for MEMS switches. 3. Results and discussion 3.1. Simulation of radiation introduced vacancies Low temperature amorphous silicon nitride constitutes a disordered material which homogeneity and stoechiometry may vary significantly across the material. Due to excess silicon, the material may consist of silicon clusters, stoechiometric silicon nitride as well as silicon nitride of varying composition that gives rise to large potential fluctuations that correspond to energy gap variations in the range of 1.6 eV, for amorphous silicon, to 4.6 eV, for Si3N4 [11]. In our case, where x ffi 1.04, reported assessment based on Raman spectra indicated that the silicon clusters size or density or both are rather small [11]. Nevertheless, the presence of potential fluctuations and silicon nanoclusters affects significantly the trapping of charges. High energy particle radiation introduces defects due to lattice atom displacement. The displaced atoms generate vacancies and interstitials. In the case of an amorphous and disordered material like SiNx, it has been shown that particle (Ar) irradiation contributes to nitrogen and silicon atoms mixing due to which the material exhibits a tendency to form a substitution of solid solution [11]. Since the atom mixing is caused by the displaced atoms it becomes obvious that the concentration of generated vacancies can be considered as a measure of the atoms mixing. In order to obtain a better insight, we calculated the introduction rate of Si (VSi) and N (VN) vacancies through the use of TRIM code [12]. The calculation was performed assuming a homogeneous SiN1.04 material since the ion projectile trajectories were not expected to be affected by the material inhomogeneity due to the small thickness of the SiN film and the high energy of alpha particles. The vacancies distribution was calculated assuming a simple MIM capacitor structure which consisted of 1000 nm Au top electrode, 300 nm SiN1.04 film and 500 nm Au bottom electrode. Calculations were also performed by taking into account the presence of the small air gap (2.5 lm) between the MEMS suspended electrode and the dielectric film. The distribution of Si and N vacancies are presented in Fig. 1. A peak observed at the front interface and the increase of vacancy introduction rate with the film thickness caused by the recoiled heavy Au atoms and the cumulative process of recoiled Si and N atoms, respectively. The effect of backscattered Au atoms at the rear interface is negligible. Finally, the extension of Si and N vacancies beyond the SiN film has to be attributed to displaced atoms that are further displaced by knock-on process (see Fig. 2).
Fig. 1. Structure of MIM capacitor.
Fig. 2. Distribution of Si and N vacancy introduction rates (V) for a MIM capacitor.
3.2. The MIM capacitors In insulators, the time and temperature dependence of polarization and depolarization processes are, in the case of dipolar polarization, driven by the competition, between the orienting action of the electric field and the randomizing action of thermal motion. The current density produced by the progressive decrease in polarization during the TSDC experiment, where time and temperature are simultaneously varied, is approximated by:
J D ðTÞ
PS ðT p Þ
s0
" # 2 EA 1 kT EA exp exp exp bs0 EA kT kT
ð1Þ
where b is the heating rate (K/s), EA the depolarization mechanism activation energy, PS (Tp) is the equilibrium polarization at the polarizing temperature Tp and s0 the corresponding infinite temperature relaxation time. In the case of space charge polarization the processes are much more complex because several mechanisms can be involved simultaneously and [13] and the TSDC spectra may show the characteristic properties of distributed processes, such as extension over a wide temperature range, etc. Since the heating rate b is constant, the TSDC spectrum allows the calculation of the stored charge:
Q¼
1 b
Z
T2
ITSDC ðTÞdT
ð2Þ
T1
The TSDC spectra for radiation fluencies up to 1011 cm2 as well as for positive and negative bias polarity applied to top electrode are plotted in Fig. 3. The TSDC spectra are asymmetric and depend strongly on the sign of polarizing bias. This behavior may be attributed to both the asymmetrical electrodes, since the work functions for Au and Ti are about 5.2 eV and 4.2 eV, respectively [14], and the dielectric film properties. The amorphous silicon-rich alloys behave like a-Si:H with expected differences that arise from nitrogen doping/incorporation and the rapid increase in the strength of the valence band tail as the degree of disorder increases with nitrogen content [15]. The current transport in silicon rich a-SiNx:H MSM structures is strongly controlled by the barriers at the metal–semiconductor interfaces. The injected electrons from cathode electrode may recombine with holes, which are ‘‘injected” from anode electrode and are either trapped at Si dangling bond midgap defects or defects that are located in valence band tail. The later gives rise to defects generation through Staebler–Wronski effect in a manner similar to a-Si:H [15], since 1.8 eV is needed to break the weakest silicon-hydrogen bonds. The probability of defect formation depends on both the number of weak Si–Si bonds and the number of Si–H bonds. These in turn
J. Ruan et al. / Microelectronics Reliability 48 (2008) 1241–1244
1243
Fig. 3. TSDC spectra and stored charge of MIM capacitor obtained before and after each successive irradiation experiment. Continuous lines correspond to bias of +20 V and dashed ones to 20 V.
depend on the amount of nitrogen and hydrogen in the material and the temperature at which it is deposited which concentration is high in HF PECVD SiN [16]. Finally, experiments performed in ni-n a-Si:H structures revealed a current-stress induced asymmetry in the device current–voltage characteristic [16–19]. The asymmetry was found to be generated when electrons are injected from the bottom electrode and attributed to the effects of impurities or the intrinsic differences in quality between first and last-grown material. Finally, the temperature was found to play a major role on the speed of trap generation [17–20]. The analysis of the TSDC spectra indicates the presence of different defects which characteristics depend on the applied bias polarity (Fig. 4a and b). This has to be attributed to effects arising from the contact asymmetry, the defects introduced during irradiation and the interaction of injected holes, which interact with as already mentioned Si midgap and band tail defects. The importance of the contribution of Si defects is also supported by the TRIM code calculations, which clearly show a larger introduction rate for Si vacancies. Finally it must be pointed out that as the presence of potential fluctuations and silicon nanoclusters are expected to further contribute to the generation of complex trapping of ionizing radiation generated charges, hence the radiation induced dielectric charging. These conclusions reveal that upon particle radiation the contribution to dielectric charging arises from space charge polarization that leads to quite important variation in the formation and release of excess carriers due to the differences in work function, blocking factor and the resulting defect interactions. The metal contacts obviously do not affect the dipolar polarization. Moreover, the effect of different metal contact work function is expected to affect in a similar way the dielectric charging hence the degradation in the MEMS switches, although for the latter the metal-dielectric contact will be not uniform. The stored charge for radiation fluencies up to 1011 cm2 is presented in Fig. 5. The dependence of stored charge on the radiation fluence seems to be linear. SRIM code [12] calculations have shown that the vacancy introduction rate is in the range of 8000 cm1 in SiN, which leads to a density of defects in the range of 1015 cm3 for the used fluence. This value is very low compared to the reported density of defects of 1018 cm2 [21]. Therefore, for a small perturbation caused by irradiation up to 1011 cm2, we can assume a linear introduction or change in concentration of stored charge, a fact that is confirmed in Fig. 3. Furthermore, since the injected charge is affected by the metal contact work function and the distribution of charged centers is not known we can calculate the degradation parameter defined from:
Fig. 4. Arrhenius plots of TSDC currents obtained before and after each successive irradiation experiment for (a) 20 V and (b) +20 V.
Fig. 5. Dependence of stored charge on radiation fluence for () positive and ( ) negative bias on top electrode.
rD ðFÞ ¼ rD;0 þ a F
ð3Þ
where rD,0 is the residual charge, a a factor related to the charging introduction rate and F the radiation fluence. The obtained values for the charging introduction rate are about 0.28 1020 C cm2 and 2.3 1020 C cm2 for positive and negative bias, respectively.
1244
J. Ruan et al. / Microelectronics Reliability 48 (2008) 1241–1244
The device degradation arises from the primarily radiation introduced vacancies and the injected charge interation/trapping during the switch down state. These effects enhance significantly the contribution of space charge polarization to the dielectric charging. In the case of capacitive switches where the dielectric and contact roughness affect the charging processes alpha particle radiation shifts the bias for minimum capacitance and varies the magnitude OFF-state capacitance. References
Fig. 6. Dependence of () change capacitance at zero bias and ( ) bias for capacitance minimum vs. radiation fluence.
3.3. The MEMS capacitive switches In MEMS capacitive switches the bias that corresponds to capacitance voltage characteristic minimum is determined by the dielectric film charge density rD [22].
V min ¼
z1 rD e0
ð4Þ
where z1 is the dielectric film thickness. If the capacitance voltage characteristic is restricted in below pull-in region, Vmin allows the calculation of residual charging while if the characteristic is extended beyond pull-in, Vmin provides information on the dielectric charging during down state, which occurs under high electric field. Finally, the capacitance for zero bias gives information on the presence of opposite polarity charges [22,23]. So, in the up-state and zero bias the switch bridge displacement, hence capacitance will be proportional to the electrostatic force. Now, if we assume a linear dependence of the additional charging caused by traps that have been introduced by radiation, the bias for minimum capacitance, see Eq. (4), is expected to vary linearly with radiation fluence. Furthermore, if we assume that the bridge displacement at zero bias is much smaller than the bridge-dielectric gap we expect the capacitance to vary as:
DCð0; FÞ / ðrD;0 þ a FÞ2
ð5Þ
The above hypothesis is supported by the plots of DC(F) and DVmin(F) vs. radiation fluence (Fig. 6) where the experimental data are fitted with a square and linear law, respectively. The degradation parameter extracted from the variation of Vmin on radiation fluence was found to range from 7 1020 C cm2 to 17 1020 C cm2, which is in reasonable agreement with the TSDC ones. 4. Conclusions In conclusion the effect of 5 MeV alpha particle irradiation effects has been investigated in both MIM capacitors and RF–MEMS capacitive switches. The experimental data show an asymmetrical charging due to both different metal work functions and asymmetric trap generation during down state charge injection.
[1] Cass S. MEMS in space. IEEE Spectrum 2001;38:56–61. [2] Yao JJ, Chien C, Mihailovich R, Panov R, DeNatale J, Studer J, et al. Microelectromechanical system radio frequency switches in a picosatellite mission. Smart Mater Struct 2001;10:1196–203. [3] Knudson AR, Buchner S, McDonald P, Stapor WJ, Campbell AB, Grabowski KS, et al. The effects of radiation on MEMS accelerometers. IEEE Trans Nucl Sci 1996;43:3122–6. [4] Lee CI, Johnston AH, Tang WC, Barnes CE, Lyke J. Total dose effects on micromechanical systems (MEMS): accelerometers. IEEE Trans Nucl Sci 1996;43:3127–32. [5] McClure SS, Edmonds LD, Mihailovich R, Jonhston AH, Alonzo P, DeNatale J, et al. Radiation effects in micro-electro-mechanical systems (MEMS): RF relays. IEEE Trans Nucl Sci 2002;49:3197–202. [6] Edmonds LD, Swift GM, Lee CI. Radiation response of a MEMS Accelerometer: AN electrostatic force. IEEE Trans Nucl Sci 1998;45:2779–88. [7] Crunteanu A, Pothier A, Blondy P, Dumas-Bouchiat F, Champeaux C, Catherinot A, et al. Gamma radiation effects on RF MEMS capacitive switches. Microelectron Reliab 2006;46:1741–6. [8] Theonas V, Exarchos M, Konstantinidis G, Papaioannou G. RF MEMS sensitivity to electromagnetic radiation. J Phys: Conf Ser 2005;10:313–7. [9] Cazaux J. A physical approach to the radiation damage mechanisms induced by X-rays in X-ray microscopy and related techniques. J Microscopy 1997;188:106–24. [10] Exarchos M, Papandreou E, Pons P, Lamhamdi M, Papaioannou GJ, Plana R. Charging of radiation induced defects in RF MEMS dielectric films. Microelectron Reliab 2006;46:1695–9. [11] Gritsenko VA, Gritsenko DV, Novikov YuN, Kwok RWM, Bello I. Short-range order, large potential fluctuations and photoluminescence in amorphous SiNx. J Exp Theoret Phys 2004;98:760–9. [12] SRIM code: http://www.srim.org/. [13] Vandershueren J, Casiot J. In: Braunlich P, editor. Topics in applied physics: thermally stimulated relaxation in solids, vol. 37. Berlin: Springer-Verlag; 1979. p. 135. [chapter 4]. [14] http://www.kayelaby.npl.co.uk/atomic_and_nuclear_physics/4_3/4_3.html. [15] Shannon JM, Deane SC, McGarvey B, Sandoe JN. Current induced drift mechanism in amorphous SiNx:H thin film diodes. Appl Phys Lett 1994;65:2978–80. [16] Papandreou E, Lamhamdi M, Skoulikidou CM, Pons P, Papaioannou GJ, Plana R. Structure dependent charging process in RF MEMS capacitive switches. Microelect Reliab 2007;47:1822–7. [17] Shannon JM, Morgan BA. Hole transport via dangling-bond states in amorphous hydrogenated silicon nitride. Appl Phys Lett 1999;86:1548–50. [18] Lau SP, Shannon JM, Sealy BJ. Changes in the Poole–Frenkel coefficient with current induced defect band conductivity of hydrogenated amorphous silicon nitride. J Non-Crystall Solids 1998;277–230:533–7. [19] Lau SP, Shannon JM. Generation and annealing kinetics of current induced metastable defects in amorphous silicon alloys. J Non-Crystall Solids 2000;266–269:432–6. [20] Nieuwesteeg KJBM, Boogaard J, Oversluizen G, Powell MJ. Current-stress induced asymmetry in hydrogenated amorphous silicon n-i-n devices. J Appl Phys 1992;71:1290–7. [21] Krick DT, Lenahan PM, Kanicki J. Electrically active point defects in amorphous silicon nitride: an illumination and charge injection study. J Appl Phys 1988;64:3558–63. [22] Papaioannou G, Papapolymerou J, Pons P, Plana R. Dielectric charging in radio frequency microelectromechanical system capacitive switches: a study of material properties and device performance. Appl Phys Lett 2007;90:233507. [23] Rottenberg X, Nauwelaers B, De Raedt W, Tilmans HAC. Distributed dielectric charging and its impact on RF MEMS devices. In: 12th GAAS symposium EuMW; 2004. p. 475–8.
