- Email: [email protected]
Quantification of steel-concrete interface in reinforced concrete using Backscattered Electron imaging technique
Construction and Building Materials 179 (2018) 420–429
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Quantification of steel-concrete interface in reinforced concrete using Backscattered Electron imaging technique Fangjie Chen a,b, Chun-Qing Li a,⇑, Hassan Baji a, Baoguo Ma b a b
School of Engineering, RMIT University, Melbourne, Australia Wuhan University of Technology, Wuhan, China
h i g h l i g h t s A accurate method for quantification of steel-concrete interface is developed. A new thresholding method to separate porous band at the interface is proposed. The size of porous band at the steel-concrete interface is not uniform around the steel bar. The water to cement ratio can significantly affect the size and the non-uniform distribution of the porous band around the steel bar.
a r t i c l e
i n f o
Article history: Received 20 March 2018 Received in revised form 23 May 2018 Accepted 28 May 2018 Available online 6 June 2018 Keywords: Reinforced concrete Steel-concrete interface Backscattered Electron imaging Microstructure
a b s t r a c t Research on quantification of the size and porosity of the interface between the reinforcing steel and concrete is limited in the current published literature. This paper presents a reliable and accurate method for quantification of steel-concrete interface using the technique of Backscattered Electron (BSE) imaging. Details of sample preparation procedure designed to minimize the damage to the steel-concrete interface due to grinding and polishing are presented. A new thresholding method to separate porous band at the interface from steel and cement pastes is proposed. It is found that there is a relatively wide porous band with large pores and voids near the steel-concrete interface. It is also found that the size of porous band around the steel bar is not uniform, and that the water to cement ratio can significantly affect the size and distribution of porous band at the interface. Quantification of steel-concrete interface as proposed in this paper is of importance in crack prediction of corrosion-affected reinforced concrete structures. Ó 2018 Elsevier Ltd. All rights reserved.
1. Introduction The interface between steel and concrete is an area with high porosity containing porous band and voids. The geometry and porosity of this interface area have a significant role in the degradation, durability and serviceability of reinforced concrete (RC) structures [1,2]. The size of this porous zone plays an important role in crack prediction of corrosion-affected RC structures [3]. Due to corrosion, the expansive corrosion products would first fill this zone. But before this area is fully filled, no pressure is exerted on concrete. This stage is commonly referred as the ‘‘free expansion” stage. Apparently larger size of this area is in favour of increasing the time to crack initiation of concrete [4,5], which leads to prolongation of service life of corrosion-affected RC structures. Furthermore, large crystals of calcium hydroxide depositing at
⇑ Corresponding author. E-mail address: [email protected] (C.-Q. Li). https://doi.org/10.1016/j.conbuildmat.2018.05.246 0950-0618/Ó 2018 Elsevier Ltd. All rights reserved.
the steel-concrete interface would affect the chloride ingression mechanisms and chloride threshold level [6–9]. In addition, corrosion products accumulated at the interface degrade the bond strength [10,11]. Despite the importance of interface area, few studies on the geometrical microstructure of the steel-concrete interface are available in the literature. With limited information, it is often assumed that the steel-concrete interface has a similar property as the interfacial transition zone (ITZ), which is the interface between cement paste and aggregates. In most of the predictive models for corrosion-induced concrete cover cracking in which the size of steel-concrete interface is considered, e.g., [5,12–14], it has been assumed that this area has uniform thickness of about 10–20 lm around the steel bar, which is nearly the same size as the ITZ investigated by other researchers [15–17]. Also, few experimental studies on the identification of geometric characteristics of the steel-concrete interface can be found in the literature. Söylev and François [18] used a video microscope with a low magnification to investigate micro-defects at the
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
steel-concrete interface. With this technique, they observed some gaps beneath the horizontal steel. However, the gap sizes at the interface were not further determined in their study. Mohammed et al. [8] reported that there is a visible gap at the steel-concrete interface forming under the steel bars oriented perpendicular to the casting direction, the size of which varies from several hundred micrometres to the order of a millimetre. The size of images used in their study is in the order of 1.0 mm. Back Scattered Electron (BSE) imaging technology is an advanced technique which has been used to observe the microstructures in cement and concrete since the work done by Scrivener and Pratt [19]. BSE imaging has the advantage of distinguishing the phases, e.g., hydrated cement pastes, anhydrous cement pastes, aggregates, pores, ITZ and cracks, of a flatpolished concrete sample. Each of the phases of concrete has its own greyscale range and brightness contrast in BSE images [17,20–23]. As a result, quantification analysis of the phases of concrete as well as reinforced concrete needs to be performed on BSE images [1,24–27]. Glass et al. [24] used 200 mm steel ribbons in concrete samples to observe the steel and concrete interface microstructure using BSE imaging. However, this size of steel ribbons cannot represent the interface in real RC structures. Horne et al. [1] used 9 mm steel bars in the study of the microstructure of steel-concrete interface, but 9 mm is usually the size of mesh. In most large-scale RC structures, the steel bars larger than 12 mm in diameter are normally required [28]. Moreover, only two locations, the top and bottom sides of the steel, are quantitatively analysed in their study. This appears to be inadequate to represent the whole steel-concrete interface around the steel bar. A thorough quantitative measurement should be conducted for the entire interface area around the steel. On the other hand, few BSE image processing techniques have been proposed for the phase determination and pore segmentation in concrete [23,26,29]. It seems that there is no standard method for these techniques. If a set of consistent rules is used for pore segmentation, the method for phase quantification in concrete can be reliable. As the phase determination based on BSE images can be very subjective [26], statistical analysis of results obtained from BSE images is also necessary. Another issue for BSE imaging is the sample preparation of reinforced concrete. With the requirement of a flat-polished surface, reinforced concrete samples must be cut into suitable size, and then ground and polished with minimum damage to the steel-concrete interface. However, the steel bars are easily deformed during sample preparation, complicating the observation of the steel-concrete interface. Therefore, a highquality grinding and polishing process is needed to remove damaged surface caused by cutting, meanwhile minimizing the deformation of steel bars caused by the grinding and polishing. Considering above discussions, it is almost certain that a reliable method is essential for observation and quantification of the steel-concrete interface. This paper presents such a method with accuracy for detail preparation of BSE samples for the measurement of porous band at steel-concrete interface. In this method, the damage due to grinding and polishing of the steel-concrete interface is minimized. Steel bars of 16 mm in diameter were used in reinforced concrete samples with a practical perspective on large RC structures. A new greyscale thresholding method is used to identify the porous band at the steel-concrete interface from BSE images. Quantitative analysis based on BSE images is conducted for the entire interface around the steel bar and the effect of w/c ratio on the steel-concrete interface is investigated. Repetition tests and statistical analysis are also carried out to verify the reliability of the quantification results from image analysis. The results for the porous band presented in this study can not only be useful for predictive models considering the microstructures
421
of the steel-concrete interface, but also be helpful for RC mesoscale structure modelling. 2. Experiment 2.1. Test specimen Concrete prism specimens of the size 150 mm 150 mm 300 mm with the three different mixes were cast. Ribbed low carbon steel bar of 16 mm in diameter was embedded in each of the specimens at the location shown in Fig. 1. The cover thickness to the bottom surface of the concrete was 20 mm. Cement used in this experiment was ordinary Portland cement. The coarse aggregates used were single sized normal weight limestone of 14 mm and the sands were medium grading river sand (<0.75 mm). Three water-cement ratios were used. The concrete mix specification is listed in Table 1. All the specimens were compacted for 10 s on vibration compacting table and water-cured for 28 days before testing. No mineral additives and chemical superplasticizers were used in this experiment, as supplementary cementitious additives like fly ashes or silica fumes give rise to some difficulties to BSE image analysis. For example, fly ashes containing carbons make it difficult to determine the boundary lines between cement pastes and pores in BSE images [30]. On the other hand, silica fumes and superplasticizers greatly affect the consistency of cement paste and slump of concrete [31,32], which also affect the steel-concrete interface significantly. Considering the spatial variability of the interfacial porous band, 10 mm thick slices at three longitudinal locations along the steel bar, as shown in Figs. 1 and 2(a), were cut from each prism specimen. These slices were numbered as 1, 2 and 3 from one end to the centre. Concrete slices were then core drilled into samples of 38 mm in diameter containing steel bar in the centre by core drill press. Fig. 2(b) shows three core drilled samples from slices as shown in Fig. 2(a) cut from Sample A. In order to have statistically reliable results for each mix design three concrete samples were cast (3 3), and three longitudinal locations in each specimen were cut to obtain core drilled samples. In total, 27 (3 3 3) core drilled samples were prepared for BSE image analysis in this experiment. Therefore, a more statistically reliable and accurate result can be produced from the experiment. 2.2. BSE sample preparation Core drilled samples were dried at a temperature of 25 °C and moulded by ultra-low viscosity epoxy. A vacuum was applied to the concrete-epoxy moulds for 20 min to force the epoxy into the pores in samples. Then, the samples were left to harden under a temperature of 25 °C for 24 h followed by oven drying below 60 °C for one hour. As the samples were prepared in the same temperature and humidity condition, the shrinkage of concrete caused by
Fig. 1. Dimensions of concrete specimen.
422
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
Table 1 Mix design of concrete specimens. w/c ratio
Maximum aggregate size (mm)
Coarse aggregate weight (kg)
Sand weight (kg)
Cement weight (kg)
Slump (mm)
A-0.50 B-0.45 C-0.40
14 14 14
937.3 937.3 937.3
795.8 742.9 695.6
396.3 455.3 509.5
115 115 110
(a) locations of slices (Sample A)
(b) core drilled samples A1, A2 and A3
Fig. 2. Core drilled sample preparation.
the drying could be ignored. In order to remove the damaged surface caused by the slice cutting process, further surface grinding and polishing operations were performed. For this purpose, an automatic polishing apparatus, which controls the polishing pressure, speed and time, was used, as shown in Fig. 3. In this process, two grinding and three polishing steps were performed. SiC papers of P220 and P500 were used to grind the sample in the grinding process. The three polishing steps were performed by using nonaqueous solutions with diamond particles of 9 lm, 3 lm and 1 lm, respectively. The final finish polishing was equivalent to 1 lm in order to obtain BSE images with a higher resolution. One concern during the grinding and polishing processes was the possible deformation of the embedded steel bars. As it is vital to reduce deformation of steel during grinding and polishing, several attempts with different grinding duration times and polishing
forces were conducted. A few extra core drilled samples were cut for the trial of grinding and polishing. It was found that the first grinding step is of paramount importance in this experiment. In this step, the damaged surface is ground off and the newly exposed surface is kept intact. Fig. 4(a) and (b) show two samples with grinding duration of 2 min and 5 min, respectively. Both were ground with P220 SiC grinding paper under a pressing force of 30 kN. The ribs of steel cannot be found in the over-ground trial sample as shown in Fig. 4(b). It shows that the overgrinding leads to the deformation of the steel and damages the steel-concrete interfacial microstructure. Apart from the grinding duration time in the first step, the pressing force applied to the sample during the following polishing steps is also important. With a few attempts of suitable polishing pressure, it was found that the polishing pressing force needs to be less than 30 kN. Fig. 5 shows the BSE images of two samples prepared under different polishing pressing forces. It can be seen that applying a pressing force higher than 30 kN leads to a nonsmooth and unclear steel edge [see Fig. 5(b)]. Applying high pressure leads to the spalling of steel surface during polishing. The spalled parts penetrate the resin filling the voids. If the pressing force is less than 30 kN, the edge of the steel is noticeably smoother with less spalling as shown in Fig. 5(a). In summary, to obtain a clear and intact interface image in this study, the first two grinding steps were set to two minutes and the grinding pressing force was maintained at 30 kN. The duration time for the three polishing steps was set to four minutes and the polishing pressing force was maintained at 25 kN. All 27 core drilled samples were prepared in this way for BSE observation. 2.3. BSE image capture
Fig. 3. The polisher used for sample preparation.
The FEI Quanta200 scanning electron microscope was used in BSE mode in high vacuum condition. All the BSE samples were carbon coated in a vacuum coating container. The observation was made on the polished surface starting from the top point of the steel (0°) and then measured in counter-clockwise along the steel-concrete interface. Each BSE image was marked by its angle
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
(a) 2 min grinding (Sample A)
423
(b) 5 min grinding (trial sample)
Fig. 4. BSE samples with different initial grinding duration time.
(a) Polishing pressure of 25 kN
(b) Polishing pressing force of 35 kN
Fig. 5. BSE images of samples with different polishing pressing force.
location. According to the casting direction, 0° is the top of the steel and 180° is the lowest point of the steel correspondingly, as illustrated in Fig. 6. At least 25 BSE images were obtained around the steel-concrete interface for each BSE sample. The images are of magnification 720 and contain 1024 943 pixels. This resolution gives the adequate information required for image analysis. In addition, 15 images of magnification 90 were also obtained for each sample. Identification of interface is based on greyscale level corresponding to each phase in concrete specimens [21,25,29,33–35]. By counting the numbers of pixels in specific greyscale range, the area percentage and size of each phase can be determined.
Fig. 6. BSE image capture process.
3. BSE image analysis 3.1. Greyscale range As the BSE image contrast varies with the mean atomic number of each phase, the phases of reinforced concrete can be identified. Fig. 7(a) shows a typical BSE image of sample A1 at the angle location of 100°. Grayscales of each phase in Fig. 7(a) were manually determined first. Fig. 7(b) shows the histograms of greyscales of each phase in the BSE image. It should be noted that the greyscale ranges of each phase selected in Fig. 7(b) are a little broader than the actual value. It is necessary to make sure that the cumulative pixel numbers of each phase are slightly greater than 100% so that all the pixels are covered without missing any information in the image. As iron has a high atomic number, it can be seen in Fig. 7(a) that steel material is shown as a bright white area with a very high greyscale value. Pores and voids are filled with epoxy which consists of carbon with a relatively low atomic number. Thus, these areas are shown as dark black in the image. A porous band with very clear dark black at the interface can be seen from Fig. 7(a). The contrasts of cement pastes and aggregate are in between these two extreme phases. Fig. 7(b) shows the histograms of greyscale range of each phase where the x-axis is the colour scale between 0 and 255 (no unit). It can be seen that there is no overlap between the greyscale ranges of steel and pores. Thus, it is easy to distinguish the boundary line between the porous band and steel edge. However, the greyscale
424
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
(a) BSE image of the steel-concrete interface
(b) histograms of the greyscale range of each phase Fig. 7. BSE image and corresponding greyscale histograms of each phase (sample A1, angle 100°).
range of hydrates in the cement paste overlaps with part of the greyscale of pores. The main issue in the BSE image analysis is to find a consistent method to segment the boundary between the cement paste and pores with a proper greyscale threshold value. With the preliminary greyscale analysis of each phase, further quantitative determination of the porous band can be performed. 3.2. Determination of porous band boundary at the interface For measuring the size of porous band at the steel-concrete interface, the boundary lines of the porous band should be identi-
fied by using a consistent and accurate method. Selection of a suitable greyscale threshold value for the porous band area is vital to the determination of interface boundary. Manual thresholding is a commonly used method, with which the features of interest regions can be shown. However, this method is highly subjective [27]. Several researchers have proposed different techniques to segment phases in the BSE image of cement paste and concrete [23,27,34]. Scrivener et al. [34] proposed a tangent-slope thresholding method to distinguish pores and cement pastes. Yang and Buenfeld [23] presented an algorithm for separating aggregate particles by a combination of greyscale thresholding. Wong et al. [26]
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
proposed a more efficient technique to segment pores from cement pastes in BSE images. However, in the case of reinforced concrete, the porous band at the interface need to be segmented from both steel and cement paste. In this paper, a new method for separating porous band from steel and cement paste at the interface is proposed. It should be noted that it is difficult to draw a conclusion on the accuracy of these methods. Nonetheless, if the method employed is consistent and the statistical analysis reaches an adequate confidence level, the phase quantification based on BSE image can be considered acceptable. The method proposed in this paper is to take a small section of 100 mm 100 mm from the BSE image of a sample (e.g. A1) at an angle of 100°, e.g., the captured image in Fig. 8(a). The image is rotated so that the porous band is vertically oriented. Then, a horizontal line of about 36 mm in length is drawn as can be seen in Fig. 8(a). This line covers the two boundaries between porous band
(a) BSE image of steel-concrete interface (100 μm ×100 μm)
425
and cement paste and those of porous band and steel. The greyscale values along this horizontal line are then extracted as shown in Fig. 8(b). Based on the greyscale ranges in the histograms of each phase, three distinguished stages can be separated as shown in Fig. 8(b). These stages represent the cement paste, the porous band and steel, respectively. Points A and B in Fig. 8(b) represent the boundary line between the cement paste and the porous band, while Points C and D represent that between the porous band and the steel bar. The threshold value for the boundaries of porous band was then determined based on the greyscale of the three stages shown in Fig. 8(b). In the proposed method of this paper, the half-height Point E on the slope line connecting Points A and B was taken as the threshold value (grey value of 42 in Fig. 8(b)). All pixels with a grey value below 42 were regarded as pores and voids. Fig. 9(a) shows the BSE image in Fig. 7(a) after thresholding with the greyscale value of 42. The porous band at the interface between steel and concrete can now be clearly seen, as well as the small pores. With the threshold grey value of 42 (for the considered image), the two boundary ends of the porous band on the drawn line were determined as Points E and F (Fig. 8(b)). The porous band thickness (length of EF) is finally measured as 22 mm. It is worth noting that there is a 1 mm distance between Points A and B and a 3 mm distance between Points C and D indicating a ±4 mm tolerance (about 18.2% measurement error) in measurement of the porous band thickness. Using the propose method, such errors can be avoided. In what follows, all the porous bands were thresholded and measured by the method demonstrated above. 3.3. Statistical analysis
(b) Greyscale value along the horizontal white line drawn in (a)
Following the thresholding, the thickness of the porous band at the steel-concrete interface can be measured. The mean value was used to represent the thickness of the porous band. To obtain a reliable average thickness, the number of recorded measurements for each porous band should satisfy a certain statistical requirement. For this purpose, student’s t-distribution [36] was used to confirm the adequacy of recorded measurements for each porous band as follows
Fig. 8. BSE image of steel-concrete interface and greyscale along the chosen white line (sample A1, angle 100°).
(a) after thresholding
t0:95;N1 ¼
y l pffiffiffiffi S= N
(b) rotation and measurements
Fig. 9. BSE image after thresholding (sample A1, angle 100°).
ð1Þ
426
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
Fig. 10. Statistical analysis of measurements of porous band in Fig. 8(b) (for Sample A1, at angle 100°).
where N is the number of required measurements for each porous
band, y is the mean value of the actual recorded measurements, S is the standard deviation of the actual recorded measurements and l is the unknown true mean value of the porous band. t 0:95; N0 1 is the t-value with the confidence level of 95%, which can be obtained from the t-value table. In this study, assuming that
Fig. 11. BSE images of porous band at the interface from sample A1.
the measured mean value y is required to be within 10% of the true
mean value, y l is taken as 0.1y. This requirement applies to all the porous band thickness measurement in this paper. For the porous band shown in Fig. 8, results of the average
thickness y along with the corresponding standard deviation S obtained from varying number of measurements (N0 ) are shown in Fig. 10. As can be seen that the increasing N0 narrows down the standard deviation S, leading to the result of average thickness of porous band closer to the true mean value l. With the given
t0:95;N1 (from t-value table) and y and S obtained from N0 measurement, the acceptable number of measurements N can be calculated from Eq. (1). If N0 is larger than N, the measured results of the average thickness of porous band are acceptable. Otherwise, more measurements are needed to achieve this statistical significance. With the mean and standard deviation from the 20 measurements (N0 = 20) in Fig. 9(b), and 95% confidence level, statistically 19 measurements (N = 19) are required according to Eq. (1). This suggests that the results obtained from 20 measurements satisfy the statistical test. Thus, the measured mean value of 21.4 mm can be considered as the representative of the porous band thickness for the image shown in Fig. 9(a). 4. Results and discussion 4.1. Spatial variability of porous band at the interface For each sample, the thickness of porous band in different angles around the perimeter of steel bar is measured. Fig. 11 shows four BSE images at the angle locations of 0°, 90°, 180° and 270° from sample A1. As it can be seen the thickness of porous band is not the same for these locations. For further analysis, Fig. 12 shows the post-thresholding BSE images of porous band at the interface from 0° to 360° around the steel bar for sample A1. For convenience in comparison of results, the porous band extracted from BSE images after thresholding are adjusted upright and parallel-arranged in the order of the angles around the steel bar. From Figs. 11 and 12, it is obvious that the porous band at the top part of the steel is slimmer than those at the bottom. This indicates that distribution of porous band at the interface around the steel is non-uniform. In Fig. 12, the average thickness of the porous band at 0° is 3.9 mm, and the value at 180° is 45.1 mm. The statistical analysis of the porous band of sample A1 in Fig. 12 is listed in
Table 2. All the measured results satisfy the statistical requirement discussed in the statistical analysis section. The non-uniformity in thickness of porous band at the steelconcrete interface is attributed to a phenomenon similar to that seen in the formation of ITZ, which is mainly governed by the wall effect [37,38]. During compacting, concrete on top of steel bar moves downwards, and as such, concrete on top becomes denser than concrete on other sides of the steel bar. The steel bar blocks the aggregates from moving downwards. This wall effect is more prominent in the case of contact between concrete (mortar) and steel bar than that between coarse aggregates and mortar (for ITZ), as steel bar is in a fixed position compared to a floating aggregate. Furthermore, it is easier for water to accumulate around steel bar as it moves faster than the aggregates. Therefore, the properties of the steel-concrete interface should not be assumed similar to those of ITZ [13]. This phenomenon shown in Figs. 11 and 12 was found in the BSE images of all nine BSE samples in the experiment. Fig. 13 shows the thickness of porous band as a function of its location represented by orientation from the top side of steel bar (moving counter-clockwise) for different slices along concrete samples. The results for each location, i.e., 1, 2 and 3 in Fig. 13 are the average value obtained from three concrete prisms of mix design A. It should be noted that the porous band at the area near the ribs of the steel bar is not included in the results shown in Figs. 13 and 14 but it is discussed separately in the next section because of the influence of the ribs. Variation of porous band size around the steel bar shows the same trend for all the slices along the steel bar. The thickness of porous band ranges from 10.1 mm to 54.4 mm based on the results of three repetition tests. Given the importance of size of porous band in the assessment of corrosion-induced cracking process, this highlights the necessity for accurate measurement of the interface.
4.2. Effect of water/cement ratio on porous band The variation of the thickness of porous band in samples A1, B1 and C1 is shown in Fig. 14 which shows the effect of w/c ratio of concrete on porous band at the interface. The porous band of sample A1 with 0.5 w/c ratio appears to be larger than those of the samples B1 and C1, especially at the bottom area of the steel. Although the thickness of porous band in sample B1 with 0.45
427
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
Fig. 12. The porous band at the interface around steel bar after thresholding from sample A1.
Table 2 Statistic analysis results of the porous band of sample A1. Angle location
Mean value (mm)
Standard deviation (mm)
t0:95;N1
Required N
Actual measurement times (N0 )
0° 90° 160° 180° 220° 270° 350°
3.9 4.2 42.1 45.1 28.7 17.6 4.5
0.82 0.95 9.09 10.6 6.45 3.74 1.03
2.093 2.074 2.086 2.069 2.080 2.093 2.074
19.6 22.1 20.3 23.5 21.9 19.8 22.4
20 23 21 24 22 20 23
60 Location 1 Location 2
Average thickness (μm)
50
Location 3
40 30 20 10 0 0
20 40 50 75 90 100 135 155 170 180 190 210 220 240 260 270 300 320 360
Angle location around steel (degrees)
(a) Effect of w/c ratio on porous band distribution around steel bar
Fig. 13. Thickness of porous band at the interface around the steel.
w/c ratio seems close to that of sample C1 with 0.4 w/c ratio, at the bottom area of the steel, namely at angles from 120° to 270°, the porous band in sample B1 is larger than that of sample C1. It is apparent that the w/c ratio can significantly affect the thickness of the porous band. High w/c ratio increases the size of the porous band as high w/c ratio favours the micro-bleeding of the cement paste and the wall-effect, especially at the bottom area of the steel bar. Fig. 14(a) also shows non-uniform distribution of the porous band around the steel due to the casting direction, i.e., the thickness of porous band on top of the steel bar is smaller than that at the bottom in the direction of compaction. However, low w/c
(b) Effect of w/c ratio on average and maximum values Fig. 14. Effect of w/c ratio on the porous band at the interface.
428
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
(a) porous band around the rib (sample A2)
(c) porous band around the rib (sample B1)
(b) corner of the rib (sample A2)
(d) porous band around the rib (sample C1)
Fig. 15. Microstructure around the rib of the steel.
ratio leads to a less the non-uniformity of the porous band around the steel. Fig. 14(b) shows the influence of w/c ratio on the average and maximum values of porous band. With the decrease of w/c ratio, both average and maximum values decline. If the w/c ratio is low, for example when w/c ratio is 0.45 and 0.4, the deceasing trend in porous band tends to fall.
4.3. Porous band near ribs of the steel bar Ribbed steel bars are widely used in RC structures to enhance the bond strength between steel and concrete. Interruption in smooth circular surface of ribbed steel bar can change the microstructures of concrete at the area near the rib, creating an interface which may be different from other parts of the steel bar surface. Fig. 15(a) shows a typical area around the rib of the steel bar from sample A2. The porous band at the interface right below the rib [denoted as Zone I in Fig. 15(a)] is larger than that at the side of the rib [denoted as Zone II in Fig. 15(b)]. The average size of the porous band formed in Zone I is 59.8 mm. It is nearly three times wider than the average size of porous band at the side of the rib, i.e., in Zone II, which is 17.1 mm. In Fig. 15(b), at the corner of the rib, a porous band can be seen, which seems to be connected to the porous band at the interface. It is also evident in the Fig. 15 (c) obtained from sample B1 that the porous area is connected to the porous area, leading to a large void at the bottom of the rib. In Fig. 15(d) obtained from sample C1 with a low w/c ratio, the porous area can also be found, however, it is not connected to the porous band. Therefore, for most corrosion-induced cracking models,
which consider the porosity or the empty porous band at the interface, the porous band and the influence of the steel ribs should also be considered. 5. Conclusions A reliable and accurate method for quantifying the geometry of porous band at the interface between reinforcing steel and concrete has been presented in this paper, using the technique of Backscattered Electron imaging. Details of the preparation procedure of reinforced concrete samples have been provided to reducing the damage to samples and deformation of steel during grinding and polishing processes. A new method for greyscale thresholding of porous band at the interface has been developed to obtain a reliable threshold value for porous band at the interface. From the analysis of the Backscattered Electron images it has been found that the size of porous band at the steel-concrete interface is not uniform around the steel bar, and that the porous band at the bottom area of the steel bar is larger than that at the top area in the direction of compaction. It has also been found that the water to cement ratio can significantly affect the size and the non-uniform distribution of the porous band around the steel bar, and that the porous band beneath the rib at the interface is larger than that at the side of the rib. It can be concluded that the microstructure of interface between steel and concrete is significantly affected by concrete mix and compaction and that further studies on influences of these factors on the steel-concrete interface are needed.
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
Conflict of interest The authors declared that there is no conflict of interest. Acknowledgments Financial support from Australian Research Council under DP140101547, LP150100413 and DP170102211 is gratefully acknowledged. The authors also acknowledge the facilities, and the scientific and technical assistance of the RMIT University’s Microscopy & Microanalysis Facility, a linked laboratory of the Australian Microscopy & Microanalysis Research Facility. References [1] A.T. Horne, I.G. Richardson, R.M.D. Brydson, Quantitative analysis of the microstructure of interfaces in steel reinforced concrete, Cem. Concr. Res. 37 (12) (2007) 1613–1623. [2] P. Monteiro, O.E. Gjorv, P.K. Mehta, Microstructure of the steel-cement paste interface in the presence of chloride, Cem. Concr. Res. 15 (5) (1985) 781–784. [3] F. Chen, H. Baji, C.Q. Li, A comparative study on factors affecting time to cover cracking as a service life indicator, Constr. Build. Mater. 163 (2018) 681–694. [4] L. Chernin, D.V. Val, Prediction of corrosion-induced cover cracking in reinforced concrete structures, Constr. Build. Mater. 25 (4) (2011) 1854–1869. [5] T. El Maaddawy, K. Soudki, A model for prediction of time from corrosion initiation to corrosion cracking, Cem. Concr. Compos. 29 (3) (2007) 168–175. [6] A. Castel, T. Vidal, R. Francois, G. Arliguie, Influence of steel–concrete interface quality on reinforcement corrosion induced by chlorides, Mag. Concr. Res. 55 (2) (2003) 151–159. [7] G.K. Glass, N.R. Buenfeld, The presentation of the chloride threshold level for corrosion of steel in concrete, Corros. Sci. 39 (5) (1997) 1001–1013. [8] T.U. Mohammed, N. Otsuki, H. Hamada, T. Yamaji, Chloride-induced corrosion of steel bars in concrete with presence of gap at steel-concrete interface, ACI Mater. J. 99 (2) (2002) 149–156. [9] C.L. Page, Mechanism of corrosion protection in reinforced concrete marine structures, Nature 258 (5535) (1975) 514–515. [10] T.A. Soylev, R. François, Quality of steel–concrete interface and corrosion of reinforcing steel, Cem. Concr. Res. 33 (9) (2003) 1407–1415. [11] G. Al-Sulaimani, M. Kaleemullah, I. Basunbul, Influence of corrosion and cracking on bond behavior and strength of reinforced concrete members, Struct. J. 87 (2) (1990) 220–231. [12] C.Q. Li, R.E. Melchers, J.-J. Zheng, Analytical model for corrosion-induced crack width in reinforced concrete structures, ACI Struct. J. 103 (4) (2006). [13] Y. Liu, R.E. Weyers, Modeling the time-to-corrosion cracking in chloride contaminated reinforced concrete structures, ACI Mater. J. 95 (6) (1998) 675– 681. [14] C. Lu, W. Jin, R. Liu, Reinforcement corrosion-induced cover cracking and its time prediction for reinforced concrete structures, Corros. Sci. 53 (4) (2011) 1337–1347. [15] A. Brough, A. Atkinson, Automated identification of the aggregate–paste interfacial transition zone in mortars of silica sand with Portland or alkaliactivated slag cement paste, Cem. Concr. Res. 30 (6) (2000) 849–854. [16] J.P. Ollivier, J.C. Maso, B. Bourdette, Interfacial transition zone in concrete, Adv. Cem. Based Mater. 2 (1) (1995) 30–38. [17] K.L. Scrivener, P.L. Pratt, Characterization of interfacial microstructure, Interfacial Transition Zone Concr. 2 (1996) 3–18.
429
[18] T.A. Söylev, R. François, Corrosion of reinforcement in relation to presence of defects at the interface between steel and concrete, J. Mater. Civ. Eng. 17 (4) (2005) 447–455. [19] K.L. Scrivener, P. Pratt, Backscattered electron images of polished cement sections in the scanning electron microscope, Proc. Int. Conf. Cem. Microsc. (1984) 145–155. [20] D. Darwin, M.N. Abou-Zeid, K.W. Ketcham, Automated crack identification for cement paste, Cem. Concr. Res. 25 (3) (1995) 605–616. [21] Y. Wang, S. Diamond, An approach to quantitative image analysis for cement pastes, MRS Online Proc. Lib. Arch. 370 (1994). [22] H.S. Wong, M. Zobel, N.R. Buenfeld, R.W. Zimmerman, Influence of the interfacial transition zone and microcracking on the diffusivity, permeability and sorptivity of cement-based materials after drying, Mag. Concr. Res. 61 (8) (2009) 571–589. [23] R. Yang, N.R. Buenfeld, Binary segmentation of aggregate in SEM image analysis of concrete, Cem. Concr. Res. 31 (3) (2001) 437–441. [24] G.K. Glass, R. Yang, T. Dickhaus, N.R. Buenfeld, Backscattered electron imaging of the steel–concrete interface, Corros. Sci. 43 (4) (2001) 605–610. [25] M.K. Head, N.R. Buenfeld, Measurement of aggregate interfacial porosity in complex, multi-phase aggregate concrete: Binary mask production using backscattered electron, and energy dispersive X-ray images, Cem. Concr. Res. 36 (2) (2006) 337–345. [26] H.S. Wong, M.K. Head, N.R. Buenfeld, Pore segmentation of cement-based materials from backscattered electron images, Cem. Concr. Res. 36 (6) (2006) 1083–1090. [27] H.S. Wong, M.K. Head, N.R. Buenfeld, Characterising the pore structure of cement-based materials using backscattered electron and confocal microscopy, in: M.S. Konsta-Gdoutos (Ed.), Measuring, Monitoring and Modeling Concrete Properties: An International Symposium dedicated to Professor Surendra P. Shah, Northwestern University, U.S.A., Springer Netherlands, Dordrecht, 2006, pp. 495–502. [28] ACI, Building Code Requirements for Structural Concrete (ACI 318-14): An ACI Standard: Commentary on Building Code Requirements for Structural Concrete (ACI 318R-14), an ACI Report, ACI Committee 318, American Concrete Institute, ACI Committee, 2015. [29] K.L. Scrivener, A.K. Crumbie, P. Laugesen, The interfacial transition zone (ITZ) between cement paste and aggregate in concrete, Interface Sci. 12 (4) (2004) 411–421. [30] S. Diamond, Considerations in image analysis as applied to investigations of the ITZ in concrete, Cem. Concr. Compos. 23 (2) (2001) 171–178. [31] B. Ma, F. Chen, B. Zhang, C. Lin, The design and application of Ballastless track enhanced concrete mineral admixture, Appl. Mech. Mater., Trans. Tech. Publ. (2014) 1427–1430. [32] H. Tan, F. Zou, B. Ma, M. Liu, X. Li, S. Jian, Effect of sodium tripolyphosphate on adsorbing behavior of polycarboxylate superplasticizer, Constr. Build. Mater. 126 (2016) 617–623. [33] K.L. Scrivener, A. Bentur, P. Pratt, Quantitative characterization of the transition zone in high strength concretes, Adv. Cem. Res. 1 (4) (1988) 230– 237. [34] K.L. Scrivener, H. Patel, P. Pratt, L. Parrott, Analysis of phases in cement paste using backscattered electron images, methanol adsorption and thermogravimetric analysis, MRS Online Proc. Lib. Arch. 85 (1986). [35] S. Diamond, J. Huang, The ITZ in concrete – a different view based on image analysis and SEM observations, Cem. Concr. Compos. 23 (2) (2001) 179–188. [36] C. Forbes, M. Evans, N. Hastings, B. Peacock, Student’s t Distribution, Statistical Distributions, John Wiley & Sons, Inc., 2010, pp. 183–186. [37] A. Bentur, M. Alexander, A review of the work of the RILEM TC 159-ETC: engineering of the interfacial transition zone in cementitious composites, Mater Struct 33 (2) (2000) 82–87. [38] K.L. Scrivener, Characterisation of the ITZ and its quantification by test methods, RILEM Rep. (1999) 3–18.
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Quantification of steel-concrete interface in reinforced concrete using Backscattered Electron imaging technique Fangjie Chen a,b, Chun-Qing Li a,⇑, Hassan Baji a, Baoguo Ma b a b
School of Engineering, RMIT University, Melbourne, Australia Wuhan University of Technology, Wuhan, China
h i g h l i g h t s A accurate method for quantification of steel-concrete interface is developed. A new thresholding method to separate porous band at the interface is proposed. The size of porous band at the steel-concrete interface is not uniform around the steel bar. The water to cement ratio can significantly affect the size and the non-uniform distribution of the porous band around the steel bar.
a r t i c l e
i n f o
Article history: Received 20 March 2018 Received in revised form 23 May 2018 Accepted 28 May 2018 Available online 6 June 2018 Keywords: Reinforced concrete Steel-concrete interface Backscattered Electron imaging Microstructure
a b s t r a c t Research on quantification of the size and porosity of the interface between the reinforcing steel and concrete is limited in the current published literature. This paper presents a reliable and accurate method for quantification of steel-concrete interface using the technique of Backscattered Electron (BSE) imaging. Details of sample preparation procedure designed to minimize the damage to the steel-concrete interface due to grinding and polishing are presented. A new thresholding method to separate porous band at the interface from steel and cement pastes is proposed. It is found that there is a relatively wide porous band with large pores and voids near the steel-concrete interface. It is also found that the size of porous band around the steel bar is not uniform, and that the water to cement ratio can significantly affect the size and distribution of porous band at the interface. Quantification of steel-concrete interface as proposed in this paper is of importance in crack prediction of corrosion-affected reinforced concrete structures. Ó 2018 Elsevier Ltd. All rights reserved.
1. Introduction The interface between steel and concrete is an area with high porosity containing porous band and voids. The geometry and porosity of this interface area have a significant role in the degradation, durability and serviceability of reinforced concrete (RC) structures [1,2]. The size of this porous zone plays an important role in crack prediction of corrosion-affected RC structures [3]. Due to corrosion, the expansive corrosion products would first fill this zone. But before this area is fully filled, no pressure is exerted on concrete. This stage is commonly referred as the ‘‘free expansion” stage. Apparently larger size of this area is in favour of increasing the time to crack initiation of concrete [4,5], which leads to prolongation of service life of corrosion-affected RC structures. Furthermore, large crystals of calcium hydroxide depositing at
⇑ Corresponding author. E-mail address: [email protected] (C.-Q. Li). https://doi.org/10.1016/j.conbuildmat.2018.05.246 0950-0618/Ó 2018 Elsevier Ltd. All rights reserved.
the steel-concrete interface would affect the chloride ingression mechanisms and chloride threshold level [6–9]. In addition, corrosion products accumulated at the interface degrade the bond strength [10,11]. Despite the importance of interface area, few studies on the geometrical microstructure of the steel-concrete interface are available in the literature. With limited information, it is often assumed that the steel-concrete interface has a similar property as the interfacial transition zone (ITZ), which is the interface between cement paste and aggregates. In most of the predictive models for corrosion-induced concrete cover cracking in which the size of steel-concrete interface is considered, e.g., [5,12–14], it has been assumed that this area has uniform thickness of about 10–20 lm around the steel bar, which is nearly the same size as the ITZ investigated by other researchers [15–17]. Also, few experimental studies on the identification of geometric characteristics of the steel-concrete interface can be found in the literature. Söylev and François [18] used a video microscope with a low magnification to investigate micro-defects at the
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
steel-concrete interface. With this technique, they observed some gaps beneath the horizontal steel. However, the gap sizes at the interface were not further determined in their study. Mohammed et al. [8] reported that there is a visible gap at the steel-concrete interface forming under the steel bars oriented perpendicular to the casting direction, the size of which varies from several hundred micrometres to the order of a millimetre. The size of images used in their study is in the order of 1.0 mm. Back Scattered Electron (BSE) imaging technology is an advanced technique which has been used to observe the microstructures in cement and concrete since the work done by Scrivener and Pratt [19]. BSE imaging has the advantage of distinguishing the phases, e.g., hydrated cement pastes, anhydrous cement pastes, aggregates, pores, ITZ and cracks, of a flatpolished concrete sample. Each of the phases of concrete has its own greyscale range and brightness contrast in BSE images [17,20–23]. As a result, quantification analysis of the phases of concrete as well as reinforced concrete needs to be performed on BSE images [1,24–27]. Glass et al. [24] used 200 mm steel ribbons in concrete samples to observe the steel and concrete interface microstructure using BSE imaging. However, this size of steel ribbons cannot represent the interface in real RC structures. Horne et al. [1] used 9 mm steel bars in the study of the microstructure of steel-concrete interface, but 9 mm is usually the size of mesh. In most large-scale RC structures, the steel bars larger than 12 mm in diameter are normally required [28]. Moreover, only two locations, the top and bottom sides of the steel, are quantitatively analysed in their study. This appears to be inadequate to represent the whole steel-concrete interface around the steel bar. A thorough quantitative measurement should be conducted for the entire interface area around the steel. On the other hand, few BSE image processing techniques have been proposed for the phase determination and pore segmentation in concrete [23,26,29]. It seems that there is no standard method for these techniques. If a set of consistent rules is used for pore segmentation, the method for phase quantification in concrete can be reliable. As the phase determination based on BSE images can be very subjective [26], statistical analysis of results obtained from BSE images is also necessary. Another issue for BSE imaging is the sample preparation of reinforced concrete. With the requirement of a flat-polished surface, reinforced concrete samples must be cut into suitable size, and then ground and polished with minimum damage to the steel-concrete interface. However, the steel bars are easily deformed during sample preparation, complicating the observation of the steel-concrete interface. Therefore, a highquality grinding and polishing process is needed to remove damaged surface caused by cutting, meanwhile minimizing the deformation of steel bars caused by the grinding and polishing. Considering above discussions, it is almost certain that a reliable method is essential for observation and quantification of the steel-concrete interface. This paper presents such a method with accuracy for detail preparation of BSE samples for the measurement of porous band at steel-concrete interface. In this method, the damage due to grinding and polishing of the steel-concrete interface is minimized. Steel bars of 16 mm in diameter were used in reinforced concrete samples with a practical perspective on large RC structures. A new greyscale thresholding method is used to identify the porous band at the steel-concrete interface from BSE images. Quantitative analysis based on BSE images is conducted for the entire interface around the steel bar and the effect of w/c ratio on the steel-concrete interface is investigated. Repetition tests and statistical analysis are also carried out to verify the reliability of the quantification results from image analysis. The results for the porous band presented in this study can not only be useful for predictive models considering the microstructures
421
of the steel-concrete interface, but also be helpful for RC mesoscale structure modelling. 2. Experiment 2.1. Test specimen Concrete prism specimens of the size 150 mm 150 mm 300 mm with the three different mixes were cast. Ribbed low carbon steel bar of 16 mm in diameter was embedded in each of the specimens at the location shown in Fig. 1. The cover thickness to the bottom surface of the concrete was 20 mm. Cement used in this experiment was ordinary Portland cement. The coarse aggregates used were single sized normal weight limestone of 14 mm and the sands were medium grading river sand (<0.75 mm). Three water-cement ratios were used. The concrete mix specification is listed in Table 1. All the specimens were compacted for 10 s on vibration compacting table and water-cured for 28 days before testing. No mineral additives and chemical superplasticizers were used in this experiment, as supplementary cementitious additives like fly ashes or silica fumes give rise to some difficulties to BSE image analysis. For example, fly ashes containing carbons make it difficult to determine the boundary lines between cement pastes and pores in BSE images [30]. On the other hand, silica fumes and superplasticizers greatly affect the consistency of cement paste and slump of concrete [31,32], which also affect the steel-concrete interface significantly. Considering the spatial variability of the interfacial porous band, 10 mm thick slices at three longitudinal locations along the steel bar, as shown in Figs. 1 and 2(a), were cut from each prism specimen. These slices were numbered as 1, 2 and 3 from one end to the centre. Concrete slices were then core drilled into samples of 38 mm in diameter containing steel bar in the centre by core drill press. Fig. 2(b) shows three core drilled samples from slices as shown in Fig. 2(a) cut from Sample A. In order to have statistically reliable results for each mix design three concrete samples were cast (3 3), and three longitudinal locations in each specimen were cut to obtain core drilled samples. In total, 27 (3 3 3) core drilled samples were prepared for BSE image analysis in this experiment. Therefore, a more statistically reliable and accurate result can be produced from the experiment. 2.2. BSE sample preparation Core drilled samples were dried at a temperature of 25 °C and moulded by ultra-low viscosity epoxy. A vacuum was applied to the concrete-epoxy moulds for 20 min to force the epoxy into the pores in samples. Then, the samples were left to harden under a temperature of 25 °C for 24 h followed by oven drying below 60 °C for one hour. As the samples were prepared in the same temperature and humidity condition, the shrinkage of concrete caused by
Fig. 1. Dimensions of concrete specimen.
422
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
Table 1 Mix design of concrete specimens. w/c ratio
Maximum aggregate size (mm)
Coarse aggregate weight (kg)
Sand weight (kg)
Cement weight (kg)
Slump (mm)
A-0.50 B-0.45 C-0.40
14 14 14
937.3 937.3 937.3
795.8 742.9 695.6
396.3 455.3 509.5
115 115 110
(a) locations of slices (Sample A)
(b) core drilled samples A1, A2 and A3
Fig. 2. Core drilled sample preparation.
the drying could be ignored. In order to remove the damaged surface caused by the slice cutting process, further surface grinding and polishing operations were performed. For this purpose, an automatic polishing apparatus, which controls the polishing pressure, speed and time, was used, as shown in Fig. 3. In this process, two grinding and three polishing steps were performed. SiC papers of P220 and P500 were used to grind the sample in the grinding process. The three polishing steps were performed by using nonaqueous solutions with diamond particles of 9 lm, 3 lm and 1 lm, respectively. The final finish polishing was equivalent to 1 lm in order to obtain BSE images with a higher resolution. One concern during the grinding and polishing processes was the possible deformation of the embedded steel bars. As it is vital to reduce deformation of steel during grinding and polishing, several attempts with different grinding duration times and polishing
forces were conducted. A few extra core drilled samples were cut for the trial of grinding and polishing. It was found that the first grinding step is of paramount importance in this experiment. In this step, the damaged surface is ground off and the newly exposed surface is kept intact. Fig. 4(a) and (b) show two samples with grinding duration of 2 min and 5 min, respectively. Both were ground with P220 SiC grinding paper under a pressing force of 30 kN. The ribs of steel cannot be found in the over-ground trial sample as shown in Fig. 4(b). It shows that the overgrinding leads to the deformation of the steel and damages the steel-concrete interfacial microstructure. Apart from the grinding duration time in the first step, the pressing force applied to the sample during the following polishing steps is also important. With a few attempts of suitable polishing pressure, it was found that the polishing pressing force needs to be less than 30 kN. Fig. 5 shows the BSE images of two samples prepared under different polishing pressing forces. It can be seen that applying a pressing force higher than 30 kN leads to a nonsmooth and unclear steel edge [see Fig. 5(b)]. Applying high pressure leads to the spalling of steel surface during polishing. The spalled parts penetrate the resin filling the voids. If the pressing force is less than 30 kN, the edge of the steel is noticeably smoother with less spalling as shown in Fig. 5(a). In summary, to obtain a clear and intact interface image in this study, the first two grinding steps were set to two minutes and the grinding pressing force was maintained at 30 kN. The duration time for the three polishing steps was set to four minutes and the polishing pressing force was maintained at 25 kN. All 27 core drilled samples were prepared in this way for BSE observation. 2.3. BSE image capture
Fig. 3. The polisher used for sample preparation.
The FEI Quanta200 scanning electron microscope was used in BSE mode in high vacuum condition. All the BSE samples were carbon coated in a vacuum coating container. The observation was made on the polished surface starting from the top point of the steel (0°) and then measured in counter-clockwise along the steel-concrete interface. Each BSE image was marked by its angle
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
(a) 2 min grinding (Sample A)
423
(b) 5 min grinding (trial sample)
Fig. 4. BSE samples with different initial grinding duration time.
(a) Polishing pressure of 25 kN
(b) Polishing pressing force of 35 kN
Fig. 5. BSE images of samples with different polishing pressing force.
location. According to the casting direction, 0° is the top of the steel and 180° is the lowest point of the steel correspondingly, as illustrated in Fig. 6. At least 25 BSE images were obtained around the steel-concrete interface for each BSE sample. The images are of magnification 720 and contain 1024 943 pixels. This resolution gives the adequate information required for image analysis. In addition, 15 images of magnification 90 were also obtained for each sample. Identification of interface is based on greyscale level corresponding to each phase in concrete specimens [21,25,29,33–35]. By counting the numbers of pixels in specific greyscale range, the area percentage and size of each phase can be determined.
Fig. 6. BSE image capture process.
3. BSE image analysis 3.1. Greyscale range As the BSE image contrast varies with the mean atomic number of each phase, the phases of reinforced concrete can be identified. Fig. 7(a) shows a typical BSE image of sample A1 at the angle location of 100°. Grayscales of each phase in Fig. 7(a) were manually determined first. Fig. 7(b) shows the histograms of greyscales of each phase in the BSE image. It should be noted that the greyscale ranges of each phase selected in Fig. 7(b) are a little broader than the actual value. It is necessary to make sure that the cumulative pixel numbers of each phase are slightly greater than 100% so that all the pixels are covered without missing any information in the image. As iron has a high atomic number, it can be seen in Fig. 7(a) that steel material is shown as a bright white area with a very high greyscale value. Pores and voids are filled with epoxy which consists of carbon with a relatively low atomic number. Thus, these areas are shown as dark black in the image. A porous band with very clear dark black at the interface can be seen from Fig. 7(a). The contrasts of cement pastes and aggregate are in between these two extreme phases. Fig. 7(b) shows the histograms of greyscale range of each phase where the x-axis is the colour scale between 0 and 255 (no unit). It can be seen that there is no overlap between the greyscale ranges of steel and pores. Thus, it is easy to distinguish the boundary line between the porous band and steel edge. However, the greyscale
424
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
(a) BSE image of the steel-concrete interface
(b) histograms of the greyscale range of each phase Fig. 7. BSE image and corresponding greyscale histograms of each phase (sample A1, angle 100°).
range of hydrates in the cement paste overlaps with part of the greyscale of pores. The main issue in the BSE image analysis is to find a consistent method to segment the boundary between the cement paste and pores with a proper greyscale threshold value. With the preliminary greyscale analysis of each phase, further quantitative determination of the porous band can be performed. 3.2. Determination of porous band boundary at the interface For measuring the size of porous band at the steel-concrete interface, the boundary lines of the porous band should be identi-
fied by using a consistent and accurate method. Selection of a suitable greyscale threshold value for the porous band area is vital to the determination of interface boundary. Manual thresholding is a commonly used method, with which the features of interest regions can be shown. However, this method is highly subjective [27]. Several researchers have proposed different techniques to segment phases in the BSE image of cement paste and concrete [23,27,34]. Scrivener et al. [34] proposed a tangent-slope thresholding method to distinguish pores and cement pastes. Yang and Buenfeld [23] presented an algorithm for separating aggregate particles by a combination of greyscale thresholding. Wong et al. [26]
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
proposed a more efficient technique to segment pores from cement pastes in BSE images. However, in the case of reinforced concrete, the porous band at the interface need to be segmented from both steel and cement paste. In this paper, a new method for separating porous band from steel and cement paste at the interface is proposed. It should be noted that it is difficult to draw a conclusion on the accuracy of these methods. Nonetheless, if the method employed is consistent and the statistical analysis reaches an adequate confidence level, the phase quantification based on BSE image can be considered acceptable. The method proposed in this paper is to take a small section of 100 mm 100 mm from the BSE image of a sample (e.g. A1) at an angle of 100°, e.g., the captured image in Fig. 8(a). The image is rotated so that the porous band is vertically oriented. Then, a horizontal line of about 36 mm in length is drawn as can be seen in Fig. 8(a). This line covers the two boundaries between porous band
(a) BSE image of steel-concrete interface (100 μm ×100 μm)
425
and cement paste and those of porous band and steel. The greyscale values along this horizontal line are then extracted as shown in Fig. 8(b). Based on the greyscale ranges in the histograms of each phase, three distinguished stages can be separated as shown in Fig. 8(b). These stages represent the cement paste, the porous band and steel, respectively. Points A and B in Fig. 8(b) represent the boundary line between the cement paste and the porous band, while Points C and D represent that between the porous band and the steel bar. The threshold value for the boundaries of porous band was then determined based on the greyscale of the three stages shown in Fig. 8(b). In the proposed method of this paper, the half-height Point E on the slope line connecting Points A and B was taken as the threshold value (grey value of 42 in Fig. 8(b)). All pixels with a grey value below 42 were regarded as pores and voids. Fig. 9(a) shows the BSE image in Fig. 7(a) after thresholding with the greyscale value of 42. The porous band at the interface between steel and concrete can now be clearly seen, as well as the small pores. With the threshold grey value of 42 (for the considered image), the two boundary ends of the porous band on the drawn line were determined as Points E and F (Fig. 8(b)). The porous band thickness (length of EF) is finally measured as 22 mm. It is worth noting that there is a 1 mm distance between Points A and B and a 3 mm distance between Points C and D indicating a ±4 mm tolerance (about 18.2% measurement error) in measurement of the porous band thickness. Using the propose method, such errors can be avoided. In what follows, all the porous bands were thresholded and measured by the method demonstrated above. 3.3. Statistical analysis
(b) Greyscale value along the horizontal white line drawn in (a)
Following the thresholding, the thickness of the porous band at the steel-concrete interface can be measured. The mean value was used to represent the thickness of the porous band. To obtain a reliable average thickness, the number of recorded measurements for each porous band should satisfy a certain statistical requirement. For this purpose, student’s t-distribution [36] was used to confirm the adequacy of recorded measurements for each porous band as follows
Fig. 8. BSE image of steel-concrete interface and greyscale along the chosen white line (sample A1, angle 100°).
(a) after thresholding
t0:95;N1 ¼
y l pffiffiffiffi S= N
(b) rotation and measurements
Fig. 9. BSE image after thresholding (sample A1, angle 100°).
ð1Þ
426
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
Fig. 10. Statistical analysis of measurements of porous band in Fig. 8(b) (for Sample A1, at angle 100°).
where N is the number of required measurements for each porous
band, y is the mean value of the actual recorded measurements, S is the standard deviation of the actual recorded measurements and l is the unknown true mean value of the porous band. t 0:95; N0 1 is the t-value with the confidence level of 95%, which can be obtained from the t-value table. In this study, assuming that
Fig. 11. BSE images of porous band at the interface from sample A1.
the measured mean value y is required to be within 10% of the true
mean value, y l is taken as 0.1y. This requirement applies to all the porous band thickness measurement in this paper. For the porous band shown in Fig. 8, results of the average
thickness y along with the corresponding standard deviation S obtained from varying number of measurements (N0 ) are shown in Fig. 10. As can be seen that the increasing N0 narrows down the standard deviation S, leading to the result of average thickness of porous band closer to the true mean value l. With the given
t0:95;N1 (from t-value table) and y and S obtained from N0 measurement, the acceptable number of measurements N can be calculated from Eq. (1). If N0 is larger than N, the measured results of the average thickness of porous band are acceptable. Otherwise, more measurements are needed to achieve this statistical significance. With the mean and standard deviation from the 20 measurements (N0 = 20) in Fig. 9(b), and 95% confidence level, statistically 19 measurements (N = 19) are required according to Eq. (1). This suggests that the results obtained from 20 measurements satisfy the statistical test. Thus, the measured mean value of 21.4 mm can be considered as the representative of the porous band thickness for the image shown in Fig. 9(a). 4. Results and discussion 4.1. Spatial variability of porous band at the interface For each sample, the thickness of porous band in different angles around the perimeter of steel bar is measured. Fig. 11 shows four BSE images at the angle locations of 0°, 90°, 180° and 270° from sample A1. As it can be seen the thickness of porous band is not the same for these locations. For further analysis, Fig. 12 shows the post-thresholding BSE images of porous band at the interface from 0° to 360° around the steel bar for sample A1. For convenience in comparison of results, the porous band extracted from BSE images after thresholding are adjusted upright and parallel-arranged in the order of the angles around the steel bar. From Figs. 11 and 12, it is obvious that the porous band at the top part of the steel is slimmer than those at the bottom. This indicates that distribution of porous band at the interface around the steel is non-uniform. In Fig. 12, the average thickness of the porous band at 0° is 3.9 mm, and the value at 180° is 45.1 mm. The statistical analysis of the porous band of sample A1 in Fig. 12 is listed in
Table 2. All the measured results satisfy the statistical requirement discussed in the statistical analysis section. The non-uniformity in thickness of porous band at the steelconcrete interface is attributed to a phenomenon similar to that seen in the formation of ITZ, which is mainly governed by the wall effect [37,38]. During compacting, concrete on top of steel bar moves downwards, and as such, concrete on top becomes denser than concrete on other sides of the steel bar. The steel bar blocks the aggregates from moving downwards. This wall effect is more prominent in the case of contact between concrete (mortar) and steel bar than that between coarse aggregates and mortar (for ITZ), as steel bar is in a fixed position compared to a floating aggregate. Furthermore, it is easier for water to accumulate around steel bar as it moves faster than the aggregates. Therefore, the properties of the steel-concrete interface should not be assumed similar to those of ITZ [13]. This phenomenon shown in Figs. 11 and 12 was found in the BSE images of all nine BSE samples in the experiment. Fig. 13 shows the thickness of porous band as a function of its location represented by orientation from the top side of steel bar (moving counter-clockwise) for different slices along concrete samples. The results for each location, i.e., 1, 2 and 3 in Fig. 13 are the average value obtained from three concrete prisms of mix design A. It should be noted that the porous band at the area near the ribs of the steel bar is not included in the results shown in Figs. 13 and 14 but it is discussed separately in the next section because of the influence of the ribs. Variation of porous band size around the steel bar shows the same trend for all the slices along the steel bar. The thickness of porous band ranges from 10.1 mm to 54.4 mm based on the results of three repetition tests. Given the importance of size of porous band in the assessment of corrosion-induced cracking process, this highlights the necessity for accurate measurement of the interface.
4.2. Effect of water/cement ratio on porous band The variation of the thickness of porous band in samples A1, B1 and C1 is shown in Fig. 14 which shows the effect of w/c ratio of concrete on porous band at the interface. The porous band of sample A1 with 0.5 w/c ratio appears to be larger than those of the samples B1 and C1, especially at the bottom area of the steel. Although the thickness of porous band in sample B1 with 0.45
427
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
Fig. 12. The porous band at the interface around steel bar after thresholding from sample A1.
Table 2 Statistic analysis results of the porous band of sample A1. Angle location
Mean value (mm)
Standard deviation (mm)
t0:95;N1
Required N
Actual measurement times (N0 )
0° 90° 160° 180° 220° 270° 350°
3.9 4.2 42.1 45.1 28.7 17.6 4.5
0.82 0.95 9.09 10.6 6.45 3.74 1.03
2.093 2.074 2.086 2.069 2.080 2.093 2.074
19.6 22.1 20.3 23.5 21.9 19.8 22.4
20 23 21 24 22 20 23
60 Location 1 Location 2
Average thickness (μm)
50
Location 3
40 30 20 10 0 0
20 40 50 75 90 100 135 155 170 180 190 210 220 240 260 270 300 320 360
Angle location around steel (degrees)
(a) Effect of w/c ratio on porous band distribution around steel bar
Fig. 13. Thickness of porous band at the interface around the steel.
w/c ratio seems close to that of sample C1 with 0.4 w/c ratio, at the bottom area of the steel, namely at angles from 120° to 270°, the porous band in sample B1 is larger than that of sample C1. It is apparent that the w/c ratio can significantly affect the thickness of the porous band. High w/c ratio increases the size of the porous band as high w/c ratio favours the micro-bleeding of the cement paste and the wall-effect, especially at the bottom area of the steel bar. Fig. 14(a) also shows non-uniform distribution of the porous band around the steel due to the casting direction, i.e., the thickness of porous band on top of the steel bar is smaller than that at the bottom in the direction of compaction. However, low w/c
(b) Effect of w/c ratio on average and maximum values Fig. 14. Effect of w/c ratio on the porous band at the interface.
428
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
(a) porous band around the rib (sample A2)
(c) porous band around the rib (sample B1)
(b) corner of the rib (sample A2)
(d) porous band around the rib (sample C1)
Fig. 15. Microstructure around the rib of the steel.
ratio leads to a less the non-uniformity of the porous band around the steel. Fig. 14(b) shows the influence of w/c ratio on the average and maximum values of porous band. With the decrease of w/c ratio, both average and maximum values decline. If the w/c ratio is low, for example when w/c ratio is 0.45 and 0.4, the deceasing trend in porous band tends to fall.
4.3. Porous band near ribs of the steel bar Ribbed steel bars are widely used in RC structures to enhance the bond strength between steel and concrete. Interruption in smooth circular surface of ribbed steel bar can change the microstructures of concrete at the area near the rib, creating an interface which may be different from other parts of the steel bar surface. Fig. 15(a) shows a typical area around the rib of the steel bar from sample A2. The porous band at the interface right below the rib [denoted as Zone I in Fig. 15(a)] is larger than that at the side of the rib [denoted as Zone II in Fig. 15(b)]. The average size of the porous band formed in Zone I is 59.8 mm. It is nearly three times wider than the average size of porous band at the side of the rib, i.e., in Zone II, which is 17.1 mm. In Fig. 15(b), at the corner of the rib, a porous band can be seen, which seems to be connected to the porous band at the interface. It is also evident in the Fig. 15 (c) obtained from sample B1 that the porous area is connected to the porous area, leading to a large void at the bottom of the rib. In Fig. 15(d) obtained from sample C1 with a low w/c ratio, the porous area can also be found, however, it is not connected to the porous band. Therefore, for most corrosion-induced cracking models,
which consider the porosity or the empty porous band at the interface, the porous band and the influence of the steel ribs should also be considered. 5. Conclusions A reliable and accurate method for quantifying the geometry of porous band at the interface between reinforcing steel and concrete has been presented in this paper, using the technique of Backscattered Electron imaging. Details of the preparation procedure of reinforced concrete samples have been provided to reducing the damage to samples and deformation of steel during grinding and polishing processes. A new method for greyscale thresholding of porous band at the interface has been developed to obtain a reliable threshold value for porous band at the interface. From the analysis of the Backscattered Electron images it has been found that the size of porous band at the steel-concrete interface is not uniform around the steel bar, and that the porous band at the bottom area of the steel bar is larger than that at the top area in the direction of compaction. It has also been found that the water to cement ratio can significantly affect the size and the non-uniform distribution of the porous band around the steel bar, and that the porous band beneath the rib at the interface is larger than that at the side of the rib. It can be concluded that the microstructure of interface between steel and concrete is significantly affected by concrete mix and compaction and that further studies on influences of these factors on the steel-concrete interface are needed.
F. Chen et al. / Construction and Building Materials 179 (2018) 420–429
Conflict of interest The authors declared that there is no conflict of interest. Acknowledgments Financial support from Australian Research Council under DP140101547, LP150100413 and DP170102211 is gratefully acknowledged. The authors also acknowledge the facilities, and the scientific and technical assistance of the RMIT University’s Microscopy & Microanalysis Facility, a linked laboratory of the Australian Microscopy & Microanalysis Research Facility. References [1] A.T. Horne, I.G. Richardson, R.M.D. Brydson, Quantitative analysis of the microstructure of interfaces in steel reinforced concrete, Cem. Concr. Res. 37 (12) (2007) 1613–1623. [2] P. Monteiro, O.E. Gjorv, P.K. Mehta, Microstructure of the steel-cement paste interface in the presence of chloride, Cem. Concr. Res. 15 (5) (1985) 781–784. [3] F. Chen, H. Baji, C.Q. Li, A comparative study on factors affecting time to cover cracking as a service life indicator, Constr. Build. Mater. 163 (2018) 681–694. [4] L. Chernin, D.V. Val, Prediction of corrosion-induced cover cracking in reinforced concrete structures, Constr. Build. Mater. 25 (4) (2011) 1854–1869. [5] T. El Maaddawy, K. Soudki, A model for prediction of time from corrosion initiation to corrosion cracking, Cem. Concr. Compos. 29 (3) (2007) 168–175. [6] A. Castel, T. Vidal, R. Francois, G. Arliguie, Influence of steel–concrete interface quality on reinforcement corrosion induced by chlorides, Mag. Concr. Res. 55 (2) (2003) 151–159. [7] G.K. Glass, N.R. Buenfeld, The presentation of the chloride threshold level for corrosion of steel in concrete, Corros. Sci. 39 (5) (1997) 1001–1013. [8] T.U. Mohammed, N. Otsuki, H. Hamada, T. Yamaji, Chloride-induced corrosion of steel bars in concrete with presence of gap at steel-concrete interface, ACI Mater. J. 99 (2) (2002) 149–156. [9] C.L. Page, Mechanism of corrosion protection in reinforced concrete marine structures, Nature 258 (5535) (1975) 514–515. [10] T.A. Soylev, R. François, Quality of steel–concrete interface and corrosion of reinforcing steel, Cem. Concr. Res. 33 (9) (2003) 1407–1415. [11] G. Al-Sulaimani, M. Kaleemullah, I. Basunbul, Influence of corrosion and cracking on bond behavior and strength of reinforced concrete members, Struct. J. 87 (2) (1990) 220–231. [12] C.Q. Li, R.E. Melchers, J.-J. Zheng, Analytical model for corrosion-induced crack width in reinforced concrete structures, ACI Struct. J. 103 (4) (2006). [13] Y. Liu, R.E. Weyers, Modeling the time-to-corrosion cracking in chloride contaminated reinforced concrete structures, ACI Mater. J. 95 (6) (1998) 675– 681. [14] C. Lu, W. Jin, R. Liu, Reinforcement corrosion-induced cover cracking and its time prediction for reinforced concrete structures, Corros. Sci. 53 (4) (2011) 1337–1347. [15] A. Brough, A. Atkinson, Automated identification of the aggregate–paste interfacial transition zone in mortars of silica sand with Portland or alkaliactivated slag cement paste, Cem. Concr. Res. 30 (6) (2000) 849–854. [16] J.P. Ollivier, J.C. Maso, B. Bourdette, Interfacial transition zone in concrete, Adv. Cem. Based Mater. 2 (1) (1995) 30–38. [17] K.L. Scrivener, P.L. Pratt, Characterization of interfacial microstructure, Interfacial Transition Zone Concr. 2 (1996) 3–18.
429
[18] T.A. Söylev, R. François, Corrosion of reinforcement in relation to presence of defects at the interface between steel and concrete, J. Mater. Civ. Eng. 17 (4) (2005) 447–455. [19] K.L. Scrivener, P. Pratt, Backscattered electron images of polished cement sections in the scanning electron microscope, Proc. Int. Conf. Cem. Microsc. (1984) 145–155. [20] D. Darwin, M.N. Abou-Zeid, K.W. Ketcham, Automated crack identification for cement paste, Cem. Concr. Res. 25 (3) (1995) 605–616. [21] Y. Wang, S. Diamond, An approach to quantitative image analysis for cement pastes, MRS Online Proc. Lib. Arch. 370 (1994). [22] H.S. Wong, M. Zobel, N.R. Buenfeld, R.W. Zimmerman, Influence of the interfacial transition zone and microcracking on the diffusivity, permeability and sorptivity of cement-based materials after drying, Mag. Concr. Res. 61 (8) (2009) 571–589. [23] R. Yang, N.R. Buenfeld, Binary segmentation of aggregate in SEM image analysis of concrete, Cem. Concr. Res. 31 (3) (2001) 437–441. [24] G.K. Glass, R. Yang, T. Dickhaus, N.R. Buenfeld, Backscattered electron imaging of the steel–concrete interface, Corros. Sci. 43 (4) (2001) 605–610. [25] M.K. Head, N.R. Buenfeld, Measurement of aggregate interfacial porosity in complex, multi-phase aggregate concrete: Binary mask production using backscattered electron, and energy dispersive X-ray images, Cem. Concr. Res. 36 (2) (2006) 337–345. [26] H.S. Wong, M.K. Head, N.R. Buenfeld, Pore segmentation of cement-based materials from backscattered electron images, Cem. Concr. Res. 36 (6) (2006) 1083–1090. [27] H.S. Wong, M.K. Head, N.R. Buenfeld, Characterising the pore structure of cement-based materials using backscattered electron and confocal microscopy, in: M.S. Konsta-Gdoutos (Ed.), Measuring, Monitoring and Modeling Concrete Properties: An International Symposium dedicated to Professor Surendra P. Shah, Northwestern University, U.S.A., Springer Netherlands, Dordrecht, 2006, pp. 495–502. [28] ACI, Building Code Requirements for Structural Concrete (ACI 318-14): An ACI Standard: Commentary on Building Code Requirements for Structural Concrete (ACI 318R-14), an ACI Report, ACI Committee 318, American Concrete Institute, ACI Committee, 2015. [29] K.L. Scrivener, A.K. Crumbie, P. Laugesen, The interfacial transition zone (ITZ) between cement paste and aggregate in concrete, Interface Sci. 12 (4) (2004) 411–421. [30] S. Diamond, Considerations in image analysis as applied to investigations of the ITZ in concrete, Cem. Concr. Compos. 23 (2) (2001) 171–178. [31] B. Ma, F. Chen, B. Zhang, C. Lin, The design and application of Ballastless track enhanced concrete mineral admixture, Appl. Mech. Mater., Trans. Tech. Publ. (2014) 1427–1430. [32] H. Tan, F. Zou, B. Ma, M. Liu, X. Li, S. Jian, Effect of sodium tripolyphosphate on adsorbing behavior of polycarboxylate superplasticizer, Constr. Build. Mater. 126 (2016) 617–623. [33] K.L. Scrivener, A. Bentur, P. Pratt, Quantitative characterization of the transition zone in high strength concretes, Adv. Cem. Res. 1 (4) (1988) 230– 237. [34] K.L. Scrivener, H. Patel, P. Pratt, L. Parrott, Analysis of phases in cement paste using backscattered electron images, methanol adsorption and thermogravimetric analysis, MRS Online Proc. Lib. Arch. 85 (1986). [35] S. Diamond, J. Huang, The ITZ in concrete – a different view based on image analysis and SEM observations, Cem. Concr. Compos. 23 (2) (2001) 179–188. [36] C. Forbes, M. Evans, N. Hastings, B. Peacock, Student’s t Distribution, Statistical Distributions, John Wiley & Sons, Inc., 2010, pp. 183–186. [37] A. Bentur, M. Alexander, A review of the work of the RILEM TC 159-ETC: engineering of the interfacial transition zone in cementitious composites, Mater Struct 33 (2) (2000) 82–87. [38] K.L. Scrivener, Characterisation of the ITZ and its quantification by test methods, RILEM Rep. (1999) 3–18.