Identification of paper and influence of its homogeneity in forensic investigations by ICP-AES/MS and other
non-invasive spectroscopic techniques
Doreen-Marie Shamon
Master’s program in Forensic Science Uppsala University
2012-06-05
Abstract
The purpose with this thesis was to find out if ordinary office paper was homogenous throughout the sheet in context of organic and elemental content. Two different types of papers were also analyzed, paper towel and corrugated paperboard. The techniques used were NIR, FT-IR and Raman spectroscopy for the evaluation of organic contents, and ICP-AES/MS for the measurement of elemental composition and concentration.
Spectrofluorometry was also utilized to establish if the content of optical brightener was homogenous in all paper types. The spectroscopic techniques didn’t require any sample preparation except for cutting the papers in pieces, according to their “geographical”
place in the sheets. The ICP-AES and ICP-MS analyses required sample preparation in form of cutting the pieces and digesting each of them with acid and hydrogen peroxide in digestion bombs. After the digestion the samples were diluted with purified water. The results showed that NIR and spectrofluorometry couldn’t differentiate samples within one sheet of all paper types, although NIR made a distinction between the office paper
samples. FT-IR on the other hand could distinguish between samples in one group from samples belonging to another group further away within same sheet. The elemental concentrations of sample pieces were also significantly different within same sheet of office paper, paper towel and corrugated paperboard. This elemental distinction could be made in both ICP-AES and ICP-MS. The results from Raman spectroscopy were
unusable due to gained broad bands as spectra instead of peaks, the reason for that is high fluorescence. Different laser intensities were used with no change in result.
Sammanfattning
Syftet med arbetet var att undersöka homogeniteten hos papper med avseende på sammansättningen av grundämnen, organiska komponenter samt optiskt vitmedel. Tre olika sorters papper användes A4 papper, hushållspapper och wellpapp. Tekniker som användes för detta syfte var ICP-AES (inductively coupled plasma-atomic emission spectroscopy) och ICP-MS (inductively coupled plasma-mass spectroscopy) för
koncentrationsbestämning av element. NIR (near-infrared spectroscopy) , FT-IR (fourier transform-infrared spectroscopy) och Raman spektroskopi användes för analys av de organiska komponenterna i pappret samt spektrofluorometri för mätning av optiskt vitmedel och andra fluorescerande molekyler i pappret. Provupparbetning krävdes för analys med ICP-AES och ICP-MS. De olika papperstyperna upplöstes med olika sammansättning av syra och väteperoxid i Uhrbergsbomber, efter upphettning kyldes proverna till rumstemperatur och späddes med milli-Q vatten. Samma prover som
användes för ICP-AES analyserna späddes med thallium internstandard innan analys med ICP-MS. De andra spektroskopiska teknikerna krävde ingen provupparbetning, annat än urklippning av pappersbitar som skulle passa provbehållaren. Resultaten visade att ICP- AES/MS kunde diskriminera olika element i olika koncentrationer genom ett helt ark.
Detta gällde för alla papperstyper. FT-IR var också diskriminernande för vissa prover inom alla papperssorter, medan resultaten från NIR och spektrofluorometri analyserna inte kunde urskilja olika prover inom ett och samma ark. Detta tyder på att dessa
instrument inte kan uppvisa någon inhomogenitet hos pappret med avseende på organisk sammansättning respektive innehållet av optiskt vitmedel.
Table of contents
Section Page
1. Introduction 2
2. Theory 4
3. Materials and methods 8
4. Results 11
5. Discussion 47
6. Conclusion 50
7. Future aspects 50
Abbreviations
ANOVA Analysis of variance
FT-IR Fourier-transform infrared spectroscopy
ICP-AES Inductively coupled plasma- Atomic emission spectroscopy ICP-MS Inductively coupled plasma-Mass spectrometer
NIR Near-infrared spectroscopy
PCA Principal component analysis
1. Introduction
Discrimination of different paper documents have become important due to many crimes committed with forged documents like wills, checks etc hence its importance to correlate, for example one piece of paper from crime scene to the rest of the sheet found elsewhere.
The most common techniques used for this task is ICP-MS because of its ability to identify and measure the concentration of different elements at the same time.
Different analyzing techniques have also been studied in this thesis to investigate the homogeneity throughout the same sheet of paper of different kinds of fiber containing materials like office papers, paper towels and corrugated paperboard. Techniques used are ICP-MS (inductively coupled plasma-mass spectrometry), ICP-AES (inductively coupled plasma- atomic emission spectroscopy), FTIR (Fourier transform infrared spectroscopy), NIR (near-infrared spectroscopy), spectrofluorometry and Raman
spectroscopy. These techniques measures different kinds of analytes throughout the paper sheets; FT-IR, Raman spectroscopy and NIR mainly reflects the organic contents in paper and ICP-AES and ICP-MS measures elemental composition and concentration. The IR spectroscopic techniques are non-invasive; therefore the samples can be stored and reanalyzed again at different occasions.
The main purpose of this forensic investigation, was to study the homogeneity within a paper and how this result could affect real crime cases involving paper found at crime scene and piece of paper found with suspect. This is important because if it shows that two pieces of paper originating from the same sheet don’t have same the same elemental composition or concentration (statistically proven different) through the whole paper then one can not be able to link the evidence from crime scene with the suspect. The statistical tool used for this comparison of different groups of samples within one paper sheets is ANOVA (analysis of variance).
Previous researches done in the area have already proven that ICP-MS can distinguish between different brands of paper and even between different reams, from the same brand by elemental concentration and composition [1]. It makes sense that different brands of paper have different elemental concentration due to different raw material sources used in the manufacturing process of paper, but in this paper the main purpose was to investigate the homogeneity within the same sheet of paper and how small pieces of paper could be used and still be able to connect the piece of paper with the rest of the sheet.
The authors in the article selected two different vendors of paper and took five reams from each brand. From every ream three sheets were chosen for analysis, taken from the top, bottom and in the middle of the reams. For every sheet five samples were taken one from each corner and one from the middle of the paper each piece with a weight of 0.029 ٕ ± 0.001g. The area of each piece was 23 *18mm. The samples were put in a quartz tube were 1.5 mL of nitric acid and 0.75 mL of hydrogen peroxide were pipetted.
The quartz tube was set into a Teflon vessel, which was filled with 11 mL water and 1
to 210°C for fifteen minutes and then hold for yet ten minutes before cooling. The samples were analyzed with ICP-MS. The results obtained showed that the papers could be discriminated into different vendors, different batches or reams from the same vendor but not between the three sheets within same ream, or within a single sheet. Different elements were discriminating between the two vendors and the reams; overall the elements that were more discriminating were Al, Ba, Sr, Mg, Mn and Ce.
In another article they chose seventeen different brands of paper (five sheets from each brand) for analysis; the papers were cut from the outer edge of each side of the paper [2].
The sample sizes were cut 3*4 cm with an approximate weight of 0.1 to 0.11 g. The specimens were each added 3 mL of sub-boiled nitric acid and 1.5 mL hydrogen peroxide, after digestion the samples were diluted to 40mL water with purified water.
The results from this study showed that different brands of paper can be discriminated by elemental composition and concentration. In this case nine elements were discriminative:
Na, Mg, Al, Mn, Sr, Y, Ba, La, and Ce. Different batches from the same manufacturer could also be distinguished by three elements Al, Zr and Mn.
According to an article about classification of papers, they could successfully distinguish eight different brands of paper by FT-IR analysis in ATR mode; eight samples were cut from each brand [3]. The spectra ranges chosen for the analysis were divided in two sets 4000-2000 and 2000-650 cm-1 but the whole spectra were taken into account for the evaluation of the results with PCA. The most discriminating region with PCA was at 4000-2000 cm-1 because this is where kaolin absorbs strongly, approximately around 3700 cm-1. Organic compounds like cellulose, hemicellulose, lignin and inorganic fillers like kaolin absorb in the infrared region, thus FT-IR is also an important technique for discrimination of different kinds of paper [4]. Raman spectroscopy discriminate also papers by their content of inorganic fillers.
Another study showed that nineteen different brands of papers could be differentiated by FT-IR [5]. Five samples from every sheet were cut for the analysis, a total of five sheets from every brand were sampled, and each side of the paper was measured. The spectra range chosen for the analysis was between 4000-650 cm-1. The papers could be
distinguished by cellulose and calcium carbonate (CaCO3) peaks in the spectrum.
Calcium carbonate is added to the paper as inorganic filler to control the characteristics of paper. Cellulose and calcium carbonate were detected in the spectra range 750-1550 cm-1. NIR is a technique commonly used in pulp and paper industries to determine the organic content and quality of the wood and pulp. The organic contents to be controlled in the industries are cellulose, lignin, pentose and the total pulp yield [6-7].
2.1 Theory
2.2 Paper manufacturing
Paper is made by firewood and consists of cellulose and hemicellulose. These fibers are held together by lignin which is the “natures glue” for the fibers. All of these organic compounds consist of carbon, hydrogen and oxygen atoms. It is not unusual that the paper also contains added kaolin as filler to improve the characteristics of paper, for instance brightness and texture after the paper pulp process [8]. Paper also contains inorganic elemental compounds because of the trees used for production of papers absorbs elements from the ground water [1].
There are two ways of producing papers either by mechanic or by chemical release of the fibers from the wood into paper pulp [9]. The method chosen depends on the type of paper to be manufactured. Mechanical release of the fibers usually consists of grinding the wood into pulp. Papers produced only by this method are darker, thinner and less strong, like for example magazine papers. Due to the lignin content the paper becomes yellowish after a while. Office papers, which are brighter and have a better printing surface are manufactured using the chemical pulping process. The wood chips are boiled in a chemical solution (either acidic or basic) under high pressure. Depending on which solution is chosen for paper pulp manufacturing; either basic or acidic, the methods are divided into sulphate and sulphite mass. When it comes to the sulphate method, sodium hydroxide and sodium sulphide solutions are added. For the sulphite method either magnesium or sodium bisulphite solution with pH 4 is added. Production of, for instance corrugated paperboard uses a semichemical method using neutral sodium sulphite solution. The chemical boiling time is shorter and the mass is chopped afterwards.
The paper pulps are thereafter dried and bleached. If the mechanical method is used, the pulp is bleached with hydrogen peroxide on the other hand if the chemical method is used the lignin is removed as much as possible. Before entering the paper machine the fibers are mixed with water to create furnish. The furnish is applied to a plastic sheeting to remove most of the water before the fibers are pressed to form paper and then further dried see figure1.
Figure 1: Picture of a paper machine, with a wet section, press section, drying section and a paper roll up section [9].
2.3 ICP-MS/ICP-AES
Inductively coupled plasma (ICP) is a technique applied in analytical chemistry to analyze metals in liquid samples [10]. The sample is introduced to the torch by a nebulizer that breaks the liquid into a fine mist of droplets, aerosol. Aerosol created is dried out due to the heat inside the torch, see figure 2. Before the sample reaches the torch, aerosol is passed through a spray chamber which first separates smaller droplets from the larger ones. Only the smaller droplets continue to the torch.
A tesla coil inside the torch gives rise to argon ions Ar+ and free electrons. The heat from the argon gas atomizes the molecules and the atoms are ionized in plasma before
introduction to interface. The ions pass through a mass filter before reaching the detector an electron multiplier. The mass filter in ICP-MS is a mass spectrometer; it separates the ions by their mass to charge ratio m/z [11]. The separation of ions occurs between four metal rods on which direct and alternating current are applied on each pair of rods. By choosing a certain AC/DC current on the rods only a specific m/z will pass through the mass filter and reach the electron multiplier. All the other ions will strike against the rods and will not be detected.
Figure 2: An overview of the ICP-MS instrument [11].
When using atomic emission spectroscopy (AES) as a detection method, the electrons in the atoms are excited to a higher energy level by the plasma, and when the electrons returns to ground state, energy is released in forms of photons. The detector, a
photomultiplier detects the emission of light released. Different elements releases energy at different wavelengths and so the elements are identified and detected. The grade of intensity indicates the concentration of the element in the sample.
When comparing the two techniques; ICP-MS is more sensitive than ICP-AES. It can detect concentration levels down to parts per trillion (ppt) whereas ICP-AES can detect
parts per billion (ppb). ICP-MS only detects ions with selected m/z, and leaves out all the rest, by doing so the noise signal is diminished.
2.4 Raman spectroscopy, FT-IR and NIR
Raman spectroscopy and Fourier transform infrared spectroscopy (FT-IR) works in a similar way, the sample is irradiated with mid-infrared laser, for Raman even near- infrared and visible light is used [12]. The sample molecule absorbs the transferred energy, and excites to a higher energy level. The molecule starts to vibrate and rotate in the same oscillating frequency as the photon that stroke it. When the molecule has a dipole moment it is IR active but Raman inactive. The IR spectra plot shows the ratio of absorbed or emitted energy from the molecule versus the wavenumber (cm-1). This relationship is proportional to the vibrational energy difference between the ground state and the excited energy level.
Samples analyzed with Raman spectroscopy must be polarizable; it means that the electron cloud surrounding the molecule will be changed when an outer electrical field is applied with two oppositely charged plates. It induces a dipole moment, due to the electrical field, see figure 3. Raman spectral plots give the same results as FT-IR spectra but reversed. Unlike other spectroscopic techniques Raman measures inelastic light scattering (Stokes scattering, how much of the energy is lost) for quantification.
Sometimes the energy is gained during irradiation and detects anti-stokes light scattering.
FT-IR and Raman spectroscopy complements each other due to FT-IR measures asymmetrical and polar analytes while Raman spectroscopy measures symmetrical bondings of non-polar analytes.
Figure 3: Induced dipole moment with Raman spectroscopy. [12]
Near-infrared spectroscopy measures overtones and combination of vibrational
transitions of the molecule irradiated with IR laser. NIR spectroscopy gives information about the atoms in the molecule like FT-IR and Raman spectroscopy. This is known as”
the near-infrared region whereas FT-IR and Raman spectroscopy are both measured in the mid-IR region.
2.5 Spectrofluorometry
When a sample is irradiated with either visible or ultraviolet (UV) light, the molecules in the samples absorbs the light and excites from ground state to a higher energy level [10].
The molecule is then in an unstable state and strives towards stability. This is when relaxation occurs which means that the molecule falls back to the ground state, when it does it emits light i.e. fluorescence. Fluorescence is measured between 500-700 nm. This spectroscopic technique works for analysis of for example optical brighteners which emits fluorescence light.
2.6 PCA and ANOVA
In this experimental work the purpose was to establish if paper samples taken near each other were statistically different from other samples at a longer distance within the same paper. Different techniques used as mentioned above give multi-variate responses from the sample analyses. The variables can be evaluated with PCA (principal component analysis) calculated with the UNSCRAMBLER software. PCA score plots are used when the numbers of variables are too extended to be evaluated in one chart [13]. New
variables (principal components) are created from the old variables and their coefficients (loadings). The coefficients maintains as much of the information from the variables as possible. Each sample result consists of score values for every component. The variation between samples is larger in the first principal component and declines with increasing number on the principal component. The scores are then plotted pairwise against each other in a scatter plot. Samples similar in composition tend to be near each other while samples different from each other are separated in the score plot. With a loading plot it is possible to find out which variables that are contributing to the differences.
However PCA gives only an indication visually if the samples are different from each other or not. To be certain of the conclusions made from the analysis a statistical calculation called ANOVA (Analysis of variance) is performed, either on the score values or the original variables. This mathematic tool compares the variances calculated for between and within groups of samples [14]. The comparison is implemented by dividing the between and within values with each other and hence forming the F value (Fishers value). The null hypothesis is that there is no true difference between groups, the F value should then not be larger than 1. If the F value is greater than the critical value or the P value (probability value) is below the significance level, in this thesis at 5% or 0,05 the null hypothesis is rejected and there is significant difference between the groups.
3. Materials and methods 3.1 Materials
Digestion bombs were used to decompose the papers into the solution before ICP analysis. The samples were diluted in pyrex tubes. Ceramic knife (Satake) and ceramic scissors (Kyocera) were used to cut the papers before digestion. Three different types of papers were used Multicopy office papers 80g/m2 (Stora enso), paper towels and
corrugated paperboard.
3.2 Chemicals
65% sub-boiled nitric acid (Merck), 30% hydrogen peroxide (JT Baker), 95-97%
sulphuric acid (Merck) and milli-Q purified water were the solutions used for sample digestion and dilution for the wet chemical techniques.
3.3 Softwares, instruments and settings
The UNSCRAMBLER and MINITAB were the softwares used for statistical calculations.
3.3.1 NIR
NIR inframatic 8620 from Percon was used for the paper analyses. Twenty filters were used and each filter measured one wavelength. The wavelengths were 2345, 2336, 2310, 2270, 2230, 2208, 2190, 2180, 2139, 2100, 2050, 1982, 1940, 1818, 1778, 1759, 1734, 1722, 1680 and 1445 nm.
3.3.2 FT-IR
FT-IR Spectrum 2000 PerkinElmer instrument was used for substances reflecting infrared spectra. The spectra were measured in the range 5000-500 cm-1 and the resolution was 2.0. The interval between each measurement was 0.5 cm-1.
3.3.3 ICP-AES
The instrument used for paper analysis was ICP-AES Spectro Ciros CCD, PerkinElmer.
The gases used in ICP-AES were nebulizer gas with a flow of 0.9 l/min, auxiliary gas with 0.9 l/min flow and coolant gas with a flow of 14 l/min. The ICP RF power was 1400W.
3.3.4 ICP-MS
ICP-MS NexION 300D from PerkinElmer was used for the elemental determination in papers. The gases and their respective flows used for ICP-MS analysis were nebulizer gas (1 l/min), auxiliary gas (1 l/min) and coolant gas (15 l/min). The ICP RF power was set to1600W.
3.3.5 Spectrofluorometry
Spectrofluorometer FP-750 from JASCO was used for measurement of fluorescence emission. The sample molecules in the papers were excited at 350 nm. The fluorescence emission was measured between 390-700 nm.
3.3.6 Raman spectroscopy
Paper samples were analyzed with Reinshaw ramascope; Leica DMLM . Sample molecules were excited with 514 nm laser. The time of sample exposure was 10s and each sample was measured once. Different laser powers were used 100, 50, 25, 10 and 1%. The spectra were collected in the range 100-3200 cm-1.
3.4 Sampling and sample preparation ICP-AES
For each paper type blanks were made containing only respective acid and hydrogen peroxide diluted with milli-Q water. Same parameters were used for both samples and blanks. Standard solution was made containing twenty two elements at 1ppm, with an exception for calcium at 50 ppm due to high concentrations in the samples. The elements measured were Al, Mn, Mg, Sr, Li, Na, K, Rb, Ca, Ba, Sc, Y, Ti, V, Cr, Fe, Co, Ni, B, Si, P and Ce.
3.4.1 Office paper
One sheet of white office paper was cut in squares 2*2cm with ceramic knife. A total of 20 pieces were cut from the paper, four taken from each corner and four from the center of the paper. These papers were weighted before sample preparation. Each sample was put in separate pyrex tube were 2mL nitric acid and 1mL of hydrogen peroxide was added before closing the digestion bomb. The heating program was set to 125°C
(29.3min) with accelerating temperature 10°C/min to a final temperature of 165°C for 90.
1 min. After cooling the samples were diluted to 10 mL with milli Q purified water.
3.4.2 Paper towels
The paper towels were cut with ceramic scissors to 4*4 cm to obtain approximately the same weight as for the office papers. A total of twenty samples were analyzed, four from each corner of the paper and four from the center of the paper. In each pyrex tube 3mL of nitric acid and 2mL of hydrogen peroxide were added before heating.
The temperature program was changed to 125°C (45min) as a start temperature and a final temperature of 168°C for 130 min. After cooling the samples they were diluted to a final volume of 10 mL with milli Q purified water.
3.4.3 Corrugated paperboard
The sample size of corrugated paper board was 2*2 cm. Altogether 16 samples were cut with ceramic knife, four from each outer edge of the corrugated paper. To each sample 3mL HNO3 and 2mL H2SO4 was added in a pyrex glass. The Pyrex glasses were heated in a heating block with a total effect of 638W for two hours. After the digestion the samples were diluted to 25mL with milli-Q water.
3.5 Sampling and sample preparation ICP-MS
Paper and paper towel samples made in section 3.4 were reanalyzed with ICP-MS.
Before the analysis the samples were diluted ten times (1mL to 10mL) with an thallium internal standard (10 ppb, 0.1% HNO3). When it comes to the corrugated paper from section 3.4.3 the samples were diluted up to 25mL, because of the higher elemental concentrations in the latter paper.
Standard solutions used were at 0, 5, 10, 20 and 30 ppb of the chosen elements; the standard solutions were also diluted with 10 ppb thallium solution. Thallium was chosen because natural materials rarely contain any thallium and will not already be present in the paper samples.
3.6 Sampling and sample preparation FT-IR analysis
3.6.1 Office paper, paper towels and corrugated paperboard
Samples of twenty pieces were cut altogether, four from each corner and four from the middle of the paper. Each paper was cut 1*1 cm to fit the sample holder of FT-IR. The spectra range was 5000-500cm-1. These are a new set of samples coming from another office paper sheet than the one used for ICP-AES and ICP-MS analysis. The sheet used for FT-IR analysis comes from the same ream as for the samples analyzed with the ICP instruments.
3.7 Sampling and sample preparation spectrofluorometry
Twenty new pieces of all paper brands were cut with a ceramic knife with the size 1.2*2 cm and analyzed with the spectrofluorometer. Excitation wavelength was set at 350 nm.
Emission of light was measured in the range 390-700 nm.
3.8 Sampling and sample preparation NIR
All the paper types were cut altogether so office paper gave 24 samples, paper towels 20 samples and corrugated paperboard 24 samples. The size of each sample was 4.5*4.5cm to fit the sample holder, because of the larger size on the sample holder new set of twenty pieces had to be cut from scratch from a new office paper. The papers were analyzed for twenty filters and twenty wavelengths.
3.9 Sampling and sample preparation Raman spectroscopy
The same sample pieces cut in section 3.6 for the FT-IR analysis were also used for Raman spectroscopy analysis. Before analysis the instrument was calibrated with a silica plate. The silica peak was corrected to 521nm if it was offset.
4. Results
4.1 Office paper 4.1.1 NIR analysis
One-way ANOVA was calculated from the results obtained from office paper analysis with groups according to figure 4. The results showed a significant group difference only for PC1 scores (P=0.002). Another possible reason for significant differences besides the samples alone is systematic errors. The NIR instrument tends to show different results of the same sample in the beginning and end of an analysis. This was avoided by
randomized order of analysis.
The papers were cut and the samples divided into columns and rows to recognize their position on the paper, see figure 4. The division makes it possible to identify differences or similarities between samples throughout the paper. The similarities or differences found can depend on the distance between the samples.
A1 B1 C1 D1 E1 F1
A2 B2 C2 D2 E2 F2
A3 B3 C3 D3 E3 F3
A4 B4 C4 D4 E4 F4
Figure 4: Division of the office paper in groups of four samples.
The score plot for the principal components one and two showed a division between samples from row one and two and row three and four, see figure 5. Figure 6 is showing an example of a NIR spectrum for sample A1.
Figure 5: Score plot of office paper samples, analyzed with NIR spectroscopy.
A1 spectrum
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
1200 1400 1600 1800 2000 2200 2400 2600
Wavelenght (nm)
Reflected intensity
Figure 6: NIR spectrum of office paper sample A1.
4.1.2 FT-IR analysis
The paper samples were analyzed on both the front and backside. Figure 7 shows that there are differences for PC2 scores between these two for the same sample. Figure 8 shows the positions where the samples were taken from the paper.
For further data evaluation the front side and backside samples were separated in the
other. Samples from column A and B are more situated on the right side of the chart, while samples from column N and O are situated on the left side on the score plot. G and H are located somewhere in between N and B. This indicates difference along the long side of the paper, see figure 9. The corresponding score plot for the backside samples shows a similar distribution of the samples as for the front side samples, see figure 10.
The A and B samples are in one group and N, O, G and H samples are grouped together in the PCA plot.
ANOVA calculations were made for principal components 1 and 2 for both front and backside of the paper. The spectrum range for these calculations was at 3710-500 cm-1. ANOVA showed significant difference for the first component of both front and backside of the office paper. All samples were divided into groups before the statistical
calculations, see table 1 below.
The difference were found between group A and B and the rest of the groups i.e. group A and B were separated from group C, D and E.
Figure 7: Score plot showing the results from FT-IR analysis of front side and backside of the same office paper. Spectrum range: 4000-500cm-1.
A1 B1 N1 O1
A2 B2 N2 O2
G1 H1 G2 H2
A10 B10 N10 O10
A11 B11 N11 O11
Figure 8: Sites on the paper from which the samples were taken.
Figure 9: Score plot of front side office paper samples, analyzed with FT-IR spectroscopy. The spectra range was set at 3710-500 cm-1.
Figure 10: Score plot of backside office paper samples, analyzed with FT-IR spectroscopy. The spectra range was set at 3710-500 cm-1.
Sample Group
A1 A
A2 A
B1 A
B2 A
A10 B
A11 B
B10 B
B11 B
G1 C
G2 C
H1 C
H2 C
N1 D
N2 D
O1 D
O2 D
N10 E
N11 E
O10 E
O11 E
Table 1: Office paper samples divided into groups before ANOVA calculation of the FT-IR results.
In figure 11 below, a spectrum of sample A1 is shown. The spectra range covers from 5000-500 cm-1.
A1 spectrum frontside
0 20 40 60 80 100 120
0 1000 2000 3000 4000 5000 6000
Wavnumber cm-1
Reflected intensity
Figure 11: FT-IR spectrum for sample A1. Front side of the office paper illustrated.
4.1.3 ICP-AES
The division of the samples within the paper is shown in figure 12. ICP-AES results of the paper samples are shown in the PCA score below, in figure 13. The responses from ICP-AES for each sample were recalculated to µg/g elemental content in one gram of paper, see appendix A. There seem to be a random distribution of the samples rather than an arranged one. Samples from different columns and rows are grouped together, except for sample A11, B11 and O11, these samples seem to belong to the same group, they were all taken from the last row in the office paper. In figure 14 the different elements contributing to the sample differences are illustrated.
One-way ANOVA was calculated on the responses, the results showed that there were significant differences between the different groups of paper samples. The discriminative elements were Na, K, Sr, Ti and V. All the elements except Sr showed a discrimination of sample A10, B10, A11 and B11 from all the rest. While Sr showed that sample A1, A2, B1, B2, N1, N2, O1 and O2 were more similar to each other from the rest of the samples.
A1 B1 N1 O1
A2 B2 N2 O2
G5 H5 G6 H6
A10 B10 N10 O10
A11 B11 N11 O11
Figure 12: Paper positions chosen for analysis with ICP-AES. Four samples taken from each corner and four samples selected approximately from the middle of the sheet, i.e. a total of five groups
Figure 13: PCA score plot of office paper analyzed with ICP-AES.
Figure 14: PCA loading plot showing the elemental contribution of the scores.
Figures 15-16 shows the concentrations of the discriminative elements in µg per one gram of the weighted paper before the sample digestion.
Elemental concentrations
-100 0 100 200 300 400 500
A1 B1 A2 B2 A10 B10 A11 B11 G5 H5 G6 H6 N1 O1 N2 O2 N10 O10 N11 O11
Sample
Concentration (µg/g)
Na 589.592 Na 588.995 K 766.491 Sr 421.552 V 311.071
Figure 15: Bar chart showing concentration (µg/g) of the discriminating elements in office paper, analyzed with ICP-AES.
Elemental concentration
0 1 2 3 4 5 6 7 8
A1 B1 A2 B2 A10 B10 A11 B11 G5 H5 G6 H6 N1 O1 N2 O2 N10 O10 N11 O11 Sample
Concentration (µg/g)
Ti 336.121
Figure 16: Bar chart showing the elemental concentration (µg/g) of Ti in office paper sheet, analyzed with ICP-AES.
4.1.4 ICP-MS
The elements, which gave too low signals to give a reliable result in ICP-AES or were not included from the beginning, were also analyzed with ICP-MS. These elements were Ce, Sc, La, Mn, V, W, Zr, Rb, Ti, Mo, Zn, Pb and Cu. After the ANOVA calculation only V, Rb and Ti were discriminative and gave a significant difference depending on the spreading of the paper. The significant difference lied in the samples A10, A11, B10 and B11. ANOVA gave no significant difference between principal component 1 and 2.
PCA score plot of all samples is shown in figure 17, the contribution of the elements are shown in figure 18. The plot shows no certain pattern for the samples; they all seem to be randomly distributed. The elemental concentrations were recalculated to ng/g per sample see appendix A. The concentrations of the discriminative elements are shown in figure 19-21. The concentrations of Ti and Rb are higher in sample A10, A11 and B11.
Figure 17: PCA score plot of office paper samples analyzed with ICP-MS. No order between the samples are shown.
Figure 18: PCA loading plot of the elements contributing to sample scores.
Elemental concentration
0 1000 2000 3000 4000 5000 6000 7000 8000
A1 A2 A10 A11 B1 B2 B10 B11 G5 G6 H5 H6 N1
N2 N10 N11 O1 O2 O10 O11 Sample
Concentration (ng/g)
Ti
Figure 19: Concentration of discriminative element Ti (ng/g), analyzed with ICP-MS.
Elemental concentration
0 100 200 300 400 500 600 700
A1
A2 A10
A11 B1
B2 B10
B11 G5 G6 H5 H6 N1
N2 N10
N11 O1 O2 O10 O11 Sample
Concentration (ng/g)
V
Figure 20: Concentration of the discriminative element V (ng/g), analyzed with ICP-MS.
Elemental concentration
0 10 20 30 40 50 60
A1 A2 A10
A11 B1 B2 B10
B11 G5 G6 H5 H6 N1
N2 N10
N11 O1 O2 O10 Sample
Concentration (ng/g)
Rb
Figure 21: Concentration of the discriminative element Rb (ng/g), analyzed with ICP-MS. Sample O11 was excluded from the figure, due to negative result.
4.1.5 Spectrofluorometer
Each paper sample was analyzed twice, including front and backside with a spectrofluorometer to observe if the content of optical brighteners were different throughout the sheet. The samples were taken from specific positions in the paper. The samples being closer to each other are considered to be in one group i.e. there are five groups with four samples in each. These groups are located in the four corners and in the middle of the paper sheet, see figure 22.
Figure 23 shows no pattern between samples taken close or further away from each other, the same resolution applies even with the frontside and backside are separated in the chart, see figure 24-25. Frontside and backside of the same samples gave different results as the first PCA score plot exhibits. ANOVA calculations on both PC1 and PC2 gave no significant difference between groups.
A1 B1 N1 O1
A2 B2 N2 O2
G1 H1 G2 H2
A10 B10 N10 O10
A11 B11 N11 O11
Figure 23: PCA score plot of office paper analyzed with spectrofluorometer. Frontside and backside of paper measured.
Figure 24: Score plot of office paper analyzed with spectrofluorometer. Frontside measured.
Figure 25: Score plot of office paper analyzed with spectrofluorometer. Backside measured.
The fluorescence spectrum of sample A1 is shown in figure 26. The peak is high which implies a high concentration of optical brighteners in the paper.
A1 spectrum
0 50 100 150 200 250 300 350 400 450
200 300 400 500 600 700 800
Wavelenght (nm) Fluorescence emission (intensity)
Figure 26: Sample A1 spectrum of office paper analyzed with spectrofluorometer.
4.2 Paper towels 4.2.1 NIR analysis
The paper towels were cut altogether with 20 samples divided into five columns and four rows pointing out their position on the paper, see figure 27.
A1 B1 C1 D1 E1
A2 B2 C2 D2 E2
A3 B3 C3 D3 E3
A4 B4 C4 D4 E4
Figure 27: Positions were samples are cut from the paper towel.
The score plot in figure 28 shows a vague separation between samples of row 1 and 2 located at the left side of the chart and 3 and 4 located on the right side. The exceptions are the samples D2, A3 and C3, which are situated on opposite sides within the chart. The one-way ANOVA calculated showed no statistical difference between the groups of samples. Figure 29 illustrates an example of a NIR spectrum the sample chosen is A1.
Figure 28: Score plot of paper towel samples analyzed with NIR.
A1 spectrum
0 0.1 0.2 0.3 0.4 0.5 0.6
1200 1400 1600 1800 2000 2200 2400 2600
Wavelenght (nm)
Reflected intensity
Figure 29: NIR spectrum of paper towel sample A1.
4.2.2 FT-IR analysis
Samples in the paper towel had a named position in the paper before the analysis, see figure 30. The results of the analysis of paper towel show differences between the front and backside. The samples illustrated in the score plot are further away from each other while they should be at the same spot, see figure 31. The range of the spectrum was cut down to 4000-500cm-1 from the original 5000-500cm-1, to remove excess information that were the same in all samples. As the figure 31 shows the samples taken from one side of the paper; from column A and B are situated in the right side of the plot while the rest are positioned at the left side of the plot.
Two more PCA plots were done for front and backside of the paper each to interpret the samples better, figure 32-33. ANOVA calculations showed significant difference for principal component 1 for both front and backside of the paper. The samples closer to each other in the paper are as usual in the same group; this gives five groups with four samples in each. The groups are situated in the four corners and in the middle of the paper. The differences were in sample group A and B which involves samples A1, A2, B1, B2 in group A and samples A4, A5, B4 and B5 in group B.
A1 B1 G1 H1
A2 B2 G2 H2
D3 E3
A4 B4 D4 E4 G4 H4
A5 B5 G5 H5
Figure 30: Different sample positions on paper towel before analysis with FT-IR.
Figure 31: PCA plot of paper towel samples analyzed with FT-IR. Spectral range: 4000-500cm-1. Both front and backside measured. Sample A5F (frontside of sample A5) and B4B (backside of sample B4) were removed because of suspected outliers.
Figure 32: PCA plot of paper towel samples analyzed with FT-IR. Spectral range; 4000-500cm-1. Only frontside measured.
Figure 33: PCA plot of paper towel samples analyzed with FT-IR. Spectral range: 4000-500cm-1. Only backside measured. Sample B4 removed because of suspected outlier.
The FT-IR spectrum of sample A1 is seen in figure 34. The spectra range covers from 5000-500 cm-1.
Figure 34: FT-IR spectrum of sample A1. Frontside of the paper towel illustrated.
4.2.3 ICP-AES
The nineteen samples chosen for ICP analysis were selected from the four corners and approximately from the middle of the sheet, to be able to view if there are similarities between samples taken near each other, see figure 35. As the results imply the samples taken closer to each other in the paper towel sheet are spread from each other in the PCA score plot, see figure 36. Same sample were analyzed several times to make sure that same results were obtained everytime. Figure 37 shows the contribution from different elements in the loading plot.
ANOVA calculations show significant difference only for Ce. The element was discriminative for the middle samples on the sheet. The concentration of Ce in every sample is illustrated in figure 38. Elemental concentrations calculated into µg/g per sample are seen in appendix A.
A1 B1 F1 G1
A2 B2 F2 G2
D3
A4 B4 D4 E4 F4 G4
A5 B5 F5 G5
Figure 35: Figure showing the paper towel samples approximate location before ICP-AES analysis.
A1 spectrum frontside
0,00 10,00 20,00 30,00 40,00 50,00 60,00 70,00 80,00 90,00
0 1000 2000 3000 4000 5000 6000
Wavenumber cm-1
Reflected intensity
Figure 36: PCA score plot of paper towel analysis with ICP-AES. Sample G2 was excluded because of suspected outlier. Samples recalculated to µg/g (µg element per g of paper).
Figure 37: PCA loading plot of paper towel analysis with ICP-AES. Showing the elemental contribution to the scores in figure 39.
Ce elemental concentration
0 0.5 1 1.5 2 2.5 3
A1 A2 A4 A5 B1 B2 B4 B5 D3 D4 E4 F1 F2 F4 F5 G1 G2 G4 G5 Sample
Concentration (µg/g)
Ce 448,691
Figure 38:Concentration (µg/g) of Ce in paper towel samples analyzed with ICP-AES.
4.2.4 ICP-MS
The sectioning of paper towel samples analyzed with ICP-MS are the same as for ICP- AES. ANOVA calculations done for the samples showed significant difference in the elements: Ce, Sc, La, Mn, V, and Cu. The differences were between the lower left corner (sample A4, A5, B4, and B5) and the rest of samples. PCA score plot done for all
samples also shows this difference, see figure 39. Sample A4, A5, B4, and B5 are separated from all other samples in the score plot. Sample D3 is a possible outlier. The elemental positions are plotted in figure 40.
The concentrations of the discriminative elements in each paper sample are shown in figure 41-42. The concentrations of Mn and Cu are low in sample A4, A5, B4, B5 and D3 while the concentrations of Ce, Sc, La and V are high in these samples. Elemental
concentrations for all samples are shown in appendix A.
Figure 39: PCA score plot of paper towel analysis with ICP-MS. Samples recalculated to µg/g (µg element per g of paper).
Figure 40: The PCA loading plot shows the elemental positions contributing to differences between samples.
Elemental concentrations
-2000 0 2000 4000 6000 8000 10000
A1 A2 B1 B2 A4 A5 B4 B5 D3 D4 E4 F1 F2 G1 G2 F4 F5 G4 G5 Sample
Concentration (ng/g)
Mn Cu
Figure 41: Elemental concentrations of Mn and Cu in paper towel analyzed with ICP-MS.
Elemental concentrations
0 1000 2000 3000 4000 5000
A1 A2 B1 B2 A4 A5 B4 B5 D3 D4 E4 F1 F2 G1 G2 F4 F5 G4 G5 Sample
Concentration (ng/g)
Ce Sc La V
Figure 42: Elemental concentration of Ce, Sc, La and V in paper towel analyzed with ICP-MS.
4.2.5 Spectrofluorometer
The paper towel samples analyzed with spectrofluorometer were taken from the four corners and the middle of the sheet, see figure 43 below.
A1 B1 G1 H1
A2 B2 G2 H2
D3 E3
A4 B4 D4 E4 G4 H4
A5 B5 G5 H5
Figure 43: Division of paper towel samples for spectrofluorometer. Showing five groups with four samples in each.
The score results are shown in figure 44. Samples positioned near each other in the sheet are spread from each other in the score plot, but there tend to be a vague pattern between samples taken from row 1-3 and with 4-5 exceptions for the samples E3, B4 and D4. The statistic ANOVA calculations done for the principal components showed no significant difference between the samples. A spectrum for sample A1 is shown in figure 45. The fluorescence peak is small due to the low content of optical brighteners in paper towel.
Figure 44: Score plot for paper towel samples analyzed with spectrofluorometer.
A1 spectrum
0 50 100 150 200 250 300 350 400
200 300 400 500 600 700 800
Wavelenght (nm) Fluorescence emission (intensity)
Figure 45: Paper towel sample A1 spectrum analyzed with spectrofluorometer.
4.3 Corrugated paperboard 4.3.1 NIR analysis
The whole corrugated paperboard was cut and the samples were labeled according to their positions on the paperboard, see figure 46. All pieces were cut because the sample holder in NIR is much larger than the other techniques used. Figure 47 shows that the samples are distributed with no similarities between samples from the same column or row. This visually determined result is also statistically proven with one-way ANOVA for each principal component and general linear model ANOVA for all the original variables from NIR analysis. Both models prove no significant difference between samples at different distances in the paper.
On occasion, the NIR instrument drifts; therefore same sample measured four times at the beginning of the analysis was reanalyzed four times in the end of analysis to ensure same results each time. Figure 48 demonstrates differences between these two sets of replicates along the PC2 axis, which mean that there are drifts in NIR spectroscopy during analysis.
Figure 49 demonstrates a spectrum for corrugated paperboard. The sample used is A1.
A1 B1 C1 D1 E1 F1
A2 B2 C2 D2 E2 F2
A3 B3 C3 D3 E3 F3
A4 B4 C4 D4 E4 F4
Figure 46: Positions were all the samples were taken from corrugated paperboard before analysis with NIR spectroscopy.
Figure 47: Score plot of corrugated paperboard samples analyzed with NIR spectroscopy. The samples A3 and B4 were removed from the calculations because of suspected outliers.
Figure 48: All samples analyzed with NIR spectroscopy. Sample D2 was analyzed several times.
Replicate 1-4 were analyzed in the beginning while replicate 5-8 were analyzed at the end of analysis.
A1 spectrum
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
1200 1400 1600 1800 2000 2200 2400 2600
Wavelenght (nm)
Reflected intensity
Figure 49: NIR spectrum of corrugated paperboard sample A1.
4.3.2 FT-IR analysis
Corrugated paperboard samples were only analyzed on one side because of the inconsistency of the backside of the paper. The samples were analyzed on the smooth surface i.e. frontside. Figure 50 shows were the samples originated from the paperboard.
The results in score plot 51-52 points out that samples taken from the left side of the paper distinguish from the rest. Samples taken from column A and B are situated at the right side in the chart while the rest are situated more at the left side.
This interpretation was proved with ANOVA calculations of principal component 1 and 2. The results from ANOVA confirmed significant difference for PC1 and samples from column A and B were distinguished from the other samples.
Samples from group A (A1, A2, B1, and B2) distinguished from all the other groups remarkably. Group B (A9, A10, B9, and B10) also distinguished from the other groups on the right side of the sheet, it was more similar to group A.
A1 B1 M1 N1
A2 B2 M2 N2
G5 H5 G6 H6
A9 B9 M9 N9
A10 B10 M10 N10
Figure 50: Positions of the sample pieces on corrugated paperboard before FT-IR analysis. There were five groups with four samples in each.
Figure 51: Corrugated paperboard samples analyzed with FT-IR at spectral range 4000-500cm-1. With all samples included.
Figure 52: Corrugated paperboard samples analyzed with FT-IR at spectral range 4000-500cm-1. Samples
The FT-IR spectrum of sample A1 is shown in figure 53. The spectrum range covers 5000-500 cm-1. The area from 4000-500 cm-1 is similar to the office paper and paper towel.
A1 spectrum
0 10 20 30 40 50 60
0 1000 2000 3000 4000 5000 6000
Wavenumber cm-1
Reflected intensity
Figure 53: FT-IR spectrum of corrugated paperboard sample A1.
4.3.3 ICP-AES
Previous attempts to find a good digestion procedure for corrugated paperboard left no paperboard to take samples from the four corners and in the middle, as with the other papers analyzed. After a digestion procedure for paperboard was found the samples were picked out from the four long edges of the paperboard instead, see figure 54. Figure 55 and 57 shows that the samples taken near each other have no correlation and are spread in the score plots. Samples within one group of paper seem to be more comparable with samples from another group at a longer distance in the corrugated paperboard. The loading plots for the scores are shown in figure 56 and 58.
ANOVA calculation showed that the elements discriminating the samples were Al and Co. The main difference of Al found, was in sample N5, N6, M5 and M6. This group of samples was significantly different from the others. While sample G9, G10, H9 and H10 were significantly different from the other groups when considering the element Co. The elemental concentrations among the samples are shown in figure 59-60. An extended list for all elemental concentrations is attached in appendix A.
G1 H1 G2 H2
A5 B5 M5 N5
A6 B6 M6 N6
G9 H9 G10 H10
Figure 54: Sample locations from corrugated paperboard, for analysis with ICP-AES.
Figure 55: PCA score plot on corrugated paperboard, analyzed with ICP-AES. The results were recalculated to µg/g (µg element per g of paper). All samples included.
Figure 56: Loading plot for the elements contributing to the score positions.
Figure 57: Score plot on corrugated paperboard, analyzed with ICP-AES. The results were recalculated to µg/g (µg element per g of paper). Sample N6 excluded because of suspected outlier.
Figure 58: Loading plot showing elements contributing to the score plot.
Elemental concentration
0 1000 2000 3000 4000 5000 6000
A5 A6 B5 B6 G1
G2 H1 H2
G9 G10 H9 H10 N5 N6 M5 M6 Sample
Concentration (µg/g)
Al 167,078
Figure 59: Al concentration of corrugated paperboard samples analyzed with ICP-AES.
Elemental concentration
3.5 3.6 3.7 3.8 3.9 4 4.1 4.2 4.3
A5 A6 B5 B6 G1 G2 H1 H2
G9 G10 H9 H10 N5 N6 M5 M6 Sample
Concentration (µg/g)
Co 238,892
Figure 60: Co concentration of corrugated paperboard samples analyzed with ICP-AES.
4.3.4 ICP-MS
Samples used for the ICP-AES analyses from section 4.3.3 were also measured by ICP- MS. First the samples were diluted ten times with milli-Q water with an addition of thallium. Thallium is used as internal standard for the samples. The distribution of the samples is shown in figure 61. The plot shows similarities between sample A5, A5, B5, B6 and G1, G2, H1 and H2. It also showed similarities between sample G9, G10, H9, H10 and M5, M6, N5 and N6; sample M6 being an outlier was excluded from the evaluation. The corresponding loading plot for the scores is shown in figure 62.
According to the ANOVA calculations the discriminating element was Mo. The calculations showed significant difference. One group containing the samples M5, M6, N5 and N6 was distinguished from all the other groups of samples. The concentration of Mo for all analyzed samples is shown in figure 63. A list of elemental concentrations for all samples is included in appendix A.
ANOVA calculated for principal component 1 and 2 didn’t prove significant difference.
Usually more differences between samples are found in the first principal components.
Figure 61: PCA plot for corrugated paperboard analyzed with ICP-MS. Sample M6 being an outlier was excluded.
Figure 62: Elemental position for the scores on corrugated paperboard, analyzed with ICP-MS. Sample was M6 excluded, because it was an outlier.
Elemental concentration
0 500 1000 1500 2000 2500 3000
A5 A6 B5 B6
G1 G2
G9 G10 H1 H2 H9
H10 M5 M6 N5 N6 Sample
Concentration (ng/g)
Mo
Figure 63: Concentration of Mo in corrugated paperboard analyzed with ICP-MS.
4.3.5 Spectrofluorometer
The sample positions on the corrugated paperboard are illustrated in figure 64. Samples taken from the same group are not situated near each other in the PCA plot showing dissimilarity in composition of optical brighteners, see figure 65. One can visually see that there is no difference between groups, which also could be verified with ANOVA calculation on the principal components 1 and 2. A spectrum of sample A1 is shown in figure 66.
C1 D1 C2 D2
A1 B1 E1 F1
A2 B2 E2 F2
G1 H1 G2 H2
Figure 64: Sample positions of corrugated paperboard before analysis with spectrofluorometer.
Figure 65: PCA plot on corrugated paperboard analyzed with spectrofluorometer. Samples B1 and C2 excluded because these samples were outliers.
A1 spectrum
0 50 100 150 200 250 300
200 300 400 500 600 700 800
Wavelenght (nm) Fluorescence emission (intensity)
Figure 66: Spectrum of corrugated paperboard sample A1, analyzed with spectrofluorometer.
