Fast and simple method for estimation and separation of radiation-Induced traps in MOSFETs devices B.Nadji 1, H.Tahi1,2 , B.Djezzar2
1
Microelectronic and Microsystems Team ,Laboratory of Electrification of industrials enterprises Faculty of Hydrocarbones and Chemistry,University of Boumerdes, Algeria 2 Microelectronics and Nanotechnology Division,Centre de Développement des Technologies Avancées (CDTA), 20 Août 1956, BP: 17,Baba Hassen, Algiers 16303
[email protected]
on charge pumping without a need of any additional technique and is applied on single transistor. However, this method underestimate the ΔNbt because the low frequency measurement. In this work, we propose a new method based on standard CP and I-V techniques to accurate estimation and separation of radiation-induced border traps in single transistor. The results of this method are confronted to those extracted using OTCP and DTBT.
Abstract - In this work, we propose a simple and fast method to estimate the radiation-induced traps in P and NMOS transistors independently. This method is based on standard current-voltage and Charge Pumping (I(V)-CP) to separate the radiation-induced border-traps (∆Nbt) and true interface-traps (∆Nit), where the radiation-induced oxidetraps (∆Not) are extracted classically by measuring the threshold voltage (∆Vth) or Mid-Gap (∆Vmg) voltage shift. The charge pumping (CP) curves are measured using the rise and fall saw-tooth signal for N-and P-MOS transistors respectively, to minimize the border-trap estimation error caused by the difference in the energy band gap scanned by standard I(V) and CP techniques. Emphasis is made on critical comparison between the radiation induced ∆Nbt extracted using I(V)-CP and classical method such as OTCP and DTBT. According to experimental data, the I(V)-CP method is more accurate than OTCP and DTBT methods, since it is more sensitive than OTCP method for the extraction of border traps and it can gives all kinds of traps for P and N-MOS transistors separately. Keywords - radiation-induced traps, P and N-MOS transistors, Border trap, Charge Pumping
II. METHODOLOGY A commonly used definition of border traps is that an oxide traps located near the Si/SiO2 interface which communicate with the underlying silicon via interface traps. The Fig.1 illustrates the electrical response of the bulk oxide, border and interface-traps with the underlying Silicon. Electrical response
I. INTRODUCTION An important challenge in Radiation Hardness Assurance (RHA) of MOS transistors for space and nuclear applications is the development of rapid, simple and reliable methods to estimate and separate the radiation-induced traps. Nowadays, a number of methods are commonly used such as: Capacitance-Voltage (CV)[1-2], Sub Threshold Slop (STS)[3], Mid Gap (MG)[4], Dual Transistor Mobility (DTM)[5], Dual Transistor Charge Pumping (DTCP)[5], Dual Transistors Border Trap (DTBT) [6] and recently Oxide Trap based on Charge Pumping (OTCP) [7]. Each of these methods offers potential advantages and disadvantages compared to the others. From these methods only C-V hysteresis, DTBT and OTCP can estimate ΔNbt. However C-V method needs a large area device, but has also the problem of extrapolation of measurement results to the actual device in the product line [1]. The DTBT method not only combines I-V and C-P measurements, but also
V
Poly-Si Poly-Si
Poly-Si
Poly-Si
SiO2
SiO2
SiO2
Not
V
Nbt
Si
Si
Nit
Si
Defect types
Oxide trap (N ot) 0,0
10
-6
10
-3
Border trap (N bt)
10
0
10
3
Interface trap (Ν it)
CP
10
6
10
9
Frequancy (Hz)
Fig. 1. Schematic illustration of electrical response defects in MOS device.
In standard I-V techniques (STS and MG), the radiation induced-interface traps ΔNit (I-V) are extracted from the stretch out of the sub-threshold slope, using equation (1)
requires both N- and P-channel transistors fabricated with the same technology. In addition, it assumes that the border-
ΔN it ( I −V ) = ΔS
trap density produced by irradiation is the same in both PMOS and N-MOS transistors. The OTCP method is based
978-1-4577-0628-8/11/$26.00 ©2011 IEEE
Swintching states
Fixed states V
469
Cox Ln ( N a , d ni ) qLn (10)
(1)
Proceedings of the 2011 IEEE ICQR
Fig.2 shows I(V)-CP extraction for N-MOS transistor irradiated at 500Krad. For accurate estimation of border traps by equation (3), the portion of band gap energy scanned by both standard CP and I-V techniques must be the same. The standard I-V technique measures ∆Nit (I-V) from mid gap to threshold voltage. In other words, it measures the traps located between Ei=qΦB (mid gap) end Einv=2qΦB (strong inversion), where ΦB is the bulk potential. Therefore the energy band gap scanned by I-V can be expressed as:
where ∆S (V/decade) is the stretch out of subthreshold slop. Cox (F/cm2) is the gate oxide capacitance per area unit, Na,d (cm-3) is the channel doping, ni (cm-3) is the intrinsic concentration and q (C) is the electron charge. Due to the low frequency measurement of standard I-V technique (~1Hz) the border traps have enough time to exchange the charge with silicon. Therefore the border traps are sensed like interface traps ∆Nit (I-V) and we can write: (2)
Δ N it ( I −V ) = Δ N it + Δ N bt
Δ N bt = Δ N it ( I −V ) − ΔN it
(3)
where the ∆Nit (cm-2) are the true radiation-induced traps extracted using standard charge pumping at high frequency (1Mhz) to avoid the electrical response of border traps in charge pumping current (ICP) [7]. Thus, ∆Nit can be written as:
Δ N it =
Δ I cp max,
(4)
h
qf h A G
ot
=
C ox ( Δ V mg ) q
-1
1,5
0
t em , h =
Vgs (V)
1
(6)
2
3
NMOS WG/LG=10/10
4
t em ,e =
Virgin Irradiated at 500Krad
-3
10
-5
10 ΔIcpmax
ΔS=S1-S0
-7
10
S1
-9
10
S0
0,6
Ids(A)
Icp x 10-8 (A)
ΔVth
-11
10
0,3
-13
10
0,0 ΔVmg
-6
-5
-4
-15
-3
-2
-1
0
1
2
(8) (9)
V th − V fb ΔV V th − V fb ΔV
tr
(10)
tf
(11)
where Vth and Vfb are the threshold and flat band voltages, respectively. tr is the rise time, tf is the fall time and ∆V is the signal amplitude . The Eem,h and Eem,h expressed by (8) and (9) can be modulated by tr and tf, respectively. It is usually assumed that the carriers emission can be neglected for sufficiently small rise and fall times. Consequently Eem,e≈ Einv for NMOS and Eem,h≈ Einv for P-MOS transistors. However, for larger rise and fall times, Eem,e and Eem,h move closer to mid gap. Subsequently, to scan the same portion of the band gap energy using I(V) and CP technique, we can use the rise and fall saw signal for N-and P-MOS transistors, respectively. Fig 3 compares the energy band scanned by CP using (fall) rise saw signal and I(V) technique for NMOS (P-MOS) transistors.
5
1,2 0,9
Eem,h = Ei + kT ln(vthσ p ni tem,h )
(5)
C ox ( Δ V th ) − ( Δ N it + Δ N bt ) q
(7)
where σp (σp) is the capture cross-section of holes (electrons), vth is the thermal velocity of carriers, tem,h and tem,e are holes and electrons emission time, respectively. In the case of trapezoidal signal tem,h and tem,e are given by:
or by: Δ N ot =
⎞ ⎟⎟ ⎠
where ∆EI-V (eV) is the band gap energy scanned by I-V, Einv(eV) is the Fermi level energy at inversion mode, Ei(eV) is the intrinsic Fermi level, k (eV/K) is the Boltzman constant and T(K) is the absolute temperature. However, in CP technique, the fraction of band gap energy, in which the interface-traps contribute to CP current, is delimited by Eem,h and Eem,e . Where Eem,h and Eem,h are the end of non-stady state emission of electron and hole, respectively, and are given by:
Eem,e = Ei − kT ln(vthσ n ni tem,e )
where ∆Icpmax,h (A) is the radiation-induced maximum charge pumping current, fh (Hz) is the high frequency measurement (1MHz) and AG (cm-2) is the gate transistor area. The radiation-induced oxide traps (∆Not) can be extracted using: ΔN
⎛ N a ,d ΔE I −V = Einv − Ei = q (2Φ B − Φ B ) = kT ln⎜⎜ ⎝ ni
10
VL (V)
Fig.2 CP and subthreshold of Ids(Vgs) curves before and after irradiation at 500krad for N-MOS transistor, illustrating I(V)-CP extraction method.
470
Proceedings of the 2011 IEEE ICQR
17
-3
16
-3
Na=1,775 10 (cm )
The CP and Ids(Vgs) measurements were made before and after each irradiation at 25°C. The irradiated devices were characterized after the latent interface-trap build-up (106s). In other words, we made a long delay between irradiation and measurements. The extraction of ∆Nit was carried out after taking out the LDD and LOCOS effects from the devices with fixed gate width and length, respectively [8]-[9].
Nd=3,910 10 (cm ) Ei qΦF (PMOS) I(V) PMOS
I(V) NMOS
Eem,e(r)
Eem,h(r) Eem,e(f)
Eem,h(f)
qΦF (NMOS)
CP
CP
IV. RESULTS AND DISCUSSION rise saw signal
fall saw signal
Fig.4 (a) and (b) display radiation-induced ΔNbt in N and P-MOS transistors extracted using DTBT, OTCP and I(V)-CP. All ΔNbt measured by OTCP are less than those extracted by DTBT and I(V)-CP. This is probably, due to the strong sensitivity of border traps to time frame measurement. Ended, In OTCP method the largest time constant that permits the characterization of border-trap is typically ranged between 10-1 - 10-2 s. Such measurements only involve a fraction of border traps in the frame of OTCP measurements; the remainder fraction of the traps is considered from an electrical viewpoint as fixed oxide trap. Whereas, the DTBT and I(V)-CP methods allow more border traps to exchange carriers with the silicon, because the effective time measurement range is typically 1s (1Hz). Hence the lower the measurement frequency is, the more important border traps. However we have observed a good correlation between ∆Nbt extracted by DTBT and the average values of ∆Nbt measured by I(V)CP for P and N-MOS transistors. The relative difference is less than 5% and 20% for fixed length and width devices, respectively. Since DTBT method can not sense the radiation induced ∆Nbt in P and N-MOS separately, I(V)-CP method is more accurate, it can estimates the radiation-induced ∆Nbt in P an N-MOS devices independently. Nevertheless, the I(V)-CP method estimate the radiation-induced ∆Nit like OTCP method, sense the radiation- induced ∆Nbt better than OTCP and give the radiation-induced ∆Not using (4) or (5).
-0,6 -0,5 -0,4 -0,3 -0,2 -0,1 0,0 0,1 0,2 0,3 0,4 0,5 0,6 E-EI (eV)
Fig.3. Energy band scanned by I(V) and CP, the rise (fall) saw-tooth
III. EXPERIMENTAL SET-UP In this work, we have investigated N-MOS and PMOS transistors fabricated on the same chip at ISiT (Institute for Silicon Technology) of Fraunhofer, Germany. The process is a conventional (soft) dual layer metal 1 μm-CMOS twin-well technology on P-type 12 μm-epi-layer on silicon substrate with 20 nm thick gate oxide layer grown in dry O2. The gate capacitance per area unit, Cox is about of 1.35x10-7(F.cm-2). The nonpackaged transistors have a gate width (length) varying from 10 μm to 1 μm with fixed gate length (width) 10 μm. The irradiation was performed at DAN (Division Applications Nuclear Laboratory) in Algeria. The radiation dose was up to 500 Krad with dose-rate of 200 rad/min, using 1.25 MeV γ-rays at room temperature and zero bias voltage. The measurements were made before and after each irradiation at room temperature.
(a)
13
10
P and N-MOS @ 500 Krad WG=10 µm
12
12
OTCP NMOS OTCP PMOS DTBT I(V)-CP NMOS I(V)-CP PMOS I(V)-CP (NMOS+PMOS)/2
-2
ΔNbt (cm )
10
-2
ΔNbt (cm )
P and N-MOS @ 500 Krad LG=10 µm
OTCP NMOS OTCP PMOS DTBT I(V)-CP NMOS I(V)-CP PMOS I(V)-CP (NMOS+PMOS)/2
10
(b)
13
10
10
10
10
10
9
10
9
10
11
10
11
10
0
2
4
LG(µm)
6
8
10
0
2
4
6 WG(µm)
8
10
Fig.4 Radiation-induced border traps (∆Nbt) extracted using I(V)-CP, OTCP and DTBT, in P and N-MOS transistors with; (a) fixed gate width and varied gate length .(b) fixed gate length and varied gate width.
471
Proceedings of the 2011 IEEE ICQR
V. CONCLUSION In this work, we have combined the standard I(V) and CP techniques to estimate the radiation-induced traps. We have shown that this fast method can gives more accurate estimation of radiation- induced border traps than DTBT and OTCP methods. REFERENCES [1]
[2] [3]
[4]
[5]
[6]
[7]
[8]
[9]
L. M. Terman, “An Investigation of Surface States at a Silicon/Silicon Dioxide Interface Employing Metal-Oxide-Silicon Diodes,” Solid-state Electron., 5,285 (1 962). S. M. Sze, Physics of Semiconductor Devices. New York:Wiley, 1981. S. Winokur, J. R. Schwank, P. J. McWhorter. P.V.Dressendorfer, and D. C. Turpin, "Correlating the Radiation Response of MOS Capacitors and Transistors,''IEEE Trans. Nucl.Sci. vol 31, pp14531453, Dec 1984. P. J. McWhorter and P. S. Winokur, "Simple Technique for Separating the Effects of Interface Traps and Trapped-Oxide Charge in Metal-Oxide Semiconductor Transistors," Appl. Phys. Lett. Vol.48, pp.133-135, Jan. 1986. D. M. Fleetwood, M. R. Shaneyfelt, J. R. Schwank, P. S. Winokur, and F. W. Sexton, "theory and application of dual-transistor charge separation analysis" Techniques," IEEE Trans. Nucl. Sci. vol 36, pp 1816-1824, 1989. D. M. Fleetwood, M. R. Shaneyfelt, W. L. Warren, J. R. Schwank, T. L.Meisenheimer, and P. S.Winokur, “Border traps: Issues for MOS radiationresponse and long-term reliability,” Microelectron. Reliab., vol. 35,pp. 403–428, 1995. B. Djezzar, S. Oussalah, and A. Smatti, “A new Oxide-Trap based on Charge Pumping (OTCP) extraction method for irradiated MOSFET devices: part I (high frequencies) ,” IEEE Trans. Nucl. Sci. vol. 51, pp. 1724-1731, Aug. 2004. B. Djezzar and H. Tahi, “ Using Oxide-Trap Charge-Pumping Method in Radiation Reliability Analysis of Short Lightly Doped Drain Transistor,” IEEE Tran. Dev. Mat. Vol. 10, pp.18-25, March 2010. H.Tahi, B.Djezzar and B.Nadji “Radiation effect Evaluation in Effective Short and narrow channel of LDD-Transistor with LOCOS Isolation Using OTCP Method,” IEEE Tran. Dev. Mat. Relia. Vol. 10, pp. 108-115, March 2010
472
