@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix skos: . vivo:departmentOrSchool "Applied Science, Faculty of"@en, "Chemical and Biological Engineering, Department of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Motiee, Sima"@en ; dcterms:issued "2013-04-13T09:12:04Z"@en, "2013"@en ; vivo:relatedDegree "Master of Applied Science - MASc"@en ; ns0:degreeGrantor "University of British Columbia"@en ; dcterms:description """It is well known that paper properties such as opacity, printing quality, brightness, and roughness are affected by the z-direction mineral filler distribution. In this study, the effect of four different parameters (filler, starch, chemical retention aids and the machine parameter (vacuum)), on the z-direction filler distribution in paper samples was investigated. Paper samples were made using an apparatus that simulates a suction box. Different levels of filler (PCC), chemical additives and vacuum were chosen based on a central composite design. The effect of these parameters on filler distribution in cross-section of paper samples was investigated by using Scanning Electron Microscopy (SEM) and Energy Dispersive X-ray analysis (EDX) which is a non-destructive technique. The area covering the thickness of the paper was divided into five layers and each layer was subjected to EDX analysis to obtain the calcium content through the paper thickness. The results indicated that the filler distribution in the z-direction of paper samples increased from the top side to the wire side. The calcium content in the layer near the wire side was significantly higher than in the other layers. The other layers had similar calcium content. The chemical additives as well as vacuum in suction box had a significant effect on the filler distribution. Increasing the filler and starch led to higher level of filler content in all layers through the sheet thickness. Increasing the vacuum and retention aids led to higher level of filler content in layers near the wire side. The effect of high charge density starch used in our study was significantly higher than the effect of retention aids in retaining the filler through the thickness of the paper. Based on the obtained data, a set of empirical models were constructed that can predict the filler distribution through the thickness of the paper. We validated our results by measuring the calcium content through applying image processing technique on the SEM cross-section images and measuring the ash content of paper samples. The results of both of these approaches confirmed the EDX measurements."""@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/44198?expand=metadata"@en ; skos:note """     EFFECT OF CHEMICAL ADDITIVES ON Z-DIRECTION FILLER DISTRIBUTION IN PAPER by Sima Motiee B.Sc., Amirkabir University of Technology, 2010 A THESIS SUBMITTED IN PARTIAL FULF,ILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in The Faculty of Graduate Studies (Chemical and Biological Engineering) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2013 © Sima Motiee, 2013   ii   Abstract It is well known that paper properties such as opacity, printing quality, brightness, and roughness are affected by the z-direction mineral filler distribution. In this study, the effect of four different parameters (filler, starch, chemical retention aids and the machine parameter (vacuum)), on the z-direction filler distribution in paper samples was investigated. Paper samples were made using an apparatus that simulates a suction box. Different levels of filler (PCC), chemical additives and vacuum were chosen based on a central composite design. The effect of these parameters on filler distribution in cross-section of paper samples was investigated by using Scanning Electron Microscopy (SEM) and Energy Dispersive X-ray analysis (EDX) which is a non-destructive technique. The area covering the thickness of the paper was divided into five layers and each layer was subjected to EDX analysis to obtain the calcium content through the paper thickness. The results indicated that the filler distribution in the z-direction of paper samples increased from the top side to the wire side. The calcium content in the layer near the wire side was significantly higher than in the other layers. The other layers had similar calcium content. The chemical additives as well as vacuum in suction box had a significant effect on the filler distribution. Increasing the filler and starch led to higher level of filler content in all layers through the sheet thickness. Increasing the vacuum and retention aids led to higher level of filler content in layers near the wire side. The effect of high charge density starch used in our study was significantly higher than the effect of retention aids in retaining the filler through the thickness of the paper. Based on the obtained data, a set of empirical models were constructed that can predict the filler distribution through the thickness of the paper. We validated our results by measuring the calcium content through applying image processing technique on the SEM cross-section images and measuring the ash content of paper samples. The results of both of these approaches confirmed the EDX measurements.   iii   Table of Contents Abstract  ....................................................................................................................................................  ii   Table of Contents  ..................................................................................................................................  iii   List of Tables  ............................................................................................................................................  v   List of Figures  ........................................................................................................................................  vi   Acknowledgements  .............................................................................................................................  viii   Dedication  ...............................................................................................................................................  ix   Chapter 1:   Review and Background  ..............................................................................................  1   1.1   Papermaking  .............................................................................................................................................  1   1.2   Using suction boxes in the forming section  ........................................................................................  3   1.3   Use of filler in the paper  .........................................................................................................................  5   1.3.1   Filler types  .............................................................................................................................................................  6   1.4   Starch  .........................................................................................................................................................  9   1.5   Retention and chemical retention aids  .............................................................................................  11   1.5.1   Chemical retention aids  ..................................................................................................................................  11   1.5.2   Interactions between dewatering, retention and formation  ................................................................  12   1.6   Z-direction filler distribution  .............................................................................................................  13   1.7   Measuring the Z- profile of paper  .....................................................................................................  16   1.7.1   Scanning electron microscope (SEM)  .......................................................................................................  17   1.7.2   Energy dispersive X-ray (EDX)  ..................................................................................................................  19   1.8   Objective of the study  ..........................................................................................................................  20   1.9   Organization of the thesis  ...................................................................................................................  21   Chapter 2:   Materials and Methodology  .......................................................................................  23   2.1   Experimental design  .............................................................................................................................  23   2.2   Handsheet preparation  ........................................................................................................................  27   2.2.1   Materials  ..............................................................................................................................................................  28   2.2.2   Starch cooking  ...................................................................................................................................................  29   2.3   The SEM/EDX technique  ....................................................................................................................  30   2.3.1   Sampling procedure  .........................................................................................................................................  30   2.4   Validating the experimental results  ..................................................................................................  33   2.4.1   Ashing  ..................................................................................................................................................................  33   2.4.2   SEM/ Image processing technique  .............................................................................................................  35   Chapter 3:   Results and Discussion  ................................................................................................  41   3.1   Replication of experiments at the central point  .............................................................................  41   3.2   Effect of experimental parameters on filler distribution  .............................................................  42   3.2.1   Effect of PCC  ....................................................................................................................................................  45   3.2.2   Effect of starch  ..................................................................................................................................................  47   3.2.3   Effect of retention aids  ...................................................................................................................................  49         iv   3.2.4   Effect of vacuum  ..............................................................................................................................................  51   3.3   Filler distribution profile  ....................................................................................................................  53   3.3.1   Effect of layer  ....................................................................................................................................................  53   3.3.2   Effect of experimental conditions  ...............................................................................................................  54   3.4   Validating the EDX measurements by ashing technique  .............................................................  55   3.5   Image processing  ...................................................................................................................................  57   3.6   Model construction and validation  ...................................................................................................  61   Chapter 4:   Conclusions and Recommendations  ........................................................................  63   References  ..............................................................................................................................................  65   Appendix A: Image processing code  ................................................................................................  73   Appendix B: 2D convolution filter  ...................................................................................................  75   Appendix C: ANOVA results for filler distribution  .....................................................................  77     v   List of Tables Table 2-1: Levels of the factors for experimental design  ............................................................................................  25   Table 2-2: Full experimental design  ...................................................................................................................................  25   Table 2-3: Concentration of materials in pulp suspension  ..........................................................................................  29   Table 2-4: SEM/EDX test conditions  .................................................................................................................................  30   Table 3-1: Standard error of mean for central runs  .......................................................................................................  42   Table 3-2: Effect of experiment parameters on filler distribution  ............................................................................  43   Table 3-3: Experiment conditions with least and most event filler distribution profile  ....................................  55   Table 3-4: The model equations to predict the filler concentration in paper layers  ...........................................  61   Table 3-5: Level of experiment factor for three compared samples  ........................................................................  62     vi   List of Figures Figure 1-1: Diagram of Fourdrinier papermaking machine [1]  ....................................................................................  2   Figure 1-2: Water drainage process in the suction box [92]  ..........................................................................................  5   Figure 1-3: Global paper filler consumption in 2004 (Adapted from [43])  .............................................................  7   Figure 1-4: Scalenohedral precipitated calcium carbonate  ............................................................................................  8   Figure 1-5: Linear amylose chain CH [20]  ..........................................................................................................................  9   Figure 1-6: Branched amylopectin molecule [20]  ............................................................................................................  9   Figure 1-7: Paper main directions (Adopted from [89])  ..............................................................................................  13   Figure 1-8: A summary of signals that will be detected when the primary beam is directed to SEM sample  ...............................................................................................................................................................................................  19   Figure 1-9: EDX process  ........................................................................................................................................................  20   Figure 2-1: Schematic of the handsheet former (Adapted from [54])  .....................................................................  28   Figure 2-2: Sampling procedure for SEM/EDX analysis, (a) handsheet, showing the four specimens, (b) SEM image showing in the z-direction of the paper specimen, and (c) SEM image showing the five areas for EDX analysis in the entire thickness of paper specimen.  ................................................................  32   Figure 2-3: Elemental composition of the viewing area determined by EDX  ......................................................  33   Figure 2-4: X-ray diffraction analysis of ash sample  ....................................................................................................  34   Figure 2-5: (a) SEM image of cross sectional of paper analyzed by EDX, (b) the extended area analyzed using image processing.  ................................................................................................................................................  35   Figure 2-6: (a) Sample image, (b) The same image after applying convolution-based filter for the purpose of edge detection.  ............................................................................................................................................................  37   Figure 2-7: (a) SEM image of cross sectional of paper (b) SEM image of cross sectional of paper after applying convolution filter.  ..........................................................................................................................................  38   Figure 2-8: (a) Image histogram of SEM image of cross sectional of paper (b) Image histogram of SEM image of cross sectional of paper after applying convolution filter (equation (3)).  ..................................  39   Figure 2-9: (a) SEM image (b) Converted image after applying convolution filter to black and white image.  ..................................................................................................................................................................................  40   Figure 3-2: Prediction profiles for filler distribution through the thickness of the paper  .................................  44   Figure 3-3: Effect of increasing PCC on filler concentration  ....................................................................................  46         vii   Figure 3-4: Effect of increasing starch on filler concentration  ..................................................................................  48   Figure 3-5: Effect of increasing retention aids on filler concentration  ...................................................................  50   Figure 3-6: Effect of increasing vacuum on filler concentration  ..............................................................................  52   Figure 3-7: Calcium content of paper samples measured by ashing and EDX  ....................................................  56   Figure 3-8: Relation between EDX calcium and ash percent  ....................................................................................  56   Figure 3-9: (a) SEM image of cross sectional of paper (b) Split cross sectional images (c) SEM images of split cross sectional of paper after applying convolution filter (d) Converted images after applying convolution filter to black and white image  ...........................................................................................................  57   Figure 3-10: Comparison of EDX and image processing results for smaller areas  ............................................  58   Figure 3-11: Correlation analysis between the data of EDX and Image processing for smaller areas  ........  59   Figure 3-12: Comparison of EDX for smaller areas and image processing results for larger areas  .............  60   Figure 3-13: Correlation analysis between the data of EDX (smaller areas) and Image processing for larger areas  .........................................................................................................................................................................  60   Figure 3-14: Experimental and predicated values for z-direction filler distribution  ..........................................  62    viii   Acknowledgements Foremost, I would like to extend my sincere gratitude to my supervisor Dr. Peter Englezos for his support, guidance and orientation. I thank him for being always patient, understanding and encouraging in times of new ideas and difficulties. He has been a steady influence throughout my master career. I have been very privileged to get to know and to collaborate with many other great people who became my friends over the past two years. I am grateful to my friends in our research group, Pulp and Paper Center and Material Engineering Department for making my graduate life a wonderful learning experience. I would like to thank the technicians without whose help I could not have completed this study. Thanks to George Soong of the UBC Pulp and Paper Centre and Jacob Kabel of the Materials Engineering Microscopy Lab for their technical support. I would like to express my deepest appreciation to my mother, my example of faith and decision. She has been a constant source of love and support and encouragement in every aspect of my life. Finally, I am really grateful to my sister, Sara. I owe a lot to her. She has been always caring, supportive and encouraging in my studies and career. I am so lucky to have her in my life.   ix   Dedication To my mother, for her love, endless support and encouragement. 1     Chapter 1: Review and Background In the following section, the related work, background information and the objectives for our study are explained. 1.1 Papermaking Paper is one of the most amazing and important inventions of all time. Papermaking was developed in China in 105 A.D. It then spread through Asia and Northern Africa to Europe in about 1150 A.D. The first modern papermaking machine was patented at the end of the 18th century in France [74]. Although the papermaking process has evolved and improved considerably through the time, the fundamentals have remained the same. To produce a sheet of paper, a dilute aqueous suspension of pulp fibres (typically wood pulp) is created. Then, a wire or mesh (called forming fabric) is used to drain the water and retain the solids. The remaining solids form a sheet, which is squeezed and dried to create the final paper sheet [7][74]. In the modern paper making the goal is to perform this process at the high speed, large capacity, low cost, low energy utilization, low water consumption, and with minimal environmental impact. Moreover, papermakers aim to produce different types of paper (from newsprint to liner board to tissue paper) with high quality and innovative properties. It is evident that the whole process should be designed and monitored carefully to achieve these goals. There are various types of pulp with different physical, optical and chemical properties such as mechanical, chemical and thermo-mechanical pulps. Using each of these types of pulps leads to papers with different properties. For example, mechanical pulps, which are cheaper than their chemical counterpart, produce weaker and less bright papers than chemical pulps [60][74]. In addition to the wood pulp, mineral fillers and various chemicals are added to the pulp suspension to improve the paper properties (performance additives) and facilitate the paper making process (functional additives). Typical chemical additives include pH control, retention aids, flocculants and drainage aids. Common mineral fillers include clay, talc and calcium carbonate [74]. The type, quantity and addition point of these additives, impact the paper formation and properties. The effects of chemical additives have been studied thoroughly in the literature and Lindstrom has provided an overview of these studies in [47]. In sections 1.3, 1.4 EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       2   and 1.5, we will review the effect of some important mineral and chemical additives on the paper making process. It is noted that 10% of paper making cost is due to use of these additives [74]. The modern papermaking machine has five main sections, shown in Figure 1-1 [24]. In the following, each section is introduced briefly: 1- Approach section: The objective of this section is to pump, mix and dilute the pulp suspension to the desired level of consistency for papermaking. The chemicals are added to the pulp suspension in this section and the solution is cleaned and screened before entering to the headbox. Headbox transfers the pulp suspension to the next section of the machine as uniform pressurized jet [74]. At this stage, the solid content of suspension is about 0.5% [88]. Figure 1-1: Diagram of Fourdrinier papermaking machine [1] 2- Dewatering section (Forming section): In this section the water of pulp suspension drains and the remaining solids form the paper. The solid content of paper increases to 20% after passing the former [18]. The forming section consists of different components such as hydrofoils, table rolls, suction boxes, dandy roll, and couch roll, which enhance the drainage the process. There are different types of forming sections with various number and order of above components. The forming section type impacts the paper properties. The simplest type, which is used in a Fourdrinier machine, has a forming fabric that is circulated under high pressure and the pulp suspension is dispersed onto it. As a result, the water drains and solid materials remain on the forming fabric. Also, some dewatering components such as wet and dry section boxes and foils are used to enhance the drainage [73]. The Fourdrinier former is similar to the former used EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       3   in this research. A more complex faster former is the Twin-wire (gap) former, which has two formic fabrics. By dispersing the pulp suspension between these two fabrics, the water is removed from both sides of the paper. As a result, the drainage speed is increased and in contrast to Fourdrinier formers, the properties of two sides of paper become more symmetric [74]. 3- Pressing section: The pressing section presses the paper between two rolls to remove more water. The excess water is absorbed by a felt that passes between the pulp mat and press rolls. The rolls can be arranged in different shapes to create papers with different properties [24]. The solid content of paper increases to 40-50% at this stage. 4- Drying section: This is the last step of water removal and is an energy intensive section. This section consists of numerous heated rolls, blowers and ventilators that enable the evaporation of the water from the fibre mat. The solid content of paper increases to 90-95% at this point [18]. 5- Finishing section: The finishing section performs functions such as calendering (compress the paper to make it thinner and smoother), finishing (applying chemicals such as starch on the paper surface to increase its strength, opacity, smoothness and brightness), winding (winding around a spool), and cutting to prepare the paper for the end users [24]. Until recently, paper was the main information and communication medium but with the advent of digital and electronic media, the role of paper has been challenged. Therefore, it is important to create high quality papers that provide innovative features for customers. To achieve this goal, papermakers require understanding of the paper structure and its relationship with paper features such as optical, mechanical and print properties. Such understanding improves the design and manufacturing the high quality paper sheets with innovative properties. 1.2 Using suction boxes in the forming section One important step in papermaking is the water drainage from the pulp suspension, which is called dewatering. In Fourdrinier former, when the pulp suspension is in the initial stage of the dewatering section (forming section), it is dilute enough that drainage occurs by gravity. The water drains through the forming fabric, which is a woven web of plastic threads. The weaving pattern of the forming fabric can impact the drainage resistance as well as other forming features such as fibre retention, formation uniformity and minimization of wire mark [65]. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       4   At the forming stage, the solid content of the suspension is at about 0.5%, and the liquid and solid parts of the suspension can move freely. However, after the forming section, as the solid content increases to 5-7%, the drainage resistance increases and gravity cannot drain more water from the suspension [73]. As a result, suction is needed to increase the drainage and consequently the solid contents. There are different methods to increase the drainage such as foils, suction (or vacuum) boxes and the suction boxes on the couch roll. The suction box, shown in Figure 3, is a trough, which is placed below the forming fabric. It has slotted or perforated covers connected to a vacuum system. When the forming fabric passes over the suction box, a suction pulse is applied to the paper mat. As a result a vacuum is created and the pressure difference between the atmosphere and the vacuum system draws water out of the paper into the suction box. By using the suction box, the solid content of the fibre suspension increases to 15% to 20% [73]. The actual increase depends on the pulp properties, suction pressure and suction time. There is usually more than one suction box in a papermaking machine. This provides a pulsing action of applied suction. By gradually increasing the applied pressure in these suction boxes, the paper sheet can be dried more efficiently. The applied pressure on the pulp suspension can be tuned by changing the vacuum pressure in each of the individual suction boxes. The applied pressure is usually between -15kPa and -50kPa [73][24]. However it can be as low as - 65kPa on extremely high-speed paper machines. Moreover, the pulse frequency must be adapted to the suction pressure [63]. For example, in lower suction pressures, the lower pulse frequencies are preferable. Also, for high weight papers and papers with fines and filler materials, the suction time should be increased because in these two cases, the permeability of the wet web is reduced due to the filing between fibres in the fibre network. As a result, the filtration resistance of paper is increased which causes the drainage rate reduction. There are other parameters that affect the suction box performance such as pulp type, freeness, consistency, and temperature [63]. As mentioned above, the suction box is used to remove the water from the pulp suspension. The water drainage in the suction box occurs in the following three steps (shown in Figure 1-2): 1- Compression of the fibre web: When the forming fabric and fibre mat pass over the suction box, the pressure difference compresses the fibre mat. As a result, the water comes out of the fibre mat and flows to the suction box through the forming fabric. The suction box efficiency EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       5   depends on the fibre mat compressibility [63]. 2- Water displacement by airflow: The pressure difference between the atmosphere and the suction box creates airflow through the fibre mat and forming fabric. This airflow displaces the water and moves it to the suction box. In order to achieve this drainage, the applied suction in the suction box should be more than the capillary pressure of the water between the pulp fibres. 3- Rewetting: When the forming fabric and fibre mat leaves the suction box, the applied pressure is removed and the fibre mat expands again. As a result, the water is drawn from the forming fabric to the fibre mat and causes a rewetting. Figure 1-2: Water drainage process in the suction box [92] 1.3 Use of filler in the paper Mineral fillers are functional additives that are used in papermaking. They are used to improve properties of paper mainly opacity and brightness. They can also improve the sheet formation (drainage, retention and permeability) by filling the voids between the fiber matrix, increase surface smoothness and improve the dimensional stability as most fillers remain inert [30][73]. Since mineral fillers are usually less expensive than wood pulp fibers [46], papermakers are motivated to replace fiber with filler. The trend in recent years has been to increase the filler content of paper from 10% to 25-30% by weight. In addition to reducing the cost of raw materials, using mineral fillers makes the paper making process more energy efficient; because it facilitates the paper drying. The drying section of the paper machine consumes 60% of the total EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       6   energy used in papermaking. Using mineral fillers in the paper expedites the drying and increases the solid content of paper that leaves the pressing section [18][39]. It is noted that the effect of mineral fillers on paper properties not only depends on the filler particle attributes, but also on the location and distribution of filler in the sheet. For example, the filler can cause the disruption of fiber network [58]. There are also disadvantages from the use of fillers. For instance, when the filler content of paper increases, the fibre mat permeability decreases which leads to inadequate drainage before the coach roll in the paper machine. Also, since accumulation of filler particle on exterior fibre surfaces interferes with the inter-fibre bonding, the paper wet strength decreases, which can result to breaking the fibre mat during the paper machine operation. The third technical problem that limits the papermakers to produce more highly filled papers is the possibility of decreasing the adhesion of the filler to fibres when the filler content increases [30][43][73]. 1.3.1 Filler types There are different types of fillers to use in papermaking. Calcium carbonate (Precipitated Calcium Carbonate (PCC) and Ground Calcium Carbonate (GCC)), Kaolin, Talc and TiO2 are the most commonly used fillers for making printing and writing papers. Figure 1-3 shows the consumption rate of different fillers in the world [32]. The type and loading level of filler are chosen based on the paper requirements and papermaking restrictions. Moreover, the interaction of fillers with other paper components such as retention aids, starch, sizing agents, dyes etc. should be taken into account when choosing the filler type. As mentioned above, the calcium carbonate (CaCO3), which has been used in papermaking since 1980’s, is one of the most common mineral fillers. Its most important feature is to increase the brightness level of paper to 90-95%. Moreover, it can improve opacity, ink receptivity, aging resistance and surface smoothness[54]. Also, since it is feasible to manufacture cheap calcium carbonate on site at paper mill, papermakers tend to use this filler to reduce papermaking costs [22]. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       7   Figure 1-3: Global paper filler consumption in 2004 (Adapted from [43]) Two types of CaCO3 filler are currently used in papermaking: 1- Ground calcium carbonate (GCC): It is typically ground limestone or calcite. It can be developed in a wet or dry process depending on desired particle size and results in a random particle shape [68]. 2- Precipitated calcium carbonate (PCC): It is developed by converting lime produced from limestone on site at the paper mill. PCC is more popular than GCC in the papermaking because of the following reasons: 1- It has fewer impurities (such as iron and silica) that can affect the paper making process. 2- It can easily be manufactured in North America due to the abundant resources of limestone. 3- Since PCC is a synthesized product, it provides the papermakers with the ability to customize the chemical and physical properties of the filler particle such as particle size, particle size distribution, surface area, and particle shape. 4- PCC provides a higher brightness level and scattering coefficient over GCC [43]. Depending on the manufacturing process parameters such as temperature, pressure, reaction speed, additives etc, PCC can have different crystalline structures such as scalenohedral (rosetteshaped), rhombohedral (cubic-shaped) or aragonite (needle-shaped). Currently, scalenohedral with a rosette shape (Figure 1-4) is the most common type of PCC in use because it provides increased caliper and bulk to the paper [58]. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       8   Figure 1-4: Scalenohedral precipitated calcium carbonate Despite the benefits of using calcium carbonate fillers, they are not suitable for use under acid and neutral papermaking conditions. The alkaline nature of calcium carbonate fillers and their poor acid-resistant properties cause pulp darkening at alkaline pH and dissolution below pH 7 [3]. Although, other fillers such as titanium dioxide and calcined clay can be used in acid and neutral papermaking, they increase the papermaking cost due to their high price. Another alternative is talc filler, which is economically efficient, but it cannot improve paper properties as calcium carbonate. Due to the important benefits of carbonate calcium papermakers have tried to control the dissolution CaCO3 at acidic conditions. It has been found that adding trace amount of phosphate lowers the solubility of PCC by 80% at pH 7.5 [21][76][77]. One solution is to use Acid tolerant (AT) PCC, which is treated chemically in manner that renders it available for the use at acid and neutral papermaking conditions [71]. One of the other important characteristics of CaCO3 is its surface charge, which can be anionic or cationic. The surface charge density depends on different factors such as the type of filler, its origin, the specific chemicals and polymers applied in the production process, the dispersing treatment, pH, and concentration [4][85]. This feature is important for selecting a suitable retention aid system. For example, the use of microparticle retention systems enables the retention of the anionic filler particles at high production speeds. On the other hand, the cationically charged fillers are not in use because they may remove the negatively charged optical brightening agents [58]. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       9   The surface charge also affects the size of aggregates. When the surface charge is low, CaCO3 is unstable and forms large aggregates, while when the surface charge is high, the aggregate formation is prevented due to the strong repulsion of CaCO3 by electrostatic forces [4][85]. 1.4 Starch As mentioned in the previous section, mineral fillers decrease the strength of paper [38][44][68]. To solve this problem, dry strength agents are added to the pulp suspension in order to promote the inter fibre bonding [87]. Fiber fines are also considered as strengthening agents because they bridge the filler-induced voids present in the fiber/fiber binding domain [44]. However, refining energy is required to produce fines. Therefore, dry strength agents such as cationic and amphoteric starch (and particularly cationic starch) are mostly used in papermaking for increasing the paper strength [9]. Starch is chemically composed of amylose and amylopectin molecules. Amylose is composed mostly of linear α-d-(1→4)-glucan units (Figure 1-5), whereas amylopectin is a highly branched α-d-(1→4)-glucan with α-d-(1→6) linkages at the branch points (Figure 1-6) [20]. Figure 1-5: Linear amylose chain CH [20] Figure 1-6: Branched amylopectin molecule [20] EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       10   Starch interacts with the other pulp suspension components and improves paper strength. Thus, it is important to choose the proper starch type, starch dosage and addition point. One evaluation criterion is the effective retention of starch in the paper [44]. The other evaluation criterion is the adsorption of starch on the pulp fibres. Efficient adsorption of starch decreases the problems associated with the un-retained starch and improves paper strength. The adsorption of cationic starch on fibres is as an ion-exchange process in which the anionic charges of the fibres are compensated by the cationic charge of starch [87]. In addition to being a strength enhancement additive, starch can also act as a retention aid [52]. It is found that significant synergism between silica and starch can improve PCC retention [68]. If starch is added to the thick stock (3-3.5% consistency), it will act as dry strength additive while if it is added to the thin stock, (0.5-1% consistency), it will act as a retention aid. Thick stock refers to the stock consistency in the blending chest and thin stock refers to the stock consistency in the approach flow system. When starch is used as a retention aid chemical, it should not be exposed to high shear forces otherwise starch molecules may shear off [20]. Starch can also aid in drainage and formation [27]. If starch is chosen properly, it does not necessitate the need for a fixation agent to deal with the dissolved and colloidal substances (DCS) problem in some cases [28]. Papermakers can select starch from different sources such as maize, potato, tapioca, waxy corn and wheat. However, the ratio of amylose and amylopectin is different in starch obtained from each of these sources. These starches are commonly used in the wet end of papermaking and are cationized with either tertiary or quaternary amine groups. Cationic starch has smaller charge density than [70] amphoteric. Modgi et al.(2006) found that tapioca starch has a higher fines/filler retention compared to potato starch. It also adsorbs more onto TMP fibres and in contrast to potato starch, it is not deactivated in the presence of DCS[52]. Therefore, while it is not required to apply anionic trash collector (ATC) when using tapioca starch, ATC should be used when using potato starch. Sang et al.(2012) also performed experiments to compare three tapioca starches with different nitrogen content. These three starches were used in conjunction with colloidal silica and a cationic polyacrylamide flocculant. They showed that the higher charge density cationic starch (S880) has a statistically significant effect on filler retention. It was also demonstrated that using EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       11   a higher charge density starch introduces more cationic charges to the system, which results in a significant flocculation and produces larger, stronger and more compact flocs [69]. 1.5 Retention and chemical retention aids Retention refers to the ratio of initially added solids to the pulp suspension that remain in the final paper sheet. If the retention is high, the cost of papermaking decreases because less amount of initial material is needed to make the paper with a given basis weight. Different parameters affect the retention such as pulp furnish, machine speed, temperature, fillers, applied suction, chemical retention aids, basis weight, consistency, particle affinity, forming fabric structure, etc. [90]. The retention ratio of each solid type depends on the difference between its size and the size of openings in forming fabric. The frame size of modern forming fabrics is about 100-400 microns. Since, the length of fibre is about 0.5-2 mm [8], it will not pass the forming fabric. On the other hand, since the size of filler particles is about 0.2-15 microns, they pass the forming fabric and are not retained by the forming factor but only by fibre mat. As a result, the filler retention is low which leads to inefficient use of fillers and high concentrations of them in the white water. To solve this problem and increase the filler retention, chemical retention aids are used in papermaking. Chemical retention aids can enhance the electrostatic attraction (attachment), which is the mechanism for retaining filler and fines particles in fiber mat. There are three other retention mechanisms: filtration, sedimentation, and flocculation. However, the main mechanism is electrostatic attraction or van der Waals electrostatic forces [75]. 1.5.1 Chemical retention aids Chemical substances that are added in a papermaking suspension to increase the retention of fiber fines, filler and other fine materials in the paper sheet are called retention aids. They are polymers with cationic or anionic charges. Some retention aids may also be non-ionic. Retention aids usually have very high molecular mass and also affect the drainage and filler distribution [36][6]. Most retention aids enhance the retention by one of the following mechanisms [84][90][10][36]: 1- Forming bridges between particles EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       12   2- Modifying the charge of suspended particles The mechanism depends on the retention aid properties. High molecular weight and high charge density retention aids help retention through bridging between particles and charge modification respectively. Both of these mechanisms increase the fibre-fibre, fibre-filler and filler-filler flocculation and create hard flocs which are able to resist hydrodynamic shear [64]. Papermakers use different types of chemical retention aids. Initially, only single-component retention aids such as alum, starch, polyamines, polyethyleneimines (PEI) or polyacrylamides were used. Later retention aids were dual-component system, which were usually based on interactions between a long-chain charged polyelectrolyte and a second polymer with the opposite charge. Microparticle systems are mostly used today. High retention, good formation, and good drainage can be achieved with these systems. They consist of cationic polymers and anionic inorganic micro or nano particles with a large surface area [20]. The following microparticulate systems are mostly used in papermaking: 1- Cationic starch and/ or cationic PAM with colloidal silica 2- Cationic coagulant and/or cationic PAM with hydrated bentonite These systems perform in a complex way. In papermaking practice, first a high molecular weight cationic polymer (e.g cationic polyacrylamid) is added to the system prior to last point of high shear and large flocs are created by bridging. The hydrodynamic shear forces degrade these flocs in the system. After the shear point, an anionic microparticle (e.g colloidal silica) which provides many negatives sites, is added to the system. The anionic sites interact with the positively charged flocs. As a result, a highly coagulated system of dense, small flocs, which dewater easily, are formed [70]. 1.5.2 Interactions between dewatering, retention and formation The main factor affecting the uniformity of sheet formation is fibre flocculation and shearing conditions in the forming section [37][40]. The higher fibre flocculation leads to poorer formation. Since retentions aids increase the fibre flocculation, they degrade the paper formation [78]. As a result, papermakers adjust the level of retention by controlling the amount of retention aids applied in papermaking process. For this purpose, a specific operating window is set in which the upper EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       13   limit for retention aid dosage is the maximum formation value, and lower limit is the minimum retention value. Although retention aids degrade the formation, they improve the drainage by collecting the fines and colloidal substances on the fibre surfaces and increasing the area for water removal. The most used retention aids for improving the drainage are polyethylenimine (PEI) and poly (DADMAC), which are low molecular mass and high charge density cationic polymers. High molecular mass, low charge density cationic polyacrylamide (CPAM) and cationic starches also enhance the drainage. The flocculation of fines increases the permeability of the fibre web by reducing the plugging of the web pore structure. 1.6 Z-direction filler distribution Z-direction filler distribution is the filler concentration through the thickness of paper sheet. The axis through the thickness of the paper is labeled with z-direction and axes along the length and width are labeled the X and Y directions. These axes are also referred to as the cross sectional, cross machine, and machine directions, respectively. Figure 1-7 shows a schematic of the paper axes. Figure 1-7: Paper main directions (Adopted from [89]) The z-direction filler distribution has a considerable impact on the characteristics and the quality of the paper. For example, it affects the paper brightness, roughness and printing properties such as ink receptivity, print quality and dusting [34]. It also impacts sheet permeability in a complex manner depending on concentration and filler type [48]. Based on the requirements of the paper users, the filler distribution can be tuned accordingly. For example, for inkjet printing, coated paper and super-calendered paper grades, the filler content should be high on the surface and low in the center [33][56]. Increasing the filler on the surface decreases the surface roughness and increases the surface brightness [61]. Also, it reduces the bending stiffness. Stiffness increases when filler is concentrated towards the centre of the paper thickness [61]. For the copy paper, the filler content should be low on the surface EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       14   and high in the center [33][56]. For offset printing paper, the filler content should be constant throughout the sheet [33]. The constant filler content increases the strength of paper sheet when the concentration on the surface increases [61]. Different factors control the filler distribution such as the type of paper machine, level of retention, headbox layering, headbox flow rate, dewatering rate, sheet basis weight, wet pressing, suction boxes and specific filtration resistance. In the following, the effect of each of these factors on filler distribution is briefly discussed. Paper machine type: It has been shown that the dewatering mechanism controls the movement of fines and fillers in the paper [33][56][79]. In a standard Fourdrinier paper machine, the water drains on the wire side of the paper through the forming fabric. As a result, the filler distribution is non-symmetrical and very different at the two sides of the paper. Also, the filler particles are removed from the wire side because of the applied pressure pulses from the suction boxes. Therefore, the filler concentration increases towards the sheet top side (away from the forming fabric) [33][56][34]. In hybrid machines, 20 to 30% of dewatering happens through the top side of the paper. Therefore, the top side filler content is decreased resulting to a more symmetric distribution with a higher filler concentration of filler in the centre of the paper [56][61]. In the twin-wire paper machines, the water is removed from both sides of the paper simultaneously providing a symmetrical distribution [56]. Also, the applied vacuum controls the filler distribution. Finally, in the hand sheet forming, which is used in laboratory experiments, since the drainage is low and only a single sustained pressure pulse is used, the filler concentration is high in the center and low in the top and wire sides [34][80]. Headbox layering: Headbox layering affects the filler distribution through the additive layering or layering of the different pulp furnishes. When layering is used, the feed streams of the pulp suspension are separated to the different layers until the forming section. Each layer has different filler percentage or chemical retention aids and increasing them enhances the filler content of the corresponding layer [56]. Headbox flow rate: Increasing the headbox flow rate increases the filler content in the top side (away from the roll former) [62]. Dewatering rate: In gap forming, increasing the blade dewatering moves filler from the center of the sheet to the surface [62]. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       15   Sheet basis weight: By increasing the sheet basis weight, more filler particles are retained by filtration. As a result, the amount of filler will be increased until it becomes constant in the middle of the paper and machine parameters will not change it largely [34]. Wet pressing: It has been shown that wet pressing changes the filler content but not the profile [80][79]. During the wet pressing, since the amount of water is decreased and the mat is compacted, the filler cannot move. As a result, the filler content will not change. Szikla and Paulapuro [79] have shown that the movement of filler or fines during the pressing happens only if the following conditions are met simultaneously. - The flow velocity in the z-direction should be at a maximum in layers that contained 30- 50%, of filler material, which means that on the wire side of the sheet. - The fibre network should be very loose. - The size of the particles should be 2.7 µm or less in diameter. - The wood fibers should be smooth and un-fibrillated. These conditions may only be met during the initial phases of wet pressing, but usually it is very unlikely that all the conditions are met. As a result, the prevailing flow in a sheet under wet pressing can move very small particles in the fiber network. So, after the dry-line the filler distribution does not change. Although this study focused on movement in a fibre mat that had a high solids content, similar principles are valid in other areas of the forming process. Retention: Tanaka et al. [80] used the adhesive tape method to investigate the effect of retention level on the z-direction distribution of clay in hand sheets. Their results showed that the the z-direction clay distribution become more even by increasing the level of retention. Another study on a pilot paper machine indicated that dosage of chemical retention aids is a major parameter in controlling the z-direction filler distribution [61]. Suction box: Early studies have shown that applied suction does not change the filler distribution [80][79]. However, later studies have shown that changing the applied suction during the forming process changes the filler distribution. When dewatering rate of loadable blades in a gap former is increased, the filler profile changes on the sheet side closest to the blades. A medium load decreased the surface and center-plane filler content whereas a high load maintained surface filler content but caused a great decrease in the center-plane [33]. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       16   A recent study has shown that although increasing the applied vacuum reduces the amount of filler on the surface, it increases the overall filler content [61]. The result of another study indicated that increasing the vacuum in suction shoes of a gap former, increases the surface filler but decreases the filler in the center plane. Also, the amount of change depends on the type of pulp [56]. The effect of applying vacuum on filler distribution was investigated by using a modified handsheet former and it was shown that cationic PPC filler content reduces in the wire side while the top side remains unchanged [54]. There is a significant paucity of data in the literature with respect to the effect of applied suction on filler content with retention aids [80] and no study investigates the effect of suction on filler distribution when using starch. Specific filtration resistance (SFR): Haggblom-Ahnger et al. [33] have shown that web compaction and increase of SFR increase the entrapment of filler. As a result, the filler distribution depends on the local SFR in a paper sheet. 1.7 Measuring the Z- profile of paper There are various techniques to measure the filler distribution in the z-direction of paper sheet. Previously, the paper sheet was manually grinded in the z-direction. Then, the ash content of remaining sheet was compared to the content of the complete sheet to measure the content of the removed layer [34]. Later, the width of the formed paper were split into an average of 8 to 10 individual strips using adhesive tapes, and their ash contents were measured independently [61]. Although these two physical destructive techniques can be used for boards or heavy weight papers, they are not suitable for lower basis weight sheets because it is difficult to get constant sheet splitting. As a result various imaging techniques have been employed such as Wavelength Dispersive Spectroscopy (WDS), X ray diffraction (XRD), X-ray fluorescence (XRF), combination of XRD and XRF, Scanning Electron Microscope (SEM), Energy Dispersive X- ray Analysis (EDX) and X-ray Synchrotron Radiation Microtomography. None of these methods are destructive and they can be used to determine the 2D z-direction distribution of fibres, fines and fillers. Wavelength Dispersive Spectroscopy (WDS) first generates a graph of the paper sample by using one wavelength at a time. It then compares the graph with the standards to determine the constituent elements. WSD can provide accurate quantitative measurement but it is very time consuming and it is difficult to use it for lighter elements [81]. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       17   X-ray diffraction is used for cellulose fibres and minerals. For this purpose, paper samples are subjected to X-ray diffraction. Then, the intensities and areas of diffraction peaks for minerals and cellulose are measured. As a result, the relationship between the ash content and X ray diffraction measurements can be determined by using linear regression [15]. This technique is challenging to use because measuring the cellulose peak areas is difficult. It has been shown that the combination of scanning electron microscope (SEM) and energy dispersion x-ray (EDX) can yield the best result when measuring the surface filler distribution, because of the following reasons: 1- The quantity and spatial distribution of elements of paper can be studied [58]. 2- The position and concentration of different elements of interest can be determined by using a mapping software. 3- The composition at the surface can be studied by tuning the SEM electron beam such that it only penetrates the first few micrometers of the paper surface [58][5]. 4- EDX can detect non-crystalline pigments rapidly which cannot be determined by the XRD system. Moreover, the efficiency of SEM/EDX has been demonstrated as in the study of binder migration in coated papers [42] and in Voillot et al.’s study to count the X_ray counts in paper samples [86]. While physical methods are destructive, non-reproducible and time consuming, the SEM/EDX technique is non-destructive and can provide the concentration of mineral elements (even light ones) in the z direction of the paper sheets with higher resolution. Also, it is not as time-consuming and expensive as other image analysis techniques. In the following sections, the SEM and EDX techniques are described in more detail. 1.7.1 Scanning electron microscope (SEM) Scanning electron microscope (SEM) provides the image of the surface of a sample at high resolution and magnification. This technique has been used in various paper analysis such as determining the effects of paper structure on printability of a paper sheet, analyzing the coating structure [11], subsequent reconstruction of surface coatings [13], determining the surface coating thickness distributions [5], specifying the paper surface characteristics of mineral-free [25] and coated papers [5]. Moreover, the combination of SEM with image processing EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       18   techniques is used to analyze the sheet fiber content in the cross section. Such analysis can provide the location of fines material [14] and sheet pore structure [16]. SEM can also be used for determining the characteristics of individual fibres in the transverse direction such as wall thickness [66]. In the following, the SEM process is described briefly. First, a primary beam of electrons, called the incident beam, is generated by heating a tungsten filament. This beam is then accelerated to a voltage between 1 and 30 KeV and directed at the sample. The interaction of the beam with the surface atoms leads to transferring its energy to surface atoms. As a result, a set of signals, shown in Figure 1-8, are released: 1- Secondary electrons: There are electrons that escape from the sample surface with energies below 50eV [31]. These are either beam electrons that have lost most of their energy or electrons on the sample surface that have received a low level of energy from the beam. They are the most used signal in SEM and their release is significant as they are yielded at the rate of 1 per primary electron. 2- X-rays: This signal is usually used in chemical analysis of the sample [31][29]. 3- Backscattered electrons: These are electrons that escape from the sample surface with a large fraction of their incident energy. Although backscattered electrons occur less than secondary electrons, they are also used in SEM imaging. Using the above signals, the SEM image is created as described below: When the primary beam moves along the sample surface, the intensity and location of detected electrons are recorded. Then, the relative intensity of the electron signals at each location is used to produce a grey scale image representing the sample surface. This image is output to the computer screen. The elemental composition of sample surface impacts the intensity of the released electrons and consequently the contrast between surface features in the image. The resolution of SEM image can be controlled by tuning the penetration depth of primary beam to the sample surface. The penetration depth can be changed between 2 to 10 nm [29][31][54][19]. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       19   Figure 1-8: A summary of signals that will be detected when the primary beam is directed to SEM sample 1.7.2 Energy dispersive X-ray (EDX) Energy dispersive x-ray analysis is used in conjunction with scanning election microscopy. It is a qualitative and quantitative analysis method that can effectively measure the elemental composition of a paper sample area over which the primary beam is projected. This technique suits the paper analysis because of complex matrix of organic and inorganic material and the importance of the material distribution. As showed by Gibbon [25], EDX can determine the various filler particles and their distribution in the paper structure. Although initially EDX could only detect the elements with atomic number 11 and heavier, it can currently detect all elements with atomic number heavier than 4 (beryllium). As a result, EDX can be used to determine the z- direction filler distribution in paper [49][50]. In the following, the process of EDX analysis is described. First, a primary electron beam is directed on the sample surface which transfers part of its energy to the atoms electrons. As a result, the electrons energy level may exceed the biding energy of their orbital shell. This leads to electron ejection from the sample surface. If an electron is ejected from a lower orbit (closer to the nucleus), the corresponding atom is in an excited state. Such atom will finally relax by one of the following relaxation methods: 1- Cathodoluminescence 2- Release of Auger electrons     Primary  Beam  X-­‐ray   S   B     S   Light   EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       20   3- Jumping an electron from an outer orbit to inner orbit to fill the inner vacancy Figure 1-9: EDX process The first two methods are not used in EDX [19][29]. In the third one, a characteristic X-ray is released during the electron jump. The X-ray energy level is equal to the energy difference between the outer and inner orbits. This energy level is different for each element and depends on the involved electron orbits. Figure 1-9 shows the direction of primary beam, ejection of K shell electron and emission of X-ray. The energy and wavelength of a characteristic X-ray is strongly correlated to the atom emitting it [31] and the number of characteristic x-rays detected in the sample area is strongly correlated with the quantity of the element present in that area [31][29]. Therefore, EDX can provide the weight percentage of sample elements by collecting the energy from characteristic X-rays and counting the number of X-rays at each energy level [19][29]. For this purpose, the primary beam moves on the sample while it focuses on a small area at a time. Then, the detector tracks the type and quantity of characteristic x-rays that it receives. Combining the obtained information at each location creates a spatial map of the elemental composition of the sample. The resolution of this map can be tuned by changing the primary beam size. 1.8 Objective of the study Z direction filler distribution which refers to differences of filler concentration such as calcium carbonate through the thickness of paper, is an important characteristic of paper affecting a number of paper properties including strength, compressibility, porosity and optical M   L   X-­‐ray   K   EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       21   properties, to name a few. Therefore, information about how filler particles are arranged in a paper structure would help the papermakers to structure the paper and characterize the paper properties according to the end use. It has been found that the main parameter to influence the z direction profile of the paper is the type of the paper machine. However, once the investment for the machine has been made, the challenge is that how to influence the paper structure in order to meet the quality demand desired by the papermaker. It is expected that chemical additives as well as the machine parameters, such as vacuum in the paper making machine can affect the z direction profile of the paper. There are some studies showing that chemical retention aids can affect the filler distribution in z direction of paper. But, how the dosage of starch or chemical retention aids changes the filler distribution has not been elaborated yet. A small body of work investigated the effect of applying suction on filler migration in paper samples. Montgomery et al. (2010) found that distribution of filler particles can be affected by applying vacuum. He carried out the study in a hand sheet former that he constructed for this purpose. However, there was no use of starch in this work and also the effect of different levels of chemical retention aids has not been studied. So, there is a limited understanding of how an applied suction affects the distribution of fillers during the papermaking in the presence of chemical additives. This provided the motivation to conduct a systematic study to understand the effect of various parameters on z direction filler distribution in paper samples produced in the simulated suction box. The parameters of interest include precipitated calcium carbonate as filler, cationic polyacrylamide and anionic silica nanoparticles as chemical retention aids, cationic tapioca starch and pressure of the vacuum chamber in the hand-sheet former. Thus, one objective of the thesis is to understand the effect of different levels of parameters on the z-direction filler distribution in hand-sheets produced by modified hand-sheet former. The second objective is to construct a predictive model to estimate the filler distribution in the paper thickness based on the parameters of interest. 1.9 Organization of the thesis The organization of this thesis is as follows: Chapter 1 introduction: Information about papermaking process, chemistry of paper, importance of controlling the z direction filler distribution in paper and available methods for EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 1       22   studying the z profile of paper are described in this chapter. Chapter 2 Materials and Methods: The experimental design to study the effect of different parameters on filler topology in paper as well as the materials and apparatus are explained. The adopted method and sampling procedure for studying the filler distribution are also presented in this chapter. Chapter 3 Results and Discussion: The results of our experiments are presented and discussed in this chapter. Chapter 4 Conclusion and Recommendations: This chapter concludes the thesis and provides a set of recommendation for future works.  23     Chapter 2: Materials and Methodology This chapter presents the experimental design, equipment and sampling procedure of our study. 2.1 Experimental design One approach for designing the experiments is to use the full factorial method in which one factor is changed at a time. This design is used wherever it is necessary to study the joint effect of several factors on the response. The primary case is to consider k factors, each at 2 levels, “high" and “low". A complete replicate of such a design requires 2k observations and hence is called 2k factorial design. The 2k factorial design is particularly useful at the early stage of experimentation when there are a lot of factors to be investigated. A complete or full factorial design is not affordable especially when k is large due to limitations in time and resources. If we reasonably assume that the high-order interactions are negligible, then the main effects of the factors can be obtained by running a fraction of the full factorial design. One special case of a fractional factorial design is a one-half fraction of a 2k factorial design. To examine more than 2 levels of a factor or to capture the curvature in the response, a 3k factorial design can be used in which each factor can have three different levels with a total of 3k observations required. However, a 3k factorial design is rarely used due to its complexity and inefficiency. The central composite design (CCD) is an alternative method which is used more often. CCD is a special design commonly used in response surface methodology (RSM). It has a significant efficiency advantage compared to the 3k design. The practical deployment of CCD often arises in sequential experimentation. In general, the 2k design does not allow an estimation of the experimental error unless some runs are replicated. A common method of including replications in a 2k design is to augment the design with several observations at the center where all factor levels are set to 0 (xi = 0, i = 1,…, k). This point is called the design center. A typical CCD consists of a full or fractional 2k factorial design with nF runs, nc centerpoint runs and 2k (axial) runs [17]. According to CCD design model, the total number of experiments can be determined using the following formula: EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2       24   Number  of  experiments = 2!!! + 2𝑘 + 𝑛! where; k = number of independent variables            𝑛  != number of repetition of experiments at the central point In our study, a fractional factorial design of the 4 factors of interest (k=4) including PCC concentration, starch concentration, CPAM-silica concentration and vacuum level was conducted. The ratio of CPAM and silica was kept fixed for all the experiments. Each of the factors had five coded levels. The minimum and maximum levels of variables were selected based on papermaking practices. The levels of factors of interest are shown in Table 2-1. To visually represent the experimental design, the values of factors are coded. The central value is coded with 0, upper level with +, maximum level with A, lower value with – and minimum value with a. For parameter    n!, 5 runs in the design center were conducted to provide a reasonably stable estimate of the experiment error. Therefore, the total number of experiments was 21. The Central Composite Design Matrix for 4 factors with fractional factorial design is shown in Table 2-2. It is usually beneficial to conduct the    n! center part runs in the early stage of experiment because it can give an estimate for the reproducibility (the random error) of the experiments. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2 25     Table 2-1: Levels of the factors for experimental design Factor Components Levels of factor studied a (-2) -1 0 +1 A (2) X 1 PCC (%) 25 30 35 40 45 X 2 Starch (kg/t) 8 9 10 11 12 X 3 CPAM-Silica1 (kg/t) 0.1 0.2 0.3 0.4 0.5 X 4 Vacuum (in Hg) 2 5 8 11 14 Table 2-2: Full experimental design Sample ID Pattern PCC (%) Starch (kg/t) CPAM-Silica (kg/t) Vacuum (inhg) 1 +++- 40 11 0.4 5 2 +--+ 40 9 0.2 11 3 -+++ 30 11 0.4 11 4 ---- 30 9 0.2 5 5 --+- 30 9 0.4 5 6 ++-- 40 11 0.2 5 7 -+-+ 30 11 0.2 11 8 +-++ 40 9 0.4 11 9 0 35 10 0.3 8 10 0 35 10 0.3 8 11 0 35 10 0.3 8 12 0 35 10 0.3 8 13 0 35 10 0.3 8 14 00a0 35 10 0.1 8 15 00A0 35 10 0.5 8 16 000a 35 10 0.3 2 17 0A00 35 12 0.3 8 18 0a00 35 8 0.3 8 19 000A 35 10 0.3 14 20 a000 25 10 0.3 8 21 A000 45 10 0.3 8                                                                                                                          1  The  ratio  of  CPAM  and  Silica  was  one  in  all  experiment  run,  i.e.  the  same  amount  of  these  two  chemicals  was  used  in  each  sample.       EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     26   This design is usually used to build a second order (quadratic) model which can describe the significant linear, interaction and second order effects [69]. where • denotes the calcium concentration in the z direction • denotes the intercept of the quadratic model • , and denote the linear, interaction, and quadratic effects, respectively • denotes the factor level for the ithfactor • k denotes the total number of factors This second-order model can capture the curvature very well through the interaction terms and quadratic terms. In the present context, the quantitative nature of the factor could fit in such setting very well. However, since the objective was to estimate the effect of the chemical additives and vacuum on the distribution of the 5 layers through the thickness of paper, it was expected that the trend for the 5 layers would change if the combination of factor levels changes. This implies that there is an interaction between layers and factors. To model the interaction, the data from the 5 layers was used separately to get different coefficient of , , and . In this way, the interaction between the layers and the factors could be demonstrated in the change of coefficient. This separating estimating method can provide a general estimation of the interaction without strong assumptions. Eventually, 5 different models for the 5 layers were obtained which could be used in prediction of calcium concentration in the 5 layers for any combination of the factor levels. These statistical models can provide a useful tool for producers to select the factor levels to produce the paper types they need. The predictive models were constructed based on the significant factors indicated by analysis of variance (ANOVA) using JMP IN 4.0 (SAS Institude INC., Cary, North Carolina). In order to validate the models 3 additional experiments (other than those used in the ANOVA) were carried out and the results were compared to the ones obtained by the models. y 0β iβ ijβ iiβ ix 0β iβ ijβ iiβ ∑ ∑∑ ≤≤≤ +++= kji k i iiijiij k ii xxxxy 1 2 1 0 ββββ EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     27   2.2 Handsheet preparation All hand sheets were formed in a laboratory former with vacuum [54]. The former consists of a 3” diameter clear acrylic circular cylinder where the pulp slurry was transferred into and exposed to the atmosphere. A forming fabric was placed between the cylinder and a horizontal flat plate and was airtight seal through using a gasket around the outside. Another circular cylinder was placed below the forming fabric. There was a flush mounted gauge pressure transducer (GP:50 Model 218-C-SZ-10-SG) in its wall measuring pressure drop across the forming fabric and fibre mat. A 20mm diameter PVC pipe and a fast acting solenoid valve connect this circular cylinder to a vacuum chamber. This setting allows the water sitting above the solenoid to transfer into the vacuum chamber. The vacuum chamber was made of ¾’’ thick PVC sheet. The desired pressure in this chamber can be achieved through the use of an attached vacuum pump. Figure 2-1 shows different parts of this former. To form a handsheet, first the cylinder between the closed valve and forming fabric was filled with distilled water. Then, the forming fabric was fitted in the place and the top cylinder was attached. The vacuum pump was turned on to provide the desired pressure level. Later pulp suspension was transferred into the top cylinder. A Labview program was used to start the experiment, which could electronically open the fast acting solenoid valve. Consequently the distilled water and pulp suspension in two cylinders were exposed to the vacuum pressure and drained through the forming fabric into the vacuum chamber. The test was completed when the Labview program automatically closed the solenoid. The handsheet remained on the forming fabric was pressed and kept in room condition prior to being analyzed. The hand sheet former constructed in the previous study [54] was modified in order to maintain the vacuum at constant level. In the previous machine, the forming fabric was placed between the two acrylic plates, and the gap that was formed between the plates because of the forming fabric was a source of leak. In the modified machine, shoulders and O-ring groove were created in order to prevent this problem. In the previous machine, for draining the process water from the box, the machine had to be disassembled in order to empty the tank. In the modified version, a hole was drilled and tapped in the suction box. As a result, the water is easily drained. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     28   Figure 2-1: Schematic of the handsheet former (Adapted from [54]) 2.2.1 Materials The following materials were used to prepare the pulp suspension: • Hydrogen peroxide bleached thermo-mechanical pulp (TMP) with the pH of 6.9±0.2 supplied by a leading pulp and paper mill in British Columbia. This pulp was a mixture of spruce, pine, and fir. • Precipitated Calcium Carbonate (PCC) filler obtained from Specialty Minerals Inc (Bethlehem, Pennsylvania, USA). It was an acid-tolerant PCC subjected to polyacrylate treatment. The density of the PCC was 2.71 g/cm3 and had a scalenohedral structure. This PCC had zeta potential value of -17.8 ± 0.3mV at a concentration of 0.002 wt% in DDW. The average particle size of PCC was 10.22µm and the brightness is 98% ISO. • Cationic tapioca starch S880 with the nitrogen content of 0.9-1.10%N and average molecular weight of 3 million Da provided by National Starch ULC (Surrey, BC, Canada). A procedure supplied by National Starch ULC was followed to cook the starch before using. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     29   • Cationic polyacrylamide (CPAM) with the average molecular weight of 10 million Da. This cationic flocculant was supplied by Eka Chemicals (Magog, QC, Canada). CPAM has a branched structure and a charge density of 2.1 meq/g. • Silica as a 8.1 wt % suspension with a mean size of 5 nm supplied by Eka Chemicals (Magog, QC, Canada) • Distilled and deionized water was used for the preparation of the pulp suspension 2.2.2 Starch cooking The starch that we used in our experiment was cooked according to the following procedure [69]: 1- The jar containing the starch (0.167 wt%) was put in a boiling water bath. 2- The starch solution was stirred until it reached the gel point. 3- For 30 minutes, a rolling boil was sustained in the bath. During this period, the solution was stirred every 10 minutes. The stirring helps to break down any un-burst grains. Also, the jar was covered to prevent the net evaporation of water. 4- After 30 minutes, the solution was cooled down to room temperature. The cooked starch should be used within 24 hours. The pulp suspension for each handsheet sample was prepared by dispersing the required amount of filler and pulp fiber. The pulp consistency was 4.28 % and was diluted to 1% consistency by adding distilled water. Then the suspension was mixed and heated to 50°C to simulate the actual papermaking process. Then the required amounts of chemicals (starch, CPAM and silica) were added to the pulp suspension in sequence and the suspension was mixed for 10 s after each addition. This sequential addition is based on industrial practice [52]. Concentrations of materials in pulp suspension are shown in Table 2-3. Table 2-3: Concentration of materials in pulp suspension Materials Filler Pulp Starch CPAM Silica Concentration% 1.02 4.28 0.167 0.011 0.006 EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     30   2.3 The SEM/EDX technique The paper samples prepared in the modified hand sheet former were analysed to determine the z-direction mineral distribution using SEM/EDX technique. In this work, Hitachi S-3000N-VP scanning electron microscope was used. The presence and locations of the elemental compositions of additives such as calcium carbonate can be determined by using this technique. SEM/EDX condition for image acquisition and EDX analysis are reported in Table 2-4. Table 2-4: SEM/EDX test conditions Setting Conditions SEM-mode used BEI Magnification 600 × Accelerating voltage 20 kv Low vacuum mode 20 kPa Working distance 15 mm EDX analysis time per scan 60 s The variable pressure mode (low vacuum) was chosen in the microscope because paper samples are not conductive to avoid the charging of the sample. The selected electron accelerating voltage is high enough to excite the desired elements for example calcium in calcium carbonate. The allocated time for The X-ray counts was 60 seconds for each scan. It should be noted that the elemental composition within a 5 µm depth from the surface is determined by EDX. The elements with atomic numbers greater than 4 are determined. Since, PCC filled paper samples were used in our study, we were interested in measuring the calcium concentration. The EDX software program considered the total elemental composition in the selected area as 100 wt% and detected the percentage of calcium present in that area in relation to other elements. As a result, it reported the calcium concentration based on the weight % along with other elemental composition. 2.3.1 Sampling procedure In order to apply the SEM/EDX technique, it was necessary to select some samples from each paper sheet. We designed our sampling procedure in a way that an orientation of 360 of filler EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     31   distribution could be covered and the filler distribution for the entire thickness of the sheet could be determined. Our sampling procedure for each paper sheet was as follows [51]: 1. Four specimens were cut at different points on the circumference of the circle of the paper using a razor blade (Figure 2-2-a). The size of each specimen was 10 mm × 5 mm. Each specimen was put in the SEM/EDX sample holder in a way that its z-direction was perpendicular to the plane of sample holder. Figure 2-2-b shows the thickness of the specimen from top side to the wire side. 2. Two viewing areas with the size of approximately (100 µm) × (paper thickness (µm)) were selected in each paper specimen using an area selector. Each area covered the thickness of the paper. Each viewing area was subjected to EDX analysis to determine the calcium concentration. We chose two viewing areas to duplicate the measurements. 3. Each viewing area was subsequently divided to five layers along the z direction (Figure 2-2-c) and each layer was subjected to EDX for determining the calcium concentration. The minimum area that can be analyzed using EDX depends on the magnification settings of machine. Since we set magnification to 600, the highest number of equal layers that could be analyzed was five. 4. Therefore, for each paper sheet, we measure the calcium concentration for four specimens and two viewing areas per specimen. Also, we considered five layers per each specimen and viewing area. As a result, we report the average of eight readings along with standard error of mean for each of the five layers of one paper sheet. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     32   (a) (b) (c) Figure 2-2: Sampling procedure for SEM/EDX analysis, (a) handsheet, showing the four specimens, (b) SEM image showing in the z-direction of the paper specimen, and (c) SEM image showing the five areas for EDX analysis in the entire thickness of paper specimen. The calcium concentration (wt. %) along with other elemental composition for a sample layer is shown in Figure 2-3. a  =  100  μm b  =  20  μm Top  side Wire  side Top  side Wire  side EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     33   Figure 2-3: Elemental composition of the viewing area determined by EDX Element Method Intensity KRatio ZAF Concentration 2 Sigma Z A F C PRZ 439.11 0.159 2.066 32.82 wt% 0.388 wt% 0.937 2.207 0.999 O PRZ 349.62 0.068 6.850 46.83 wt% 0.636 wt% 0.988 6.931 1.000 Mg PRZ 0.00 0.000 1.000 0.00 wt% 0.000 wt% 1.000 1.000 1.000 Si PRZ 0.00 0.000 1.000 0.00 wt% 0.000 wt% 1.000 1.000 1.000 Ca PRZ 1537.30 0.183 1.110 20.35 wt% 0.139 wt% 1.131 0.982 1.000 2.4 Validating the experimental results 2.4.1 Ashing As it was explained before, EDX cannot detect the elements with atomic numbers less than 4 such as H2. As a result, the amount of calcium indicated in the EDX is higher than the actual amount of calcium in the paper sheet. Therefore, independent measurements are required to calibrate the SEM/EDX measurements. In this research, the amount of calcium indicated in the EDX was correlated with the ash content of paper samples measured through low temperature ash method. The paper samples with known dry weight were put in a muffle furnace at 525°C for 12 hours. The ash content is reported in %. Ash, % = (W1/W2) x 100 EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     34   Where: W1 = weight of ash, and W2 = initial weight of 105°C dried sample All the ash in the hand sheet was assumed to be CaCO3. Therefore, we could easily calculate the calcium content in the sheet based on the molecular formula of CaCO3. Moreover, the phase analysis was performed with a Rigaku Multiflex X-ray diffractometer using the Cu-Ka radiation on ash samples. Comparing the XRD result with JCPD number 47- 1743card showed that the ash was a single phase composed of crystalline calcium carbonate. Considering the characterization limit of XRD, which is 5wt%, it was concluded that the other elements were not in crystalline form or they consisted less than 5wt% of the ash. Therefore, the calcium amount in each hand-sheet measured by EDX could be compared with the calculated calcium content obtained by ashing method. Figure 2-4 shows the X-ray diffraction analysis of a ash sample. 20 25 30 35 40 45 50 55 60 0 1000 2000 3000 4000 5000 6000 7000 8000 (122)(211) (116)(024) (202)(113) (110) (006) (012) (104) 2Θ In te ns ity (C PS ) Figure 2-4: X-ray diffraction analysis of ash sample EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     35   2.4.2 SEM/ Image processing technique In our study, one limitation of the sampling procedure was the size of the areas selected to be analyzed for their filler content. With the magnification that we used, the area chosen to cover the thickness of the paper was approximately (100 µm) × (paper thickness (µm)). One question would be what if we chose a larger area. For example, the whole area of the image obtained by the SEM. Would the measured concentrations still be the same? We tried to answer this question by using the image processing technique. Figure 2-5 shows the areas analyzed by EDX (a) and the layers processed using image processing technique (b). (a) (b) Figure 2-5: (a) SEM image of cross sectional of paper analyzed by EDX, (b) the extended area analyzed using image processing.   Top  side Wire  side   Top   side Wire   side EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     36   The purpose of image processing was to demonstrate that the data obtained by EDX analysis from a particular area of sample (Figure 2-5-a) is in reasonable agreement with image processing results. Also, it is shown that this particular area can be a reasonable representative of a larger area of the sample which was not measured by EDX (Figure 2-5-b). First, the areas measured by EDX were analyzed using image processing technique and the results were compared. The SEM image was divided to a certain number of layers, consistent with the areas used for EDX (Figure 2-5-a). The images of each different layer of a sample were imported to a program (Appendix A) as raw data and were converted to 2D matrices with values based on the gray scale level of each pixel. This would normalize the pixel values to a number between 0-1. The image processing technique that is chosen for our purpose should be able to identify the particles (which are Calcium Carbonate particles) in the whole image. To achieve this goal, an appropriate image filter should be used to amplify the particles’ signals compared to the other signals (i.e. those related to fibers and voids). Most useful image processing operators are area based. Area-based operations calculate a new pixel value based on the values in local, usually small, neighbourhood pixels. This can typically be implemented via either linear or nonlinear filtering operation with a finite-sized operator (i.e., a filter).In general, consider a centered and symmetric 5×5 neighbourhood of the image pixel at position i and j, with the image value of f i, j . An area-based transformation can be expressed as: ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) 2, 2 1, 2 , 2 1, 2 2, 2 2, 1 1, 1 , 1 1, 1 2, 1 , 2, 1, , 1, 2, 2, 1 1, 1 , 1 1, 1 2, 1 2, 2 1, 2 , 2 1, 2 2, 2 f i j f i j f i j f i j f i j f i j f i j f i j f i j f i j g i j T f i j f i j f i j f i j f i j f i j f i j f i j f i j f i j f i j f i j f i j f i j f i j − − − − − + − + −⎛ ⎞ ⎜ ⎟ − − − − − + − + −⎜ ⎟ ⎜ ⎟= − − + + ⎜ ⎟ − + − + + + + + +⎜ ⎜ − + − + + + + + +⎝ ⎠ ⎡ ⎤ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎟ ⎢ ⎥⎟ ⎣ ⎦ (1) Where 𝑔 is the output image resulting from applying transformation T to the 5×5 input image  𝑓. It should be noted that the spatial dimensions and geometry of the neighborhood are generally determined by the needs of the application. An example of image processing region- based for our purpose is particle detection however it has many other applications like edge detection, image enhancement, edge sharpening and many other applications for many other specific purposes. Linear image filtering using convolution is one the most common methods of EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     37   processing images. To achieve a desired result one must specify an appropriate convolution kernel. Tasks such as smoothing, sharpening, edge finding, and zooming are typical examples of image processing tasks that have convolution-based implementations. Based on our main purpose (particle detection) and the images that we have, a very accurate convolution-based filter can be developed to identify and capture those signals related to the particles that we have in our images. For example, Figure 2-6 shows the application of a convolution-based filter with an edge detector kernel, to detect the edges of image (a). Image (b) is the same image after applying the edge-detector. Figure 2-6: (a) Sample image, (b) The same image after applying convolution-based filter for the purpose of edge detection. In the convolution filters, the basic idea is to combine a kernel signal (in our work since our signal is a 2D image, the kernel is a 2D matrix) with the main signal i.e. the image. The convolution of a kernel 𝐾! with a list 𝑓! has the general form 𝐾!𝑓!!!! . For example, for the edge detection purpose in Figure 2-6, the following edge detector kernel (Sobel Kernel) is convolved with the whole images in a mathematical function, which is presented as following (Figure 2-6, Image (b) is the result): 1 0 1 1 . 2 0 2 4 1 0 1 r xK − −⎛ ⎞ ⎜ ⎟= −⎜ ⎟ ⎜ ⎟−⎝ ⎠ and 1 2 1 1 . 0 0 0 4 1 2 2 r yK − ⎛ ⎞ ⎜ ⎟= ⎜ ⎟ ⎜ ⎟− − −⎝ ⎠ (2) (a) )a   (b) )a   EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     38   ( ) ( )22, ,r x r yEdges Conv img K Conv img K− −= + For our purpose another kernel has been developed and used for detecting the particles of the SEM images. As it is shown in Figure 2-7(a), the particles in our image can be observed by very small discrete bright (white) dots. The following kernel is used to identify the filler particles in the image. Appendix-B explains why this kernel can capture the filler particles. 1 1 1 1 1 1 1 1 1 1 1 1 24 1 1 1 1 1 1 1 1 1 1 1 1 rK − − − − −⎛ ⎞ ⎜ ⎟− − − − −⎜ ⎟ ⎜ ⎟= − − − − ⎜ ⎟ − − − − −⎜ ⎟ ⎜ ⎟− − − − −⎝ ⎠ ( )rParticles Image Conv img,K= (3) The result of the convolution kernel applied to our data is shown in Figure 2-7-b. Figure 2-7: (a) SEM image of cross sectional of paper (b) SEM image of cross sectional of paper after applying convolution filter. Figure 2-8 shows the image-histograms of images before and after applying the convolution filter. These two diagrams show how the developed filter separates the signal of the particles with other signals (for example fibers). a   b   EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     39   Figure 2-8: (a) Image histogram of SEM image of cross sectional of paper (b) Image histogram of SEM image of cross sectional of paper after applying convolution filter (equation (3)). After applying the filter and identifying the particles, a threshold value is determined with a method which is based on Ridler and Calvard [67]. The pixels were compared with threshold value and set to 0 (black) or 1(white). So the resulting binary images were converted into black and white images where the black pixels present the areas of fiber and voids and the white pixel show the areas covered with filler particles (Figure 2-9-b). Number of pixels which contain white spots were summed up and divided by the total number of pixels in the image. This number shows the percentage of the surface covered by the filler. The same procedure was done for the lager areas of the sample (Figure 2-5-(b)). In section 3.5 the result of image processing analysis is presented to show its compatibility with EDX results. Also, it is illustrated how well the results of EDX analysis of a small area represents the filler concentration of a larger area. a b EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 2     40   Figure 2-9: (a) SEM image (b) Converted image after applying convolution filter to black and white image. • Comparing the results of image processing with SEM/EDX The space filling obtained from image processing technique is the ratio of the area covered by the white pixels to the total pixels. EDX gives the weight percent of calcium in a particular area. Following formula was used to compare the results of these two techniques to each other. !!"!!"!#$ = !!"!#!  !!"!#!!.!∗!!"#$%  !!"!#$ Where !!"  !!"!#$ and  !!"!#!!!"!#$ were obtained by SEM/EDX and image processing receptively. The density of paper is just the ratio of basis weight and thickness of paper: d b =ρ Volume fraction was calculated based on the area ratio obtained by image processing to the power of 1.5. a   b    41     Chapter 3: Results and Discussion In this section the results of our study are presented. We first discuss the reproducibility of the measurements followed by explaining the effects of experimental parameters on the z- direction calcium (filler) distribution. Then, the filler distribution profiles of our samples are analyzed and compared. We have also measured the calcium content of paper samples by ashing and compared the results to EDX measurements. Image processing technique is implemented on SEM images of samples to further validate the EDX results. Finally, we present the model developed from experimental data and validate it using a new data set. 3.1 Replication of experiments at the central point As explained in section 2-1, in order to have an estimate for the reproducibility of the experiment, 5 central runs were performed in the early stage of experiment. To this end, 5 paper samples with the same level of factors namely PCC, starch, CPAM, silica and vacuum were produced and analyzed for their z direction calcium distribution by using SEM/EDX technique and the sampling procedure explained in section 2.3.1. 10 15 20 25 30 35 40 PCC:35% Starch:10 Kg/t Floc:0.3 Kg/t Silica:0.3 Kg/t Vacuum:8 in Hg WSTS ED X C a (% ) Sheet Thickness Figure 3-1: Variation of calcium content in z direction of paper samples at the central point It should be noted that since the PCC filled paper samples were used in our study, we were interested in measuring the calcium concentration. The EDX software program considered the total elemental composition in the selected area as 100 wt% and detected the percentage of EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     42   calcium present in that area in relation to other elements. As a result, it reported the calcium concentration based on the weight % along with other elemental composition. The CaCO3 percent value can be obtained by stoichiometric conversion. Figure 3-1 shows the variation of calcium content through the thickness of the paper samples (central runs) from top side to the wire side (the side which is in contact with the forming fabric). In this graph, each point is the average of 40 measurements. For each paper sample, the calcium concentration for four specimens and two viewing areas per specimen was measured (4 specimens× 2 viewing areas). The error bars represent the standard error of mean. The error values are reported in table 3.1. As the error values are quite small, we can consider our experiment, as a reproducible one meaning large variation across replicates was not found. Table 3-1: Standard error of mean for central runs Layer Standard error of mean 1 0.555 2 0.615 3 0.738 4 0.787 5 1.454 3.2 Effect of experimental parameters on filler distribution The purpose of our study was to understand the effect of different factors on the trend of filler distribution and to construct a predictive model based on the significant factors. To this end, 21 paper samples having five different levels of chemical additives and vacuum were produced using the modified hand sheet former (section 2.2). The profile of filler distribution through the thickness of the paper samples from top side to the wire side were obtained by the technique elaborated in section 2.3.1. To analyze the effect of chemical additives and vacuum on the filler distribution in each layer, we employed the analysis of variance (ANOVA). The four factors (independent variables) were PCC, starch, retention aids and vacuum, each with five value levels. The five dependent variables were the amount of filler in five layers of paper samples. Table 3-2 (Refer to Appendix C for ANOVA details) shows the factors that had significant, very EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     43   significant and extremely significant main effect (linear effect), interaction effect and quadratic effect on the dependent variable (filler distribution). As seen in the table, the main effects of all parameters were statistically significant in layer 4 and 5. The effect of starch on filler concentration in all layers except the layer near the wire side was found to be extremely significant while the effect of retention aids is significant only in layers near the wire sides. Also, the quadratic effect of starch is extremely significant in layers near the top side and the interaction effect between PCC and vacuum was found to be extremely significant in all layers except the layer near the wire side. Information obtained from this table was used to build the predictive models, which are illustrated in section 3.4. Table 3-2: Effect of experiment parameters on filler distribution Effect type Main Interaction Quadratic Fa ct or PC C St ar ch C PA M V ac uu m PC C *S ta rc h PC C *C PA M PC C *V ac uu m St ar ch *C PA M St ar ch *V ac uu m C PA M *V ac uu m PC C *P C C St ar ch *S ta rc h C PA M *C PA M V ac uu m *V ac uu m Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 Extremely significant (p-value < 0.0001) Very significant ( 0.0001 < p-value < 0.01 ) Significant (0.01 < p-value < 0.05) The analysis of variance also provided the prediction profiles and the 95% confidence intervals for filler concentration in each layer from top side to the wire side of the paper samples based on parameters of interest. The prediction profiles are shown in Figure 3-2. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     44   Figure 3-2: Prediction profiles for filler distribution through the thickness of the paper As seen in the figure, PCC and starch had a positive effect on the filler concentration for all the layers. The higher the loading level of PCC and starch the more filler would be retained through the thickness of the paper. However, the slopes of the prediction curves for starch in La ye r1 39.58 11.16 17.0525 La ye r2 43.83 10.87 18.015 La ye r3 41.06 13.14 18.65375 La ye r4 37.4 11.567 22.49875 La ye r5 56.47 15.079 29.2925 PCC 25 4535 Starch 8 1210 CPAM/Silica 0. 1 0. 5 0.3 Vacuum 2 148 EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     45   layers near the top side were greater than that of layers near the wire side. Also, increasing the level of retention aids has increased the filler concentration in layers near the wire side while the layers near the top side have not been affected significantly. In addition, vacuum had a positive effect on the filler concentration in layers near the wire side and a small negative effect in layers near the top side. Therefore, PCC, starch, chemical retention aids and vacuum have played significant role in changing the filler concentration in layers near the wire side (layer 4 and layer 5) as seen from the steepness of the prediction curves. For layers near the top side (layer 1-3), increasing the level of starch extremely increased the filler concentration while increasing the level of the other factors did not have significant effect on the response. We have performed further analysis to investigate the effect of experimental factors on filler distribution. In the following, the effect of these factors namely PCC, starch, retention aid and vacuum is discussed in more detail. 3.2.1 Effect of PCC This section explains the effect of different loadings of PCC on z directional filler distribution in hand sheets. As it was seen in the prediction profiles, increasing the level of PCC resulted to more filler in the paper samples. More specifically, the slopes of the prediction curves for PCC in layers near the wire side were greater than layers near the top side. To further analyze the effect of PCC, we picked three paper samples with three different PCC loadings (25%, 35%, 45%), but the same value for starch (10 kg/t), retention aids (0.3 kg/t) and vacuum (-8 inHg). Each sample provided 8 (4 specimens × 2 viewing areas) filler concentration measurements for each layer of paper sample. Figure 3-3(a) shows the filler profile of these three samples. Each point in this graph is the average of 8 filler concentration measurements. The 95% confidence intervals for the average values are shown by error bars above and below each point. Error bars represent the standard error of mean. A one-way between subject ANOVA followed by a Tukey post hoc test was conducted to compare the effect of three loadings of PCC on filler concentration. The result showed that increasing the level of PCC from 25% to 35% did not change the filler amount in any of the EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     46   layers significantly but changing the level of PCC from 25% or 35% to 45% increased the amount of filler in all the layers significantly. These effects can be observed in Figure 3-3(a). We also compared the effect of layer on filler distribution when these three loadings of PCC were used. For this purpose, a two way between subject ANOVA followed by a Tukey post hoc2 test was conducted where the layer and PCC were the independent variables and filler concentration was the dependent variable. The result showed that the effect of layer and PCC is significant while their interaction is not. The post hoc test indicated that the filler distribution in layers near the top side (layer 1, 2 and 3) were similar while the filler concentration in layer near the wire side (layer 4 and 5) was significantly higher than top side layers. (Except that layer 4 and 3 was not significantly different). Also, layer 5 had a significantly higher filler concentration than layer 4. Figure 3-3 (b) illustrates these effects. In summary, the filler concentration is maximized in all layers by using the highest level of PCC indicating an increased retention (as expected) that is observed across the z-direction. Layer five has the highest filler concentration independent of the level of PCC. 10 15 20 25 30 35 40 45 50 Filler:45% Filler:35% Filler:25% WSTS ED X C a (% ) Sheet Thickness (a) (b) Figure 3-3: Effect of increasing PCC on filler concentration The observed effect of PCC on filler concentration can be explained by the mechanisms that filler particles are retained during the hand sheet forming. The filler particles are retained through attachment to the fibres with the aid of the retention aids and also through filtration by                                                                                                                          2  While  the  ANOVA  indicates  whether  the  groups  in  sample  differ,  the  Tukey  test  determines  which  group  differs.   15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 E D X C a (% ) PCC (%) EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     47   the fiber mat. In the first mechanism, the anionic PCC filler particles attach to the cationic retention aids through electrostatic attraction. As a result, the filler particles form agglomerates that are large enough to be filtered by the fibre mat. As the suspension drains through the forming fabric, agglomerated fillers are deposited and entrapped in the pulp mat through the filtration mechanism whereas the unbound fillers are passed through the forming fabric [23]. During the formation of hand sheets in our modified former, applying suction increases the flow rate of drainage and drags the agglomerate filler toward the bottom layers i.e layers near the wire side which have the highest specific filtration resistance (SFR)[33]. This explains why we observed the highest filler concentration on the layer near the wire side. Moreover, we observed the most increase of filler concentration for the highest level of PCC. As the PCC content increases, more and larger flocs (agglomerated fillers) are formed because increasing PCC facilitates the particle collision and more efficient flocculation occurs [45]. These high density agglomerated fillers are retained through filtration along the fibre mat. The fiber mat is formed during the initial stage of formation and acts as a sieve for agglomerated fillers. 3.2.2 Effect of starch This section explains the effect of five different loadings of starch on z directional filler concentration through the thickness of the paper samples. As it was shown in the prediction profiles, starch had a positive effect on the filler concentration in all the layers. The higher the loading level of starch the more filler was retained through the thickness of the paper. However, the slopes of the prediction curves for starch in layers near the top side were greater than the layers near the wire side. To further analyze the effect of starch, we selected three paper samples with three different loadings of starch (8, 10, 12 kg/t), but the same value for PCC (35%), retention aid (0.3 kg/t) and vacuum (-8 inHg). Each sample provided 8 (4 specimens × 2 viewing areas) filler concentration measurements for each layer of paper sample. Figure 3-4 shows the filler profile of these three samples. Each point in this graph is the average of 8 filler concentration measurements. The 95% confidence intervals for the average values are shown by error bars above and below each point. Error bars represent the standard error of mean. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     48   A one-way between subject (samples) ANOVA followed by a Tukey post hoc test was conducted to compare the effect of three loadings of starch on filler concentration. The result showed that increasing the level of starch from 8 to 10 did not change the filler amount in any of the layers significantly but changing the level of starch from 8 or 10 to 12 increased the amount of filler in all the layers significantly. These effects can be observed in Figure 3-4(a). 10 15 20 25 30 35 40 45 50 starch:12 kg/t starch:10 kg/t starch: 8 kg/t WSTS ED X C a (% ) Sheet Thickness (a) 6 8 10 12 14 0 5 10 15 20 25 30 35 40 45 Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 E D X C a (% ) Starch (kg/t) (b) Figure 3-4: Effect of increasing starch on filler concentration We also compared the effect of layer on filler distribution when these three loadings of starch were used. For this purpose, a two way between subject ANOVA followed by a Tukey post hoc test was conducted where the layer and starch were the independent variables and filler concentration was the dependent variable. The result showed that the effect of layer, starch and their interaction is significant. The pairwise comparison indicated that increasing the level starch from 8 and 10 to 12 increases the filler concentration significantly in all the layers and this increase in layer 5 is significantly higher than other layers. Figure 3-4(b) illustrates these effects. In summary, the filler concentration is maximized in all layers by using the highest level of starch. Also, layer five has the highest filler concentration independent of the level of starch. As explained in section 1.4, besides increasing the paper strength and eliminating the need for a fixation agent to deal with the dissolved and colloidal substances, starch improves the retention [51]. In our study, cationic tapioca starch S880 was selected to be used as it contributes EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     49   to stronger paper, better drainage and more compact flocs compared to other types of starch because of its high charge density [69]. Although, both starch and retention aids improve the filler concentration, the effect of starch was more significant than the effect of retention aids in our study as it increased the filler concentration in all the layers significantly; however, retention aids only enhanced the filler amount in layer 5. The reason is that the starch type we used plays a more significant role in retention that CPAM. It was shown in [69] that when starch S880 was used, increasing the starch enhanced the filler concentration more compared to increasing the level of CPAM. However, when starch S858 was used, increasing CPAM significantly enhances the filler amount. The reason of these effects is that starch S880 has a higher charge density than S858 and introduces more cationic charge to the flocculation system. As a result, more flocs are created and retention improves. 3.2.3 Effect of retention aids This section explains the effect of five different loadings of retention aids on z directional filler concentration through the thickness of the paper samples. As it was shown in the prediction profiles, increasing the level of retention aids only increases the filler concentration in layers near the wire side. To further analyze the effect of retention aids, we picked three paper samples with three different loadings of retention aids (0.1, 0.3, 0.5 kg/t), but the same amount of dry mass PCC (35%), starch (10 kg/t) and vacuum drainage set to -8 inHg. Each sample provided 8 (4 specimens × 2 viewing areas) filler concentration measurements for each layer of paper sample. Figure 3-5(a) shows the filler profile of these three samples. The 95% confidence intervals for the average values are shown by error bars above and below each point. Error bars represent the standard error of mean. A one-way between subject ANOVA followed by a Tukey post hoc test was conducted to compare the effect of three loadings of retention aids on filler concentration. The result showed there is no statistically significant difference between the three levels of retention aids in layers 1,2,3 and 4. (Except that there was a significant decrease in filler concentration in layer 2 when the retention aids was increased from 0.1 to 0.5). Only in layer 5, increasing the retention aid EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     50   from 0.1 and 0.3 to 0.5 increased the filler concentration significantly. These effects can be observed in Figure 3-5 (a). We also looked at the filler % in each layer when these three loadings (0.1, 0.3 and 0.5) of retention aids were used. For this purpose, a two way between subject ANOVA followed by a Tukey post hoc test was conducted where the layer and retention aids were the independent variables and filler concentration was the dependent variable. The result showed that there is a significant interaction between layer and retention aids. The pairwise comparison indicated that there is significant difference between the retention aid levels of 0.1 and 0.5 as well as 0.3 and 0.5 in layer 5. However, there is no significant difference between different levels of retention aids in layer 1, 2, 3 and 4. (Except that there was a significant decrease in layer 2 when changing the retention aids from 0.1 to 0.5). These effects are shown in Figure 3-5 (b). (a) (b) Figure 3-5: Effect of increasing retention aids on filler concentration In summary, the increase in filler concentration was observed only in layer 5 when the highest level of retention aids was used. The retention aids used in our study were cationic PAM (CPAM) and anionic colloidal silica. The high molecular mass, low charge density CPAM helps bind the micro particles to each other and to the larger fibres through macromolecular bridging. This increases the flocculation and the average filler particle size grows from 10.22µm to 160.20µm [69]. As a result, the small filler particles are kept from being drained from the wet web of the paper during the dewatering process and more filler and fine particles retain in the final sheet of paper. As the dosage of 5 10 15 20 25 30 35 40 45 50 CPAM-silica: 0.5 kg/t CPAM-silica: 0.3 kg/t CPAM-silica: 0.1 kg/t WSTS ED X C a (% ) Sheet Thickness 0.0 0.2 0.4 0.6 0 5 10 15 20 25 30 35 40 45 Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 ED X C a (% ) Retention aids (kg/t) EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     51   retention aids in pulp suspension increases, more bridges between the filler particles are created which induces better flocculation. The increased size of filler agglomerate increases the probability of filtration retention. The effect of increasing the amount of retention aids was mostly observed on the layers near the wire side where the flocs are dragged under the vacuum drainage. In layers near the top side there is no the variation of filler concentration by increasing the amount of retention aids because the larger flocs which are created by having more level of retention aids were dragged toward the bottom side of the paper sample. 3.2.4 Effect of vacuum In this section the effect of increasing the dewatering rate in suction box on z directional filler content through the thickness of the paper samples is examined. As it was shown in the prediction profiles, increasing vacuum (applying more suction) led to a slight decrease in filler content in layers near the top side, however, it has a positive effect on the filler concentration in layers near the wire side and the slope of the prediction curves in layers near the wire side were much greater than layers near the top side. To further explore the effect of vacuum, we picked three paper samples with three different levels of vacuum in suction box (-2, -8, -14 inHg), but the same amount of dry mass PCC (35%), starch (10 kg/t) and retention aids set to 0.3 kg/t. Each sample provided 8 (4 specimens × 2 viewing areas) filler concentration measurements for each layer of paper sample. Figure 3-6 shows the filler profile of these three samples. The 95% confidence intervals for the average values are shown by error bars above and below each point. Error bars represent the standard error of mean. A one-way between subject ANOVA followed by a Tukey post hoc test was conducted to compare the effect of three levels of vacuum on filler concentration. The result showed there is no statistically significant difference between the three levels of vacuum in layers 1,2 and 3. (Except that there was a significant decrease in filler concentration in layer 2 when the vacuum was increased from 2 to 8). In layer 4 and 5, increasing the vacuum from 2 and 8 to 14 increased the filler concentration significantly. These effects can be observed in Figure 3-6 (a). EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     52   A two way between subject ANOVA followed by a Tukey post hoc test was conducted where the layer and vacuum were the independent variables and filler concentration was the dependent variable. The result showed that the effect of vacuum, layer and their interaction is significant. The pairwise comparison indicated that there is significant difference between the vacuum levels of 2 and 14 as well as 8 and 14 in layer 4 and 5. Also, vacuum level of 2 was different from level of 8 in layer 5. However, there is no significant difference between different levels of vacuum in layer 1,2 and 3. (Except that there was a significant decrease in layer 2 when changing the vacuum from 2 to 8). These effects are shown in Figure 3-6 (b). 10 15 20 25 30 35 40 45 50 55 vacuum:-14 inHg vacuum:-8 inHg vacuum:-2 inHg WSTS ED X C a (% ) Sheet Thickness (a) 0 5 10 15 0 5 10 15 20 25 30 35 40 45 Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 E D X C a (% ) Vacuum (inHg) (b) Figure 3-6: Effect of increasing vacuum on filler concentration In summary, the increase in filler concentration was observed in layers 4 and 5 when the highest level of vacuum was applied. On the other hand, the applied vacuum did not change the filler distribution on layers near the top side. It is hypothesized that the formation of top side layers (in x-y plane) occurs with a relatively uniform flow field compared to the wire side layers. The uniform flow field is due to the fact that as we move a distance upstream from an object in a flow, the object has less and less effect on the flow field, until it is not noticeable. So, since it is the forming fabric that makes the flow non-uniform, at a distance upstream of the fabric the flow will return to unformity. In the higher levels of vacuum, the velocity gradients increases in layers near the wire side where there is the smallest open area and highest specific filtration resistance (SFR). Consequently, the filler distribution is more affected in layers near the wire side than layers near the top side where uniform flow field governs the layers formation. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     53   Although some previous studies showed that applying vacuum does not impact the filler distribution, more recent studies have confirmed the effect [54]. The author of [54] compared the effect applying vacuum vs gravity drainage and showed that using vacuum reduces the filler concentration at the wire side. When suction is applied, the average flow rate increases. This increase is highest at the wire side because it has the smallest open area and thus highest local surface filtration resistance [11]. The increased flow rate drags filler particles more intensely. Specifically, the filler particles that were retained by attachment to pulp fibres are largely affected by this increased flow and a larger number of them are detached from the fibres with increased suction. However, there was no use of starch in preparing the paper samples of this study, which could lead to formation of weaker filler agglomerates. Therefore, it is speculated that the filler agglomerates created in our study were stronger and filler particles did not detach from the flocs. As the suction increased, more filler agglomerates are dragged from the top side to the wire side. As a result, applying suction in our study increased the filler content in layer near the wire side due to clogging of the screen. 3.3 Filler distribution profile In this section, we explain how the filler concentration has changed through the paper sample layers. For this purpose, first, the effect of layer on filler concentration is explained and second, the effect of experiment condition on filler profile is discussed. 3.3.1 Effect of layer To compare the effect of paper sample layer on filler concentration, a one-way between subject ANOVA was conducted, where layer was considered as independent variable and filler concentration was the dependent variable. The total data set size was 680 as for each of five layers, there was 136 data samples (17 runs × 8 viewing areas). The results indicated that the filler concentration differed significantly across the five layers, F(4, 675) = 95.263, p < 0.0001. Tukey post-hoc comparison of the five layers indicated that the filler concentration in layer 5 (M=33.88) was significantly higher than other layers, p < 0.0001. The same situation applies for layer 4 except the filler concentration in this layer was not significantly different from layer 3. On the other hand, the filler concentration in layer 1, 2 and 3 did not differ significantly, p > 0.05. In summary, the filler concentration was significantly higher in layer 4 and 5 than other layers while the filler concentration in layer 1,2,3 was similar. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     54   The lowest layers of the fiber mat close to the forming fabric (layer 4 and 5) are formed during the initial stages of formation and thus captured particles for the longest time, while the top layers have formed in the final stages and are subjected to the shortest capture time [12]. Therefore, more filler particles are retained in layers near the wire side rather than layers near the top side. Also, as it was mentioned previously the formation of top side layers occurs with a relatively uniform flow field. Moreover, during the hand sheet formation, the first layers of deposited particles tend to form over open areas. Higher flow rate over open areas cause higher rate of deposition. This characteristic of formation tends to smooth out any existing flow flied non- uniformity as more layers are deposited. Hence the top side of the sheet would become more uniform. 3.3.2 Effect of experimental conditions To further analyze the effect of experimental parameters on filler distribution, we investigated which experiment condition produced a more even filler distribution, i.e. the filler concentration was similar through the thickness of the paper. We used the following procedure to identify such experiment conditions: 1- Since there were 17 distinct experiment conditions in our study, seventeen between subject one-way ANOVAs were conducted to compare the effect of layer on filler concentration in each condition. The layer was the independent variable and filler concentration was dependent variable. The effect of layer in all seventeen experiments conditions was statistically significant. 2- The post-hoc tukey test was conducted for each condition. 3- The number of significantly different layers in each condition was counted. 4- The conditions with the least number of significant layers were considered as conditions with most even filler distribution. Table 3-3 shows the conditions with most and least even filler distribution profile according to the above procedure. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     55   Table 3-3: Experiment conditions with least and most event filler distribution profile Filler distribution profile Sample ID PCC (%) Starch (kg/t) CPAM-Silica (kg/t) Vacuum (inHg) Most even 3 6 14 30 40 35 11 11 10 0.4 0.2 0.1 11 5 8 Least even 15 19 21 35 35 45 10 10 10 0.5 0.3 0.3 8 14 8 As seen in table 3-3 the least even profiles were observed when the highest level of retention aids, PCC and vacuum were applied. This can be related to the particle flocculation where at higher level of retention aids and PCC, larger filler agglomerates are created which can lead to uneven filler distribution through the paper thickness. It is noted that Tanaka et al reported that the z direction filler distribution profile becomes more even as retention increases [80]. The dosages of retention aids used in this study were 0.05, 0.1 and 0.2% CPAM. Therefore, it can be concluded that increasing retention aids dosage beyond certain level reduces the evenness of the z directional filler distribution. 3.4 Validating the EDX measurements by ashing technique As explained in section 2.4, we have employed the ashing technique to determine the filler retention of paper samples and calibrate the EDX measurements. The calcium content data of paper samples obtained by EDX and ashing are illustrated in Figure 3-7. As shown in the diagram, the EDX calcium is higher from the calculated calcium measured from ash technique Assuming that the solid remained in the ash sample is CaCO3, the calcium amount was calculated based on the stoichiometric conversion. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     56   15 20 25 30 35 40 45 50 5 10 15 20 25 30 EDX calcium Calculated calcium C a (% ) PCC (%) Figure 3-7: Calcium content of paper samples measured by ashing and EDX The correlation between calcium content measured by EDX and the ash content is given by the following equation: Y = 0.704 X - 1.1609 where Y = ash in paper (%) and X = EDX calcium (%). 10 15 20 25 30 35 15 20 25 30 35 40 45 As h (% ) EDX Ca (%) y  =  0.704x  -­‐  1.1609   R²  =  0.8297     Figure 3-8: Relation between EDX calcium and ash percent EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     57   In Figure 3-8, the value of R2 indicates that there is a reasonably good correlation between EDX calcium (%) and ash (%) in paper samples. 3.5 Image processing Image processing of the SEM images was conducted. For this purpose, each image has been divided into five layers and the mass ratio is measured by EDX and calculated based on the image processing methodology. Figure 3-9 shows a sample result. Figure 3-9: (a) SEM image of cross sectional of paper (b) Split cross sectional images (c) SEM images of split cross sectional of paper after applying convolution filter (d) Converted images after applying convolution filter to black and white image EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     58   In Figure 3-9 (a) is a portion of an actual picture, which has been taken by SEM and can be analyzed by EDX. For the purpose of the image processing, first we split the image into five different layers in the thickness direction (b). After splitting the image, the code applies the convolution kernel (Eq. 3) on the whole image to identify all the particles (filler) in the whole image (c) and finally a threshold is applied to the different layer images for binarizing images and getting them ready for mass ratio calculations. The results of the image processing compared to the EDX results are shown in Figure 3-10 for all layers of all images. Figure 3-10: Comparison of EDX and image processing results for smaller areas This shows that the results of the EDX compared to those of our image processing technique for some cases are in a reasonable agreement and for some cases they are not following the reasonable trend. To our knowledge the source of errors and discrepancies would be because of the following reasons: 1. The first reason for the observed discrepancies (sources of the errors) is because of using image processing convolution filter which makes changes in the real image to identify the particles. 2. Using threshold which binarize images i.e. changes the actual signal to 0 and 1 (binary) signal so that some part of the actual data change. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     59   3. The images are 2D whereas the observed particles are 3D. Computing the mass ratio from the area ratio in 2D images which is calculated by image processing needs the shape of the particles to be considered as spherical objects which are actually not. This is just an assumption to let the problem mathematically tractable. However this propagates some error. 4. The quality of some images are low because of the imaging issues so the filter can’t be applied equally to all of the Images. This can also be considered as a source of error. For more clarification, the correlation analysis between the data of EDX and Image processing for each sample is calculated. Figure 3-11, shows the results of the correlation. Figure 3-11: Correlation analysis between the data of EDX and Image processing for smaller areas As we can see in Figure 3-11, the values of the correlation function between EDX and Image processing data for each sample are close to 1 for some cases and for some other cases they do not show a reasonable agreement. After validating our technique, we decided to do image processing for the larger Images which EDX was not been able to calculate the mass ratio. The same procedure has employed to analyze the large images. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     60   The results of the image processing in terms of the mass ratio compared to those of EDX in limited area are in a good agreement with each other for some samples. This means that the mass ratio of the smaller area can be considered as a good representative of the mass ratio of the whole area for samples that they are in a good agreement (Figure 3-12). The correlation data in Figure 3-13 confirms this idea. Figure 3-12: Comparison of EDX for smaller areas and image processing results for larger areas Figure 3-13: Correlation analysis between the data of EDX (smaller areas) and Image processing for larger areas EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     61   3.6 Model construction and validation As it was explained in section 2.1, the data obtained from five layers through the thickness of the hand sheet were used separately in order to fit a second order model. Regression models were constructed based on the statistical analysis to predict the z directional filler distribution in paper samples. The equations indicate the estimated parameters for all main, interaction and quadratic effects. The different magnitude of the effects and potentially different significant effect for five layers are the result of interaction between the layers and the factors. The filler amount for any combination of the factor levels can be predicted using these models. The regression models are shown in Table 3-4. The x1,x2,x3,x4 parameters were defined in Table 2-1. Table 3-4: The model equations to predict the filler concentration in paper layers Layer Equation 1 y1 =1190.91−30.32x1 − 22.09x2 +104.61x3 −142.37x4 − 0.24x1x2 +1.74x1x3 + 4.06x1x4 −16.54x2x3 +1.95x22 2 y2 = 85.57+ 0.70x1 − 20.90x2 +84.84x3 − 2.13x4 − 0.24x1x2 +1.91x1x3 + 0.18x1x4 −15.16x2x3   −0.41x2x4 + 2.04x22 3 y3 =125.96+ 0.39x1 −31.34x2 +121.07x3 − 7.31x4 + 0.21x1x4 −12.18x2x3 +1.93x22 4 y4 = −36.46− 0.39x1 +160x3 − 4.56x4 + 6.85x2 + 0.12x1x4 −16x2x3 + 0.085x42 5 y5 = −21.11+−0.79x1 + 7.93x2 − 9.93x4 + 209.31x3 + 0.17x1x4 −19.03x2x3 + 0.20x42 The predictive capability of the regression models was examined by comparing the model predicted values with the results of 3 additional experiments (other than those used in the ANOVA). Table 3-5 shows the experimental conditions of paper samples that were used for model validation. Figure 3-14 shows the experimental and predicted values for filler content through the thickness of the paper samples. It can be seen that the model can predict the trend of filler distribution with good agreement compared to experimental data. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 3     62   Table 3-5: Level of experiment factor for three compared samples Diagram   PCC(%)   Starch(kg/t)   Floc(kg/t)   Vacuum(inhg)   A   25   10   0.3   8   B   35   10   0.3   8   C   45   10   0.3   8     5 10 15 20 25 30 35 40 45 Model Experiment WSTS ED X C a (% ) Sheet Thickness 5 10 15 20 25 30 35 40 45 Model Experiment WSTS ED X C a (% ) Sheet Thickness (A)   (B) 5 10 15 20 25 30 35 40 45 Model Experiment WSTS ED X C a (% ) Sheet Thickness (C) Figure 3-14: Experimental and predicated values for z-direction filler distribution 63     Chapter 4: Conclusions and Recommendations Even though the importance of the type of former on z direction filler distribution has been noted in prior research, less attention has been paid to the role and use of chemical additives as a tool to control the filler distribution through the thickness of paper. Therefore, the scope of this thesis was to investigate the effect of chemical additives and a machine parameter (vacuum), on the z direction filler distribution in paper. The paper making process was simulated by using a modified hand sheet former, which provides vacuum. Paper samples with different levels of filler, starch, chemical retention aids and vacuum were formed and analyzed for their z direction filler distribution by using SEM/EDX technique. It was shown that all the studied parameters had statistically significant effect on the filler concentration in the layers near the wire side (close to the forming fabric) of the paper and only starch had significant effect on filler concentration in the layers near the top side. Increasing the amount of filler, starch and retention aids led to efficient flocculation. As a result, more and larger agglomerates of fillers were created and the probability of filtration retention increased. During the vacuum drainage, the agglomerated fillers are dragged toward the layers near the wire side where we observed the significant effect of the parameters. The high charge density starch used in this study had a significant role in filler retention through the paper thickness as it produces strong compact flocs. Using the highest level of PCC and starch dosage maximized the filler concentrations in all layers. However, using the highest level of retention aids and vacuum led to an increase in filler amount only in layers near the wire side. The increase of filler amount in layer near the wire side at the highest level of vacuum (more suction) indicates that the filler agglomerates created did not detach from the flocs by applying suction. This highlights the effective role of chemical additives in retaining the filler particles. The trend of z direction filler distribution of paper samples was found to be increasing from the top side to the wire side. This is due to the longest capture time for the layers near the wire side formed during the initial stage of formation. Also, the filler concentration did not differ significantly in layers near the top side. Therefore, it is hypothesised that the formation of top side layers occurs with a relatively uniform flow field. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER CHAPTER 4     64   By using the highest level of PCC, retention aids or vacuum, the least even z direction profiles were obtained which is due to the uneven distribution of filler agglomerates. To validate our calcium measurement by EDX, we independently measured the ash content of the paper. The results showed that there is a reasonably good correlation between EDX calcium and ash content of the paper samples. To mitigate the limitation of EDX in analyzing the large area of SEM images, we applied the image processing technique. A very accurate convolution-based filter was developed to capture the signals related to filler particles. The technique was initially validated with the experimental results obtained for small areas through EDX and subsequently was used for the larger areas of the paper samples. The results illustrated that weight percent reported by EDX for the small area of the sample provides an accurate estimate for the total area of the image. Based on the experimental results, a set of empirical models are developed that can predict the filler distribution in different layers through the thickness of the paper under various conditions. For future work, additional experiments can be carried out to further investigate the effect of chemical additives on z direction filler distribution in paper. In our study, we only used the chemicals that are currently used in industry. Moreover, the levels of chemicals were chosen based on industrial practice. We made these choices to simulate the current paper making process. To continue this research the effect of other chemicals such as other types of filler (e.g talc, clay) and other starches (e.g potato starch) on z direction profile of paper can be studied. Moreover, studying the effect of additional levels of chemicals used in this study is another area for future research. In our study, one type of forming fabric was used. The effect of the geometry of forming fabric on filler distribution through using different types of the fabrics can be investigated with the same procedure and technique implemented in this work. The shear rate of mixing the pulp suspension can be varied to study the effect of shear rate on filler flocs formation and filler distribution in paper samples.  65     References [1] http://en.wikipedia.org/wiki/Paper_machine, March 2013. [2] Ahmed J. et al., Starch-Based Polymeric Materials and Nanocomposites: Chemistry, Processing, and Applications CRC Press, 2012. [3] Ain, R. L., and Laleg, M. “Mill experiences with AT™ precipitated calcium carbonate (PCC) in papers containing mechanical pulp: New technology stabilizes carbonate and pH at neutral”, Pulp Paper Canada 98(12), pp. 172-176, 1997. [4] Alince, B., Cationic latex as a multifunctional wet-end additive in highly filled paper, TAPPI Journal, vol 3, pp. 16-18, 2004. [5] Allem, R., “Characterization of Paper Coatings by Scanning Electron Microscopy and Image Analysis”, Journal of Pulp Paper Science, vol. 24, no. 10, pp. 329-336, 1998. [6] Asselman, T., Alince, B., Garnier, G., Van de Ven, T.G.M., “Mechanism of polyacrylamide-bentonite - microparticulate retention aids”, Nordic Pulp & Paper Research Journal, vol. 15, pp. 515-519, 2000. [7] Blanco et al., Papermaking Chemistry, Fapet Oy, 1999. [8] Bown, R., “Particle size, shape and structure of paper fillers and their effect on paper properties”, Paper Technology, vol. 39, no. 2, pp. 44-48, 1998. [9] Brannvall, E., Eriksson, M., Lindstrom, M.E., Wagberg, L., Fibre surface modifications of market pulp by consecutive treatments with cationic and anionic starch, Nordic Pulp & Paper Research Journal 22, pp. 244-248, 2007. [10] Cadotte, M. et al., “Effects of various retention aids on fiber flocculation, filler retention and drainage”, PAPTAC 91st Annual Meeting: A31-A37, 2005. [11] Chinga, G. and Helle, T., “Three-Dimensional Reconstruction of a Coating Layer Structure”, Journal of Pulp Paper Science, vol. 29, no. 4, pp. 119-122, 2003. [12] Clark, Andrew Clark, “ Investigation of factors contributing to the deposition of contaminant on dryer cylinders”, Georgia Institute of Technology, May 2007 [13] Chinga, G. and Helle, T., “Structure characterisation of pigment coating layer on paper by scanning electron microscopy and image analysis”, Nord. Pulp Paper Res. Journal, vol. 17, no. 3, pp. 307-312, 2002. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER REFERENCES     66   [14] De Silveira, G. et al., “Location of Fines in Mechanical Pulp Handsheets Using Scanning Electron Microscopy”, Journal of Pulp Paper Science, vol. 22, no. 9, pp. 315-320, 1996. [15] Delgado, H.M.S, and Gomes, C.S.F, “X-ray diffraction for Quantifying Calcium Carbonate Fillers in Printing and Writing Papers”, Tappi, vol. 78, no. 4, pp. 135- 139, 1995. [16] Dickson, A.R., “Quantitative analysis of paper cross-sections”, Appita Journal, vol. 53, no. 4, pp. 292-295, 2000. [17] Douglas, C., Montgomery, J., Design and Analysis of Experiments, 2008, John Wiley & Sons, Inc. [18] Dong, C., Song, D., Patterson, T., Ragauska A. and Y. Deng, “Energy Saving in Papermaking through Filler Addition”, Ind. Eng. Chem. Res., vol. 47, no. 21, pp. 8430 - 8435, 2008. [19] Egerton, R.F., Physical Principles of Electron Microscopy. New York: Springer, 2005. [20] Eklund, et al., Paper chemistry An Introduction, DTPaper Science publication, Grankulla, Finland ; 1991. [21] Evans, D. B., Drummond, D. K., and Koppelman, M. H. “PCC fillers for groundwood papers”, Proceedings from 1991 Papermakers Conference, pp. 321- 330. [22] Evanse, B., “PCC fillers for groundwood papers”, Papermakers conference, 1991. [23] Gaudreault, R., Di Cesare, N., van de Ven, T.G.M., Weitz D., The Structure and Strength of Flocs of Precipitated Calcium Carbonate Induced by Various Polymers used in Papermaking, Advances in Pulp & Paper Research, The 14th FRC Symposium, St-Anne’s College, Oxford, UK, 13-18 September, 2009, 1193-1219. [24] Gavelin, G., Paper Machine Design and Operation, Vancouver: Angus Wilde Publications, 1998. [25] Gibbon, D. et al., “New electron and light optical technique for examining papermaking”, Tappi, vol. 72, no. 10, pp. 87-91, 1989. [26] Gill, R. and Scott, W., “The relative effects of different calcium carbonate filler pigments on optical properties”, Tappi Journal., vol. 70, no. 1, pp. 93-99, 1987. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER REFERENCES     67   [27] Glittenberg, D., Hemmes, J.L., Bergh, N.O., Cationic starches in systems with high levels of anionic trash, Paper Technology 35, pp. 18-27, 1994. [28] Glittenberg, D., R.J. Tippett, P. Leonhardt, “Wet-end starches-still up to date for high speed papermaking”, Paper Technology 45, pp. 27-33, 2004. [29] Goldstein, J., Scanning Electron Microscopy and X-ray Microanalysis, New York: Kluwer Acadmeic/Plenum Publishers, 2003. [30] Griggs, W.H., “Use of fillers in papermaking”, Tappi Journal. vol. 71, no. 4, pp. 77-81, 1988. [31] Goodhew, P.J. et al., Electron microscopy and analysis. New York: Taylor and Francis, 2001. [32] Haller, T., PCC Application Strategies to Improve Papermaking Profitability. Part I. Thick Stock Precipitated Calcium Carbonate Addition, papermaker conference 2001. [33] Haggblom-Ahnger, U.M. et al., “Conventional and stratified forming of office paper grades”, Tappi Journal, vol. 81, no. 5, pp. 149-157, 1998. [34] Hansen, E., “The Distribution of Filler in Paper”, Tappi Journal, vol. 34, vol. 4, pp. 180-185, 1951. [35] Holmstad, R., Methods for paper structure characterisation by means of image analysis, PHD Thesis, Norwegian University of Science and Technology, 2004 [36] Hubbe, M.A., “Mechanistic aspects of microparticle systems”, Tappi Journal, vol. 4, pp. 23-28, 2005. [37] Huber, P., Pierre, C., Bermond, C., Carre, B., “Comparing the fiber flocculation behaviour of several wet-end retention systems”, Tappi Journal, vol. 3, pp. 19-24, 2004. [38] Ionides, G., Adding value for improved market competitiveness of mechanical printing papers, Pulp & Paper-Canada 105, pp. 40-44, 2004. [39] Karlsson, M., “Papermaking part 2: Drying,” in Tappi Press, Atlanta, 2000. [40] Kiviranta, A., Dodson, C.T.J., “Evaluating fourdrinier formation performance”, Journal of Pulp and Paper Science, vol. 21, pp. 379-383, 1995. [41] Kontturi, K.S., Tammelin, T., Johansson, L.S., Stenius, P., Adsorption of cationic starch on cellulose studied by QCM-D, Langmuir 24, 4743-4749, 2008. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER REFERENCES     68   [42] Krishnagopalan, A., and Simard, G.L, “An Improved Technique for Studying Binder Migration in Coated Paper”, Tappi, vol. 59, no. 12, pp. 96-99, 1976. [43] Laufmann, M., Pigments as filler , Handbook of paper and board, 2006. [44] Li, L., Collis, A., Pelton, R., “A new analysis of filler effects on paper strength”, J.ournal of Pulp Paper Science. 28, pp. 267-273, 2004. [45] Liimatainen, H., “Investigations between fibres, fines and fillers in papermaking” University of Oulu, 2009. [46] Liimatainen, H. et al., “Effect of PCC filler on dewatering of fiber suspensions”, Tappi Journal, vol. 5, no. 11, pp. 11-16, 2006. [47] Lindstrom, T., “Some Fundamental Chemical Aspects on Paper Forming”. Fundamentals of Papermaking. Ed. Baker, C.F. and Punton, V.W., London: Mechanical Engineering Publications, pp. 311-412, 1989. [48] Modgi, S.B. et al., “Retention of Calcium Carbonate in Mechanical Pulps with PEO/Cofactor/Coagulant and the Role of Hydrogen Bonding Interactions”, Journal of Pulp Paper Science, vol. 30, no. 11, pp. 293-298, 2004. [49] Modgi, S. et al. “SEM/EDX analysis: A technique for Z-directional mineral topography in paper”, Nord. Pulp Paper Res. Journal, vol. 21, no. 5, pp. 659-664, 2006. [50] Modgi, S. et al., “SEM/EDX analysis of Z-direction distribution of mineral content in paper along the cross-direction”, Pulp Paper Canada, vol. 107, no. 5, pp. 124- 127, 2006. [51] Modgi S, McQuaid ME, Englezos R, “Effect of cationic tapioca and potato starches on TMP fibres loaded with PCC”, Journal of Pulp and Paper Science, vol. 33, pp. 156-162, 2007. [52] Modgi, S., Interaction of fibres and additives in mechanical pulp suspensions, Doctoral dissertation, University of British Columbia, Vancouver, BC, Canada, 2007. [53] Modgi, S., Lauter, J., McQuaid, M.E., Englezos, P., Assessment of the need for a coagulant (poly DADMAC or PAC) with tapioca or potato starch in mechanical paper filled with PCC, Pulp and Paper Canada 111, pp. 22-22, 2010. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER REFERENCES     69   [54] Montgomery, J., The role of suction boxes on forming section retention and filler migration, M.A.Sc. Thesis, The University of British Columbia, 2010. [55] Newbury, D.E. et al., Advanced Scanning Electron Microscopy and X-Ray Microanalysis, Plenum Press, New York, NY, USA; 1986. [56] Odell, M.H., “Paper structure engineering”, Appita Journal, vol. 53, no. 5, pp. 371- 377, 2000. [57] Omya Paper Handbook, Klauser Weibel Design, Zürich 2005. [58] Ormerod, D., Simon, G., and Gibbon, D., “Paper mill additive and contaminant distribution mapping using energy dispersive X-ray analysis”, Tappi Journal, vol. 71, no. 4, pp. 211 - 214, 1988. [59] Park, S., Vendetti, R. A., Jameel, H., and Pawlak, J., “Hard-to-remove Water in Cellulose Fibers Characterized by Thermal Analysis: A Model for the Drying of Wood-based Fibers”, Tappi Journal, vol. 90, no. 6, pp. 10 - 16, 2007. [60] Peel, J. D., Paper Science and Paper Manufacture, Vancouver: Angus Wilde Publications, 1999. [61] Purrtinen, A., “Wet end Control of Filler Distribution Using Additive Layering Principles”, Paper Technology, vol. 44, no. 4, pp. 33-39, 2003. [62] Puurtinen, A., “Controlling filler distribution for improved fine paper properties”, Appita Journal, vol. 57, no. 3, pp. 204-227, 2004. [63] Raisanen, K., “The effects of retention aids, drainage conditions and pretreatment of slurry on high vacuum dewatering: a laboratory study”, in Papermakers Conference, pp. 463 - 467, San Francisco, 1994. [64] Raisanen, K. et al., “Vacuum dewatering optimization with different furnishes”, Paperi Ja Puu, vol. 78, no. 8, pp. 461-467, 1996. [65] Ramaswamy, S., “Vacuum Dewatering During Paper Manufacturing,” Drying Technology, pp. 685-717, 2003. [66] Reme, P.A., et al., “Assessment of Fibre Transverse Dimensions using SEM and Image Analysis”, Journal of Pulp Paper Science, vol. 28, no. 4, pp. 122-128, 2002. [67] Ridler, T.W. and Calvard, S., “Picture thresholding using an iterative selection method”, IEEE Trans. Syst. Man Cybern., vol 8, no. 8, pp. 630-632, 1978. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER REFERENCES     70   [68] Sang, Y., McQuaid, M., Englezos, P., Optimization of chemical use for highly filled mechanical grade papers with precipitated calcium carbonate, Bioresources 6, pp. 656-671, 2011. [69] Sang, Y., Application of nano-structured silica technology and modified starch biopolymers into highly precipitated calcium carbonate filled mechanical grade paper, University of British Columbia, 2012-09-19. [70] Scott, William E., Principles of wet end chemistry, Tappi Pr, 1996 [71] Shen et al. Acid-tolerant PCC in papermaking”, BioResources Journal, vol 4, no. 3, pp. 1178-1189, 2009. [72] Shirazi, M., Ven, T.G.M. van de, Garnier, G., Adsorption of modified starches on pulp fibers, Langmuir 19, pp.10835-10842, 2003. [73] Singh, R., Lavrykov, S., and Ramarao, B. V., “Permeability of pulp fibre mats with filler particles,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, vol. 333, pp. 96 - 107, 2009. [74] Smook, G.A., Handbook for Pulp & Paper Technologists, Second Edition Edition, Angus Wilde Publications, Vancouver, 1992. [75] Solberg, D. And Wagberg, L., “On the mechanism of GCC filler retention during dewatering – New techniques and initial findings”, Journal of Pulp Paper Science, vol. 28, no. 6, pp. 183-188, 2002. [76] Subramanian, R., Fordsmand, H., Paulapuro, H., Precipitated calcium carbonate (PCC) -Cellulose composite fillers: effect of PCC particle structure on the production andproperties of uncoated fine paper, Bioresources Journal, vol 2, pp. 91-105, 2007. [77] Subramanian, R., Fordsmand, H., Paltakari, J., Paulapuro, H., A new composite fine paper with high filler loading and functional cellulosic microfines, Journal of Pulp and Paper Science, vol 34, pp. 146-152, 2008. [78] Swerin, A., Sjodin, U., Odberg, L., “Flocculation of cellulosic fibre suspensions by model microparticulate retention aid systems: Effect of polymer charge density and type of microparticle”, Nordic Pulp & Paper Research Journal, vol. 8, 389-398, 1993. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER REFERENCES     71   [79] Szikla, Z., and Paulapuro, H., “Z-directional distribution of fines and filler material in the paper web under wet pressing conditions,” Pap. Puu, vol. 68, no. 9, p. 654, 1986. [80] Tanaka, H. et al., “How retention aids change the distribution of filler in paper”, Tappi Papermakers Conference, pp. 129-144, 1982. [81] Thomas, W.L, and Heitur, H.I., “X-Ray Fluorescence Spectroscopy, Its Use in Paper and Coating Analysis”, Tappi, vol. 61, no. 5, pp. 57-60, 1978. [82] Vandesteeg, H. G. M., , De Keizer, A., Cohen Stuart, M. A., Bijsterbosch, B. H., “Polyelectrolyte adsorption: a subtle balance of forces”, Langmuir, vol. 8, no. 10, pp. 2538-2546, 1992. [83] Vandesteeg, H.G.M., Dekeizer, A., Stuart, M.A.C., Bijsterbosch, B.H., Adsorption of Cationic Potato Starch on Microcrystalline Cellulose, Colloids and Surfaces A- Physicochemical and Engineering Aspects 70, pp. 91-103, 1993. [84] Vanerek, A. et al., “Interaction of Calcium Carbonate Fillers with Pulp Fibres: Effect of Surface Charge and Cationic Polyelectrolytes”, Journal of Pulp Paper Science, vol. 26, no. 9, pp. 317-322, 2000. [85] Vanerek, A. et al., “Collidal Behaviour of Ground and Precipitated Calcium Carbonate Fillers: Effects of Cationic Polyelectrolytes and Water Quality”, J. Pulp Paper Sci.26, no 4, pp. 135-139, 2000. [86] Voillot, C. et al., "The Use of X-ray Image Processing to Analyze the Z-Direction distribution of Fillers and Pigments", Tappi, 73(5): 191-193 1990. [87] Wagberg, L., and Kolar, K., “Adsorption of Cationic Starch on Fibres from Mechanical Pulps”, Ber.Bunsenges Physical Chemistry 100, no. 6, pp. 984-993, 1996. [88] Wahlstrom, P.B., “Paper Machine – Forming Section”. Handbook of Pulp and Paper Technology. Ed. Britt, K.W., New York: Van Nostrand Reinhold Company, pp. 401-426, 1970. [89] Xia, Q.S., Boyce, M.C. & Parks, D.M. A constitutive model for the anisotropic elastic-plastic deformation of paper and paperboard. Int. J. Solids Struc. Vol. 39, no. 15, pp. 4053–4071, 2002. EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER REFERENCES     72   [90] Yan, Z. and Deng, Y., “Cationic microparticle based flocculation and retention systems”, Chem. Eng. Journal, vol. 80, pp. 31-36, 2000. [91] Yoshizawa, J., Isogai, A., and Onabe, F. “Analysis and retention behaviour of cationic and amphoteric starches on handsheets,” Journal of Pulp and Paper Science, vol 24, no. 7, pp 213-218, 1998. [92] Zhang, X., The Dynamic Compressibility of Fibre Sheet during Suction Pulse with a Laboratory Suction Box, Master Thesis, Division Papermaking, STFI-Packforsk AB, Stockholm, Sweden, 2006.  73     Appendix A: Image processing code EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER APPENDIX     74    75     Appendix B: 2D convolution filter This appendix explains how a 2D convolution filter is applied on a 2D matrix and how the kernel, that we used for our image processing purpose, affects a 2D data structure. The definition of 2D convolution and the method for convolving in 2D are explained in the following by using an example. Imagine there are a 3×3 matrix and a kernel as defined in Eq. (1): 10 10 10 10 10 10 10 10 10 M ⎡ ⎤ ⎢ ⎥= ⎢ ⎥ ⎢ ⎥⎣ ⎦ (Matrix) 1 1 1 1 8 1 1 1 1 K − − −⎡ ⎤ ⎢ ⎥= − −⎢ ⎥ ⎢ ⎥− − −⎣ ⎦ (Kernel) (1) The origin of the kernel function is always centered (located in i = 0 and j = 0 by definition). In the 2D convolution filter, the center element of the kernel is placed over the source pixel. The source pixel is then replaced with a weighted sum of itself and the nearby pixels. This means that for convolving kernel K into matrix M, the following formula should be used: [ ] [ ] [ ] [ ] [ ] i , j M ' i, j M i, j .K i, j M i m, j n .K i m, j n= + + + + +∑ (2) While m = −1, 0, 1  and n = −1,0,1 (m and n refer to all the points around the source pixel). As a sample calculation for the center point of matrix M we have: [ ] [ ] [ ] [ ] [ ] [ ] [ ]( )0 0 0 0 0 0 1 1 1 1 11 11 0M' , M , .K , M , .K , ... M , .K ,= + − − − − + + = . After convolving kernel K on matrix M, the result will be as following (considering that the values outside of the boundaries of the matrix M are set to zero): [ ] 50 30 50 30 0 30 50 30 50 Conv M ,K M ' ⎡ ⎤ ⎢ ⎥= = ⎢ ⎥ ⎢ ⎥⎣ ⎦ The above example shows that this type of kernel can predict the gradient around the source pixel. The gradient of the center pixel with respect to one layer surrounding pixels is zero (because all the members of the matrix is 8) however large values are observed in boundaries (as predicted by the filter) because the values of the ghost cells, which can contribute to convolve the kernel into matrix, are considered as zero. Therefore, it is concluded that these types of kernels (gradient based kernels) can predict the gradient in data. Since replacing a pixel with the weighted sum (values after convolution) of its neighboring pixels can frequently result in much larger pixel values, dividing the weighted sum can scale EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER APPENDIX     76   back the value of the effect and ensure that the initial value of the image is maintained. This procedure is called normalization. In our images, a large gradient is observed between the signals related to the filler particles and other materials such as fibers. As it is shown in Figure 2-5 parts of the paper that contain filler are very bright and close to white color (i.e. the value of them are close to 1.0). This feature cannot be observed in the other parts of images, so there is a gradient between those parts of the images that contain filler and those parts that do not have any filler. Based on this information, it is concluded that these types of kernels are reasonable to show the presence of filler particles in our images. The kernel that we used is 5×5. This indicates that the effect of second layer surrounding pixels is taken into account to calculate the convolution for the source pixel. However, this may raise a question about the accuracy of the analysis. The SEM images, which are analyzed, are 634×128 pixels. Based on the scale of images (100 microns in 30 microns), the size of one pixel is 0.15×0.23 micrometers. The kernel that we chose for our analysis is a 5×5 two-dimensional matrix. This means that the local source pixel of our image uses the information of two layers of pixels surrounded by, to build up the final value in the output matrix. Since the filler particle sizes are in the order of 10 micrometer, this kernel can identify the filler particles in the images to some extent.  77     Appendix C: ANOVA results for filler distribution Layer 1: A) Actual by Predicted Plot B) Effect Tests Source DF Sum of Squares F Ratio Prob > F Comments PCC 1 82.88773 9.3438 0.0028 Very significant Starch 1 746.11905 84.1086 <.0001 Extremely significant CPAM/Silica 1 0.31683 0.0357 0.8504 - Vacuum 1 67.50180 7.6093 0.0067 Very significant PCC*PCC 1 27.14222 3.0597 0.0828 - Starch*PCC 1 63.42683 7.1500 0.0085 Very significant CPAM/Silica*PCC 1 38.44883 4.3343 0.0395 Significant Vacuum*PCC 1 236.08517 26.6134 <.0001 Extremely significant Starch*Starch 1 336.97262 37.9863 <.0001 Extremely significant CPAM/Silica*Starch 1 199.39534 22.4775 <.0001 Extremely significant Vacuum*Starch 1 10.58019 1.1927 0.2770 - CPAM/Silica*CPAM/Silica 1 0.37534 0.0423 0.8374 - Vacuum*CPAM/Silica 1 0.26169 0.0295 0.8639 - Vacuum*Vacuum 1 0.24878 0.0280 0.8673 - RSquare = 0.654914; RSquare Adj: 0.614654; Root Mean Square Error = 2.978406; Mean of Response = 19.41615; Observations (or Sum Wgts) = 135; 10 15 20 25 30 35 40 La ye r1 A ct ua l 10 15 20 25 30 35 40 Layer1 Predicted P<.0001 RSq=0.65 RMSE=2.9784 EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER APPENDIX     78   Layer 2: A) Actual by Predicted Plot B) Effect Tests Source DF Sum of Squares F Ratio Prob > F Comments PCC 1 77.03005 6.6008 0.0114 Significant Starch 1 661.44096 56.6795 <.0001 Extremely significant CPAM/Silica 1 0.64594 0.0554 0.8144 - Vacuum 1 14.32567 1.2276 0.2701 - PCC*PCC 1 2.09652 0.1797 0.6724 - Starch*PCC 1 90.31352 7.7390 0.0063 Very significant CPAM/Silica*PCC 1 50.05580 4.2893 0.0405 Significant Vacuum*PCC 1 191.91784 16.4456 <.0001 Extremely significant Starch*Starch 1 361.87296 31.0092 <.0001 Extremely significant CPAM/Silica*Starch 1 145.74782 12.4893 0.0006 Very significant Vacuum*Starch 1 70.84717 6.0710 0.0152 Significant CPAM/Silica*CPAM/Silica 1 18.61145 1.5948 0.2091 - Vacuum*CPAM/Silica 1 26.46204 2.2676 0.1347 - Vacuum*Vacuum 1 23.68034 2.0292 0.1569 - RSquare = 0.628346; RSquare Adj: 0.584986; Root Mean Square Error = 3.416116; Mean of Response = 20.12511; Observations (or Sum Wgts) = 135; 10 15 20 25 30 35 40 45 La ye r2 A ct ua l 10 15 20 25 30 35 40 45 Layer2 Predicted P<.0001 RSq=0.63 RMSE=3.4161 EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER APPENDIX     79   Layer 3: A) Actual by Predicted Plot B) Effect Tests Source DF Sum of Squares F Ratio Prob > F Comments PCC 1 213.26672 18.7143 <.0001 Extremely significant Starch 1 795.84079 69.8356 <.0001 Extremely significant CPAM/Silica 1 0.28971 0.0254 0.8736 - Vacuum 1 31.46435 2.7610 0.0992 - PCC*PCC 1 45.86442 4.0246 0.0471 Significant Starch*PCC 1 35.09662 3.0798 0.0818 - CPAM/Silica*PCC 1 38.72491 3.3981 0.0677 - Vacuum*PCC 1 259.36144 22.7592 <.0001 Extremely significant Starch*Starch 1 329.54657 28.9179 <.0001 Extremely significant CPAM/Silica*Starch 1 122.44109 10.7443 0.0014 Very significant Vacuum*Starch 1 20.04315 1.7588 0.1873 - CPAM/Silica*CPAM/Silica 1 2.13220 0.1871 0.6661 - Vacuum*CPAM/Silica 1 0.01680 0.0015 0.9694 - Vacuum*Vacuum 1 0.35532 0.0312 0.8601 - RSquare = 0.630183; RSquare Adj: 0.587038; Root Mean Square Error = 3.375784; Mean of Response = 21.09111; Observations (or Sum Wgts) = 135; 15 20 25 30 35 40 La ye r3 A ct ua l 15 20 25 30 35 40 Layer3 Predicted P<.0001 RSq=0.63 RMSE=3.3758 EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER APPENDIX     80   Layer 4: A) Actual by Predicted Plot B) Effect Tests Source DF Sum of Squares F Ratio Prob > F Comments PCC 1 408.97664 29.7706 <.0001 Extremely significant Starch 1 217.43042 15.8274 0.0001 Very significant CPAM/Silica 1 66.54839 4.8443 0.0297 Significant Vacuum 1 375.92954 27.3650 <.0001 Extremely significant PCC*PCC 1 2.30637 0.1679 0.6827 - Starch*PCC 1 62.79851 4.5713 0.0345 Significant CPAM/Silica*PCC 1 16.38650 1.1928 0.2769 - Vacuum*PCC 1 58.47092 4.2563 0.0413 Significant Starch*Starch 1 0.95301 0.0694 0.7927 - CPAM/Silica*Starch 1 207.61623 15.1130 0.0002 Very significant Vacuum*Starch 1 8.36461 0.6089 0.4367 - CPAM/Silica*CPAM/Silica 1 21.72043 1.5811 0.2110 - Vacuum*CPAM/Silica 1 0.64063 0.0466 0.8294 - Vacuum*Vacuum 1 51.32661 3.7362 0.0556 - RSquare = 0.56224; RSquare Adj: 0.511168; Root Mean Square Error = 3.706427; Mean of Response = 23.03667; Observations (or Sum Wgts) = 135; 15 20 25 30 35 La ye r4 A ct ua l 15 20 25 30 35 Layer4 Predicted P<.0001 RSq=0.56 RMSE=3.7064 EFFECT OF CHEMICAL ADDITIVES ON Z DIRECTION FILLER DISTRIBUTION IN PAPER APPENDIX     81   Layer 5: A) Actual by Predicted Plot B) Effect Tests Source DF Sum of Squares F Ratio Prob > F Comments PCC 1 430.38181 9.2185 0.0029 Very significant Starch 1 238.82765 5.1155 0.0255 Significant CPAM/Silica 1 417.06355 8.9332 0.0034 Very significant Vacuum 1 397.51052 8.5144 0.0042 Very significant PCC*PCC 1 0.09592 0.0021 0.9639 - Starch*PCC 1 35.40850 0.7584 0.3856 - CPAM/Silica*PCC 1 108.25990 2.3189 0.1304 - Vacuum*PCC 1 141.97218 3.0409 0.0837 - Starch*Starch 1 117.56840 2.5182 0.1152 - CPAM/Silica*Starch 1 280.62971 6.0109 0.0157 Significant Vacuum*Starch 1 192.83208 4.1303 0.0443 Significant CPAM/Silica*CPAM/Silica 1 2.90417 0.0622 0.8035 - Vacuum*CPAM/Silica 1 14.31298 0.3066 0.5808 - Vacuum*Vacuum 1 281.23375 6.0238 0.0155 Significant RSquare = 0.465267; RSquare Adj: 0.402882; Root Mean Square Error = 6.832777; Mean of Response = 31.91778; Observations (or Sum Wgts) = 135; 15 20 25 30 35 40 45 50 55 60 La ye r5 A ct ua l 15 20 25 30 35 40 45 50 55 60 Layer5 Predicted P<.0001 RSq=0.47 RMSE=6.8328"""@en ; edm:hasType "Thesis/Dissertation"@en ; vivo:dateIssued "2013-05"@en ; edm:isShownAt "10.14288/1.0071969"@en ; dcterms:language "eng"@en ; ns0:degreeDiscipline "Chemical and Biological Engineering"@en ; edm:provider "Vancouver : University of British Columbia Library"@en ; dcterms:publisher "University of British Columbia"@en ; dcterms:rights "Attribution 3.0 Unported"@en ; ns0:rightsURI "http://creativecommons.org/licenses/by/3.0/"@en ; ns0:scholarLevel "Graduate"@en ; dcterms:title "Effect of chemical additives on Z-direction filler distribution in paper"@en ; dcterms:type "Text"@en ; ns0:identifierURI "http://hdl.handle.net/2429/44198"@en .