The use of diluted acid preparations of biological samples in microwave digestion

The use of diluted acid preparations of biological samples in microwave digestion

Astri Göransson

Bachelor’s Programme in Chemistry Degree project C in Chemistry

2019-11-04 – 2020-01-10 Department of Chemistry - BMC

Uppsala University Supervisor: Jean Pettersson

1

Abstract

To make it possible to analyse trace elements in biological samples, the elements need to be in solution. Microwave digestion is a fast and effective method suitable for this kind of sample preparation, normally performed with concentrated acids. For the analysis itself, the Inductively Coupled Plasma Atomic Emission Spectrometry instrumentation is common as it is a powerful and fast technique, and it is also used in these experiments. The problem with the Inductively Coupled Plasma Atomic Emission Spectrometry instrumentation is that concentrated acids might do harm to it, making dilution prior to the measurement important.

These experiments investigate the possibility to do microwave digestions in diluted acids in order to avoid the dilution of the samples later on. This is interesting as it opens up the possibility to analyse low element concentrations in the sample and minimize the waste of sample and reagents.

The trace elements in eight reference samples (Bone Meal SRM 1486, Bovine Liver SRM 1577a, Bovine Muscle BCR-CRM-184, Cabbage IAEA-359, Citrus Leaves SRM 1572, Pig Kidney BCR-CRM-186, Tuna Fish IAEA-350 and Wheat Flour SRM 1567a) were analysed.

Bovine liver and wheat flour were chosen for further analysis due to their content of trace elements. Sample sizes at 25, 50 and 100 mg were used, and digestions were carried out in HNO3 with concentrations at 2.5, 5 and 10% as well as with concentrated HNO3. H2O2 was used as an auxiliary oxidation agent and the elements analysed were Mg, P, S, Ca, Mn, Fe, Cu and Zn. The method was then evaluated by calculating analysis of variance, limit of detection, limit of quantification and by examining residual carbon content. Many element concentrations were below limit of detection and limit of quantification for the small amount of sample, especially for wheat flour which also resulted in large relative standard deviation values. For bovine liver the results were better, with more of the calculated concentrations within the confidence intervals. Microwave digestion in diluted acids is possible and with more research a good balance between sample size and acid concentration will probably be found.

2

Table of content

Abstract ... 1

Table of content ... 2

Abbreviations ... 3

1. Introduction ... 4

2. Theory ... 5

2.1. Microwave digestion ... 5

2.2. Inductively Coupled Plasma Atomic Emission Spectrometry ... 6

2.3. Samples, reagents and standards ... 6

2.3.1. Internal Standard... 6

3. Experimental ... 7

3.1. Instrumentation ... 7

3.1.1. Microwave digestion ... 7

3.1.2. Inductively Coupled Plasma Atomic Emission Spectrometry ... 7

3.2. Samples, reagents and standards ... 7

3.2.1. Internal standard ... 8

3.2.2. Calibration standards ... 8

3.2.3. Reference Samples ... 8

3.2.4. Start-up ... 9

3.2.5. Changing parameters ... 9

3.2.6. Method evaluation ... 11

4. Results and discussion ... 11

4.1. Sample reagents and standards ... 11

4.1.1. Calibration standards ... 12

4.1.2. Start-up ... 12

4.1.3. Changing parameters ... 13

4.1.4. Method evaluation ... 27

5. Conclusion ... 28

5.1. Future aspects ... 29

Acknowledgments ... 29

References ... 30

Appendix ... 32

3

Abbreviations

ANOVA Analysis of Variance

CI Confidence Interval

ICP-AES Inductively Coupled Plasma Atomic Emission Spectrometry

LOD Limit of Detection

LOQ Limit of Quantification

ppm Parts Per Million

RCC Residual Carbon Content

RSD Relative Standard Deviation

SD Standard Deviation

(v/v) volume/volume

4

1. Introduction

When analysing trace elements in biological samples, Inductively Coupled Plasma Atomic Emission Spectrometry is a suitable method. A fast and efficient method for sample preparation is microwave digestion, in which concentrated acids are normally used. Fusion, dry ashing and extraction are other sample preparation methods, where extraction is suitable for organic analytes and fusion for an inorganic matrix [1]. Dry ashing is a slower method with large risk of contamination and losses and require expensive parts but is safe, need small amounts of reagents and larger sample sizes can be used [2]. As concentrated acids may do harm to the instrumentation in Inductively Coupled Plasma, it is not possible to run samples with the high amount of acid used in microwave digestion. The samples therefore need to be diluted before an analyse takes place. This makes it hard to discover elements with low concentrations in the sample as well as it results in higher amounts of waste, both when it comes to samples and reagents [3], making it an important part of green chemistry [4-11]. It was investigated if it is possible to do microwave digestions with diluted acids and still maintain a high efficiency.

In 1998 Lamble and Hill looked at different methods for microwave digestion, as matrix composition [12]. An early article on the subject of diluted acid in microwave digestions was Araújo et al., who in 2002 used diluted acids for the digestion of plant materials [13]. They found that it is possible to obtain an efficient digestion with the use of diluted acid and that the presence of hydrogen peroxide plays an important role. Several other articles came to the same conclusion when digesting biological samples [4-11, 14, 15] and all of them used nitric acid with hydrogen peroxide as an auxiliary oxidant agent. Many of the articles discusses the importance of oxygen [4, 7-10, 15], where Gonzalez et al. came to the conclusion that the oxygen present in the vessels is sufficient to obtain an efficient digestion [10]. Some of the articles carried out the digestions in oxygen pressurized vessels [4, 7-9], which made it possible to lower the acid concentration even further without compromising the efficiency of the microwave digestion. The residual carbon content was investigated in some prior work [4- 6, 8-10, 13-15], which was used to see how much of the sample that has been digested.

Barbosa et al. also mentions that digestions of products with a high protein and fat content should be more difficult [6], and Gonzalez et al. that concentrated nitric acid decompose even those samples at 160°C [10]. In previous work sample sizes ranging between 100 and 500 mg were used together with digestion volumes between 3 and 8 mL [3-15]. Tarantino et al. found that an efficient digestion was obtained at an acid concentration of 4.5 mol/L when 250 mg of rice was digested at 200°C in 7 mL together with 1 mL H2O2 [5]. Barbosa et al., on the other hand, found that an acid concentration at 2.1 mol/L gave an effective digestion with the same set up, but with soya beans and a temperature at 210°C [6]. Gonzalez et al. used 200 mg of bovine muscle and found that when digesting that sample amount in 3 mL at a temperature at 180-200°C acid concentration as low as 2 mol/L gave satisfying results [10].

Eight biological reference samples were used in these experiments, namely Bone Meal SRM 1486, Bovine Liver SRM 1577a, Bovine Muscle BCR-CRM-184, Cabbage IAEA-359, Citrus Leaves SRM 1572, Pig Kidney BCR-CRM-186, Tuna Fish IAEA-350 and Wheat Flour SRM 1567a, choosing bovine liver and wheat flour for deeper studies. The microwave efficiency was investigated after both the amount of sample and acid concentration were changed to try to find an efficient digestion at as low acid concentration as possible. The digestions were performed with 2.5%, 5%, 10% and concentrated nitric acid with 0.6 mL hydrogen peroxide

5

as auxiliary oxidant agent. The amounts of samples used were 25, 50 and 100 mg and Inductively Coupled Plasma Atomic Emission Spectrometry was then used to evaluate the efficiency of the digestions. The trials were evaluated with analysis of variance, limit of detection, limit of quantification and residual carbon content. It was found that it is possible to do microwave digestions with the use of diluted acid, even though there is still much to investigate and evaluate.

2. Theory

Inductively coupled plasma atomic emission spectrometry (ICP-AES) is a commonly used analyse method since it is fast and able to detect several analytes in one single run [1, 12]. The samples need to be in a liquid state when injected, but concentrated acids might do harm to the instrumentation, making dilution prior to the injection essential when using microwave digestion as the sample preparation method [8-10]. This does not only result in large amount of waste [5, 9, 11], but the dilution also makes it harder to detect analytes that exist in low concentrations in the sample.

2.1. Microwave digestion

Sample preparation is a very important part of analytical chemistry to make sure that the analytes of interest are detectible in a representative manner. Many analytical instrumentations require liquid samples and to get an accurate analysis, a correct sample preparation is crucial. To be able to analyse trace elements with ICP-AES, the analytes need to be completely released into the solution. Microwave digestion is a fast and effective method developed in the 1970’s. In a microwave digestion the sample is decomposed by being boiled with concentrated acids under high temperatures. A closed system, called bomb digestion, is preferred as higher temperature and pressure can be obtained. Other advantages with a closed system are that a smaller amount of sample can be used, as there is no sample loss by vaporization, and the risk of contamination is also less. The vessels used for microwave digestion is usually made of Teflon as it is a material resistant to the acids commonly used in digestions as well as it is thermally stable and transparent to microwaves [1, 12, 15, 16].

For the trace elements to be dissolved into solution, the matrix needs to be decomposed [10, 12, 15]. Nonoxidizing acids can be used in microwave digestions, but for materials that does not dissolve in them, an oxidizing acid is needed [1]. One commonly used oxidizing acid is HNO3, often with H2O2 as an auxiliary oxidant agent, giving reactions according to Equation 1 and Equation 2 [9, 15]:

2𝐻₂𝑂₂ (𝑙) → 2𝐻₂𝑂 (𝑙) + 𝑂₂ (𝑔) (Eq. 1)

(𝐶𝐻₂)_𝑛+ 𝐻𝑁𝑂₃(𝑎𝑞) → 𝐶𝑂₂ (𝑔) + 𝑁𝑂 (𝑔) + 𝐻₂𝑂 (𝑙) (Eq. 2)

Equation 2 shows that oxidation is an important part of the digestion, the presence of oxygen in the microwave digestion contributes to an effective digestion, and hence the product can be oxidized more efficiently. The NO then reacts further to regenerate HNO3, see Equation 3-5 [9, 15], making the presence of O2 important [15].

2 𝑁𝑂 (𝑔) → 𝑂₂ (𝑔) + 2 𝑁𝑂₂ (𝑔) (Eq. 3) 2 𝑁𝑂₂ (𝑔) + 𝐻₂𝑂 (𝑙) → 𝐻𝑁𝑂₃(𝑎𝑞) + 𝐻𝑁𝑂₂ (𝑎𝑞) (Eq. 4)

6

2 𝐻𝑁𝑂₂ (𝑎𝑞) → 𝐻₂𝑂 (𝑎𝑞) + 𝑁𝑂₂ (𝑔) + 𝑁𝑂 (𝑔) (Eq. 5)

If the matrix decomposition is incomplete, the analyte concentration might not correspond to the actual analyte concentration in the sample, as metals might still be in complex. An incomplete decomposition can be shown by the solution not being completely clear [10, 11].

The matrix also affects how efficient the digestion can be and how high the temperatures and acid concentrations need to be. Naturally, stronger chemical bonds are harder to oxidize than those bonded more loosely. The oxidative power for HNO3 increases with temperature and in order to obtain as an efficient digestion with diluted acids as for digestions with concentrated acid, the temperature therefore needs to be higher [13].

2.2. Inductively Coupled Plasma Atomic Emission Spectrometry

ICP-AES is an atom spectroscopy analyse method [1, 16] that can determine several trace elements in a sample in one single, fast measurement [1, 12]. As for all atomic spectroscopic techniques, the analytes need to be atomized and in gas phase to make the analyse possible.

The ICP technique uses a plasma, i.e. a gas with high concentrations of cations and electrons, to atomize the analytes [16]. The plasma for ICP-AES consists of the inert Ar-gas and the instrumentation of three quartz tubes leading up to the plasma. The inner quartz tube is for sample, the middle one for the plasma gas flow and the outer for Ar-gas used as a cooler.

Around the top of the outer quartz tube, an inductive coil is wrapped. The ionization of the plasma Ar-gas is initiated by a spark from a Tesla coil and the ions are then accelerated to the plasma by the magnetic field applied to the inductive coil. This process where free electrons collide with atoms, forwarding the energy to the plasma, makes it possible for the plasma to maintain temperatures at up to 10 000 K [1, 16].

The samples in ICP-AES are introduced by a flow of Ar-gas into the plasma [16] where the analyte gets into excited state by collisions in the hot plasma. When returning to their ground state, photons are emitted with a wavelength equivalent to the energy difference and an intensity proportional to the element concentration [1]. Different elements emit photons with wavelengths characteristic for that specific element, as they have individual energy differences between the ground and excited state. When running an analyse based on AES, wavelengths for the analyte of interest needs to be chosen, with ICP-AES having the advantage of being able to measure on several wavelengths in one run [16].

2.3. Samples, reagents and standards

Nitric acid is commonly used in microwave digestions because it is a strong oxidizing agent and is easy to purify [10, 15]. It is also a strong oxidizing agent with the possibility to dissolve most metals [16]. Certified reference materials are used for evaluation of a method and to see if the calculated concentration is within the confidence interval (CI) [1].

2.3.1. Internal Standard

An internal standard is an element, other than any of the analytes and therefore not present in the sample, added to all samples and standards. The added concentration is known and the same for all solutions being measured on. As the signal intensities are proportional to the concentrations, the intensities representing the concentration of the internal standard should therefore be the same for all measurements. The analyte intensity is then corrected from the intensity of the internal standards for each element by calculating the ratio between them. The internal standard method is used because the signal intensities in some instrumentation might

7

differ slightly between runs. The method therefore corrects for eventual instrumental drift or other uncertainties in the measurement. Variations in the method can thus be reduced, such as sample introduction [1, 16].

3. Experimental

3.1. Instrumentation 3.1.1. Microwave digestion

For the microwave digestions in this experiment a Titan MPS 16 Position Microwave Sample Preparation System oven with 75 mL closed vessels was used with the digestion volume in all bombs being 6 mL. After all reagents were added to the bombs, they were left to rest in about 15 minutes before the lid and cap were put on and they were transferred into the microwave.

The temperature program settings used is shown in Table 1.

Table 1: The microwave digestion settings used for the digestions.

Step T [°C] P [bar] Ramp Hold P [%]

1 165 30 5 10 80

2 190 30 1 10 90

3 50 30 1 10 0

After the temperature programs were finished, the lid on the microwave was open, in order to let the bombs cool down, before the cap and lid were removed and the samples poured into 15 mL Falcon tubes.

3.1.2. Inductively Coupled Plasma Atomic Emission Spectrometry

The ICP-AES used had a plasma with an axial view and was of the brand Spectro Ciros CCD.

The samples were introduced with a flow rate at 0.9 L/min, the Ar-gas flow rate for the plasma was also set to 0.9 L/min and the Ar-gas flow rate used as a coolant was set to 14 L/min. The inductive coil has a 40 MHz electric field with a power of 1 400 W applied.

Wavelengths were chosen after what trace elements to expect in the reference samples. The chosen wavelengths were 279.553, 280.270 and 285.213 nm for Mg, 396.847, 393.366 and 317.933 nm for Ca, 257.611 and 259.373 nm for Mn, 259.941, 238.204 and 244.451 nm for Fe, 324.754, 327.396 and 224.700 nm for Cu, 213.856 and 206.191 nm for Zn, 177.495 and 178.287 nm for P and 180.731 and 182.034 for S.

3.2. Samples, reagents and standards

All the samples were weighed in weighing boats with a five decimals Mettler AE 240 scale located in a scale room. For dilution of all samples, distilled and deionized MQ-water was used deionized with a Millipak® 0.22 μm membrane 40 filter. The nitric acid used was a 68%

solution and the hydrogen peroxide a 30% solution, both from VWR Chemicals manufactured by VWR International S.A.S., Fontenay-sous-Bois, France, with the nitric acid being sub- boiled before use.

When different parameters were about to be changed, two reference materials were chosen after the amount of trace elements possible to analyse, those being Bovine Liver 1577a and Wheat Flour 1567a. The weighing boats were weighed before and after the samples were added to the bombs to obtain the net weight.

8

As the calibration curves were created from 10% HNO3 calibration standards, a 1 ppm solution with 2.5% HNO3 was made from the 5 ppm calibration standard. This to tell whether the acid concentration effects the determinations or not. The 5 ppm solution was diluted five times to obtain a 1 ppm solution with 2.5% acid. 2 mL of the 5 ppm solution was pipetted into a 15 mL Falcon tube, adding 400 μL of the internal standard solution in order to maintain the Sc concentration. The 1 ppm solution was then measured on at repeated times during analysis.

The concentrations of those measurements were then calculated from the calibration curves together with the means. All values from measurements in 2.5% HNO3 has been corrected from the mean values of the control standard.

Between every experiment, the bombs and their caps were cleaned with MQ-water at least three times and then left to dry.

3.2.1. Internal standard

Scandium was selected to be the internal standard as no reference sample was certified for Sc and because it shows clear peaks from the ICP-AES analysis. The solution used was a 996±3 ppm scandium solution with SS-1555 from Spectrascan distributed by Teknolab AB, Kungsbacka, Sweden.

As an internal standard, a Sc-solution was added to all samples to obtain a Sc concentration of 5 ppm. The Sc-solution was prepared by pipetting 5 mL of the 1 000 ppm Sc-solution into a 50 mL Falcon tube. 0.5 mL HNO3 was added and the Falcon tube was then made up to mark.

0.3 mL of that standard solution was added to all sample solutions to obtain the Sc- concentration of 5 ppm.

3.2.2. Calibration standards

Solutions for the calibration standards were prepared from a multi-element 50.00 ppm solution from Inorganic Ventures, Inc, Christiansburg, Virginia, USA, with 5.0% HNO3 (v/v) containing Ag, Al, As, B, Ba, Ca, Cd, Co, Cr3, Cu, Fe, K, Mg, Mn, Na, Ni, Pb, Pd, Pt, Sb, Se, Sr, Ti, V, Zn and Zr and a diluted solution containing 50 ppm P and S. The phosphorous solution used was a 999±4 ppm solution with SS-1544 and the sulphuric solution was a 1 009±3 ppm solution with SS-1552, from Spectrascan distributed by Teknolab AS, Ski, Norway and Teknolab AB, Kungsbacka, Sweden, respectively.

For preparation of the calibration standards, 0, 0.2, 1, 5, 10 and 25 mL of both a 50 ppm phosphorous/sulphur solution and a multi-element 50 ppm solution were added to six 50 mL Falcon tubes. The two-element 50 ppm solution, containing phosphorous and sulphur, was prepared from the 1 000 ppm solutions in a 50 mL Falcon tube with 1 mL HNO3 added before it was made up to mark with MQ-water. To all the calibration standards, 2.5 mL of the internal standard were added before they were made up to mark with MQ-water. Before each measurement, the calibration standards were remeasured.

3.2.3. Reference Samples

Certified reference samples were used for evaluation of the method. They were Bone Meal SRM 1486, Bovine Liver SRM 1577a, Bovine Muscle BCR-CRM-184, Cabbage IAEA-359, Citrus Leaves SRM 1572, Pig Kidney BCR-CRM-186, Tuna Fish IAEA-350 and Wheat Flour SRM 1567a. They were all purchased dried and homogenized and some of them were quite old. All samples were prepared in current condition, i.e. they were not dried again before use

9

and therefore not corrected for humidity taken up from the air. For the certified reference concentrations in the samples and their confidence interval (CI), see Table 2.

Table 2: The reference concentrations of the chosen elements in the certified reference samples.

ppm Bone

Meal 1486 [17]

Bovine Liver 1577a [18]

Bovine Muscle No 184 [19]

Cabbage IAEA- 359 [20]

Citrus Leaves 1572

[21]

Pig Kidney

No 186 [19]

Tuna Fish IAEA- 350 [22]

Wheat Flour 1567a [23]

Mg 4 660

± 170

600

± 15

2 160

± 52.1

5 800

± 300

400

± 20 P 123 000

± 1 900

11 100

± 400

1 300

± 200

1 340 ± 60

S 7 800

± 100

4 070

± 90

1 650

± 20 Ca 265 800

± 2 400

120

± 7

18 500

± 510

31 500

± 1 000

100, CI:

72-191

191 ± 4

Mn 9.9

± 0.8

0.334

± 0.028

31.9

± 0.56

23

± 2

8.5

± 0.30

0.60, CI:

0.52-0.74

9.4

± 0.9

Fe 99

± 8

194

± 20

79

± 1.98

148

± 3.9

90

± 10

299

± 9.87

72.1, CI:

66.7-77.3

14.1

± 0.5

Cu 158

± 7

2.36

± 0.059

5.67

± 0.18

16.5

± 1.0

31.9 ± 0.41

2.83, CI:

2.55-3.10

2.1

± 0.2

Zn 147

± 16

123

± 8

166

± 2.99

38.6

± 0.67

29

± 2

128

± 2.3

17.4, CI:

16.6-18.5

11.6

± 0.4

3.2.4. Start-up

Before different parameters were changed, experiments were carried out to test the repeatability of the method. The eight reference samples were digested in 10% HNO3 and then analysed with ICP-AES. 25 mg of each sample was weighed in weighing boats and then added to the bombs. 4.5 mL MQ-water, 0.6 mL HNO3, 0.6 mL H2O2 and 0.3 mL of the internal standard solution were then added to the bombs. Three replicates of each sample were made, where the MQ-water was added prior to the sample in one of them. Blanks were prepared in the same manner.

3.2.5. Changing parameters

For evaluation of the method different parameters were changed and analysed. Three replicates were done in almost all tries.

Acid concentration

The same procedure was carried out as in the start-up, 25 mg of bovine liver and wheat flour were weighed and added to the bombs, six with liver and six with flour. To three of the liver bombs and three of the flour bombs, 4.8 mL MQ-water and 0.3 mL HNO3 was added to obtain an acid concentration at 5%. To the other six bombs, 4.95 mL MQ-water and 0.15 mL HNO3 were added to obtain an acid concentration at 2.5%. H2O2 and internal standard were added in the same volumes as in the start-up. Two blanks of each acid concentration were prepared.

After the digestion with 25 mg of samples in 2.5% and 5% HNO3, the bombs were cleaned with 5 mL HNO3 and 1 mL H2O2.

10

The digestion efficiency with different acid concentrations was also investigated for 50 mg and 100 mg of sample. 50 and 100 mg of bovine liver and wheat flour samples were digested first in 2.5% HNO3 with the same procedure as for the 25 mg samples, and then in concentrated acid. The samples digested in concentrated acid were digested in 5 mL HNO3, 1 mL H2O2 and 0.3 mL internal standard solution. Those samples were also diluted prior to the analysis with ICP-AES. 1 mL of the digested samples were pipetted into 15 mL Falcon tubes, 400 μL Sc-solution was added, to keep the Sc concentration in all solutions constant, before they were made up to 10 mL with MQ-water. Two replicates were prepared in this case, together with two blanks with an acid concentration at 2.5% and four blanks with concentrated acid.

Sample amount

To see how much sample is possible to digest with an acid concentration at 2.5%, other than the 25 mg samples, digestion with 50 and 100 mg bovine liver and wheat flour were prepared.

The reagents were added in the same volumes as for the 25 mg samples. Four blanks were also prepared in the same manner.

Temperature program

Investigation of temperature program impact was performed by repeating the experiment carried out with 50 and 100 mg of sample digested in 2.5% HNO3. The only parameter changed was the settings of the temperature program, see Table 3.

Table 3: The temperature setting for the run with a longer temperature program.

Step T [°C] P [bar] Ramp Hold P [%]

1 165 30 5 15 80

2 190 30 1 25 90

3 50 30 1 10 0

Wheat flour sample preparation

An experiment was carried out where a different sample preparation was used for wheat flour.

25 mg of sample was weighed, a few droplets of water were added to the weighing boats and then mixed with the sample with a glass rod. The paste was then rinsed down the bomb with 4.95 mL MQ-water, 0.3 mL of the internal standard, 0.6 mL H2O2, and 0.15 mL HNO3. In this trial, the weighing boats were also weighed after the sample had been rinsed from them, as there was still some sample remaining. They were left to air dry, then weighed, then wiped to remove excess remains and finally weighed again. Two blanks were prepared.

Hydrogen peroxide

As oxidation is an important part of the digestion, the importance of hydrogen peroxide was also investigated for the wheat flour sample. 25 mg of wheat flour was added to the bombs using the paste sample preparation technique described in the previous example. The same amount of internal standard and HNO3 were also used, but 3 mL H2O2 and 2.55 mL of MQ- water.

11 3.2.6. Method evaluation

ANOVA

For method evaluation, analysis of variance (ANOVA) was used at a 95% confidence level for the changing parameter experiments. One-way ANOVA was used when investigating the acid concentration, sample amount, wheat flour sample preparation and the importance of hydrogen peroxide while two-way ANOVA was performed when looking at the effect of temperature program at different sample sizes and acid concentration at different sample sizes.

LOD, LOQ

Limit of detection (LOD) and limit of quantification (LOQ) were also calculated to evaluate the method. In total, seven blanks were prepared with an acid concentration of 2.5%. The standard deviation (SD) of the blanks at every wavelength was calculated. LOD was then calculated by multiplying each SD value with three, and LOQ by multiplying SD with ten, see Equation 6 and Equation 7.

𝐿𝑂𝐷 = 𝑆𝐷 ∗ 3 (Eq. 6)

𝐿𝑂𝑄 = 𝑆𝐷 ∗ 10 (Eq. 7)

Thus, LOD is the lowest signal possible to distinguish from the blanks, i.e. a signal most positive not being a part of the signal to noise. More noise therefor contributes to larger LOD values. When studying the LOD and LOQ, the element concentrations in sample calculated from the calibration curves were used for each element at the wavelengths chosen after how well the calculated concentrations corresponded with the given reference concentrations.

RCC

To see how effective the digestions were, the residual carbon content (RCC) for bovine liver and wheat flour were also investigated. The RCC was roughly estimated by looking at two carbon wavelengths, C 193.091 and C 247.856. As the two wavelengths gave about the same result, one of them were chosen for the evaluation, namely C 193.091. The carbon signals were corrected from the internal standard before evaluation. To see whether the ICP-AES only measured the carbon in carbon chains or also the carbon in CO2, one sample from the digestion with a larger amount of H2O2 was degassed with N2 for about ten minutes. It was then compared with one of the other replicates from the same experiment.

4. Results and discussion

4.1. Sample reagents and standards

All results obtained from ICP-AES were calculated to the element concentration in the sample, i.e. they were all calculated from the background corrected intensities and corrected for the intensities from the internal standard. When plotting the results into diagrams, the mean value and the relative standard deviation (RSD) of the replicates were calculated.

As, in some cases, only two replicates were done, the mean value of the two measurements were calculated to obtain a third value. This to make the ANOVA calculations possible when three replicates were evaluated together with two.

12 4.1.1. Calibration standards

One calibration curve for each wavelength was created. The emission lines with the best linearity over the calibrated area were used. For examples of calibration curves, see Appendix 2.

4.1.2. Start-up

After the microwave digestions, almost all samples were clear, and no colour could be observed in the Falcon tubes, with the only exception being the citrus Leaves. One of the citrus leaves’ replicates had a weak yellow colour, and two of the replicates seemed to have some particles in them and an orange ring of dirt left in the bombs. Whether the orange rings appeared from this experiment or a previous one is not clear as citrus leaves should be quite easy to digest, and the bombs were not controlled before the experiment took its start.

From the calibration curves the element concentration of the eight reference samples were calculated and analysed. The concentrations were calculated from the linear equations to see what emission line gave the values closest to the reference sample values. Bone muscle does not have a single of its concentrations within the range of CI, with all but one value for manganese being below. Two of the concentrations for pig kidney are inside of CI, with rest of them being too low. Bovine liver, wheat flour and tuna fish have nine, seven and eight concentrations within CI, respectively. Bovine liver has one value above and wheat flour four, with the rest of the concentrations for the three samples being below CI. For bone meal and citrus leaves, three values are inside CI and for cabbage one, with rest of the calculated concentrations being below. For the calculated concentration and their RSD values, see Appendix 3.

When looking at LOD and LOQ, all samples have concentrations above them for magnesium, calcium, phosphorous and sulphur. The manganese concentrations for cabbage and citrus leaves are below LOQ and the rest of the samples have concentrations below LOD. Pig kidney and bovine liver are above LOQ for iron and all other samples have concentrations below LOQ, with wheat flour also being below LOD. All samples have element concentrations below LOQ except for bovine liver for copper and zinc, and wheat flour, tuna fish, cabbage and citrus leaves are below LOQ. The rest of the samples have concentrations also below LOD, see Table 4.

13

Table 4: Display of elements below LOD (red), below LOQ (orange) and above LOQ (green). The grey cells represent that the sample does not have a certified reference of that element.

Mg Ca Mn Fe Cu Zn P S

Bovine

muscle LOD LOQ LOQ LOD

Pig

kidney LOD Above LOQ LOD

Bovine

liver Above Above LOD Above Above LOD Above Above

Wheat

flour Above Above LOD LOD LOQ LOQ Above Above

Tuna

fish Above LOD LOQ LOQ LOQ

Bone

mMeal Above Above LOQ LOD Above

Cabbage Above Above LOQ LOQ LOQ LOQ

Citrus

leaves Above Above LOQ LOQ LOQ LOQ Above Above

4.1.3. Changing parameters

From the start-up experiment, it was found that the P-line at 213.618 nm overlapped with another element. After running the samples, it seemed like the wrong peak was chosen and it was therefore removed from further analysis. As bovine liver and wheat flour contained many of the same trace elements and had quite many calculated concentrations within CI, they were chosen for further analysis.

From the calculations of the mean concentrations, one wavelength for each element was chosen from how many values it had within the given CI. The chosen wavelengths were Mg at 279.553 nm, Ca at 317.933 nm, Mn at 259.373 nm, Fe at 238.204 nm, Cu at 224.700 nm, Zn at 213.856 nm, P at 178.287 nm and S at 182.034 nm. Some of the wavelengths did correspond very well with the given reference values, while other did not. Overall, the calculated element concentration for bovine liver better corresponded with the given reference concentrations than the ones for wheat flour, which had few values within CI and large RSD values, see Appendix 4 and Appendix 5.

Acid concentration

When looking at the solutions from when 25 mg of sample were digested with an acid concentration ranging from 2.5%, 5%, 10% to concentrated acid, the solutions all looked clear at first sight. A closer look at them, the solutions containing bovine liver did have a barely visible hint of yellow in the solutions containing 2.5% and 5% HNO3. The very weak hint of yellow did not differ between the two acid concentrations though, if any, the colour seemed less visible from the digestion with an acid concentration at 5%. The bovine liver solutions with 25 mg sample and 10% acid were transparent and no colour was visible. For the wheat flour, the 2.5%, 5% and 10% seemed completely clear. For the samples digested in concentrated acid, both the wheat flour and bovine liver solutions might have had a vague yellow colour, which seems a bit strange.

14

Before the digestion of 25 mg sample in 2.5% and 5% acid, it was discovered that two of the bombs had orange rings in them, with one of the bombs being the same that had an orange ring after the start-up experiment. The bombs were used as blanks to see if the remains affected the analysis. The blank values from the bombs were very low, and when compared with the other two blanks, the ring did not seem to have any effect on the result. It is hard to tell when the rings arose as it was early on in the experiments, and due to inexperience with the equipment, the bombs were probably not controlled as they should have been.

For the bovine liver, many of the calculated concentrations are inside of the CIs. Of 20 wavelengths, 14 of the calculated concentrations are inside of the CI when using 2.5% acid.

When 5%, 10% and concentrated acid were used, 14, 7 and 12 of the calculated concentrations are inside the CI range, respectively. Most of the values outside CI are being too low, something that could indicate that digestions were incomplete. Except for one Ca concentration for digestion in 10% acid, Zn at 206.191 nm for 5% and 10% is the only element that has a concentration higher than the given reference values, see Appendix 4.

When comparing the results from the calculated element concentrations at different acid concentrations with 25 mg of bovine liver, it is hard to see any clear trends as they seem to differ from element to element. For magnesium and calcium, the element concentration seems to increase with increasing acid concentration, while for manganese and iron the trend might be decreasing and zinc, phosphorous and sulphur show a clear decreasing trend. For copper, no clear trend is visible, see Figure 1. That the concentrations for P and S reduces with a higher acid concentration is surprising as more acid present should be able to drive the digestion further. If the digestion was complete with lower amount of acid, then no change in the calculated concentrations should be observed with higher acid concentration. If the digestion was not complete with the lower acid concentration, the calculated element concentrations should instead increase with higher acid concentrations, as for the case with Mg and Ca.

Figure 1: The concentrations of trace elements in 25 mg bovine liver at different acid concentrations and the RSDs of the replicates given as error bars. Three replicates were made for the different acid concentrations except for the digestion in concentrated acid where only two replicates were made. The concentrations and RSDs are normalized from the solution with 2.5%

HNO3 to make a comparison possible.

0 0,2 0,4 0,6 0,8 1 1,2

Mg 279.553 Ca 317.933 Mn 259.373 Fe 238.204 Cu 224.700 Zn 213.856 P 177.495 S 182.034

Normalized concentration

Bovine liver: Acid concentration, 25 mg

2.5% 5% 10% Conc.

15

When doing a statistical evaluation with ANOVA, calcium and iron do not show a significant difference, while all the other elements do, see Appendix 6. This might be a bit surprising, since all the calculated sample concentrations for manganese were inside the CI, but manganese also has quite big RSD values, especially for the samples digested in concentrated acids.

The 25 mg samples with wheat flour, digested in the four different acid concentrations, are harder to analyse since the values have a wider spread and hence the RSDs are a lot bigger, see Figure 2. It looks like there is one trend present though, that the concentrations for zinc, phosphorus and sulphur seem to decrease with increasing acid concentration.

Figure 2: The concentrations of trace elements in 25 mg wheat flour at different acid concentrations and the RSDs of the replicates given as error bars. Three replicates were made for the different acid concentrations except for the digestion in concentrated acid where only two replicates were made. The concentrations and RSDs are normalized from the solution with 2.5%

HNO3 to make a comparison possible.

A lot fewer values are found within the range of the CIs, with, at some wavelengths, the spread results in values both below and above. For calcium, manganese and iron it appears that there are no existing trends. This is also confirmed by the ANOVA calculations showing no significant differences, see Appendix 7. The ANOVA calculations do show significant differences for the other elements. For magnesium, zinc, phosphor and sulphur the calculated concentrations seem to decrease with increasing acid concentration, in contrary to what is expected. It is hard to do any evaluation of copper since the calculations for the concentrated acid digestion really stands out. Two of the three Cu wavelength gave much higher concentrations than expected, with one of them being 224.700. This could be because of the dilution prior to the ICP-AES analysis give analyte concentrations far below the detection limit. Another striking feature is that many of the calculated element concentrations from the digestions carried out in 5% acid are a bit higher. This might be an effect from the calibration standards being prepared in 10% HNO3 and no correction from a 5% control standard was made.

The bovine liver samples digested in 2.5%, 5% and 10% HNO3 all have concentrations above LOQ for all elements except for manganese and zinc. Manganese has concentrations below

0 0,5 1 1,5 2 2,5 3

Mg 279.553 Ca 317.933 Mn 259.373 Fe 238.204 Cu 224.700 Zn 213.856 P 177.495 S 182.034

Normalized concentration

Wheat flour: Acid concentration, 25 mg

2.5% 5% 10% Conc.

16

LOD, and zinc has for 2.5% and 5% concentrations below LOQ and for the 10% digestion concentrations below LOD. For the sample digested in concentrated acid, only the concentrations for phosphorous were above LOQ. Calcium and sulphur have concentrations below LOQ and the rest of the elements have concentrations below LOD, see Table 5.

Table 5: Display of elements below LOD (red), below LOQ (orange) and above LOQ (green) for 25 mg of bovine liver digested with different acid concentrations.

Bovine

liver Mg Ca Mn Fe Cu Zn P S

2.5% Above Above LOD Above Above LOQ Above Above

5% Above Above LOD Above Above LOQ Above Above

10% Above Above LOD Above Above LOD Above Above

Conc. LOD LOQ LOD LOD LOD LOD Above LOQ

The samples of wheat flour digested in concentrated acid have concentrations below LOD for all elements but for calcium and phosphorous, where they are below LOQ. Many elements also have concentrations below LOD for the 2.5%, 5% and 10% digestions of wheat flour.

Copper and zinc are below LOQ for the 10% acid digestion, but manganese, calcium, phosphorous and sulphur all have concentrations above LOQ, see Table 6.

Table 6: Display of elements below LOD (red), below LOQ (orange) and above LOQ (green) for 25 mg of wheat flour digested with different acid concentrations.

Wheat

flour Mg Ca Mn Fe Cu Zn P S

2.5% Above Above LOD LOD LOD LOD Above Above

5% Above Above LOD LOD LOD LOD Above Above

10% Above Above LOD LOD LOQ LOQ Above Above

Conc. LOD LOQ LOD LOD LOD LOD LOQ LOD

When comparing the results of using 2.5% and concentrated acid for digestion of a larger amount of sample, the 2.5% solutions containing 50 mg of bovine liver sample hade a vague yet distinct yellow colour, and the wheat flour solutions with the same sample size were completely transparent. The 100 mg bovine liver solution had a clear yellow colour and the solution containing 100 mg of wheat flour had a slight touch of yellow. For the concentrated solutions, the one with 50 mg bovine liver had minimal colour and the one with 100 mg a weak yellow colour while the wheat flour 50 mg samples also had minimal colour and the 100 mg ones might have had a slight shade of yellow. The yellow colour indicates that the digestions were not entirely complete, and the calculated element concentrations could therefore also be lower.

After the digestion with larger samples and different acid concentrations, orange rings were discovered in the bombs where digestion with bovine liver had taken place at an acid concentration of 2.5%. As the bombs had been cleaned prior to the experiment, the ring most likely arose from this trial.

Many of the calculated element concentrations for 50 mg and 100 mg bovine liver digested in 2.5% and concentrated acid are again within the range of CI, with some of the Zn values above and the rest of the values below, see Appendix 4. Some of the trends observed when digesting 25 mg of bovine liver in different acid concentrations are visible when digesting a

17

larger amount of sample as well; the concentrations for magnesium appear to increase with increasing acid concentration, and decrease with increasing acid concentrations for iron, zinc, phosphorous and sulphur. There also seem to be a slight increase in element concentration for calcium with increasing acid concentration. For manganese the trend shows the opposite, for the larger sample sizes, the element concentration increases with increasing acid concentration, see Figure 3.

Figure 3: The concentrations of trace elements in 50 mg and 100 mg bovine liver at different acid concentrations and the RSDs of the replicates given as error bars. Three replicates were made for the digestions in 2.5% HNO3 and two replicates were made for the digestions in concentrated acid. The concentrations and RSDs are normalized from the solution containing 50 mg sample with 2.5% HNO3 to make a comparison possible.

For method evaluation two-way ANOVA was used, both with respect to acid concentration and sample size. For magnesium, the differences are significant, looking at acid concentration, sample size and the correlation between them. For iron, zinc and phosphorous the differences are significant when looking both at acid concentration and sample size but there is no significant correlation between them. For calcium, manganese and sulphur there are no significant differences for the different sample sizes, but for the different acid concentrations. There are no significant correlations for those three elements between acid concentration and sample size. Compared to the other elements, copper has a significant correlation between acid concentration and sample size. There is also a significant difference with respect to acid concentration, but not for sample size.

For the larger sample amounts of bovine liver digested in 2.5% acid, most element concentrations are above LOQ. The only element that is not is manganese where the 100 mg sample is below LOQ and the 50 mg below LOD. For the digestions in concentrated acid, the phosphorous and sulphur concentrations are above LOQ for both sample sizes, with magnesium also being above LOQ for 100 mg. Calcium is below LOQ for 50 mg, and the rest of the elements for 50 mg are below LOD. Calcium, iron and copper are below LOQ for the 100 mg of sample and manganese and zinc are below LOD, see Table 7.

0 0,2 0,4 0,6 0,8 1 1,2

Mg 279.553 Ca 317.933 Mn 259.373 Fe 238.204 Cu 224.700 Zn 213.856 P 177.495 S 182.034

Normalized concentration

Bovine liver: Acid concentration, 50 mg and 100 mg

2.5%, 50 mg 2.5%, 100 mg Conc., 50 mg Conc., 100 mg

18

Table 7: Display of elements below LOD (red), below LOQ (orange) and above LOQ (green) for 50 mg and 100 mg of bovine liver digested with different acid concentrations.

Bovine liver Mg Ca Mn Fe Cu Zn P S

50 mg 2.5% Above Above LOD Above Above Above Above Above

100 mg 2.5% Above Above LOQ Above Above Above Above Above

50 mg conc. LOD LOQ LOD LOD LOD LOD Above Above

100 mg conc. Above LOQ LOD LOQ LOQ LOD Above Above

Once again, the analyse of the wheat flour samples is more difficult. And, once again, the element concentrations of samples digested in 5% HNO3 does show a trend of being higher than expected. The 2.5% acid digestions with 100 mg of sample does not have a single calculated element concentration within the CI range, see Appendix 5, instead all the concentrations are higher than the CI range. When the same sample sizes instead were digested in concentrated acid, only one value is found within CI while most of the others are found below, or even far below. The results of 25 mg of wheat flour digested in 2.5% acid show that for half of the wavelengths, the element concentrations are within CI, while those outside of CI are found to be both above and below. The 50 mg sample digested in concentrated acid differs a lot depending on element, with copper again giving way higher values than probable. As the values from digestion in 5% acid are higher than expected, the results are widely spread, giving large RSD values, and does not seem coherent, it is hard to tell the accuracy of the trends. But it appears that for all elements, the concentration decreases with increasing acid concentration, see Figure 4. The result does seem odd as the presence of more acid should make it possible to digest a larger amount of sample.

Figure 4: The concentrations of trace elements in 50 mg and 100 mg wheat flour at different acid concentrations and the RSDs of the replicates given as error bars. Three replicates were made for the digestions in 2.5% HNO3 and two replicates were made for the digestions in concentrated acid. The concentrations and RSDs are normalized from the solution containing 50 mg sample with 2.5% HNO3 to make a comparison possible.

The two-way ANOVA evaluation shows that magnesium does not show a significant difference for acid concentration nor sample size, but indicates a significant difference between the two. For calcium and manganese there is no significant difference when looking

0 0,5 1 1,5 2

Mg 279.553 Ca 317.933 Mn 259.373 Fe 238.204 Cu 224.700 Zn 213.856 P 177.495 S 182.034

Normalized concentration

Wheat flour: Acid concentration, 50 mg and 100 mg

2.5%, 50 mg 2.5%, 100 mg Conc., 50 mg Conc., 100 mg

19

at sample size, but significant differences with respect to acid concentration, and also significant correlation between acid concentration and sample size. The ANOVA for iron, copper and sulphur show that there is no significant difference when changing sample size, but a significant difference with digestion in different acid concentrations. None of the three elements has a significant correlation between acid concentration and sample size. Zinc has no significant differences, whether looking at acid concentration, sample size or the correlation between them. For phosphorus it is the opposite, with significant differences with respect to all three parameters, see Appendix 9.

Most concentrations from digestion of larger wheat flour samples are below LOD for manganese, iron, copper and zinc, with manganese for the 100 mg sample digested in 2.5%

being the only exception. Magnesium, calcium, phosphorous and sulphur all have concentrations above LOQ. The magnesium concentrations for both samples in concentrated acid are below LOD, the phosphorous concentrations above LOQ and the sulphur concentrations below LOQ. The 50 mg concentrated sample has calcium concentrations below LOQ and the one with 100 mg above, see Table 8.

Table 8: Display of elements below LOD (red), below LOQ (orange) and above LOQ (green) for 50 mg and 100 mg of wheat flour digested with different acid concentrations.

Wheat flour Mg Ca Mn Fe Cu Zn P S

50 mg 2.5% Above Above LOD LOD LOD LOD Above Above

100 mg 2.5% Above Above LOQ LOD LOD LOD Above Above

50 mg conc. LOD LOQ LOD LOD LOD LOD Above LOQ

100 mg conc. LOD Above LOD LOD LOD LOD Above LOQ

When looking at RCC for the 25 mg samples digested in different acid concentrations, the trend is the same for both bovine liver and wheat flour, the carbon signals are higher for digestions carried out in 10% HNO3 than for those in 2.5% and 5% acid, which have very similar carbon signals. The concentrated acid digestions are a bit harder to read, since there are only two replicates and the blank values are a bit high, but it seems like they are quite similar to the 10%, see Figure 5 and Figure 6.

Figure 5: Residual carbon content (RCC) at different acid concentrations for 25 mg of bovine liver.

Figure 6: Residual carbon content (RCC) at different acid concentrations for 25 mg of wheat flour

Blank 10% Blank 10% 10% 10% 10% Blank 5% Blank 5% 5% 5% 5% Blank 2.5% Blank 2.5% 2.5% 2.5% 2.5% Blank conc. Blank conc. Conc. Conc.

0 0,1 0,2 0,3 0,4 0,5 0,6

0 5 10 15 20

Intensity

Bovine liver: Acid conc., 25 mg

Blank 10% Blank 10% 10% 10% 10% Blank 5% Blank 5% 5% 5% 5% Blank 2.5% Blank 2.5% 2.5% 2.5% 2.5% Blank conc. Blank conc. Conc. Conc.

0 0,05 0,1 0,15 0,2 0,25 0,3 0,35

0 5 10 15 20

Intensity

Wheat flour: Acid conc., 25 mg