0018-9499 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TNS.2018.2825644, IEEE Transactions on Nuclear Science Abstract— Engineers usually test multiple printed circuit boards (PCB’s) and integrated circuits (ICs) in the neutron beam to improve the error statistics. The attenuation of an atmospheric-like neutron spectrum by materials commonly used in printed circuit boards is evaluated experimentally and with simulations of the neutron transport. Measurements were performed during neutron irradiation experiments of multiple boards using a beam monitor to correlate simulations with experimental results. Index Terms— Neutron radiation effects, Nuclear measurements, Particle beams I. INTRODUCTION ow critical charge for memory cells fabricated at the advanced technology nodes has necessitated neutron tests for estimating soft error vulnerability of electronic systems operating in the terrestrial environment. The limited worldwide availability and cost of atmospheric neutron beam- time to yield statistically significant error counts forces test engineers to test multiple ICs/circuits simultaneously in the neutron beam. This is usually achieved by placing multiple Printed Circuit Boards (PCB) one after the other in the neutron beam with the assumption that attenuation in neutron beam flux due to PCB is negligible. In this work Monte-Carlo simulations of neutron transport through PCB materials show attenuation of the neutron flux and modulation of the neutron spectra. These factors, attenuation and modulation of the beam, must be taken into consideration when calculating Failure-in-Time rates for electronic circuits. In particular, this work focuses on the effect of the most common materials in electronic boards, the fiberglass of the PCB (FR-4) and silicon. This attenuation This work has been submitted to IEEE TNS on 28 September 2017. Carlo Cazzaniga and Christopher D. Frost are with ISIS Facility, STFC, Rutherford Appleton Laboratory, Didcot OX11 0QX, UK. (e-mail: carlo.cazzaniga@stfc.ac.uk; christopher.frost@stfc.ac.uk) Bharat Bhuva is with Department of EECS, Vanderbilt University, Nashville, TN 37235 (e-mail: bharat.bhuva@Vanderbilt.Edu) Marta Bagatin and Simone Gerardin are with DEI - Padova University, via Gradenigo 6B 35131 Padova, Italy (e-mail: marta.bagatin@dei.unipd.it; simone.gerardin@dei.unipd.it) Nicolò Marchese is with University of Calabria, Via P. Bucci, Arcavacata di Rende (CS), Italy (e-mail: n.marchese@dimes.unical.it) becomes obvious in a reduction of the Single Event Error Rate (SER). This effect was measured using an SRAM-based beam monitor for quantifying the reduction in SER for a test IC. Measurements have been performed at ChipIr, a beamline of the ISIS spallation neutron source at the Rutherford Appleton Laboratory (UK). It is dedicated to the irradiation of microelectronic system with an atmospheric-like neutron spectrum. ChipIr is specifically tailored to the study of single event effects and its design is therefore optimized to extract a neutron spectrum as similar as possible to the atmospheric one with intensity increased by a factor up to 10 9 depending on configuration [1,2]. II. SIMULATIONS Simulations have been carried out to evaluate the attenuation of neutrons of different energies by fiberglass (FR- 4) and silicon. For simulations, the atmospheric neutron spectrum (JESD89A) [3] is used as input. The choice of using this spectrum allows for a more general result that can be applied on ChipIr and on other atmospheric–like neutron facilities [4-6]. Differences between JESD89A spectrum and spectrum at different facilities are relatively small and are not the object of this paper. However, the agreement of experimental results and simulations presented in this work will already be a demonstration that ChipIr spectrum is in fact representative of the atmospheric one. Monte Carlo simulations of the neutron transport are performed with MCNPX [7]. The source is a collimated square beam of 7 cm × 7 cm. The result is scored in an area of 2 cm × 2 cm after a lump of material of a defined thickness. With MCNPX it is possible to determine the flux energy distribution (the spectrum). The simulations were run for different thicknesses of the selected materials: fiberglass FR-4 (density 2.64 g/cm 3 ), Silicon (density 2.33 g/cm 3 ), Aluminium (density 2.7 g/cm 3 ), Copper (density 8.96 g/cm 3 ) and Polyethylene (density 0.96 g/cm 3 ) to estimate the neutron fluxes and spectra. Though this is not a complete list of materials that will be present in the beam line (IC socket material, metal lines on the PCB, cables, etc.), these simulations will provide an estimate of the attenuation in flux and changes in beam spectrum. Figure 1 shows on the top the neutron spectrum φ(E) after 0 mm, 19.2 mm and 38.4 mm of FR-4 material. It is clear that the intensity of the neutron spectrum is reduced by the Atmospheric-like Neutron Attenuation during Accelerated Neutron Testing with Multiple Printed Circuit Boards Carlo Cazzaniga, Bharat Bhuva, Marta Bagatin, Simone Gerardin, Nicolò Marchese and Christopher D. Frost L