Theoretical design of an XRF system for environmental measurements of Mercury in fiber banks

(1)

Theoretical design of an XRF system for

environmental measurements of Mercury in

fiber banks

Runpeng Yu

Final Project

Main field of study: Electronics.

Credits: 30

Semester/Year: VT2020

Supervisors: Börje Norlin, Siwen An

Examiner: Göran Thungström

Course code/: EL038A

(2)

This thesis demonstrates the advantages of using the Energy-dispersive X-ray fluorescence (ED-XRF) system to quantify the mercury content in fiber banks at first. The Monte Carlo N-Particle (MCNP) code was then used to simulate the XRF system model with suitable parameters such as the input X-ray energy level, the detector material, and the environmental factor (water depth). The SNR results of the mercury spectrum when applying different parameters were obtained. Then, the limit of detection (LOD) and limit of quantification (LOQ) based on the SNR approach are considered. Finally, system parameters were determined in order to obtain more accurate qualitative and quantitative analysis results for future environmental measurements.

(3)

4

Acknowledgements

First of all, thanks to my supervisor Börje Norlin, who helped me to clarify and explore the research direction in the thesis, appreciate his comments on the thesis.

Secondly, thanks to Siwen An for introducing MCNP software, guiding me in the lab and sorting out my ideas. I could not finish this thesis without her advice during this period.

Besides, thanks to my friend Xi Ma for his encouragement and help. In addition, thanks to my friend Huaiyang Mu for her critical questions in presentation and thanks to Xiaolei Xia for responding to my first questions about the thesis.

(4)

5

Terminology

1.1 Acronyms

MCNP Monte Carlo N-Particle Transport Code

ICP-MS Inductively coupled plasma mass spectrometry

XRF X-ray fluorescence

ED-XRF Energy-dispersive X-ray fluorescence WD-XRF Wavelength-dispersive X-ray fluorescence

SNR Signal-Noise-Ratio

LOD Limit of Detection

LOQ Limit of Quantification

1.2 Mathematical notation

Symbol Description

𝜆 Input X-ray wavelength ρ Density of the sample

d Thickness of the sample

𝐸_𝑠𝑐 scattered photon energy 𝑚_𝑒𝑐2 electron mass energy 𝑃_{𝑠𝑖𝑔𝑛𝑎𝑙} Power of signal

(7)

8

1 Introduction

1.1 Background and problem motivation

The latest geological survey[1] in Sweden found that the water content in the Västernorrlands, especially near Sundsvall Bay, is higher than elsewhere. The reason was not only the discharge of the wood processing industry in the past, but also the deposition of a layer of wood fiber rich in mercury, dioxins, PAH and other heavy metals on the seabed, and the existence of more than 20,000 barrels of discarded mercury waste in the nearby seabed.

In the 1960s, paper mills built near the Sundsvall Bay discharged process wastewater into nearby waters, with fiber banks accumulating at the discharge point and its surrounding areas. According to known data, Fiber bank knows that its fiber properties adsorb heavy metal elements such as mercury. Besides, the mercury will eventually enter the human body through biomagnification, causing human death eventually. Therefore, it is significant to quantitative analyse the heavy metal elements contained in the fiber bank sample, especially the mercury content.

In order to quantitatively analyse the amount of mercury in the fiber bank sample, XRF is a feasible method. In the actual measurement of the outdoor environment, in order for researchers to easily and quickly confirm the level of mercury in the fiber bank samples salvaged on-site, it is necessary to design a portable XRF system. In order to solve this problem, using the simulation method to design an XRF system in advance can determine limitation. At present, MCNP is very popular used in the XRF system design and evaluation for its flexibility, general characteristics, and powerful functions. Therefore, using MCNP to simulate the XRF system is an excellent solution.

1.2 Overall aim

The overall aim of this project is to use the MCNP6 code to simulate the X-ray fluorescence detection system. Focus the parameters of essential components, such as the level of the X-ray source energy, the material of the detector, and the environment outside the fiber bank sample. Analyse the simulation result. Conclude the design of an X-ray fluorescence detection system for

(8)

9

1.3 Scope

This thesis focuses on the establishment of an X-ray fluorescence detection system by MCNP6 code. Since no other compounds were added to the sample except mercury, the matrix effect caused by other elements was not

considered.

1.4 Concrete and verifiable goals

This thesis has an objective to respond to the following questions: P1: Which chemical composition quantitative analysis method is more suitable for measuring the heavy metal element mercury in fiber banks? P2: For designing an XRF detection system to measure mercury in fiber banks, what parameters have been selected for better results?

In this thesis, the main target of discussion is the input X-ray energy level, environmental factors (different water depth), and detector materials in XRF detection systems.

1.5 Outline

In the rest of this thesis: Chapter 2 introduces the basic theories used in XRF systems. Chapter 3 compares three chemical composition quantitative analysis methods, explains why the ED-XRF method was used to measure mercury, and finally gives a basic model using the ED-XRF method. Chapter 4 gives different designs for different parts of the system based on the basic model given earlier. Chapter 5 presents the results of different simulation models and analyses their output results. Chapter 6 gives the theoretically better setting parameters of the XRF system for quantification of mercury element in fiber bank and gives suggestions for controlling environmental factors. In the end, future work section, methods for improving the model and areas for further exploration are proposed.

(9)

10

2 Theory

2.1 X-ray tube

Figure 2.1 General set-up of a X-ray tube. [2]

An X-ray tube is an energy converter that converts electrical energy into radiant energy. As shown in the figure above, an X-ray tube is a relatively simple electronic device that usually contains two main elements: a cathode and an anode. When current flows from the cathode to the anode, electrons with high kinetic energy will impact the anode material and cause energy loss, thereby generating X-rays.

The design and construction of the X-ray tube can maximize the generation of X-rays and dissipate heat as quickly as possible. It receives electrical energy and converts it into two other forms: x-radiation and heat, where heat is a byproduct.

(10)

11

In MCNP simulation software, the simplest modelling of an X-ray source is to directly set the energy value of the photons to a specific value.

2.2 Interaction of X-rays with matter

Figure 2.2 Three main interactions of X-rays with matter[3]

There are three main interactions when X-rays contact matter: Fluorescence, Compton scatter and Rayleigh scatter like figure 2.2.

(11)

12

2.2.1 X-Ray fluorescence

Figure 2.2.1.1 The principle of the X-ray fluorescence radiation[4]

The principle of the X-ray fluorescence radiation is shown simplified in the figure above.

Atoms contain atomic nucleus and electrons. When X-ray strikes the

electrons, the electrons in the inner and outer layers may be released, causing an unstable state. The energy generated by the complementary energy step is fluorescence. If the electrons are released, it is possible to be supplemented by the electrons of the L layer, the M layer, or the N layer. The fluorescent light supplemented by the L layer is 𝐾α , and the M layer is supplemented by 𝐾β . If the L layer electrons are released, the M-layer electrons are referred to as 𝐿α , and if N-layers are referred to as 𝐿β .

(12)

13

Figure 2.2.1.2 Major lines and their transitions[3]

In this process, the falling electron emits a characteristic fluorescent X-ray photon with an energy which is equal to the difference between the binding energies of those two shells. According to characteristic fluorescent radiation of each element, it is easy to identify the elemental composition of measured material.

Each fluorescence radiation energy of mercury was shown in appendix B.

2.2.2 Rayleigh and Compton scatter

Figure 2.2.2.1 Compton scatter[3]

(13)

14

in which the electrons are hit. This type of scattering is called Compton or incoherent scattering.

The energy of the scattered photon can be calculated as:

𝐸

_𝑠𝑐

=

𝐸0

1+( 𝐸0

𝑚𝑒𝑐2)(1−𝑐𝑜𝑠𝜃)

[2.1]

where 𝐸0 is the original photon energy, 𝐸𝑠𝑐 is the scattered photon energy, 𝑚𝑒𝑐2 is the electron mass energy, and 𝜃is the scattering angle.

Figure 2.2.2.2 Rayleigh scatter[3]

Another phenomenon is the Rayleigh scatter. This phenomenon happens when photons collide with strongly bound electrons. That means the

electrons start oscillating at the frequency of the incoming radiation without bounced off as Compton scatter. The electrons would emit radiation energy as the incoming radiation energy when it starts oscillating at the same frequency of the incoming radiation. Usually, it happens like the atom reflects the

incoming radiation. This type of scattering is called Rayleigh or coherent scatter.

2.3 Absorption edge

(14)

15

Edge keV Edge keV

K 83.1023 M1 3.4811

L-I 14.8393 M2 3.7817

L-II 14.2087 M3 4.3548

L-III 12.2839 M4 5.1987

Table 2.3 absorption edges of mercury [5].

From the table we can see, the absorption edge of mercury’s K line is

83.1023 keV, which means the energy source should be larger than this value if we want to detect mercury’s characteristic radiation spectrum of K line.

2.4 Escape Peaks

Escape peaks are artifact peaks due to the absorption of some of the energy of a photon by atoms in the detector [6].

In our case that using CdTe detector,

𝐸_{𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑} = 𝐸𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡– 𝐸𝐶𝑑 where 𝐸𝐶𝑑= 23.17 keV or 𝐸𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑 = 𝐸𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡– 𝐸𝑇𝑒 where 𝐸𝑇𝑒= 27.47 keV

2.5 SNR

Signal-to-noise ratio is defined as the ratio of the power of a signal (meaningful information) to the power of

background noise (unwanted signal) [7]: SNR = 𝑷𝒔𝒊𝒈𝒏𝒂𝒍_𝑷

𝒏𝒐𝒊𝒔𝒆 [2.2]

In this case, the signal will be the mercury fluorescence peaks intensity, while the background noise will be produced by water scattering energy.

If we want to calculate the SNR, we should get background noise, amount signal, and the signal after subtraction of background noise.

2.6 LOD & LOQ

In individual analytical procedure, the LOD (Limit of Detection) is the lowest amount of analyte in a sample which can be reliably detected. the LOQ (Limit of Quantification) is the lowest amount of analyte in a sample which can be reliably quantified with suitable precision and accuracy.

(15)

16

2.7 Transmittance

The transmittance, in this case, refers to energy loss by absorption and scattering from water,

T = 𝑃x

𝑃0 [2.3]

(16)

17

3 Methodology

In order to achieve a quantitative analysis of chemical components, the use of modern instrumental analysis methods has become more popular because of its sensitivity and simplicity. Regarding the measurement of heavy metal elements represented by mercury in sediments, common analysis methods include ICP-MS (Inductively coupled plasma mass spectrometry) and XRF (X-Ray Fluorescence). XRF is divided into two methods: WD-XRF and ED-XRF. The following chapters will compare ICP-MS and XRF, WD-XRF and ED-XRF, and explain the reasons for choosing the simulation modelling for the ED-XRF method. In the end, the simulation model of the ED-XRF system used as the basic model is given.

3.1 Inductively coupled plasma mass spectrometry & X-ray

fluorescence method comparison

XRF is an X-ray optical analysis technology that is based on the spectral detection of the fluorescence of atoms excited by X-rays. This is an elemental analysis technique that can tell us the concentration of different elements in the sample. Therefore, this technology can distinguish atoms with different atomic numbers. ICP-MS (Inductively Coupled Plasma Mass Spectrometry) is an elemental analysis technique based on mass spectrometry. The two

technologies are different from a technical point of view, but at the same time the final results and applications of the two technologies are also very similar, that is, the atomic quantitative analysis of the elemental composition of any type of material [8].

The comparison between those methods has been shown in table 3.1, Method Resolution Sensitivity Portability Costs Time

ICP-MS H H L H H

XRF H H H L L

Table 3.1 Comparison of ICP-MS and XRF

(17)

18

method, which was known as non-destructive, less waste generation, and faster overall speed. [10]

3.2 Wavelength-dispersive X-ray fluorescence & Energy-dispersive

X-ray fluorescence

The main difference between the two methods, WD-XRF

(Wavelength-dispersive X-ray fluorescence) method and ED-XRF (Energy-(Wavelength-dispersive X-ray fluorescence) method is that the WD-XRF adds an analysing crystal to the system for different energy disperse. Detector in the WD-XRF system is used to detect the wavelengths emitted by different elements. This detector make the systems have better sensitivity and detection limits for light elements. The detector in the ED-XRF system is used to detect the energy emitted by the atoms of different elements, and there may be a relatively high spectral overlap. [11]

The comparison between those methods has been shown in table 3.2, Method Detection

limit

Sensitivity Resolution Costs WD-XRF Good for Be and all heavier elements Reasonable for light elements Good for heavy elements Good for light elements Less optimal for heavy elements Relatively expensive ED-XRF Less optimal for light elements Good for heavy elements Less optimal for light elements Good for heavy elements Less optimal for light elements Good for heavy elements Relatively inexpensive

Table 3.2 Comparison of WD-XRF and ED-XRF [3]

(18)

19

heavy metal element mercury, so it should be selected ED-XRF method, which have a better resolution and lower cost.

3.3 ED-XRF system simulation model

Figure 3.3 Typical ED-XRF system schematic diagram [12]

A typical ED-XRF system is shown in the figure above. The main components include:

1. Tested sample

2. X-ray source for irradiate samples

(19)

20

Figure 3.3 Simulation model built for XRF system

Based on the MCNP code shown in Appendix A, a simulation model of the XRF detection system as shown in the figure 3.3, is used as the basic model. In this XRF detection system model, the outside of the sphere is defined as a vacuum, the inside is air. The yellow 1*1*0.05 cm cuboid is the detector. The red dots represent the X-ray source. The black 1*1*1 cm square is the test sample, the grey shaded cuboid is water covering the sample.

(20)

21

4 Design

This design part modified the basic XRF model established in the previous chapter. Different comparison experiments are designed for environmental factors (different water depth), X-ray source energy, and detector materials to obtain the best system settings.

4.1 Sample

The fiber bank, the sediment of the fiber layer at the bottom of the river, is usually covered with a certain depth of water after it has been collected without going through the sample preparation process (e.g., drying and grinding).

In order to investigate the effect of water depth on the detection of mercury by the XRF system, the water depth above the sample can be simulated by adjusting the thickness of the model water layer.

Besides, the material of the detector in the XRF system is set to CdTe, the energy source is set to 100 keV, the measured sample is set to pure mercury, and the thickness of the water layer gradually changes from 0 to 14 cm. After modifying the model, a total of 3 models are established, as shown in the diagram below.

(21)

22

The first one is to build a model with only water layers but no mercury to get the background noise.

Figure 4.1.2 Simulation model with mercury and water layers

The second one is to build a model with mercury and water layers to get the total count, and then subtract the background noise to get the signal to calculate the SNR.

(22)

23

The third one is to build a model with only mercury that is no water layer covered to calculate the transmittance of mercury atoms at different water depths.

4.2 X-ray source

As described in section 2.1 of the theory, in an actual XRF measurement system, X-ray is generated by an X-ray tube. However, in the simulation system using the MCNP code, the source will be directly defined by the code [ERG = X].

The characteristic of fluorescent radiation of mercury has several values. It is usually used to observe the spectral position of mercury in the L and K layers of mercury. The energy position of the L layer is 9.98 keV, and the energy position of the Ka layer is 70.8 keV.

In the results, because of the error, the spectral position of mercury can usually be observed near these two positions.

According to the absorption edge shown in Theory 2.3, if we want to observe the spectral position of the L layer of mercury, the energy source can be set to 30 keV. If we want to observe the spectral position of the K layer, the energy source can be set to 100 keV.

Test the samples directly in the external environment, and determine whether the size of the source energy is 30 keV or 100 keV or higher, which can be discussed from the transmittance and signal-to-noise ratio of mercury atoms in water.

4.2.1 Transmittance of mercury by applying different source energy

After modifying the model, there are two kinds of simulations model was established. The first one is to observe the attenuation rate of mercury atoms at different water depths when the energy source is 30 keV with an X-Ray detector using Si-Pin sensor. The second one is applied the 100 keV energy source with an X-Ray detector using CdTe sensor.

4.2.2 Impact of source energy on SNR

(23)

24

4.3 Detector

Si-Pin and CdTe detector, each of which has advantages in certain applications.

In order to explore which detector is better when the energy source is set to 100 keV, the two detectors are compared.

(24)

25

5 Results

5.1 The SNR and transmittance of mercury spectrum while applying

100 keV source energy

The following three figures show the total signal, background noise, and effective signal of the mercury element spectrum, respectively.

Figure 5.1.1 Total signal at different water depths when using a 100 keV energy source

From the output of the MCNP simulation software, we can find that the relative counts have higher values at 44 keV, 48 keV, 69.1 keV, 71.1 keV, 71.7 keV.

The peak at the position of 69.1 keV is Mercury 𝐾α2 Line, the peak at the position of 71.1 keV is Mercury 𝐾α1 Line, and the peak at the position of 71.7 keV is the background noise.

According to the formula [2.1] 𝐸𝑠𝑐 =

𝐸0 1+( 𝐸0

in theory 2.2.2, when the original photon energy 𝐸0 is 100 keV, scattering angle 𝜃 is 180°, it can be calculated that scattered photon energy 𝐸𝑠𝑐 = 71.87 keV. Hence, it can be proved that the peak at 71.7 keV consist of scattered photon energy.

44keV 48keV

69.1keV 71.1keV

(25)

26

Escape peaks will occur according to the theory in chapter 2.4, 𝐸𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡 The dominant peak is the scattered source photon energy at 71.7 keV, which causes escape through the formula:

𝐸_{𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑} = 𝐸𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡– 𝐸𝐶𝑑 where 𝐸𝐶𝑑= 23.17 keV and 𝐸𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑 = 𝐸𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡– 𝐸𝑇𝑒 where 𝐸𝑇𝑒= 27.47 keV

It can be calculated that 𝐸𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑 are 48.53 keV and 44.23 keV, so it can be inferred that the peaks appearing at 48 keV are Escape Peaks of Cd and the peaks appearing at 44 keV are Escape Peaks of Te.

Figure 5.1.2 comparison between Mercury Kα1 peaks and background noise using a

100 keV source.

It can be clearly seen from this figure that at the spectral position of 71.1 keV, the peak of mercury is largely affected by the background scattering.

It is worth noting that the small size bin can be set in the simulation software, which means that used a high resolution detector. However, in the real world, CdTe detector has a resolution energy of 1.5 keV, which means that when the mercury Kα1 peaks to the background noise is less than the distance 1.5 keV, these two peaks would be overlap so that there will be indistinguishable. Hence, the CdTe detector is not reliable in the situation when the source energy is 100 keV.

71.1keV

(26)

27

Figure 5.1.3 Background noise at different water depths when using a 100 keV energy source

When detecting water at different depths separately, there is a lot of

background noise gathered at 71.7 keV. This is due to the scattering effect of the water.

Figure 5.1.4 Effective signal at different water depths when using a 100 keV energy source

After subtracting the background noise from the total signal, the radiation fluorescence of mercury can be clearly seen at the spectral position of 71.1 keV

44keV 48keV

(27)

28

Chart 5.1.5 Relative counts at 71.1 keV at different water depths

The above line chart shows the trend of the total signal, background noise, and effective signal at different water depths when using a 100 keV emission source and a CdTe sensor.

We can see that the background noise slowly increases with the depth of the water, however, the total amount of signal decreases in the water depth range of 1-7 cm, and rises in the water depth range of 7-14 cm. Because the sum of noise and signal is the total amounts, the effective signal decreases rapidly.

0,00E+00 5,00E-06 1,00E-05 1,50E-05 2,00E-05 2,50E-05 3,00E-05 3,50E-05 4,00E-05 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Relative counts at 71.1KeV at different water depths

(28)

29

Chart 5.1.6 SNR and transmittance of pure mercury at different water depths when the source energy is 100 keV

The chart above shows the SNR and transmittance of pure mercury at different water depths when the source energy is 100 keV. The line chart shows the SNR change trend and the bar chart shows the change in

transmittance. As can be seen from the above figure, as the depth of water increases, both the SNR and transmittance decrease rapidly.

Table 5.1 SNR and Transmittance of mercury atoms at different water depth

0,00% 20,00% 40,00% 60,00% 80,00% 100,00% 120,00% 0,00 10,00 20,00 30,00 40,00 50,00 60,00 70,00 80,00 90,00 100,00 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

SNR and transmittance of mercury at different water

depths when the source energy is 100KeV

(29)

30

The above table shows the SNR value of mercury fluorescence peaks intensity and the transmittance value of mercury atoms at a water depth of 1 cm-14 cm using a 100 keV energy source.

When the water depth is 1 cm, the SNR is 24.67, and the transmittance is 80.65%

When the water depth is 2 cm, the SNR declines to 13, and the transmittance is drops to 56.68%.

When the depth is greater than 5 cm, the SNR and transmittance become extremely small, the SNR is less than 1.7, and the transmittance is less than 17.17%.

According to the theory of LOD and LOQ, when the SNR is larger than 3, the mercury in the fiber bank sample can be detected in the real situation. For quantifying the mercury in the sample, SNR should be larger than 10. Hence, it is better to keep the water depth lower than 2 cm since SNR is 13 in the actual measurement.

5.2 The SNR and transmittance of mercury spectrum while applying

110 keV source energy

The following three figures show the total signal, background noise, and effective signal of the mercury element spectrum, respectively.

Figure 5.2.1 Total signal at different water depths when using a 110 keV source

(30)

31

Figure 5.2.2 comparison between Mercury Kα1 peaks and background noise using a

110 keV source.

The peak at the position of 69.1 keV is Mercury 𝐾α2 Line, the peak at the position of 71.1 keV is Mercury𝐾α1 Line, and the peak at the position of 71.7 keV is the background noise.

According to the formula [2.1] 𝐸𝑠𝑐 =

𝐸0 1+( 𝐸0

in theory 2.2.2, when the original photon energy 𝐸0 is 110 keV, scattering angle 𝜃 is 180°, it can be calculated that scattered photon energy 𝐸𝑠𝑐 = 76.90 keV. Hence, it can be proved that the peak at 77.1 keV is the scattered photon energy.

According to the theory in chapter 2.4, 𝐸𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡 at this time is scattered photon energy is 77.1 keV, through the formula:

𝐸_{𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑} = 𝐸𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡– 𝐸𝐶𝑑 where 𝐸𝐶𝑑= 23.17 keV and 𝐸𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑 = 𝐸𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡– 𝐸𝑇𝑒 where 𝐸𝑇𝑒= 27.47 keV

It can be calculated that 𝐸𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑 are 53.93 keV and 49.63 keV, so it can be inferred that the peaks appearing at 50 keV are Escape Peaks of Cd and the peaks appearing at 54 keV are Escape Peaks of Te.

Comparing the picture 5.1.2, it can be found that the scattered energy of water which represents background noise have moved to the right at position

77.1 keV.

(31)

32

The distance between two peaks is larger than resolution energy of CdTe 1.5 keV, overlap of the two peaks would not happen, hence, CdTe detector is more reliable in situation using 110 keV energy source than 100 keV energy source.

Figure 5.2.3 Background noise at different water depths when using a 100 keV energy source

Compared with the picture 5.1.2, the value of the background noise at the position of 71.1 keV has become very small, and is an order of magnitude lower.

Figure 5.2.4 Effective signal at different water depths when using a 110 keV energy source

69.1keV 71.1keV

(32)

33

After subtracting the background noise from the total signal, the radiation fluorescence of mercury can be clearly seen at the spectral position of 71.1 keV.

Chart 5.2.5 Relative counts at 71.1 keV at different water depths

The above line chart shows the trend of total signal, background noise, and effective signal at different water depths when using a 110 keV emission source and a CdTe sensor.

We can see that as the depth of water increases, the background noise increases is extremely slow, and the total amount of signals are gradually decline. Therefore, the effective signals also show a gradual decline in the depth of 1-14cm. 0,00E+00 5,00E-06 1,00E-05 1,50E-05 2,00E-05 2,50E-05 3,00E-05 3,50E-05 4,00E-05 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Relative counts at 71.1KeV at different water depths

(33)

34

Chart 5.2.6 SNR and transmittance of mercury at different water depths when the source energy is 110 keV

The chart above shows the SNR and transmittance of mercury atoms at different water depths when the source energy is 110 keV. The line chart shows the SNR change trend and the bar chart shows the change in transmittance. As can be seen from the above chart, as the depth of water increases, both the SNR and transmittance are rapidly decreasing.

Water depth(cm) 0 1 2 3 4 5 6 7 SNR ∞ ∞ 200.0 0 76.00 31.67 12.40 9.60 5.25 Transmitta nce 100.00 % 84.80 % 58.48 % 44.44 % 27.78 % 18.13 % 14.04 % 6.14 % Water depth(cm) 8 9 10 11 12 13 14 SNR 2.00 1.22 1.13 0.58 0.69 0.24 0.14 Transmitta nce 4.68% 3.22% 2.63% 2.05% 2.63% 1.17% 0.88%

Table 5.2 SNR and Transmittance of mercury atoms at different water depth

0,00% 10,00% 20,00% 30,00% 40,00% 50,00% 60,00% 70,00% 80,00% 90,00% 100,00% 0,00 20,00 40,00 60,00 80,00 100,00 120,00 140,00 160,00 180,00 200,00 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

SNR and transmittance of mercury at

different water depths when the source

energy is 110KeV

(34)

35

The above table shows the SNR of mercury fluorescence peaks intensity and the specific transmittance of mercury atoms at a water depth of 1 cm-14 cm using a 110 keV energy source.

We can see that when the depth of water is 0 cm or 1 cm, the SNR tends to infinity because the background noise is zero. It is worth noting that this is not possible in actual situations, because in reality, noise between electronic components is ubiquitous, and MCNP simulations only include physical noise but exclude electronic noise.

When the water depth is 1 cm, the SNR is infinite due to the noise is zero, the transmittance declines to 80.65%.

When the water depth is 2 cm, the SNR is 200 and the transmittance is 58.48%. When the water depth increase to 5 cm, the SNR drop to 12.4 and the

transmittance declines to 18.13 %. When the water depth increase to 7 cm, the SNR continually drops to 5.25 and the transmittance declines to 6.14 %. When the depth is greater than 10 cm, the SNR become extremely small, the SNR is less than 1. At the 6 cm or greater, transmittance is less than 10%. According to the theory of LOD,, in this case, the water depth should be controlled within 7 cm during actual measurement for detecting the mercury contents in fiber bank sample.

According to the theory of LOQ,, in this case, the water depth should be controlled within 5 cm during actual measurement for quantifying the mercury contents in fiber bank sample.

5.3 Source energy change on the SNR

Chart 5.3.1 SNR curve with different energy source

0,00 20,00 40,00 60,00 80,00 100,00 120,00 140,00 160,00 180,00 200,00 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

SNR curve with different energy source

(35)

36

From the above line comparison chart, we can explore how the energy of the energy source will affect the SNR of mercury fluorescence peaks intensity. The SNR obtained when using a 110 keV energy source is significantly better than using a 100 keV energy source.

At water depths of 0 and 1 cm, the SNR at 110 keV is positive infinity, which means that in this case it is not affected by any scattering energy of mercury fluorescence, while the SNR at 100 keV is 91 and 24, subject to some scattering impact of the source energy.

At water depths of 2 cm and 3 cm, the SNR of the former is 15 and 8 times that of the latter, respectively. It can be seen that increasing the amount of energy is very significant for increasing the signal SNR.

5.4 The transmittance of mercury atoms with different water depth

when the source energy is 30 keV

By modifying model, a pure mercury cube and 30 keV energy source was applied, the relative counts can be measured at 10 keV which represents scattering energy from L shell of mercury.

(36)

37 Water depth(mm) 0 0.1 0.2 0.3 0.4 0.5 0.6 Relative counts (cps) 3.70E-06 2.00E-07 2.00E-07 2.00E-07 2.00E-07 1.00E-07 1.00E-07 Transmittance 100.00% 5.41% 5.41% 5.41% 5.41% 2.70% 2.70%

Table 5.4 Transmittance of mercury atoms at different water depth

Chart 5.4 Transmittance of mercury at different water depths when the source energy is 30 keV

As can be seen from these charts, in the case where the energy source is 30 keV, and the distance between the detector and sample is 15 cm, while the thickness of water was increasing to 0.1 mm, the transmittance of mercury atoms is greatly reduced to 5.41%. As the thickness of the water deepens, the transmittance of mercury atoms becomes lower and lower, and when the thickness of water is 0.5 mm, the transmittance of mercury atoms is already close to zero.

Hence, we can conclude that the energy of the emission source set to 30 keV is very weak for penetrating water.

0,00% 20,00% 40,00% 60,00% 80,00% 100,00% 120,00% 0,00E+00 5,00E-07 1,00E-06 1,50E-06 2,00E-06 2,50E-06 3,00E-06 3,50E-06 4,00E-06 0mm 0.1mm 0.2mm 0.3mm 0.4mm 0.5mm 0.6mm

Transmittance of mercury at different water depths when

the source energy is 30KeV

(37)

38

5.5 Compare two kind of material of detector

Figure 5.5 Si-Pin vs. CdTe sensor when the source energy is 100 keV

Chart 5.5 Si-Pin vs. CdTe sensor when the source energy is 100 keV

When the energy source is 100 keV and the water depth is 0 mm, we can find that the number of particles emitted by the K-line of the mercury atom that can be received by the detector using CdTe is 3.67E-05, and the number of particles received by the Si detector is only 1.80E-06, which is 20 times lower than using CdTe.

As is clear in principle the same simulation time, CdTe detectors will have better efficiency in this situation. Therefore, the XRF system should use CdTe detector when detecting K-line of the mercury atom spectrum.

1,80E-06 3,67E-05 0,00E+00 5,00E-06 1,00E-05 1,50E-05 2,00E-05 2,50E-05 3,00E-05 3,50E-05 4,00E-05 Si-Pin CdTe

(38)

39

6 Conclusion

Based on the results and analysis of the previous chapter, the following conclusions can be summarized:

When setting the energy source as 30 keV, it can be seen that the relative count of mercury atoms obtained at the spectral position of 10 keV is

extremely low. That means, the energy radiated by the mercury atom L-line is weak, and it is very difficult to penetrate the water layer. It is impossible to obtain reliable data for samples with water or moisture.

When setting the energy source as 100 keV, it can be seen that the energy radiated by the mercury Kα1 line is relatively high. Considering the standard of LOQ, in the case of water depth of 0-2cm, a good SNR larger than 13 can be maintained. However, the resolution of the CdTe detector is not high enough to distinguish mercury signal and background noise at this energy.

When setting energy source as 110 keV, due to the less background noise overlap at the energy radiated by the mercury Kα1 line, the mercury signal get less affected by background noise. Besides, due to the right shift of the background noise, the CdTe detector is reliable to distinguish mercury signals and background noise at this energy.

The SNR of mercury atoms at different water depths is improved compared to the energy source set to 100 keV. Considering the standard of LOQ, in the case of water depth of 0-5 cm, a better SNR larger than 12.4 can be

maintained. Hence, the water depth should be controlled within 5 cm during actual measurement for quantification in this situation.

In summary, in order to obtain better accuracy when the XRF system

(39)

40

6.1 Future work

The sample used in this thesis is 100% mercury, it can be inferred that when the concentration of mercury is much lower than 100% in the actual sample, the radiant energy will be further weakened, which will make it more difficult to detect the mercury content. Some future work can be further discussed: 1. Set mercury samples with different concentrations to simulate more

realistic conditions.

2. Apply filters to reduce the scattering effect caused by water and improve the XRF system.

(40)

41

References

[1] Fredrik Israelsson, ”Sundsvallsbukten - Sveriges mest

kvicksilverförgiftade område”, SVT Västernorrland, 2019-04-08, https://www.svt.se/nyheter/lokalt/vasternorrland/sundsvallsbukten-sveriges-mest-kvicksilverforgiftade-omrade, retrieved 2019-09-15. [2] Björn Wiese, “The Effect of CaO on Magnesium and Magnesium

Calcium Alloys”, PhD Thesis, Clausthal University of Technology, 2016, DOI: 10.21268/20170504-133828.

[3] Peter Brouwer, ”Theory of XRF – Getting acquainted with the principles”, PANalytical B.V., 2010.

[4] ”X-Ray Fluorescence Spectroscopy (XRF) – Basics”, Helmut Fischer, commercial webpage https://xrf-spectroscopy.com, retrieved 2019-11-30.

[5] Ethan A Merritt, ” X-ray Absorption Edges”,

http://skuld.bmsc.washington.edu/scatter/data/Hg.html, retrieved 2019-10-12.

[6] Pete Palmer, ”Interpretation of XRF Spectra”, Introduction to XRF- An

Analytical Perspective, LibreTexts, retrieved 2019-12-15.

[7] Wikipedia, ”Signal-to-noise ratio”,

https://en.wikipedia.org/wiki/Signal-to-noise_ratio, retrieved 2019-10-05.

[8] An, S. , Norlin, B. , Hummelgård, M. & Thungström, G., “Comparison of Elemental Analysis Techniques for Fly Ash from Municipal Solid Waste Incineration using X-rays and Electron Beams”. I IOP Conference

Series : Earth and Environmental Science. (2019),

https://doi.org/10.1088/1755-1315/337/1/012007

[9] ResearchGate online discussion thread, ”Is it possible to measure Mercury using ICP MS?”,

(41)

42

[10] Jacqueline Q. McComb, Christian Rogers, Fengxiang X. Han, and Paul B.Tchounwou, ”Rapid screening of heavy metals and trace elements in environmental samples using portable X-ray fluorescence

spectrometer, A comparative study”, Water Air Soil Pollut. 2014 Dec; 225(12): 2169, doi: 10.1007/s11270-014-2169-5

[11] ”Wavelength Dispersive X-ray Fluorescence (WDXRF)”, XOS, commercial webpage https://xos.com/WDXRF, retrieved 2020-01-15. [12] ”XRF Spectroscopy”, HORIBA, commercial webpage

(42)

43

Appendix A: MCNP program code

Study of energy deposition in four different semiconductors C **************** Cell cards ****************************************

1 1 -5.85 1 -2 3 -4 5 -6 imp:p,e 1 $ CdTe 2 3 -13.6 1 -2 3 -4 -9 10 imp:p,e 1 $ Hg 3 2 -0.001 -7 #1 #2 #5 imp:p,e 1 $ Air

4 0 7 imp:p,e 0 $ Outer void 5 6 -1 12 -13 14 -15 16 -17 #2 imp:p,e 1 $ Water H2O C **************** Surface cards ************************************* 1 PX -0.5000 2 PX 0.5000 3 PY -0.5000 4 PY 0.5000 5 PZ 3.0000 6 PZ 3.0500

7 SO 20.000 $ Rest of the universe 8 PZ 2.0000 9 PZ -15.0000 10 PZ -16.0000 11 PZ 2.0000 12 PZ -16.000 13 PZ -1.000 14 PX -8.000 15 PX 8.000 16 PY -8.000 17 PY 8.000 C ***************** Data cards *************************************** MODE P E $ Particle for tracking C *** Monodirectional area source at centre ********************* SDEF POS=0.0000 0.0000 0.000 DIR=1 ERG=0.100 par 2 vec 0 0 -1 C CUT:P J 1.0e-4

C CUT:E J 1.0e-4

C *** Tally of the deposited energy in each cube ********************* F8:p 1

(43)

44

C *** Atomic number of the material ********************************** m1 52000.14p 0.5 $Te 48000.14p 0.5 $Cd m2 8000.14p 0.2 $O2 7000.14p 0.8 $N4 m3 80000.14p 1.0 $Hg m6 1000.14p 2 $H2 8000.14p 1 $O nps 1e7 print 110

C ptrac file=asc write=all C PRDMP 1

C *** Simulation time/number of simulated particles ****************** C ctme 40

(44)

45

Appendix B: X-Ray Data Booklet Table

Peak position (keV)

Theoretical design of an XRF system for environmental measurements of Mercury in fiber banks