Image analysis for the quantification of dislocations in hemp fibres

https://doi.org/10.1016/j.indcrop.2004.03.001Get rights and content

Abstract

When a natural fibre is subjected to compression stress in the longitudinal direction, local misalignments of cellulose microfibrils, so-called dislocations, may form in the cell wall. For most uses, dislocations have a negative influence on the performance of fibre-based products, and a means of characterising fibre resources with regard to dislocations would be valuable. In the present study, we have developed a procedure for semi-automatic quantification of the amount of dislocations in elementary hemp fibres. The procedure is based on polarised light microscopy and simple image analysis tools. Results from a similar approach have been published earlier by others, but no details of the method used were reported. In the present study, the relative dislocation area is determined from two digital images captured using polarised light microscopy. One image is optimised for the detection of the fibre edge, the other is optimised for the detection of dislocations. The area of the dislocations (obtained from the second image) relative to the total fibre area (obtained from the first image) gives a figure that expresses the relative dislocation area. The method is sensitive to the light intensity of the microscope and to the angle between the longitudinal direction of the fibre and the vibration direction of the polariser. It follows from this that strict standardisation is important if results from different fibres are to be compared.

Introduction

Dislocations are local misalignments of cellulose microfibrils in the cell wall of natural fibres such as wood, flax and hemp. Depending on the severity of the deformation and on the context, such zones have also been called nodes, kinks, kink bands, slip planes, misaligned zones or microcompressions. In this presentation the term “dislocations” is chosen in accordance with the review by Nyholm et al. (2001). Cellulose crystal chains of microfibrils in dislocations have a different orientation than those of the undisturbed cell wall. This is why dislocations may be revealed by polarised light microscopy.

Dislocations are formed as a result of subjecting a fibre to compression stress in the longitudinal direction (Robinson, 1920), and it appears that dislocations very often occur already in the living plant due to for example wind load or growth stress. However, several studies have shown that dislocations are easily introduced into wood samples during the preparation of slides for microscopy (for example Hartler, 1969 and Hoffmeyer, 1990), so the results from earlier studies on the amount of dislocations in living trees should be interpreted with care. For pulp wood, a number of publications have shown that the process of isolating the fibres from the stems as well as other steps during pulping give rise to new dislocations and/or to the enhancement of those already present (Nyholm et al., 2001 and references therein).

When natural fibres are used as reinforcement in composites, the interphase between the matrix and the fibres shows stress concentrations adjacent to dislocations (hemp–epoxy composites studied by Hughes et al., 2000), indicating that these zones may lead to crack initiation and/or debonding between matrix and fibre. In another study, Davies and Bruce (1998) found a negative correlation between the tensile strength of flax and nettle fibres and the extent of dislocations in the fibres. Bos et al. (2002) reported that the tensile strength of elementary flax fibres isolated by hand was 1834±900MPa, while it was 1522±400MPa for scutched and hackled fibres. The fibres isolated by hand were reportedly “virtually free of kink bands”. Based on observations of 63 spruce latewood fibres constrained by a vinylester matrix, Ljungqvist et al. (2002) used Weibull statistics to calculate that the strain to failure was approximately 12% larger for a theoretical 0.1 mm long fibre without visible dislocations than for a corresponding fibre with 70 dislocations per millimetre.

A number of studies have shown that dislocations negatively influence the properties of chemical pulp (for example Terziev et al., 2003 and references given by Nyholm et al., 2001). However, the relationship between dislocations and pulp properties is not black-and-white, and not yet fully understood. It appears that the effects of the dislocations on the pulp properties are related to the stage of the pulping process at which dislocations are induced (Hartler, 1969, Hakanen and Hartler, 1995, Ander et al., 2003), and that the severity of the effects are confounded with the type of pulping process (sulfite/sulfate, Hartler, 1969). Hartler (1969) and Hakanen and Hartler (1995) stated that dislocations only constitute weak locations if they are exposed to high temperature in alkaline or acidic medium, i.e. only if lignin is removed after the dislocation was created. Contrary to this, Kibblewhite (1974) and Kibblewhite and Brookes (1975) concluded that the wet strength of kraft pulps could be increased if the fibres were kinked before bleaching. The relationship between dislocations and pulp properties is presently being studied by Ander and co-workers (Swedish Agricultural University, Uppsala, Sweden).

The studies mentioned above indicate that the amount of dislocations could be an important characteristic of natural fibres and serve as a potential parameter when evaluating the quality of different fibre resources or when evaluating the damaging effect of a specific step in the production process, no matter whether the end product is a fibre composite or a sheet of paper. It could also help illuminate the role of dislocations in the pulping process. In spite of this, documented attempts to develop an automatic or semi-automatic method for the quantification of dislocations in fibre resources are scarce.

In almost all studies published hitherto, visualisation of dislocations is achieved by use of a polarising light microscope, i.e. a light microscope equipped with devices capable of producing and detecting plane polarised light (Preston, 1974). Between the light source and the sample stage the light is plane polarised using a plate called a polariser. Between the sample stage and the eyepieces, the analyzer, i.e. a second identical plate is inserted. The plates are normally polaroids. They are sometimes called polars or Nicols. If the polariser and/or the analyzer are rotated so that their vibration directions are perpendicular to each other, the set-up is denoted crossed polars. When plane polarised light passes through a birefringent material at a direction normal to the surface, the light will split into two components vibrating in two perpendicular planes, and the two components will travel through the material at two different speeds (a slow and a fast component). When a birefringent material such as crystalline cellulose is rotated under crossed polars, it will disappear (become dark) when one of the two vibration directions of the material parallel the vibration direction of one of the polars (i.e. at four different positions 90° apart during a complete rotation of the stage). These positions are called the extinction positions. The maximum brightness is observed at the four positions rotated 45° away from the extinction positions. When two birefringent materials are superimposed, interference will occur between the light components. Especially, if the two materials are identical and placed so that the two slow components are at right angles to each other, they cancel out, and complete compensation is achieved, i.e. no light is observed. Further information on polarised light microscopy may be found in textbooks, for example Preston (1974) or McCrone et al. (1978) or at www.microscopyu.com.

The above implies that a plant fibre with microfibrils running helically round the cell will appear dark when observed under crossed polars, if the fibre is oriented so that its longitudinal direction is parallel to the vibration direction of either the polariser or the analyser. This is because complete compensation will take place when the light passes through the two superimposed birefringent cell walls. However, this is not the case along the outermost edges of the fibre, and a thin line along the edges may therefore be visible even at these positions. The orientation of a dislocated zone is perpendicular to that of the microfibrils in the undisturbed cell wall (Hoffmeyer, 1990). Because microfibrils are most often oriented at an angle to the longitudinal direction of the fibre, dislocations form in a spiral pattern. Light passing through a dislocation in the first cell wall is therefore not countered when passing through the second cell wall, and accordingly it can be made visible as a bright zone. Crossed polars is therefore an excellent tool for visualisation of dislocations if two-dimensional information suffices.

A number of studies have employed manual counting of the number of dislocations per fibre or per mm as seen using polarised light microscopy, e.g. Kibblewhite (1976), Hoffmeyer (1993), Mohlin et al. (1996), Ellis et al. (1997), Allison et al. (1998) and Ljungqvist et al. (2002). It appears that none of these studies differentiated between “wide” and “narrow”, or between “short” or “long” dislocations, but counted each dislocation seen as “one”. Kibblewhite (1976) assigned the dislocations to one of two classes according to their severity: “wall fractures” and “zones of dislocation”.

For quantification of dislocations that are so severe that they result in bends of the fibre axis, Kibblewhite (1974) introduced the “Kink index” calculated as (N10–20 + 2N21–45 + 3N46–90 + 4N91–180)/total fibre length, where the subscripts give the angle of the bends seen using a microscope. Thus, the index gives the number of fibre bends per mm, but bends larger than 20° are multiplied by arbitrary weights according to their angle. This kink index was used by for example Mohlin and Alfredsson (1990). Also other indices have been introduced for fibre curl and/or kink, as reviewed by Page et al. (1985).

Pihlava (1998) studied the relationship between the number of dislocations in Kraft pulp fibres (Norway spruce and Scotch pine) and pulp properties. She introduced the term “dislocation number” as a means of obtaining a quantitative characteristic of the amount of dislocations in pulp fibres. Each fibre was given a grade from 0 to 4 according to a subjective evaluation of the dislocation frequency as seen with polarised light microscopy (0 corresponded to “almost no dislocations”, and 4 to “full of dislocations”). 100 fibres per sample were evaluated and their grades summed up, resulting in a dislocation number that in theory could vary between 0 and 400. However, when a technician determined the dislocation number twice for a set of four pulp samples, results were systematically higher at the second determination because the 0–4 scale was used differently in the two cases. This result illustrates the inherent weakness of all attempts of quantification that depend on subjective grading.

All of the quantification methods mentioned above rely on manual observations of fibres directly in the microscope or on micrographs. The only study published hitherto that employs image analysis was presented by Davies and Bruce (1998). They measured the proportional area of the bright regions of the fibres in images from polarised light microscopy. Compared to counting this approach has the advantage of implicitly including the width and length of the dislocations as seen in the two-dimensional image. However, Davies and Bruce (1998) did not give any details on their approach.

In the present work, we have studied the effects of light intensity and of the fibre angle relative to the polariser for the method of Davies and Bruce (1998), and have developed a procedure for semi-automatic determination of the proportion of dislocations for single hemp fibres. The method is based on simple image analysis tools. As already pointed out by Davies and Bruce (1998), their approach does not assess the severity of the dislocations and the method therefore remains semi-quantitative.

Section snippets

Material and methods

The fibres used for this study were hemp (Cannabis sativa L., var. Futura) elementary fibres, isolated from fibre material produced from stems that were water retted for 4 days at 35 °C and subsequently mechanically decorticated. 32 elementary fibres were included in the study. The fibres were provided by Claus Felby, Plant Fibre Laboratory, The Royal Veterinary and Agricultural University, Denmark.

The microscope used was a Leitz light microscope (orthoplan, 30.5.16.8 FSA/GW-/402a LILOM)

Fibre area

In order to calculate the relative dislocation area, the fibre area must first be determined. This implies that the fibre is separated from the image background. The greyscale images in Fig. 1 show that a threshold greyscale value is not feasible for this purpose for any of the tested combinations of light intensity and fibre angle, as parts of the fibre are as dark as the background in all images. However, except for modes 1, 4 and 7 (i.e. those for fibre angle 0°) the two edges of the fibre

Conclusions

Results from fibre damage quantification using polarised light microscopy were published by Davies and Bruce (1998). In the present work, we have studied their method, and have documented that reliable results require standardisation with regard to light intensity and fibre angle relative to the polarization filters. The procedure presented here is semi-automatic, but nevertheless gives robust results. The method implies that two greyscale images are taken per fibre—one with the fibre parallel

Acknowledgements

This study was carried out within the project ‘High performance hemp fibres and improved fibre network for composites’ financed by the Danish Research Council. The digital camera used was funded by the H + H Foundation. Thygesen wishes to thank Anders Nielsen for valuable advice regarding the microscope used.

References (23)

  • Allison, R.W., Ellis, M.J., Kibblewhite, R.P., Duffy, G.G., 1998. Effect of mechanical processes on the strength of...
  • Ander, P., Burgert, I., Frühmann, K., 2003. The possible relationship between dislocations and mechanical properties of...
  • H.L. Bos et al.

    Tensile and compressive properties of flax fibres for natural fibre reinforced composites

    J. Mater. Sci.

    (2002)
  • G.C. Davies et al.

    Effect of environmental relative humidity and damage on the tensile properties of flax and nettle fibers

    Textile Res. J.

    (1998)
  • M.J. Ellis et al.

    Fibre deformation during medium consistency mixing: role of residence time and impeller geometry

    Appita

    (1997)
  • A. Hakanen et al.

    Fiber deformations and strength potential of kraft pulps

    Paperi ja Puu

    (1995)
  • N. Hartler

    Misaligned zones in cellulosic fibres. Part 1: survey

    Norsk Skogindustri

    (1969)
  • Hoffmeyer, P., 1990. Failure of wood as influenced by moisture and duration of load. Thesis from the State University...
  • P. Hoffmeyer

    Non-linear creep caused by slip plane formation

    Wood Sci. Tech.

    (1993)
  • M. Hughes et al.

    An investigation into the effects of micro-compressive defects on interphase behaviour in hemp–epoxy composites using half-fringe photoelasticity

    Compos. Interfaces

    (2000)
  • R.P. Kibblewhite

    Effects of fiber kinking and pulp bleaching on wet web strength

    Tappi

    (1974)
  • Cited by (0)

    View full text