The radiation hardness assurance facility at INFN-LNS Catania for the irradiation of

electronic components in air Mauro Menichelli, Member, IEEE, Behcet Alpat, Member, IEEE , Andrea Papi, Reno Harboe

Sorensen, Pablo Cirrone, Francesco Ferrera, Pierpaolo Figuera, Paolo Finocchiaro, Marcello Lattauda, Danilo Rifuggiato, Fabrizio Bizzarri, Diego Caraffini, Marco Petasecca, Francesca Renzi, Haluk

Denizli, Ozge Amuktan.

� Abstract— This paper describes the beam flux monitoring system that has been developed at the Superconducting Cyclotron at INFN-LNS (Istituto Nazionale di Fisica Nucleare, Laboratori Nazionali del Sud, Catania, Italy) in order to monitor the beam parameters such as energy, flux, beam profile, for SEE (Single Event Effects) cross-sections determination and DD (Displacement Damage) studies. In order to have an accurate and continuous monitoring of beam parameters we have developed fully automatic dosimetry setup to be used during SEE (with heavy ions) and DD (with protons up to 60 MeV ) tests of electronic devices and systems. The final goal of our activity is to demonstrate that the operation of such a system in air is not detrimental to the accuracy on controlling the beam profile, energy and fluence delivered onto the DUT (Device Under Test) surface, even with non relativistic heavy ions. We have tested our beam monitoring system with the "Reference SEU monitor" developed by ESA/ESTEC The results are discussed here.

Index Terms— dosimetry, heavy ion accelerators, radiation effects, radiation tests.

Manuscript received September 10, 2009 This work was supported in part by INFN and part by MAPRad s.r.l.

Mauro Menichelli is with the Istituto Nazionale di Fisica Nucleare Sezione di Perugia, Via Pascoli s.n.c., 06100 Perugia (Italy) (corresponding author phone: +390755852764; fax: +390755852764; e-mail: mauro.menichelli@pg.infn.it).

Behcet Alpat and Andrea Papi are with with the Istituto Nazionale di Fisica Nucleare Sezione di Perugia, Via Pascoli s.n.c., 06100 Perugia (Italy)

Reno Harboe Sorensen is with ESA/ESTEC, Keplerlaan 1 Postbus 299, 2200 AG Noordwijk The Netherlands

Pablo Cirrone, Francesco Ferrera, Pierpaolo Figuera, Paolo Finocchiaro, Marcello Lattauda and Danilo Rifuggiato, are with the Laboratori Nazionali del Sud dell’INFN, Via S.Sofia 62, 95125, Catania, (Italy).

Fabrizio Bizzarri, Diego Caraffini, Marco Petasecca, Francesca Renzi are with MAPRAD S.r.l., Via C.Colombo 19/I, 06127, Perugia, (Italy).

Haluk Denizli is with Abant Izzet Baysal University, Department of Physics, 14280, Bolu, (Turkey)

Ozge Amuktan is with Ortadogu Teknik Universitesi, Physics Department, 06531 Ankara, (Turkey).

I. INTRODUCTION HE LNS Superconducting Cyclotron (CS), is a compact, strong focusing three-sector machine. The pole radius is 90 cm and the magnetic field at the center ranges from 2.2

to 4.8 T. This is obtained with superconducting Nb-Ti coils cooled down to 4.2 K in an LHe bath. The expected maximum energies of the machine are of 20 MeV/amu for the heaviest ions, like 238U38+, and 100 MeV/n for fully stripped light ions. The bending limit is Kb=800 and the focusing limit is Kf=200. The relatively high energy of the beams (for this study 20 MeV/nucleon) allows the irradiation of components in air which is also used as a degrader. The selection of the ion species we use in SEE studies is done by taking into account the easiness of beam changing operation and at the same time the necessity to cover uniformly a LET interval as large as possible. Hence, selected for operation four noble gases beams (20Ne, 40Ar, 84Kr, 129Xe) all with 20 MeV/n energy. A careful evaluation of energy loss in air and of the energy spread at DUT surface is carried out through a full Monte Carlo simulation of the test setup and comparing the results with data. The DD studies are performed using protons of 60 MeV in kinetic energy.

II. BEAM FLUX MEASUREMENT

The measurement of beam flux and uniformity is one of the ingredients for the calculation of SEE cross-section. According to the ESA standard ESCC 25100 these measurements should be done with an accuracy of �10%.

At LNS we perform extraction in air of ion beams and our flux dosimetry setup resides in air; additionally we use air as a degrader in order to adjust LET values. A picture of the dosimetry measurement setup is shown in Fig. 1.

T

446

978-1-4577-0493-2/09/$26.00 ©2009 IEEE

Fig.1 Picture of beam flux measuring setup. Thin scintillator is mounted on two arms attached to the beam flange. The supporting frame holding a DUT is also shown.

The setup consists in a thin scintillator counter (NE102A of

50, 100 or 2000 μm thickness) read by a photomultiplier just after (5 cm distance, this distance is referred as Air1) the beam exit into air. The scintillator is inserted into a metallic box with variable size beam hole to allow the adjustment of the beam size. The DUT is placed onto a supporting frame (DUT holder) at a variable distance from the scintillator (this distance may vary from 5 to 30 cm and is referred as Air2). The DUT holder is capable to move in XYZ directions and to rotate around a vertical axe. On the same structure close to DUT holder a double sided, 1.5 mm thick microstrip detector (500 μm readout pitch and 3x3 cm2 active area) is mounted for energy, fluence and beam profile measurements. Because of its thickness the silicon detector absorbs the entire energy of all ions we use. The stopping range in silicon is, for all ions we use, at least 30 μm, in accordance to the minimum penetration depth required by the ESA standard ESCC-25100 [1]. The dosimetry and SEE testing setup includes two additional important features. A laser device is used to measure the distances in Z (beam) direction (namely Air2) with 200 μm position accuracy. Such level of accuracy in measurement of the relative distances (i.e. silicon detector surface to delidded DUT surface, DUT surface to beam exit in air etc.) is important to minimize the systematic error on LET value in silicon. The other feature of the setup is a custom module (SELDP, Single Event Latchup Protector and Detector [2]) built specifically to monitor the current drawn on power line of a DUT. The SELDP cuts the power supply to DUT for an adjustable time duration whenever the current drawn by DUT exceeds a preset current limit. This both protects DUT from burnout because of Single Event Latchup (SEL) effect and registers the number of SEL occurred by incrementing a counter. The measurement strategy is the following. Prior to beam period, according to the requested LET interval, a detailed

simulation of beam parameters is done with Geant4 and a list containing the beam species, distances and tilting angles are prepared. At the beam site, for each energy point present in this list, before starting the DUT irradiation, we make a calibration run of the beam and during irradiation we measure the beam parameters. During calibration run we place under the beam the scintillation counter and the silicon detector. The scintillator acts as a counter of ions crossing the detector. Careful adjustment of the counting threshold is performed in order to have maximum efficiency.

From the charge collected in the detector we can measure the energy of each ion hitting this detector. The beam uniformity, profile, time stamp for each event are measured as well. Typical beam profiles from online monitoring and from reconstructed data are shown in fig.2.

During calibration runs both detectors are on beam and we perform the following measurements:

� Energy of the ion by measurement of charge

collection of each ion hitting the silicon strip detector. This measurement is important to determine the uncertainties on LET measurement and the degree of fragmentation of the beam.

� Beam profile: we can measure the flux using information on the number of counts per strip per unit time and the position of each hit. From this measurement we can determine the flux and the fluence during the calibration run

� Comparative measurement of the counting rate in the scintillation counter and in the silicon detector. We perform this measurement by recording the counting rate of the scintillator counter and in the silicon detector in order to have a correlation between the flux on the silicon sensor and the counting rate of the thin scintillator that acts as flux-meter when the DUT will be on the beam in place of the Silicon strip detector.

During the actual irradiation run on the various DUTs we

use the results of beam profile and comparative (scintillator-silicon) measurements in order to calculate the flux and the fluence and the energy measurement to have an estimation of the errors on LET.

The energy spread provided by the simulation and silicon detector data are in good agreement and the spread is well below the 10% of the target energy value indicating the good beam purity and full control of the systems parameters (see Table I).

After each calibration run we started irradiation. During irradiation we use the counts on the calibrated scintillation counter in order to determine the fluence. The scintillator counter gives average fluence of those ions passing through its active are hence the fluence value given by scintillator is normalized and corrected by using the beam profile available at DUT surface .

447

Fig.2 On Top: The online monitoring of beam parameters. The beam profile is shown in detail. Each channel corresponds to 0.5 mm (strip pitch of double sided silicon detector). On Bottom: The beam profile reconstructed from data. Schematic of beam flux measuring setup.

TABLE 1: THE ION AND CORRESPONDING LET VALUES AND RELATIVE ERRORS.

Ion/LET (MeV/mg/cm2) Error on LET (MeV/mg/cm2)

Neon-20/3.7 0.1 Argon-40/13.13 0.2 Krypton-84/30.6 0.7 Xenon-129/52.9 0.8

III. SYSTEMATIC ERROR EVALUATION FOR SEE CROSS-SECTION MEASUREMENTS

A careful evaluation of systematic errors is necessary to estimate overall error introduced on LET and on fluence values. Definition of LET and cross section is given in Eq. 1 where � is silicon density.

;)(;1)/( 22

��

� CosFluenceN

cmdXdEmgcmMeVLET SEE

���

(1)

Systematic errors on LET values are; � Distance (air thickness) measurements. This is done with a 200 μm accuracy laser system only once during the initial calibration phase. All other positions are relative to that point with submicron precision 4-D stage (X,Y,Z, Theta)

� Fragmentation in air. It is negligible (i.e. <10-4 for 20 MeV/n 40Ar after 15 cm of air) according to Geant4 simulation performed using “binary light ion cascade” and “Wilson abrasion” models. Fragmentation is also measured during calibration. The amount of fragmentation measured is in agreement with monte-carlo predictions.

� The determination of energy deposited and range in DUT by Geant4 simulation; the energy distribution from Monte Carlo has less than 3% error at FWHM (see Table I), and there is a good correlation with the charge measured by the silicon detector, which provides the possibility to convert the charge values to deposited energy. The correlation between charge collected in silicon and simulated energy for different air thicknesses for 40Ar is given in Fig.3.

Figure 3: Correlation curves between charge deposited in silicon(ADC counts) and simulated energy release (MeV). The two data sets and curves refer to different ways to estimate energy from Monte Carlo distribution: Black triangles use the most robable value taken from energy histogram, red upside-down triangles use the mean from a gaussian fit.

� Positioning of beam spot to the center of DUT and correction of non uniformity of beam over the beam spot. Fig.4 shows a typical beam spot and DUT dimensions. This error includes a 180 μm contribution from the silicon detector spatial resolution, as well as a ~300 μm one from the accuracy of the DUT mounting on the holder frame; the two contributions are summed in quadrature since they are independent.

The detailed analysis of errors have shown that the overall error on determination, both of fluence and of LET, is less than 4% which is well below the upper limit (10%) required by ESCC-22500. The I lists the overall systematic errors on LET values for different ion species at 10 cm of Air2.

448

Figure 4: Beam spot is centred over the DUT surface. Form this profile one can extract the real fluence distribution over DUT surface. The beam spot positioning accuracy is at most 400 μm.

IV. TEST PROCEDURE FOR DD MEASUREMENTS The displacement damage are non ionizing effects caused

mostly by protons and neutrons on silicon lattice structure by dislodging or displacing the atoms from their lattice sites. The DD studies at LNS are carried out using 60 MeV protons. Here the testing procedure is simpler than that of SEE since tests are performed at a fixed Air1 and Air2 distances. The DUTs under proton irradiation are usually unbiased (unless otherwise is requested). The critical parameters of DUT are tested before and after irradiations . In the case of combined Total Ionizing Dose and displacement damage measurements (with DUT biased) the corresponding dose value are given by:

Dose (rad(Si)) = LET *Fluence*1.610-5 (2) In this kind of test, the main problem is that test equipment materials, do activate after prolonged exposition to the proton beam. The supporting frame was designed so that only the DUT and a minimal amount of test board is directly under the beam, in order to reduce activations as much as possible. To avoid putting the silicon detector under the beam for long times, it is possible to use it to monitor the intensity of the beam tails after an initial intercalibration phase with a gas chamber put in front of the beam outlet; the chamber can then be removed during actual DUT irradiation. The electrical connections of the DUT’s pins to power supply, bias line and measurement apparatus must be designed so that it is possible to (dis)connect them remotely to avoid getting close to a possibly activated DUT.

V. COMPARISON WITH A BENCHMARK SETUP

In order to measure the consistency of our measurements with other measurements done in other beams we used a benchmark setup developed by ESA/ESTEC [3] named SEU monitor. This setup measures the number of SEU events on a Atmel AT60142F-DC1 SRAM which is a component that has a well known SEU cross section versus LET curve (measured in several facilities) and is Latchup immune. The software of this test setup once the data on LET and fluence are inserted calculates immediately the cross section versus LET and display the result on a graph where previously measured values are displayed. Fig.5 shows our results compared with these values measured in other facilities. As shown our results are in good agreement with the previous measurements.

Fig.5 Cross-section versus LET of Atmel AT60142F-DC1SRAM measured with our setup (blue dots) and with several other beam facilities (red symbols).

VI. CONCLUSIONS In Europe there is a limited number of accelerator sites

whose delivered ion beams fulfil the requirements of ESCC standards for SEE testing. We have developed a fully automatic dosimetry system to demonstrate the validity of beam characteristics of LNS sites for SEE and DD tests as well as to accurately measure the parameters relevant to perform detailed SEE and DD studies. With energies available at LNS and with four selected gaseous beams it is possible to perform SEE studies from few up to 110 MeV/mg/cm2 of LET. The beam changing time is relatively short (few hours) and beam size and fluence are stable in time. Furthermore, the presence of air gives possibility to reach “fine-tuned” LET values by adjusting the air thickness accordingly. Last but not least, operating the setup in air has its obvious advantages of reducing setup time and complexity.

449

VII. ACKOLEDGEMENTS

The Scientific and Technical Research Council of Turkey (TUBITAK) is acknowledged for the granting of Haluk Denizli postdoctoral study in the framework of TUBITAK-BIDEB 2219 grant.

REFERENCES [1] SINGLE EVENT EFFECTS TEST METHOD AND GUIDELINES

ESCC Basic specification 25100. Issue 1 2002 edited by ESA available at: https://escies.org/

[2] SELDP: An instrument for the automatic detection and protection from SEL during beam tests. Proceedings of RADECS Workshop 2002 pp.4

[3] R. Harboe-Sorensen, F.-X. Guerre and A. Roseng Design, Testing and Calibration of a "Reference SEU Monitor" System. Proceedings of RADECS

450