Soft X-rays spectroscopy with a commercial CMOS image sensor at room temperature

https://doi.org/10.1016/j.radphyschem.2019.108354Get rights and content

Abstract

Besides their application in point and shoot cameras, webcams, and cell phones, it has been shown that CMOS image sensors (CIS) can be used for dosimetry, X-ray and neutron imaging applications. In this work we will discuss the application of an ON Semiconductor MT9M001 CIS, in low energy X-ray spectroscopy. The device is a monochromatic front-side illuminated sensor, very popular in consumer electronics. In this work we introduce the configuration selected for the mentioned sensor, the image processing techniques and event selection criteria, implemented in order to measure the X-ray energy in the range from 1 to 10 keV. Several fluorescence lines of different samples have been resolved, and for first time the line resolution have been measured and analyzed. We achieved a FWHM of 232 eV at 6.4 keV, and we concluded that incomplete charge collection (ICC) of the charge produced by the X-ray contributes to the resolution, being this effect more important at higher X-ray energies. The results analyzed in this work indicate that the mentioned CIS are specially suitable for X-ray applications in which energy and spatial resolutions are simultaneously required.

Introduction

Since their discovery in 1895 by Roentgen, X-rays have found a countless number of everyday applications in medicine, industry, and have been a powerful tool in material science and biology. Common technologies for their detection and energy measurement, are silicon drift detectors (SDD) and ad hoc detectors. In the field of big area X-ray radiography, due to their low cost compared to thousand of dollars flat panels, radiographic films still been the preferred option.

Nowadays CMOS image sensors (CIS) can be found in several consumer electronics products, like picture cameras and cell phones. They are mass-produced, have a low cost, and thanks to the integration in the same chip of the analog and digital reading electronics, can be easily integrated in embedded systems. In the last two decades, they have been used as charged particles and gamma photons detectors. For example in high resolution X-ray transmission imaging of small objects (Lane, 2012) (Hoidn and Seidler, 2015) (Castoldi et al., 2015) (Alcalde Bessia et al., 2018). Also, previous works have demonstrated the potential of these sensors for X-ray spectroscopy without analyzing the resolution capabilities of this devices (Lane, 2012) (Hoidn and Seidler, 2015) (Holden et al., 2018) (Nachtrab et al., 2009), which is the main objective of this work.

In Section 2 we present a brief overview of CIS physics, from the interaction of the X-rays with the silicon to the collection of charge in the pixels, which is necessary to interpret the experimental results introduced in this work. In Section 3 we will describe the employed sensor and the image processing technique followed to detect the X-ray lines emitted from the fluorescence of different materials. The response of the CIS at different X-ray energies, as well as the experimentally obtained energy resolution ere shown in Section 4. Finally, in Section 5, the read-out noise, energy resolution, dynamic range and efficiency of the detector will be analyzed.

Section snippets

X-ray spectroscopy with CIS

The most probable interaction mechanism of soft X-rays with silicon is the photoelectric effect, in which one atomic electron of the silicon absorbs the total energy of the incoming photon. The K-shell electrons, because of their proximity to the Silicon nucleus, have the highest probability of interacting with the photons (Grupen and Shwartz, 2008). When this interaction occurs, the K-shell electron is ejected from the Si atom with a kinetic energy ExEb, where Ex is the energy of the X-ray

Experimental method

An MT9M001 CIS manufactured by On Semiconductors was chosen for this work (ONSemi). It is a front-illuminated monochrome image sensor with 1280×1024 pixels of 5.2×5.2 μm2 each one, covering a total area of 6.66×5.32mm2. A previous work in our group estimated the epitaxial layer thickness in 2μm (Bessia et al., 2018), together with its pixel size, a good spectral resolution is expected. In this work, the pixel voltage was read using with the internal 10-bit analog to digital converter (ADC). A

Measurement of the X-ray spectrum

To extract the cluster of pixel that correspond to an X-ray interaction, we applied the same method used in CCD particle detection experiments (Chavarria et al., 2015). First, seeds of single or neighboring pixels that have a value more than four times the image noise, this is 4 σ, are identified in the acquired frame. Then, all the pixels around the event seed with more than 3σ are added to form a cluster of pixels that compose the event. The charge of the event is obtained adding all the

Gain and noise

Correlating the energy of the X-ray and the obtained μ parameter, a calibration factor of 90.3 ADU/keV was obtained. The resulting fit is shown in Fig. 12, where a high linearity in the measurement of the X-ray energy is observed.

Taking in account that 3.64 eV are required in average to generate an electron-hole pair in silicon, the 90.3 ADU/keV conversion gain imply that each ADU corresponds to 3.04 electrons in our configuration with an analog gain of ×15. Considering the pixel saturation of

Conclusions

We have demonstrated the capability of a commercial and low cost CIS to resolve several X-rays peaks in the range from 1.7–9 keV, without cooling the detector. For first time the energy resolution of this devices was measured and analyzed for soft X-rays. The contribution of the read-out noise, Fano noise and ICC, to the resolution was studied, being the ICC the dominant factor at higher energies. A resolution comparable to a professional SDD was obtained. CIS are therefore a valid option as

Acknowledgment

The authors would like to thank to Ignacio Artola for his help with the experimental setup, to Jorge Pelegrina for the use of X-ray sources, and to Sergio Suarez for provide the samples.This work was supported by the ANPCyT under Project PICT-2014-1966, Project PICT-2015-1644 and Project PICT-2015-2128, by UNCuyo under Grant C018, and in part by CONICET under Project PIP-2011-0552 and Project PIP-2013-0077.

References (25)

  • A. Castoldi et al.

    Commercial cmos image sensors as x-ray imagers and particle beam monitors

    J. Instrum.

    (2015)
  • M. Esposito et al.

    Geant4-based simulations of charge collection in cmos active pixel sensors

    J. Instrum.

    (2017)
  • Cited by (24)

    • Performance of a focal plane detector for soft X-ray imaging spectroscopy based on back-illuminated sCMOS

      2022, Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment
    • Evaluation of a Commercial Off The Shelf CMOS Image Sensor for X-ray spectroscopy up to 24.9 keV

      2020, Radiation Physics and Chemistry
      Citation Excerpt :

      In the case of the Kα line of Zr, it is possible to observe that the peak obtained with one-pixel events is narrower than the acquired considering all the registered events. Therefore, it can be seen that it is possible to achieve a better energy resolution taking into account only the one-pixel events, which is consistent with the results presented in (Haro et al., 2019). Fig. 5 shows the calibration curve of the sensor, the values of energy expressed in ADU were obtained fitting all of the measured peaks with Gaussian functions and estimating their mean values.

    View all citing articles on Scopus

    Fully documented templates are available in the elsarticle package on CTAN.

    View full text