the role of surface recombination in

Impact of surface recombination on drain charge collection for a moderate LET ion attack in the channel. Spatial charge collection distributions from finite-element simulations without surface recombination and a high SRV.

Dissertation organization

The validated model is used to demonstrate how single-event charge collection is affected by surface recombination. Gamma radiation is used to manipulate surface recombination in situ to illustrate the impact that surface recombination can have on single event phenomena.

Radiation environments

The potential gradient of the junction (the electric field) will be redistributed according to the conductivity of the ion track [29], visualized in fig. Simulations and experiments can be used to determine how much of the charge will be collected at critical circuit nodes based on where it is deposited [30], [31].

Fig. 3. Parker Solar Probe measurements of the solar energetic particle environment [7]

Mechanisms of radiation energy deposition

Photon energy deposition

Particle energy deposition

In order for an ion to collide with a nucleus, it must overcome the repulsive force between the positive charge of the ion and the positive charge of the nucleus (e.g. the Coulomb barrier). MRED is useful for calculating the energy deposition produced in sensitive volumes by particle tracks, including those of secondary reaction products, but cannot describe the physics of the device (e.g., the collective motion of charges acted upon by electric fields).

Fig. 12. Particle-matter interactions. The top panels illustrate three types of particle interactions that result in energy deposition

Mechanisms of single-event charge collection

Thermalized carrier assumption

In this work, we assume that the excess carrier concentration does not change appreciably while crossing the multiscattering gap during carrier relaxation (e.g., the range of electrons with 10 eV < E < 100 eV is negligible compared to the unit functions). Following this assumption, band structure-based methods can be used to approximate a thermalized carrier distribution. Synopsis The Sentaurus TCAD package is used to perform FEM of single events and other unit physics in this thesis.

Drift and diffusion of electrons and holes are calculated from the electric field, carrier mobility (derived using the effective mass approximation), and carrier concentration gradients. A Gaussian distribution of electron–hole pairs with a radial standard deviation of 50 nm sufficiently captures track structure and carrier relaxation effects for modeling subsequent drift and diffusion processes.

Fig. 19. Electron and hole thermalization after ion strike. In the top figures a dense column of energetic electron-hole pairs crosses an n + p junction

Junction charge collection

The rate of recombination occurring along the device is visualized at the bottom of the figure. Prior to the stop, a depletion region is formed between the n+ and p-sides of the diode. Recombination takes place along the entire length of the device, especially in the heavily doped n-type region.

An ion attack in the drain of the off-state nmos device causes the output of the inverter (tied to both the nmos and pmos drains) to drop as the transient current is collected. The current flowing at the output is limited by the current drive of the pmos device and exhibits a plateau where the faulty state is maintained [34], [35].

Fig. 21. Single-event charge collection in an n + p junction. The electron and hole carrier concentrations are illustrated extending outward from the conduction and valence band, respectively

Total ionizing dose mechanisms

TID conditions for increasing surface recombination

An important early finding in research on TID effects was that unless there is a significant electric field across the oxide, many of the electron-hole pairs are created during irradiation. The electric field must be directed to drive positively charged holes and protons towards the interface with the semiconductor in order to maximize the creation of interface traps. In addition to the presence of an electric field in the oxides during irradiation, it is also important that the dose rate is sufficiently low [67], [68].

The effect of the electric field on the recombination of electron-hole pairs in SiO2 from different sources of ionizing radiation. Although dose rate and oxide electric field are clearly the most important irradiation conditions for TID effects, temperature during irradiation and annealing time after irradiation have also been demonstrated to be important considerations.

Fig. 26. The effect of electric field on recombining electron-hole pairs in SiO 2 from various ionizing radiation sources

Recombination physics

Surface recombination velocity measurements

They measured the SRV of thermally grown and chemically etched Si/SiO2 gate oxide interfaces using a large alternating current signal conduction technique, before and after annealing steps. They found that the SRV of passivated surfaces increased linearly with doping greater than 1018 per cm3 from 102 to 5 x 104 cm/s. The measured values presented in this section are compared with new measurements of an irradiated buried oxide interface in Chapter VI.

The active silicon and buried oxide thickness for ultra-thin body and BOX (UTBB) FDSOI devices can be 10 nm or less while the technology used in this thesis is PDSOI with silicon thickness of 250 nm and a 200 nm thick buried oxide . The perimeter of the electrically sensitive semiconductor region is composed of a buried oxide layer and shallow trench insulation (STI).

Fig. 30. Small and large signal AC conductance techniques for investigating interface defects [82]

Leakage paths

Generation leakage current

For surface generation, the generated current is proportional to the depleted surface and hence the drain voltage. 34, the drain current is plotted as a function of gate voltage for different surface generation rates. As the excitation velocity increases, more leakage at the drain is measured at low gate voltages.

Floating body effects

The accumulated holes in the body mimic the positive voltage applied to the base of a bipolar transistor and current flows from source to drain (emitter to collector). The parasitic bipolar structure in an SOI device is shown in the upper part of Fig. 35 illustrated, the potential distribution (left) and hole concentration (right) across the body of the SOI device are plotted at three times: at equilibrium before an ion strike, during an ion strike, and 85 ps after the ion strike.

Holes introduced in the body region modulate the potential and preset the source-body (emitter-base) junction during ion impact, engaging the bipolar operation of the structure. Holes introduced in the body region forward the source-body (emitter-base) junction, which turns on the bipolar element.

Fig. 35. Parasitic bipolar enhancement in SOI. Holes introduced in the body region forward-bias the source-body (emitter-base) junction turning on the bipolar element

Capacitive influence

Capturing bipolar enhancement in SOI devices is important for quantitative SEE modeling and is further illustrated in Section III.5. After further experimentation, they found that the spatial variation in charge collection was due to the non-uniform passivation of interface defects (Fig. 37 bottom). Bottom: Effective passivation near the edges results in a larger depletion region and higher charge accumulation.

As discussed in Section II.4, unpassivated interface defects can be charged and their charge serves to screen the applied bias; therefore, unpassivated defects reduce the size of the depletion region and the amount of charge collected.

Fig. 37. Charge induced as a function of strike location across a capacitor during microbeam testing [98]

Radiation-hardened SOI

BUSFET features

39, the backchannel leakage for a deep source (previously shown in Figure 33) is compared to the leakage with a shallow source using 2D FEM modeling. The shallow source reduces rear channel leakage by increasing the length and resistance of the rear channel leakage path. The body contact is used to control the potential in the body area and prevent floating body effects.

The body contact and shallow source allow holes to easily escape the body area and reduce parasitic bipolar amplification. The placement of body ligaments has a direct impact on their efficacy, with tighter body ligaments offering less resistance in maintaining body potential [89].

BUSFET optimization

The modeled cross sections are calculated using MRED with sensitive volumes derived from 3D TCAD charge collection modeling of the component transistors. Quantitative modeling of charge collection in a single event requires an accurate physical description of the device functions and valid physics-based models. Calculations of doping-dependent mobility, Auger recombination, SRH recombination, and surface recombination were performed using customized Sentaurus physics models.

The next section discusses the Sandia device model and IV characteristics, including the effect of surface recombination. The next section discusses simulations of charge collection of heavy ions and the effect of surface recombination on charge collection.

Fig. 41. Top: Experimental SEE cross-section measurements on Sandia’s CMOS7 performed at the Texas A&M cyclotron

Generation leakage current in SNL SOI

The experimentally selected device has a particularly large port width to aid microbeam analysis; however, this also allows the generation leakage current to be measured in correlation with surface recombination. The width of the depletion region and hence the generating area increases as the square root of the drain voltage. As a result, the increase in generation leakage current with drain voltage is sublinear.

Based on an SRV of 104 cm/s for buried oxide interfaces measured by Nakamura in Section II.5.1, the off-state leakage due to generation along the buried oxide interface is approximately ~1 pA.

Fig. 43. Generation leakage current with different drain voltages and an SRV of 10 6 cm/s

Single-events in SNL SOI

Single-handed transient current (top) and accumulated charge (bottom) in source, drain, and body contacts for an impact site in the channel. As the hole current collects in the body, the charge collection tail in the drain falls off. The peak flow in the drain as a function of impact location is shown in fig.

A high peak current is measured when ion strikes occur within the channel, especially in the drain body depletion region. When charge is deposited directly into the heavily doped deep drain, much less charge is collected.

Fig. 45. Single-event transient current (top) and collected charge (bottom) in source, drain, and body contacts for a strike location in the channel

Impact of surface recombination on charge collection

Effect of surface recombination on drain charge collection for high LET ion impact in the channel. 54, the amount of charge lost due to increasing surface recombination is plotted as a function of impact location. In the above image, surface recombination along the buried oxide and the shallow trench isolation is simulated.

55 shows the amount of charge lost due to surface recombination along all insulating interfaces for a higher ion LET (LET = 50 MeV-cm2 per mg). Spatial distribution of charge lost due to surface recombination along all insulator interfaces (top) and only along the STI (bottom).

Fig. 49. Impact of surface recombination on drain charge collection for a high LET ion strike in the channel

Enhanced surface recombination velocity

As discussed in Chapters II and III, off-state leakage can be attributed to either interface generation in junction depletion regions or back-channel/side-channel leakage pathways. The amount of leakage measured after irradiation is consistent with a surface generation rate of 106 cm/s. To confirm that the electrostatic influence of oxide-trapped charge does not create significant off-state leakage, a substrate bias sweep up to +10 V (bottom inset in Fig. 63) was performed and resulted in none.

Top: The increase in OFF-state leakage current occurs after TID and corresponds to interface generation from insulation interfaces observed in the modeling results. Bottom: Linear I(V) and gate current do not show pronounced TID effects or back channel conduction.

Fig. 62. Experimental high-speed device packaging with substrate contact. An electric field is created across the isolation oxide during irradiation with an applied substrate bias

Impact of surface recombination on charge collection measurements

The mean charge accumulation distribution exhibits an immediate drop after TID but is stable with accumulated ion flux. Correlating the reduction of charge collection after gamma irradiation during experiments with the simulated results for charge collection as a function of SRV provides an estimate of the degraded quality of the interface. The simulated shock location is resolved with the same charge accumulation as the experimental pre-rad mean.

Spatial charge collection distributions for a.) IBICC measurements before and after gamma rays b.) finite element simulations with low surface recombination, corresponding to pre-rad, and high surface recombination (106 cm/s) corresponding to post-rad interface conditions. Our results indicate that the SRV of TID-degraded insulation oxides can be 2 to 20 times greater, in the range 105 – 106 cm/s, and can significantly impact the charge collection in single events.

Fig. 65. TRIBICC charge collection distribution for PDSOI device pre- and post-rad. (Inset) The mean of the charge collection distribution exhibits an abrupt drop after TID but is stable with accumulated ion fluence

Technology trends

BUSFET device structure with shallow source and body contact
Left: Back-channel leakage in SOI with a shallow or deep source. Right: Leakage in Sandia’s
Charge collection from a single event in a normal SOI MOSFET compared to the BUSFET
Top: Experimental SEE cross-section measurements on Sandia’s CMOS7 performed at the
Generation leakage current along buried oxide interface in Sandia SOI
Generation leakage current with different drain voltages. SRV of 10 6 cm/s
Single-event ion strike in SNL SOI device channel
Single-event transient current (top) and collected charge (bottom) in source, drain, and body
Peak drain current as a function of ion strike location for an LET of 7 MeV-cm 2 per mg with
Single-event charge collection in source, drain, and body contacts as a function of ion strike
Finite-element modeling of interface effects. On the top, a simulated ion strike generates excess
Impact of surface recombination on drain charge collection for a high LET ion strike in the

Weller et al., “A general framework for predicting the degree of single-event effects in microelectronics,” IEEE Trans. King et al., "Effect of delta-beams on single-event interference in highly scaled SOI SRAMs," IEEE Trans. Ryder et al., “Polarization Dependence of Pulsed Laser-Induced SEEs in SOI FinFETs,” IEEE Trans.

Detcheverry et al., “Critical charge SEU and the sensitive region in submicron CMOS technology,” IEEE Trans. Loveless et al., “On-chip single-event transient measurement in 45 nm silicon-on-insulator technology,” IEEE Trans.