04033819

IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 53, NO. 6, DECEMBER 2006 3223

Modeling Total-Dose Effects for aLow-Dropout Voltage Regulator

V. Ramachandran, Student Member, IEEE, B. Narasimham, Student Member, IEEE, D. M. Fleetwood, Fellow, IEEE,R. D. Schrimpf, Fellow, IEEE, W. T. Holman, Member, IEEE, A. F. Witulski, Senior Member, IEEE,

R. L. Pease, Senior Member, IEEE, G. W. Dunham, Member, IEEE, J. E. Seiler, andD. G. Platteter, Senior Member, IEEE

AbstractTotal ionizing dose effects in a low-dropout voltageregulator are explained based on experimental data and circuit sim-ulations. Transistor gain degradation is shown to be the dominantcause of the circuit degradation at lower dose rates. In addition, col-lector-to-emitter leakagecurrent in one of the NPN transistors of thebandgap reference part of the circuit is responsible for increasingthe postirradiation output voltage at high dose rates. Parametricchanges in the bandgap, differential amplifier, and output pass tran-sistor circuit blocks are identified that are responsible for variousaspects of the observed circuit degradation. The different annealingcharacteristics of oxide-trap and interface-trap charge are respon-sible for the complex postirradiation recovery of the output voltage.

Index TermsEnhanced low-dose-rate sensitivity, IC radiationresponse, linear bipolar ICs, modeling and simulation, radiation-induced leakage, voltage regulator.

I. INTRODUCTION

ALOW-DROPOUT (LDO) voltage regulator supplies adesired load current at lower dropout voltages comparedto standard regulators. The dropout voltage is the minimumvoltage across the regulator required to maintain normal opera-tion. LDO voltage regulators use an output pass transistor (PNPor NPN) between their input and output terminals, as shown inFig. 1. The magnitude of the collector-to-emitter voltageof the pass transistor determines the dropout voltage. Sincethis value is typically less than or equal to 0.2 V, LDO voltageregulators dissipate low power and dominate battery-poweredapplications, especially in space.

The large output pass transistor conducts most of the load cur-rent, and hence dissipates most of the power in a LDO voltageregulator. The operational amplifier (op amp) is a differentialtransistor pair that acts as an error amplifier. Its role in the circuitis to maintain equal voltages at its input terminals. The outputvoltage is given by:

(1)

Manuscript received July 14, 2006. This work was supported in part by theU.S. Defense Threat Reduction Agency (DTRA) and the U.S. Navy.

V. Ramachandran, B. Narasimham, D. M. Fleetwood, R. D. Schrimpf, andW. T. Holman are with the Department of Electrical Engineering and ComputerScience, Vanderbilt University, Nashville, TN 37235 USA (e-mail: vishwa. [email protected]).

A. F. Witulski is with the Institute for Space and Defense Electronics (ISDE),Vanderbilt University, Nashville, TN 37235 USA (e-mail: [email protected]).

R. L. Pease is with RLP Research, Los Lunas, NM 87031 USA (e-mail:[email protected]).

G. W. Dunham, J. E. Seiler, and D. G. Platteter are with NAVSEA Crane,Crane, IN 47522 USA (e-mail: [email protected]).

Digital Object Identifier 10.1109/TNS.2006.885377

Fig. 1. LDO voltage regulator block schematic diagram.

where is the bandgap reference voltage internally gener-ated in the LDO voltage regulator by the bandgap-reference cir-cuitry. This bandgap circuitry is used to reduce the tempera-ture dependence of the LDO voltage regulator over its operatingrange. The reference voltage is fed to one of the input ter-minals of the operational amplifier. Resistances andrepresent external feedback resistors, while represents theexternal load resistance. This may be either an active (currentsource) or a passive (resistive) load.

Several types of LDO voltage regulators have been shownto exhibit enhanced low-dose-rate sensitivity (ELDRS) [1],[2], and complex circuit responses to total ionizing dose.While operational amplifiers and comparators typically becomeworse after annealing due to gain degradation associated withincreasing interface-trap formation after radiation exposure,some LDO voltage regulators instead recover during annealing[3]. These issues prompted a detailed evaluation of totaldose effects for LDO voltage regulators from a circuit-levelperspective.

In this paper, the circuit elements are identified that are re-sponsible for the irradiation and annealing response of a pos-itive LDO voltage regulator, the MIC29372 from Micrel. TheLDO voltage regulator is investigated under four bias condi-tions during irradiation: bias with load, bias without load, allpins grounded, and the shutdown mode of operation. Here wefocus primarily on the bias without load condition, except fora discussion of the effects of loading on the observed annealingresponses.

Two significant effects are observed that explain the postir-radiation changes in regulator output voltage vs. inputvoltage and load current . First, degradation intransistor gain dominates the circuit response for low-dose-rateand elevated temperature irradiations. Most significantly, this

0018-9499/$20.00 2006 IEEE

3224 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 53, NO. 6, DECEMBER 2006

degradation is manifested as a decrease in the output voltageat high load currents. The decrease in for a givenis caused by an increase in the offset voltage of theoperational amplifier, coupled with gain degradation in theoutput pass transistor. Second, simulations indicate that acollector-emitter (C-E) leakage path in the bandgap circuitrycauses an increase in for a given over the corre-sponding preirradiation values at high dose rates. The presenceof C-E leakage at high dose rates is verified experimentally byisolating the relevant transistor in the bandgap circuitry andmeasuring its electrical characteristics directly. Failure dosesand mechanisms are discussed for each case.

II. EXPERIMENTS AND CIRCUIT EXTRACTION

A. Total-Dose Experiments

Ground-based total-dose experiments were performed insupport of the NASA LWS (Living With a Star) Space Environ-ments Testbed (SET) project [4]. One of the aims of the LWSSpace Environments Testbed project is to analyze and model ra-diation-induced performance degradation of components usedin spacecraft. The pre- and postirradiation data obtained fromthe ground-based experiments were used as calibration pointsfor circuit simulations that we have performed to facilitate anunderstanding of the circuit response. Three types of irradiationexperiments on LDO voltage regulators have been carried outas part of this project. Low dose rate experiments were done at10 mrad(SiO )/s to a total dose of 50 krad(SiO ) at room tem-perature (RT) in a Co-60 room source. High dose rate experi-ments were carried out at 100 rad(SiO )/s to 100 krad(SiO ) atRT, while elevated temperature irradiation (ETI) experimentswere performed at 100 C and 5 rad(SiO )/s to 50 krad(SiO ).The high dose rate and ETI tests were performed in NordionGammacell 220 and Shepherd 484 irradiators, respectively.Samples were irradiated for each of the four bias conditionsmentioned above, for each type of exposure. Annealing mea-surements were made after 16 and 74 days of RT annealingafter the high dose rate irradiations. The same procedures werecarried out after 14 and 35 days for the ETI parts, which wereannealed at 100 C. In all of the annealing experiments, thedevices were biased under the same conditions used during theirradiation experiments.

B. Electrical Characterization

The MIC29372 is a positive LDO voltage regulator withan output voltage that is programmable from to V,with an option to disable the output by providing an externalinput, called the shutdown input [5]. Electrical character-ization during the ground-based tests included conventionalline and load regulation analyses before and after radiationexposure, and through postirradiation anneal. Line regulationwas determined by measuring at different values of ,at a constant mA, while load regulation wasdetermined by measuring at different values, at aconstant V. All of the electrical measurements useda pulsed current of 5 ms duration for at the output, from aminimum value of 5 mA to the maximum value of the desired

. The nominal used for all measurements was V.All of the above tests were carried out at the irradiation andtesting facility at NAVSEA Crane.

C. Circuit ExtractionBased on a die photomicrograph of the MIC29372,

the detailed circuit schematic was extracted, as shown inFig. 2(a) and (b).

In particular, three main circuit blocks are important in under-standing the radiation response and postirradiation behavior ofthe LDO voltage regulator. The first block consists of Brokawbandgap transistors and , with having an emitter area10 times that of transistor . The other two blocks are theoperational amplifier, including the differential transistor pair,

and , and the output pass transistor that occupiesabout 40% of the LDO voltage regulator die. Simulations wereperformed with device dimensions (perimeters for LPNP tran-sistors and areas for NPN transistors) extracted from the diephotos. The forward current gain , the forward base-emitterrecombination current , and the forward base-current idealityfactor were varied in PSPICE models to describe the radi-ation-induced degradation over a sufficient range to verify thatthe estimates used, on the basis of previous knowledge of sim-ilar circuits built in this technology, were sufficient to describethe circuit response. Pre- and postirradiation models were ob-tained for the different bias conditions mentioned above mod-eling approach.

III. CIRCUIT ANALYSIS

A. Modeling Preirradiation CharacteristicsA slight droop in with increasing is typically seen

in voltage regulators. However, as shown in Fig. 3, in the preir-radiation load regulation experiments performed here, in-creased with increasing at currents above 400 mA. Thisphenomenon is typically associated with increased die temper-ature due to the high load current.

In an effort to quantify this effect, load regulation data wereobtained using an active (pulsed current) source and comparedto the results obtained using a passive (load resistor) source.The pulsed current source has an on-time of 5 ms during whichpower is dissipated, while the resistive load dissipates power ona continuous basis. There was negligible difference in the resultsobtained using the different current sources, owing to appro-priate heat-sinking measures incorporated in the circuit sety-up.However, these measures are not able to prevent local heatingdue to the instantaneous high currents that flow in the circuit.Compared to the output pass transistor, which conducts mostof the load current, the other transistors in the circuit conductmuch smaller amounts of current. Hence, it was concluded thatthe increase in was primarily due to heating of the outputpass transistor at higher . Since SPICE does not allow oneto increase the operating temperature of individual transistors,the global temperature settings were varied at every above400 mA to model the experimental data, as shown in Fig. 3. Wefound this procedure to be satisfactory for the purposes of cal-ibrating our circuit models and using them to help understandthe observed results.

RAMACHANDRAN et al.: MODELING TOTAL-DOSE EFFECTS FOR A LOW-DROPOUT VOLTAGE REGULATOR 3225

Fig. 2. (a) Die photo of MIC29372, (b) Extracted circuit schematic of the MIC29372 LDO.

Fig. 3. Preirradiation load regulation characteristics of the LDO.

B. Operation of Critical Transistor Blocks1) Bandgap Circuit: The Brokaw bandgap circuit in the

LDO voltage regulator sets up the as well as , as seen

from (1). The collector currents of transistors and canbe written as

(2)

(3)

Here, and are the saturation currents of and , re-spectively. Since the emitter of is 10 times larger than that of

it follows that is 10 times larger than in magnitude,while has a higher than . Dividing equation Q (3)by (2) and rearranging, we get,

(4)

where is the thermal voltage.From (4), it can be seen that radiation-induced changes in

collector current or directly affect the value of .This in turn affects the bandgap voltage , which depends

on the difference between the base-emitter voltages of and


Fig. 4. Preirradiation and postirradiation experimental line regulation char-acteristics at low dose rate, when the LDO is biased without load duringirradiation.

. Equation (1) describes how this change affects the LDOvoltage regulator output.

2) Op-Amp Circuit: As shown in Fig. 1, the operational am-plifier circuit is designed to maintain equal voltages at its inputterminals. This means that is of the order of a few mVin practical cases, assuming minimal input bias currents. Sincethis circuit is made up of a pair of identical PNP BJTs connectedas differential transistors, their base currents correspond to thebias currents of the operational amplifier. As the base currentsof the transistors increase due to gain degradation during irradi-ation, the bias currents of the operational amplifier, and hence its

, also increase. Thus, the operational amplifier is no longerable to maintain tight feedback, and its performance can dete-riorate significantly, depending upon the amount of gain degra-dation in the transistors. Also, since one of its inputs is ,deterioration in the bandgap circuitry affects its performance.

3) Output Pass Transistor: The output LPNP pass transistorhas about 650 emitters connected in parallel. Because atthe output of the LDO voltage regulator is almost equal to thecollector current of this pass transistor, radiation-induced gaindegradation (and the corresponding increase in base current)affects .

IV. EXPERIMENTAL RESULTSThe ground-based experiments were carried out at three dif-

ferent dose rates with four bias conditions at each dose rate. Wefocus on some of the most significant issues from these exper-iments in this paper. Unless stated otherwise, all results are forirradiations in which the circuit is biased without a load. Resultsfor other irradiation bias conditions are discussed elsewhere [3].

A. Response at Low Dose RateFig. 4 shows the low dose rate (LDR) line regulation response

for the bias without load case for a dose of 50 krad(SiO ).Considering only the postirradiation curve, it is seen

that there is a degradation of regulation (slope change) inover the entire range of . In addition, the postirradiationvalues are lower than corresponding preirradiation values over

Fig. 5. Preirradiation and postirradiation experimental load regulation charac-teristics at low dose rate, when the LDO is biased without load during irradia-tion. Postirradiation output voltages beyond I of 300 mA are not plotted.

the entire range of , and the difference between them is muchmore than those allowed by the manufacturer line regulationspecifications [5]. Thus, there was regulation as well as func-tional failure in the LDO voltage regulator after exposure to50 krad(SiO ). Of the four bias conditions at low dose rate irra-diated to the same dose of 50 krad(SiO ), the grounded caseshowed the largest degradation in both postirradiation andregulation.

The low dose rate load regulation for the bias without loadcase for a dose of 50 krad(SiO ) is shown in Fig. 5. In thiscase too, the postirradiation values are lower than corre-sponding preirradiation values. Also, the LDO voltage regulatorshows functional failure for postirradiation above 300 mAof . The trends in load regulation were similar for allbias conditions irradiated to 50 krad(SiO ) at low dose rates,with the worst-case degradation occurring for the groundedcase, where the LDO voltage regulator stopped regulating above

mA at 40 krad(SiO ). Thus, in the grounded casein load regulation, the LDO voltage regulator failed functionallyas well as parametrically after 50 krad(SiO ) at low dose rate.

B. Elevated Temperature Irradiation

Fig. 6 shows the pre- and postirradiation line regulationcharacteristics for parts irradiated at elevated temperature to50 krad(SiO ) for the bias without load condition.

Similar to the low dose rate line regulation trends above, thepostirradiation exhibits degradation in regulation (slopechange) over the entire range of . Here too, the postirradi-ation values of are lower than corresponding preirradiationvalues over the entire range of , with the difference greaterthan manufacturer line regulation specifications [5]. Anothersimilarity was that the grounded bias case showed the worst

and regulation degradation of all four irradiation bias con-ditions at elevated temperature for a dose of 50 krad(SiO ).

The difference between pre- and postirradiation line reg-ulation values is smaller than that observed in the low dose ratecase for the same bias without load bias condition and dose.This is also true when one compares the difference in pre- and


Fig. 6. Preirradiation and postirradiation experimental line regulation charac-teristics at elevated temperature, when the LDO is biased without load duringirradiation.

Fig. 7. Preirradiation and postirradiation experimental load regulation charac-teristics at elevated temperature, when the LDO is biased without load. Postir-radiation output voltages beyond I of 300 mA are not plotted.

post-irradiation line regulation values for all other bias con-ditions for irradiation to 50 krad(SiO ). While there was bothdegradation of regulation as well as functional failure (at high

) in the LDO voltage regulator after an ETI exposure to50 krad(SiO ), the degradation was lower when compared tothat at low dose rate for the same total dose.

Fig. 7 shows the pre- and postirradiation load regula-tion trends for the same bias condition. Again, the postirradia-tion trend observed is similar to that observed in the cor-responding low dose rate case. The postirradiation valuesare lower than corresponding preirradiation values and the LDOvoltage regulator functionally failed above an of 300 mA.Similar trends are observed in line and load regulation forthe other three bias conditions.

Annealing at 100 C showed almost complete recovery ofline-regulation to preirradiation characteristics for all biasconditions. In load-regulation, there was complete recoverytowards preirradiation characteristics at lower values of ,but at higher values, the recovery was not complete, for

Fig. 8. Preirradiation and postirradiation experimental output line regulationcharacteristics at high dose rate, when the LDO is biased without load duringirradiation.

Fig. 9. Preirradiation and postirradiation experimental output load regulationcharacteristics at high dose rate, when the LDO is biased without load duringirradiation Postirradiation output voltages beyond I of 580 mA are notplotted.

all bias conditions, but was within 2% of corresponding preir-radiation values. In both cases, the slopes recovered completelyafter irradiation, again for all bias conditions.

C. Response at High Dose RateFig. 8 shows the pre- and postirradiation high-dose-rate line

regulation characteristics for the bias without load case, for adose of 100 krad(SiO ). A unique feature observed is that thepostirradiation values of are greater than correspondingpreirradiation values over the entire range of . Consideringonly the postirradiation curve, the change in over theentire range of was within the manufacturers line regulationspecifications [5]. However, when compared to correspondingpreirradiation values, the change in postirradiationexceeds the manufacturers line regulation specifications [5]significantly.

Fig. 9 shows the load regulation response for a dose of100 krad(SiO ). Here, starts out at a value higher thancorresponding preirradiation value and then decreases withincreasing before failing functionally above an


of 580 mA. Both of the above trends in line and loadregulation were also observed for the other three bias con-ditions at high dose rates, irradiated to the same dose, i.e.,100 krad(SiO ), with the grounded bias condition showingthe worst response. Of all the parts irradiated at high dose rateto identical doses of 100 krad(SiO ), those irradiated underbias with load conditions exhibited the greatest amountof recovery in line and load regulation during annealing, towithin 2% of corresponding preirradiation values. Those partsirradiated under the grounded conditions recovered the leastduring annealing.

V. ANALYSIS OF THE RADIATION RESPONSE

The responses of the LDO voltage regulator show that theLDR and ETI irradiation conditions produce similar degrada-tion in postirradiation (values lower than preirradiationvalues) for the same dose of 50 krad(SiO ), with the ETI degra-dation less than that observed for LDR irradiation. The degra-dation following HDR irradiation is different from the LDR andETI results in that the postirradiation actually increasesabove corresponding preirradiation values. This suggests thatthe circuit-level degradation mechanisms are different for HDRirradiation; this is confirmed through simulations and experi-ments on isolated transistors. The results are discussed in detailin the following sections.

A. Gain Degradation

As seen from Figs. 5, 7, and 9, decreased slowly and con-tinually with increasing during postirradiation load-reg-ulation characterization for all dose rates and bias conditions.This was caused by gain degradation in the LDO voltage regu-lator circuit, particularly in the operational amplifier and outputpass transistor circuit blocks.

There are two mechanisms in the operational amplifier cir-cuit block of the LDO voltage regulator that account for theshape of the postirradiation output voltage vs. load current plotshown in Fig. 9. After irradiation, the operational amplifier is nolonger able to maintain tight feedback because of an increase inits since its bias currents increase. The bias current in-creases due to increased base-emitter recombination currents inthe differential pair transistors, and , which constitutethe op amp circuit block. Coupled with the current gain degra-dation in pass transistor , which is driven by the operationalamplifier through the driver transistors, the LDO voltage regu-lator is no longer able to sustain the supply of higher , and

starts to decrease slowly as increases.Fig. 10 shows the simulation results of the degradation

in the operational amplifier circuit block for the high dose ratebias with load case. A large increase in is seen aroundan of 500 mA, which is when the LDO voltage regulatorfunctionally fails for irradiation at this bias condition. Simula-tion results corresponding to Fig. 10 also show that saturationof the collector current of occurs around the same value of

. Along with the operational amplifier degradation, this ac-counts for the functional failure of the LDO voltage regulator.

The fact that the LDO voltage regulator failed completely forLDR irradiation at a dose of 40 krad(SiO ) in the grounded

Fig. 10. Pre- and post-irradiation offset voltages of the op-amp block showhow the droop in the post-irradiation output voltage can be modeled as gaindegradation in the differential amplifiers and the output pass transistor.

case while it did not in the ETI and HDR cases shows greatergain degradation in the LDR case, which is consistent with theELDRS effect reported for this LDO voltage regulator in an ear-lier work [2]. This was also in agreement with the modeling re-sults where the gain for the grounded bias condition at LDRwas lower than the corresponding value at HDR, while the valueat ETI was intermediate.

B. Collector-Emitter LeakageThe increase in during the high dose rate line regulation

experiments was caused by degradation in the Brokaw-bandgapcircuit, since the internally-generated bandgap-voltageinfluences directly, as seen from (1). Previous work [6]identified collector-emitter (C-E) leakage in the VNPNtransistor ( in this case) of an identical Brokaw-bandgap cir-cuit, observed only at high dose rates (up to 200 krad(SiO ) inthat study) to be the main reason for an increase in of an-other voltage regulator. We now show that this C-E leakage af-fects these circuits significantly.

Fig. 11 shows how the C-E leakage phenomenon was mod-eled using a C-E leakage resistor between the collector andemitter terminals of [6]. The bandgap voltage is thesum of and the voltage given by (4), multiplied bya factor corresponding to the voltage division between resis-tances R and R as shown in Fig. 11. Hence, any in-crease in increases the value of . An increase in theleakage current drives the collector current of Q down, therebyincreasing of transistors Q and Q , and thus driving up

An increase in correspondingly increases . Sim-ulations including a C-E leakage path resistor demonstrated thatthis mechanism can account for the increase in observedexperimentally after high dose rate exposure, as shown in Fig. 8.

The C-E leakage mechanism occurs mainly due to the buildupof radiation-induced positive oxide-trap charge over the base of

. This creates an inversion layer between the collector andemitter. While both Q and share a common base with met-allization over it, the emitters and collector of are placed atsome distance from each Fig. 2(a). This contributes to a greaterC-E leakage in the than in the .


Fig. 11. Modeling of C-E leakage by inserting a high resistance between col-lector and emitter terminals of the 10x bandgap transistor Q .

When bias is applied to the LDO voltage regulator, it inducesa larger positive electric field in the p-base in as comparedto the case when all of the LDO pins are grounded during ir-radiation, leading to the formation of more oxide-trap chargein the case when is biased. Hence, has a higher valuepostirradiation in the case where the LDO was biased during ir-radiation as compared to corresponding values in the case whenall of its pins were grounded during irradiation. More descrip-tion on the dependence of changes for different irradiationbiases is given in [3]. The starting value of in the postir-radiation load regulation plot of Fig. 9 is also higher than itscorresponding preirradiation value due to the C-E leakage.

The existence of the C-E leakage phenomenon was verifiedby direct measurement of the pre- and postirradiation Gummelcurves of the transistor in question, i.e., . Bare MIC29372die were procured from a Micrel vendor. Transistor was iso-lated from the rest of the LDO voltage regulator circuit usinga focused ion beam (FIB). Gummel curves were obtained bymicro-probing the isolated and using a HP 4156 parametricanalyzer. The base-emitter voltage was swept from 0 V to0.8 V, while maintaining the collector-emitter voltage at2 V. This procedure was carried out successfully for three diesamples, with similar results obtained in each case.

A fresh MIC29372 LDO die was biased through its input padsto +15 V using a micro-probe, and then irradiated at this biaswith a 10-keV X-ray source at Vanderbilt University. No loadwas connected at the output of the die, corresponding to thebias with no load case in the experiments. The dose rate andtotal doses were 100 rad/s(SiO ) and 200 krad(SiO ), respec-tively. Transistor was then isolated and characterized.

Fig. 12 illustrates the pre- and post-irradiation results. Thepostirradiation collector current of shows a leakage current

Fig. 12. Pre- and post-irradiation and post-anneal characterization of the 10xbandgap transistor Q .

of about 1 A at lower values of , clearly showing the effectof C-E leakage. This leakage current does not exist when theparts are irradiated at lower dose rates or at elevated temperaturebecause of in-situ annealing. The post-irradiation anneal of theC-E leakage is due to the annealing of oxide-trap charge, asdiscussed further in Section VI below.

The postirradiation anneal of base current at RT to close tocorresponding preirradiation values might suggest that the baseof probably goes back to accumulation once the oxide-trapcharges have annealed. This would decrease its value dueto lack of carriers and thus shift the post-anneal curvedownwards.

VI. ANNEALING MECHANISMS

A. Annealing After High Dose Rate IrradiationHigh dose rate line and load regulation characteristics were

measured after 16 and 74 days of room-temperature annealing.Considering the two extreme cases, the bias with load devicehad a die temperature of 100 C because of the flowing atthe output, while the grounded device remained at RT duringanneal.

Fig. 13 shows that the former bias condition had a near-com-plete recovery in load regulation toward corresponding preir-radiation characteristics after both anneals [3]. At higher

, the post-anneal load regulation does not recover to itscorresponding preirradiation values, while it does so completelyat lower values of , which is consistent with annealing ofoxide-trap charge during either lower-rate irradiation (in situ an-nealing) or postirradiation annealing [7].

A significant annealing effect was also seen in the isolated-Qexperiment, as shown in Fig. 12, where the postirradiation C-Eleakage went down to near the preirradiation values after RTannealing. This is also consistent with enhanced annealing ofoxide-trap charge for the loaded devices irradiated at higherdose rates, owing to the accelerated rate at which oxide-trapcharge anneals or is neutralized by compensating electron trap-ping at elevated temperature [8].


Fig. 13. Experimental postirradiation anneal for high dose rate, bias with load,load regulation characteristics. The second anneal curve is not plotted since itoverlaps the first one.

Fig. 14. Experimental postirradiation anneal for high dose rate, grounded loadregulation characteristics. The second anneal curve is not plotted since it over-laps the first one.

In line regulation, there was near-complete recovery towardscorresponding preirradiation characteristics after bothanneals [3]. In contrast, Fig. 14 shows that, for the groundedcase, the load regulation slope does not show recoverythislikely is because there is less annealing of oxide trap charge,since the die never experienced high temperatures (either dueto or external environment) during anneal.

In addition, there was no recovery in either the slope or mag-nitude of towards corresponding preirradiation values afterboth anneals, in line regulation [3]. A probable cause for theparallel downward shift of the load regulation curve could beun-annealed interface traps that remain after all of the oxide-trapcharges anneal out.

B. Annealing After Elevated Temperature IrradiationIn the ETI line and load regulation experiments, annealing

was done at 100 C. Again, annealing measurements were car-ried out after 14 and 35 days. There was no load currentflowing at the output, since the die temperature was already at

Fig. 15. Experimental postirradiation anneal for elevated temperature irradia-tion, bias without load, load regulation characteristics. The second anneal curveis not plotted since it overlaps the first one.

100 C. Fig. 15 shows how all measured parameters recover sig-nificantly [3]. There was complete recovery of both slope andmagnitude, even in line regulation [3].

Both the line and load regulation results are consistentwith the annealing of interface traps at 100 C. Earlier studies[9][11] have found similar results at low electric fields atsimilar temperatures, suggesting that interface-trap annealingis responsible for the observed recovery.

VII. CONCLUSIONSimulations and experimental results show that collector-

emitter leakage of a critical NPN transistor in the Brokawband-gap circuit, associated with enhanced oxide-trap chargeduring high dose rate exposures, is responsible for the manysignificant changes in circuit response for the MIC29372 LDOvoltage regulator under various bias conditions at a high doserate. The collector-emitter leakage phenomenon is thus thelimiting factor in determining the circuit response at highdose rates. For lower dose rates, a steady decrease in outputvoltage with increasing load current is observed. Modeling thisdecrease as gain degradation in the differential pair transistorsforming the operational amplifier circuit block and that ofthe output pass transistor emulates the total ionizing dosefunctional failures observed in the LDO voltage regulator.The circuit response at low dose rates is thus limited by gaindegradation in the LDO circuit.

A near-complete recovery of output and regulation char-acteristics at lower load currents is observed after annealingafter high dose-rate and elevated temperature irradiation. Thisrecovery is associated with the annealing of interface trapsat elevated temperatures associated either with the ambienttemperature during annealing, or with the power dissipationassociated with loaded circuit operation. The behavior of thisLDO voltage regulator contrasts with earlier studies that haveshown that operational amplifiers and comparators becomeworse after annealing due to increased postirradiation interfacetrap formation, thus accounting for the unique response ofthe LDO voltage regulator. The development of calibrated


circuit models greatly facilitates the understanding of the LDOvoltage regulator response over a wide range of experimentalconditions. These models should enable further insight into theresponses of these and similar devices in the space radiationenvironment.

ACKNOWLEDGMENTThe authors would like to thank J. Rowe, A. Kalavagunta,

and A. Karmarkar of Vanderbilt University for assistance withthe characterization experiments. The authors would also like tothank H. Barnaby from Arizona State University for stimulatingtechnical discussions.

REFERENCES[1] R. L. Pease, W. E. Combs, A. Johnston, T. Carriere, C. Poivey, A.

Gach, and S. McClure, A compendium of recent total dose data onbipolar linear microcircuits, Proc. IEEE Radiation Effects Data Work-shop Rec., pp. 2837, 1996.

[2] R. L. Pease, S. McClure, A. H. Johnston, J. Gorelick, T. L. Turflinger,M. Gehlhausen, J. Krieg, T. Carriere, and M. Shaneyfelt, An updatedcompendium of enhanced low dose rate sensitive (ELDRS) bipolarlinear circuits, Proc. IEEE Radiation Effects Data Workshop Rec., pp.127133, 2001.

[3] R. L. Pease, G. Dunham, and J. Seiler, Total dose and dose-rate re-sponse of low-dropout voltage regulators, presented at the NSREC06 Data Workshop, 2006.

[4] NASA website [Online]. Available: http://lws.gsfc.nasa.gov/docu-ments/mowg_june2003/set_report.pdf

[5] MIC29372 data sheet, Micrel Inc.[6] H. J. Barnaby, H. J. Tausch, R. Turfler, P. Cole, P. Baker, and R. L.

Pease, Analysis of bipolar linear circuit response mechanisms for highand low dose rate total dose irradiations, IEEE Trans. Nucl. Sci., vol.43, no. 6, pp. 30403048, Dec. 1996.

[7] R. D. Schrimpf, R. J. Graves, D. M. Schmidt, D. M. Fleetwood, R. L.Pease, W. E. Combs, and M. DeLaus, Hardness assurance issues forlateral PNP bipolar junction transistors, IEEE Trans. Nucl. Sci., vol.42, no. 6, pp. 16411649, Dec. 1995.

[8] D. M. Fleetwood, S. L. Miller, R. A. Reber, P. J. McWhorter, Jr., P.S. Winokur, M. R. Shaneyfelt, and J. R. Schwank, New insights intoradiation-induced oxide-trap charge through thermally-stimulated-cur-rent measurement and analysis, IEEE Trans. Nucl. Sci., vol. 39, no. 6,pp. 21922203, Dec. 1992.

[9] R. L. Pease, D. G. Platteter, G. W. Dunham, J. E. Seiler, H. J. Barnaby,R. D. Schrimpf, M. R. Shaneyfelt, M. C. Maher, and R. N. Nowlin,Characterization of enhanced low dose rate sensitivity (ELDRS) ef-fects using gated lateral PNP transistor structures, IEEE Trans. Nucl.Sci., vol. 51, no. 6, pp. 37733780, Dec. 2004.

[10] M. R. Shaneyfelt, J. R. Schwank, D. M. Fleetwood, R. L. Pease, J. A.Felix, P. E. Dodd, and M. C. Maher, Annealing behavior of bipolarlinear devices with enhanced low-dose-rate sensitivity, IEEE Trans.Nucl. Sci., vol. 51, no. 6, pp. 31723177, Dec. 2004.

[11] D. M. Fleetwood, P. V. Dressendorfer, and D. C. Turpin, A reevalua-tion of worst-case postirradiation response for hardened MOS transis-tors, IEEE Trans. Nucl. Sci., vol. NS-34, no. 6, pp. 11781183, Dec.1987.

Documents

04033819