Influence of optical fiber diameter on thermoluminescence response

Author’s Accepted Manuscript Influence of Optical Fiber Thermoluminescence Response

Diameter

on

Ghafour Amouzad Mahdiraji, Mostafa Ghomeishi, Faisal Rafiq Mahamd Adikan, D.A. Bradley www.elsevier.com/locate/radphyschem

PII: DOI: Reference:

S0969-806X(17)30392-4 http://dx.doi.org/10.1016/j.radphyschem.2017.04.001 RPC7511

To appear in: Radiation Physics and Chemistry Received date: 1 November 2016 Revised date: 3 April 2017 Accepted date: 5 April 2017 Cite this article as: Ghafour Amouzad Mahdiraji, Mostafa Ghomeishi, Faisal Rafiq Mahamd Adikan and D.A. Bradley, Influence of Optical Fiber Diameter on Thermoluminescence Response, Radiation Physics and Chemistry, http://dx.doi.org/10.1016/j.radphyschem.2017.04.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Influence of Optical Fiber Diameter on Thermoluminescence Response Ghafour Amouzad Mahdiraji1,2*, Mostafa Ghomeishi3, Faisal Rafiq Mahamd Adikan2,3, D.A. Bradley4,5 1

School of Engineering, Taylor’s University, 47500 Subang Jaya, Malaysia Flexilicate Sdn. Bhd., University of Malaya, 50603 Kuala Lumpur, Malaysia 3 Integrated Lightwave Research Group, Department of Electrical Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia 4 Department of Physics, Surrey University, UK 5 Sunway University, Institute for Healthcare Development, 46150 PJ, Malaysia * Correspondence author. [email protected] 2

Abstract: Detailed investigation is made of the thermoluminescence (TL) response of various sizes of optical fiber, fibers being developed with either a fixed core-to-cladding ratio or various core-to-cladding ratios. Two Ge-doped optical fiber preforms have been used to reconfirm the experimental findings. For further clarification, a pure silica rod has also been used to fabricate different diameter rods. Experimental investigations show the main TL signal to be generated from the fiber core within which the Ge is doped, the fiber cladding producing insignificant TL signal. Prior to normalization, the results show that in doubling the fiber diameter the TL signal quadruples. Conversely, subsequent to normalizing the different sizes of optical fiber to their mass or fiber cross sectional area, the smaller diameter fibers show slightly greater sensitivity compared to the larger diameter fibers. Relating to the fiber drawing-down process, this is due primarily to the greater shearing effect that the smaller fibers experience compared to the larger fibers within the fiber preform neck-down region. Keywords: optical fiber dosimeter, thermoluminescence based dosimeter, fiber size, fiber fabrication

1. Introduction Over the past several years interest has grown significantly in optical fiber dosimetry, a passive form of dosimetry, initially at least, based on the luminescence yield that the doped fibers present, either promptly upon irradiation (radioluminescence, RL) or stimulated from the stored energy of trapped electrons (thermoluminescence, TL or optically stimulated luminescence, OSL), responding to the needs of different ionizing radiation applications. The inherent advantages of these doped fibers include their high spatial resolution, immunity to electromagnetic interference, and capability to be used in real time or in offline monitoring systems, as well as other interesting dosimetry characteristics such as linear response over a wide range of applied dose, high sensitivity and energy independence within the MV external-beam radiotherapy regime. They are also temperature independent, below through to above room temperature, reusable providing reproducible results, suffer low thermoluminescence (TL) signal fading, are impervious to water, are of significantly lower cost compared to current commercially available passive dosimeters and are able to detect a variety of radiations, from fast to heavy particles. The irradiation performance of different types of commercially available optical fibers such as standard single mode fibers (SMFs) (Mahdiraji et al., 2015b) and multimode fibers (MMFs) (Girard et al., 2013) doped with different rare earth materials, for instant germanium (Benabdesselam et al., 2013, Mahdiraji et al., 2015b, Girard et al., 2013), phosphorus (Girard et al., 2006, Paul et al., 2009, Ghosh et al., 2011, Girard et al., 2013, Lu et al., 2000), aluminum (Polf et al., 2004, Yaakob et al., 2011), Ge-boron (Mahdiraji et al., 2015a), fluorine (Alessi et al., 2012b), have all been demonstrated. Performance comparison between the SMFs and MMFs in terms of sensitivity towards radiation dose detection suggests the greater performance of the MMFs to be due mainly to the larger fiber core area; however, currently there does not exist any classical report that clearly evaluates the performance of such optical fibers, neglecting the influence of the fiber manufacturing process and the elemental doping concentrations since most of those reports have used optical fibers fabricated by different manufacturers using different recipes. Zahaimi et al. (2014) compared the thermoluminescence (TL) response of Ge-doped optical fibers of different core diameters, from 8 to 50 µm, showing the larger core fibers to generate greater TL; however, the optical fibers used in that study were fabricated by different manufacturers, almost certainly with different Ge concentrations, leaving unclear the effect of the manufacturing process and elemental concentrations. Reports of the effect of fiber drawing condition on characteristics of optical fibers date back to the 1970s, as an example showing absorption induced at wavelengths of 215 nm and 248 nm (Lee et al., 1998), 630 nm (Hibino and Hanafusa, 1986, Hanafusa et al., 1987), and 1530 nm (Sakaguchi et al., 1985) at higher

fiber drawing tension. Also shown was refractive index reduction with increase of residual stress (or drawing tension) (Hibino et al., 1989). Conversely, to-date a detailed report on the influence of the fiber manufacturing process on the irradiation characteristics of optical fibers for dosimetry applications has not been forthcoming. Recently, Girard and Alessi et al., demonstrated the influence of manufacturing process on radiation induced attenuation (RIA) of optical fibers considering both the preform fabrication using modified chemical vapor deposition (MCVD) process and the fiber drawing conditions (Alessi et al., 2011, Alessi et al., 2012a, Girard et al., 2006, Girard et al., 2012). In (Girard et al., 2006), Girard et al. observed a slight reduction in RIA at 1550 nm in reducing the fiber drawing tension and further observed that by lowering the preform deposition temperature during MCVD process the RIA at 1310 nm and 1550 nm reduces for short periods of time after irradiation. In (Alessi et al., 2011, Alessi et al., 2012a, Girard et al., 2012), the authors have shown there to be an insignificant effect of drawing parameters on RIA of optical fibers within a range of fiber drawing parameters used for special fiber fabrications. However, these studies are limited to RIA at high radiation doses, in the range of kGy – MGy. Furthermore, the relationship between irradiation dose sensitivity of optical fibers against fiber core diameter/area has yet to be made clear. Although, Zahaimi et al. (2014) have shown a linear response of y = 24x or 25x, with x representing dose, this relationship would not be a sufficiently accurate prediction since the optical fibers are made with different Ge concentrations and fabricated by different manufacturers. In previous work, we have shown the influence of fiber structure and materials on sensitivity of optical fibers. In (Mahdiraji et al., 2015a, Mahdiraji and Adikan, 2014), we have compared dose detection sensitivity of eight different optical fibers, four different materials, i.e., ultra-pure silica, pure-silica, Ge-doped, and Ge-B-doped fibers, and two different structures, i.e., capillary and flat optical fibers. We have shown that while a capillary optical fiber might have very low dose detection sensitivity, its sensitivity can be increased, in our investigations by up to 31 times when its structure is collapsed down during the fiber drawing process. The sensitivity of the Ge-B-doped flat fiber has been shown to be even greater than that of commercially available TLD-100 dosimeters (doped LiF media). In (Bradley et al., 2015), we have compared TL response of flat fiber with photonic crystal fiber (PCF). It has been shown that when fabricated from the same material a PCF is more sensitive than a flat fiber in terms of dose detection. In yet another study (Dermosesian et al., 2015), we have shown that the TL yield in a PCF can be further increased by collapsing down all of the holes in the fiber structure during the drawing process. The TL characteristics of a Ge-doped PCF and collapsed-hole-PCF have been compared against that of a Ge-doped rod made from the same preform (Mahdiraji et al., 2015c). In (Ghomeishi et al., 2015), the TL properties of Ge-flat fiber in comparison with other fiber types have been investigated in detail. In (Mahdiraji et al., 2016), the dose detection sensitivity of different optical fibers have been compared considering the fabrication process and dosimeter characteristics. Very recently, we have investigated the influence of dose history and background dose on TL response of Ge-doped optical fiber dosimeter (Moradi et al., 2017a) and the angular dependence of TL based fiber dosimeters (Moradi et al., 2017b). To the best of our knowledge there is no previous report investigating in detail the influence of fiber size on TL performance of optical fibers. In this present study, the effect of different fiber diameters, core-to-cladding ratio, and fiber drawing conditions on performance of radiation dose detection sensitivity are demonstrated. However, the relationship between the sizes of such tailor-made optical fibers have yet to be studied. 2. Methodology 2.1 Optical Fiber Fabrication In this study, two Ge-doped fiber preforms and one pure silica rod have been used to observe the influence of fiber diameter on fiber dose detection sensitivity. The germanium-silicate core fiber preforms have been developed by Multimedia University and TM R&D group using the MCVD process. The average concentration of germanium doped in the fiber preform core were 5.78 and 8.0 mole percent (for simplicity, from here onwards they will be referred to as 6 and 8 mole%, respectively). The preforms had an approximate length of around 15 cm and 30 cm with average core/cladding diameter of 2.48/14.98 and 2.34/14.92 cm for the 6 and 8 mole% preforms, respectively. Figures 1 and 2 show the physical parameters of the Ge-doped preforms including the core and cladding diameter, Ge-concentration variation, and the refractive index difference between the core and cladding over the fiber preforms length. The other preform, pure silica rod, was a Heraeus HQS300 silica rod with outer diameter of 21 mm. The pure silica rod is used to reconfirm the results observed from the doped preforms.

Figure 1: 6 mole% Ge-doped fiber preform physical parameters, (left) core and cladding diameter and (right) refractive index difference.

Figure 2: 8 mole% Ge-doped fiber preform physical parameters, (left) core and cladding diameter and (right) refractive index difference profile. All preforms were subsequently pulled into optical fibers to obtain five different diameters, as listed in Table 1, use being made of the fiber drawing facilities in the Department of Electrical Engineering, University of Malaya. The nominal fiber cladding diameters for the Ge-doped fibers were 120, 241, 362, 483, and 604 µm, which was planned for fiber core diameters of 20, 40, 60, 80, and 100µm, respectively. However, the actual fiber sizes are as shown in Table 1, where the fiber cladding diameters listed in this table are the average value measured from SEM images. The fiber core diameters are estimated based on the average core/cladding ratio of the fiber preform. For simplicity in discussion and illustrations, the nominal fiber cladding diameters have been used. A variation of up to around ± 5% have been observed in the fiber cladding diameters. All optical fibers have been pulled with a low drawing speed of 2-3 m/min, relative temperature of 2000 °C (for the smallest fiber size) up to 2080 °C (for the largest fiber size), and a relatively fixed drawing tension of around 45 ±10 g. Figure 3 shows examples of fiber cross sectional images for five different sizes taken by SEM from 8 mole% Ge-doped fibers. The Ge concentration in the core of 6 and 8 mole% Ge-doped optical fibers have been measured using EDX and photon-induced x-ray emission (PIXE) (using the PIXE facility at the Ion Beam Centre of the University of Surrey). The average Ge concentration measured by EDX shows 8.2 and 12.1 weight% for 6 and 8 mole% fibers, respectively. These concentrations approached agreement with the micro PIXE measurement, details of the measurements being available in (Sani et al., 2014).

Table 1: List of optical fiber sizes pulled from different preforms. The average fiber cladding diameters are measured from SEM images, while the core diameters are estimated from the fiber preform core/cladding ratio (for 6 and 8 mole% fibers). Average fiber size (µm) No. Preform type 1 2 3 4 5 1 6 mole% Ge-doped (core/cladding) 18/110 40/240 60/360 80/480 96/580 2 8 mole% Ge-doped (core/cladding) 20/130 38/245 57/365 75/480 95/610 3 Pure silica rod 150 200 250 300 350

Figure 3: Optical fiber cross sectional images for five different diameters taken from 8 mole% Ge-doped fibers. The actual fiber diameter in this figure is (a) 130 µm, (b) 240 µm, (c) 379 µm, (d) 482 µm, and (e) 602 µm. 2.2 Sample Preparation and Irradiation Prior to the irradiation process, the fibers were cleansed by cotton cloth containing methyl alcohol to remove any impurities on the fibers and then were manually cut into lengths of 5 ± 0.5 mm using a diamond cone point cutter. Afterward, the samples were annealed at 400 ˚C using a Thermo Electric furnace for 1 hour. The annealing process is carried in order to standardize the thermal history within the fiber and to remove any prior TL signal (e.g., from triboluminescence, due to mechanical deformations, abrasions etc), eliminating to large extent the unstable glow curve at low-temperature (Hashim et al., 2009). After cooling (to room temperature), 10-15 pieces of each sample were placed in small plastic bags to ensure homogeneity during irradiation and easy handling purposes. The Ge-doped fiber samples were exposed to 6 MV x-ray beam produced by a Varian Model 2100C linear accelerator (Varian Medical System, Palo Alto, USA) at the University of Malaya Medical Centre, with doses ranging from 0.5 to 8 Gy. Field size of 20 × 20 cm2, source to skin distance (SSD) of 100 cm and applicator size of 20 × 20 cm2 were used for all the irradiations. The samples were placed at the surface of a solid-waterTM phantom. The pure silica fiber dosimeter samples were exposed to a dose of 10 Gy dose using a Cobalt-60 gamma irradiator (Gammacell 220, GC-220) located at the Department of Physics, University of Malaya.

2.3 TL measurement and normalization After exposing the fiber samples, the TL yield of the fibers were read out using a Harshaw 3500 TL reader. The time temperature profile (TTP) of the TL reader was set to provide a preheat temperature of 50 °C, an acquired temperature ramp rate of 25 °C/s and acquisition time of 20 s. The mass of the fiber samples were measured by means of an accurate electronic balance and then the TL responses were normalized to fiber mass; in this study the mass of 10-15 fiber samples per species were measured and the average mass was used for normalization. 2.4 Experiments In this work five sets of samples from 6 mole% Ge-doped fibers with different cross sectional size have been compared against each other. The experiment has been repeated using another five series of samples with almost the same fiber diameters but this time with the fibers made from another preform, 8 mole% Ge-doped fiber preform, acting as a double check of the size comparison. Finally, a pure silica rod has been used to fabricate five different sized rod, seeking to ensure validity of the observation from the Ge-doped fibers regarding the effect of fiber size. TL yield and glow curve, linearity and sensitivity are used herein to discuss the observed phenomenon. 3. Results and Discussion

TL yield (nC)

This study has been performed principally to investigate the influence of cylindrical optical fiber diameter on TL yield and dose detection sensitivity. Since fabrication of optical fibers with different diameters requires different fiber drawing parameters, including temperature, drawing speed, and/or preform feed-rate, the output fibers might not be directly comparable. On the other hand, any changes in the drawing speed and temperature have direct affect on the drawing tension, in which the tension increases by either increasing the drawing speed or reducing the temperature, while the other factor is fixed. Since the tension is the outcome interaction between drawing speed and temperature, the tension is kept fixed for all optical fibers fabricated for this study. However, during the fabrication process, sometimes, it was somewhat difficult to obtain the desired tension with small variation. As stated earlier in Section 2.1, the tension variation between different optical fibers fabricated for this study has been observed to be ±10 g. To understand the effect of tension variation on TL yield, a Gedoped optical fiber with fix diameter of 250 ± 10 µm has been fabricated with six different tensions, from 15 g up to 100 g. The tension has been changed by keeping the fiber drawing speed relatively constant (small change was necessary in controlling the fiber size at different temperature) while changing the furnace temperature from high to low temperature. Figure 4 shows the result of TL yield measured from the fibers with six different drawing tensions. The results do not reveal any significant difference between the TL yield at 15 g tension compared to that at 100 g. The highest variation is observed at a tension of 60 g, with a TL yield some 10% lower than the average TL yield amongst the different tensions. This variation is expected to be due to the variation in fiber diameter, there being an approximate ± 10 µm variation in the fiber diameter used for this experiment.

500

250

0 15

30

45 60 Drawing tension (g)

80

100

Figure 4: TL yield comparison between Ge-doped fibers fabricated with six different drawing tensions. In another experiment, the TL contribution from fiber cladding has been experimentally investigated. This has been done by etching the cladding of the Ge-doped fiber. Figure 5 illustrates the etching steps over different times (left-hand side

illustration) and the resulting fibers experienced different etching times (right-hand side series of illustrations). Two sets of 604 µm Ge-doped fibers, five pieces from 6 mole% and another five pieces from 8 mole% were selected. The fiber samples were placed inside the HF (hydrofluoric) acid solution and each of them were removed from the acid after a fixed period of 15 min. Figure 6(a) shows the TL response of the optical fibers that have been subjected to different etching times. This shows that by removing the cladding of the Ge-doped fibers, the TL response of the fiber does not change significantly. This implies insignificance of fiber cladding in generating TL signal compared to the TL yield from the fiber core. While it is not a true conclusion to imply there is no TL signal generated from the cladding, nevertheless compared to the fiber core it provides a negligible contribution. It should be noted that this statement would be valid only for optical fibers with pure silica cladding and doped elements in the core. However, for the optical fibers in which the cladding is doped with a material, usually boron, and the core is pure silica, or any other type of optical fiber in which the cladding includes some dopant element, this observation might not be valid. The results in Figure 6(a) also implies that regardless of the fiber cladding diameter, if the core diameter of different sizes of optical fibers are the same, the TL response of all fibers should be about equal. In other words, normalizing TL yield to fiber core area results in similar sensitivity as shown in Figure 6(a), as the fiber core areas are the same. However, if the TL yield is normalized to the fiber cladding area (the area here represents the fiber mass as the length of the fibers are about the same, 5 ± 0.5 mm), the results show different sensitivity for different fiber sizes. In this case, of course, the smaller fiber will be more sensitive as it has lower mass. Also, the result suggests a second order polynomial relationship which is in agreement with diminished surface size (π r2) for circular cross section fiber. This second order rate of decay in TL signal normalized to the fiber area shows that although all optical fibers have the same TL response before normalization, their sensitivity varies from the smaller sample to the larger sample.

Figure 5: (left) Schematic of etching fiber cladding, (right) cross section image of 8 mole% Ge-doped optical fibers experienced different etching time.

Figure 6: (a) TL yield from 6 and 8 mole% Ge-doped fiber samples subjected to different etching periods or fiber samples with different cladding thicknesses, and (b) TL yield normalized to the fiber area for 8 mole% Ge-doped fiber. Figure 7 shows the glow curves (GC) of 6 and 8 mole% Ge-doped fibers of different sizes as well as for different applied doses. The optical fibers fabricated from the two preforms show very similar GC shape, the GCs in 6 mole% having slightly narrower peaks compared to that obtained at 8 mole%. Another observation is that at 6 mole%, the GCs trailing edge is not as sharp as the rising edge and reduces more gradually compared to the 8 mole%, as illustrated in the Figure 7(b).

Figure 7: Glow curve (GC) of 6 mole% (top) and 8 mole% (bottom) Ge-doped fibre samples. (a) and (c) GC comparison for different doses measured from 604 µm fiber. (b) and (d) GC comparison for different fiber sizes measured at 8 Gy dose. Figures 8 show the TL yield of the Ge-doped fibers for both 6 mole% (a) and 8 mole% (b) for the five different sizes over five different dose applied before normalization to the fiber sample mass. Increase in the TL yield of samples over different sizes shows an exponential relation between TL yield and fiber size. Since the density of the samples in each fiber type is the same, having the same dopant percentage, it can be concluded that the TL response has a direct relation with the root square of the mass or volume of the samples. This pattern is repeated in almost all the doses from 0.5 Gy to 8 Gy for both groups of 6 and 8 mole%. Although there is some deviation from the pattern, this could be due to errors during the experiment, sample preparation, and/or the variation during the fiber preform and fiber fabrication. The error bars are evidence of these variations, showing the standard deviation between the experimental samples.

Figure 8: TL yield of optical fibers with five different diameters and five different applied doses using (a) 6 mole% and (b) 8 mole% Ge-doped fibers before normalization to the sample mass. For an improved understanding of the relationship between the fiber size and their TL yield, further analysis is shown in Figure 9. Figures 9(a) and (b) show the TL yield over fiber core diameters for 6 and 8 mole% Ge-doped fibers exposed to various doses, respectively. Almost all results suggests that this relationship is a second order polynomial, meaning that by doubling the fiber core diameter, the TL response increased by four times. For reconfirming this observation, the TL yields are plotted versus fiber core area as depicted in Figures 9(c) and (d) for the 6 and 8 mole% for different applied doses, respectively. These diagrams show a linear response between the TL yield and fiber core area. In some cases in Figure 9, the experimental results not well matched with the fitted curves, as in graphs number (a2) and (c2), (b1) and (d1), and (b4) and (d4). These variations would be from different sources. One of the main source of variation could be from the fiber preform. Referring to Figure 2(a), the fiber preform core diameter and the Ge-concentration show variations of up to ± 10% from one side of the preform to the other. Additionally, variations of up to approximately ±10% is expected to be induced due to the variation in the length of fiber samples as they were cut manually to 5 ± 0.5 mm. These two variations cause significant uncertainty in the TL response of resulting fibers. Both of these variations can be suppressed in real practice, one by automatically cutting the sample and another by repeating the preform fabrication process in MCVD, as the fiber preforms used in this study were custom made.

Figure 9: (a) and (b) TL response versus fiber core diameter for 6 mole% and 8 mole% Ge-doped fibers over five different applied doses. (c) and (d) TL yield versus fiber core area for 6 mole% and 8 mole% Ge-doped fibers over five different applied doses. (e) TL response versus silica rod diameter and (f) TL response versus silica rod cross section area for 8 Gy applied dose.

To reduce these variations, a pure silica rod has been used to reconfirm the above observations. It should be noted that the silica rod used in this experiment is not as pure as the Suprasil F300 material used in the cladding of Ge-doped fibers. Since there is not any dopant in this pure silica rod, the variations due to fiber preform should be reduced here. This experiment is performed by drawing 5 different sizes of silica rods and exposed to a fixed dose of 8 Gy using gamma irradiation. The TL yield from silica rods versus rods diameter is shown in Figure 9(e) and the TL yield versus the rod cross section area is shown in Figure 9(f). A regression of R2 > 99% is observed in both graphs confirming an exponential relationship between TL response and fiber size or a linear response between TL response with fiber core area. Finally, the TL response of the optical fiber samples have been normalized to their fiber mass, as shown in Figure 10. It should be noted that if optical fibers of different size are fabricated from the same fiber preform, then the fiber core-tocladding ratio is expected to be the same. As such, there should be no difference in normalizing against fiber mass, core area, or volume. However, if two different groups of optical fiber with different core-to-cladding ratio are to be compared, then the fiber core area should be considered as the main reference for normalizing different fibers (with the understanding that the cladding of the fiber is undoped and the assumption that the fiber core generates the predominant TL signal). Thus, two groups of optical fiber with different core-to-cladding ratio cannot be fairly compared if they are normalized to their mass or cross sectional area (core plus cladding area); such comparison would be invalid. The TL response of 6 mole% Ge-doped fibers with five different sizes and five applied doses are normalized to the fiber mass (Figure 10(a) and (b)) and show that the TL sensitivity slightly reduces from the small diameter fiber to the larger fiber samples. This can be seen from almost all five applied doses. This observation is even more obvious from the results of 8 mole% Ge-doped fibers, as shown in Figure 10(c) and (d). However, since in both 6 and 8 mole% Ge-doped fibers, some variations in TL yield were observed due to the two main sources of variations explained above, i.e., variation from the fiber preform and the variation from the length of fiber sample; therefore, pure silica rod is used here as a third reference to validate the result. Since in pure silica there is no any doped element, the variation induced by doping concentration and fiber core area does not exist. However, the variation due to the cutting of fiber sample may still remain. Figure 10(e) shows the TL response of the silica rods with different diameters exposed under 8 Gy dose and normalized with their mass. The result shows a reducing TL response from the fiber with smaller diameter to the fiber with larger diameter. The results observed from the pure silica rod in Figure 10(e) is in agreement with the results observed in the doped fibers presented in Figure 10(a) to (d) in which, the fibers with smaller diameter generates greater TL signal compared to a fiber with larger diameter if both are fabricated from the same preform.

Figure 10: TL response of optical fibers with different sizes and different applied doses normalzied to their mass for (a) and (b) 6 mole% Ge-doped fibers, (c) and (d) 8 mole% Ge-doped fibers, and (e) pure silica rod. The reason for such behaviour existing in optical fibers of different sizes fabricated from the same fiber preform has not been well resolved in any previous report, including those using the same batch of optical fibers, 6 and 8 mole% (Wahib et al., 2015, Begum et al., 2013). In (Wahib et al., 2015), the reason why a smaller optical fiber generates higher TL signal compared to a larger fiber is explained, it being concluded that this should be due to the loss of the TL signal traveling from the fiber cladding. Since the larger fiber has thicker cladding compared to the smaller fiber, thereby its TL yield is lower. This statement has not been experimentally validated. Further, the results of present study (Figure 6) do not support the statement. In Figure 6, we have observed that the fiber TL yield does not change when the fiber cladding is removed by etching. This suggests that even if the TL signal generated in the fiber core experiences some loss when passing from the cladding thickness, it is not sufficient to explain that observed in Figure 10. The main reason for the sensitivity difference between the different fiber sizes is suggested to be due to the fiber shearing effect in the fiber neck-down, as depicted in Figure 11. Drawing a fiber to smaller diameter results in a greater degree of shearing and thereby greater defects production in the fiber when compared to a larger diameter fiber fabricated from the same fiber preform.

Figure 11: Schematic diagram illustrating the shearing effect at the neck-down region in drawing down two different sizes of optical fibers from the same preform, (a) larger diameter fiber experiences less shearing, and (b) smaller fiber diameter experiences comparatively greater shearing. Conclusion The influence of fiber size on TL response of optical fibers is demonstrated in detail. Various techniques are used to analyze and evaluate the effect of fiber size on TL yield. With the aid of HF acid, different sizes of optical fibers with different coreto-cladding ratio have been generated and their characteristics have been discussed. It is shown that the main TL signal is generated from the fiber core of a Ge-doped fiber, while the cladding produces insignificant TL compared to the fiber core. This is proven by showing that the TL signal of optical fibers remained constant before and after removing part of the fiber cladding. On the other hand, five different diameters of optical fibers with the same core-to-cladding ratio have been fabricated from a fiber preform with 6 mole% Ge concentration. The result of this fibers is reconfirmed by drawing another five sizes of optical fibers fabricated from another Ge-doped preform, i.e., 8 mole% Ge concentration. For further clarifications, a pure silica rod has also been used to fabricate five different sizes of silica rod to validate the observations from the two Ge-doped fibers. Results show that the TL signal quadrupled when the fiber diameter is doubled before normalization. However, when the fibers were normalized to their mass or cross sectional area, unexpectedly, greater sensitivity is observed from the fibers of smaller diameter compared to the fibers of larger diameter. The reason why a smaller optical fiber become more sensitive that a larger fiber when both of them are fabricated from the same preform has been suggested be due to the shearing effect in the preform neck-down region. A smaller diameter fiber experiences greater shearing in the neck-down region compared to the larger diameter fiber, thereby inducing higher defects production in the fiber, resulting in greater sensitivity. Acknowledgements The authors would like to acknowledge the MOHE grant number FRGS/1/2016/TK04/UM/01/3 to support this project as well as the UM-MOHE High Impact Research Grant numbers A000007-50001 and UM.C/625/1/HIR/33, supporting the Gedoped fibers project, and also the Multimedia University MCVD group for fabricating the Ge-doped preforms with full specification. References ALESSI, A., GIRARD, S., CANNAS, M., AGNELLO, S., BOUKENTER, A. & OUERDANE, Y. 2012a. Influence of drawing conditions on the properties and radiation sensitivities of pure-silica-core optical fibers. Journal of Lightwave Technology, 30, 1726-1732. ALESSI, A., GIRARD, S., MARCANDELLA, C., AGNELLO, S., CANNAS, M., BOUKENTER, A. & OUERDANE, Y. 2011. X-ray irradiation effects on a multistep Ge-doped silica fiber produced using different drawing conditions. Journal of Non-Crystalline Solids, 357, 1966-1970.

ALESSI, A., GIRARD, S., MARCANDELLA, C., VACCARO, L., CANNAS, M., BOUKENTER, A. & OUERDANE, Y. 2012b. Influence of the Manufacturing Process on the Radiation Sensitivity of Fluorine-Doped Silica-Based Optical Fibers. Nuclear Science, IEEE Transactions on, 59, 760-766. BEGUM, M., RAHMAN, A. K. M. M., ABDUL-RASHID, H. A., ABDULLAH, W. S. W., NOOR, N. M., ZULKIFLI, M. I., MUHAMAD-YASIN, S. Z., MAT-SHARIF, K. A., TAMCHEK, N., NAWI, S. N. M., WAHIB, N., AMIN, Y. M. & BRADLEY, D. A. Thermoluminescence response of Ge-doped optical fiber dosimeters with different core sizes. 2013 IEEE 4th International Conference on Photonics (ICP), 28-30 Oct. 2013 2013. 291-293. BENABDESSELAM, M., MADY, F., GIRARD, S., MEBROUK, Y., DUCHEZ, J. B., GAILLARDIN, M. & PAILLET, P. 2013. Performance of ge-doped optical fiber as a thermoluminescent dosimeter. IEEE Transactions on Nuclear Science, 60, 4251-4256. BRADLEY, D. A., MAHDIRAJI, G. A., GHOMEISHI, M., DERMOSESIAN, E., ADIKAN, F. R. M., RASHID, H. A. A. & MAAH, M. J. 2015. Enhancing the radiation dose detection sensitivity of optical fibres. Applied Radiation and Isotopes, 100, 43-49. DERMOSESIAN, E., MAHDIRAJI, G. A., MAHAMD ADIKAN, F. R. & BRADLEY, D. A. 2015. Improving thermoluminescence response through the fabrication of novel microstructured fibers. Radiation Physics and Chemistry, 116, 135-137. GHOMEISHI, M., MAHDIRAJI, G. A., ADIKAN, F. R. M., UNG, N. M. & BRADLEY, D. A. 2015. Sensitive fiber-based thermoluminescence detectors for high resolution in-vivo dosimetry. Scientiric Reports, 5, 1-10. GHOSH, S., DAS, S., PAUL, M. C., DASGUPTA, K., BOHRA, D., CHAUDHARY, H. S., PANWAR, L., BHATNAGAR, P. K. & VAIJAPURKAR, S. G. 2011. Evaluation of the performance of high phosphorous with germanium codoped multimode optical fiber for use as a radiation sensor at low dose rates. Applied Optics, 50, E80-E85. GIRARD, S., MARCANDELLA, C., ALESSI, A., BOUKENTER, A., OUERDANE, Y., RICHARD, N., PAILLET, P., GAILLARDIN, M. & RAINE, M. 2012. Transient Radiation Responses of Optical Fibers: Influence of MCVD Process Parameters. Nuclear Science, IEEE Transactions on, 59, 2894-2901. GIRARD, S., MARCANDELLA, C., MORANA, A., PERISSE, J., DI FRANCESCA, D., PAILLET, P., MACE, J. R., BOUKENTER, A., LEON, M., GAILLARDIN, M., RICHARD, N., RAINE, M., AGNELLO, S., CANNAS, M. & OUERDANE, Y. 2013. Combined high dose and temperature radiation effects on multimode silica-based optical fibers. IEEE Transactions on Nuclear Science, 60, 4305-4313. GIRARD, S., OUERDANE, Y., BOUKENTER, A. & MEUNIER, J. P. 2006. Transient radiation responses of silica-based optical fibers: Influence of modified chemical-vapor deposition process parameters. Journal of Applied Physics, 99. HANAFUSA, H., HIBINO, Y. & YAMAMOTO, F. 1987. Formation mechanism of drawing-induced defects in optical fibers. Journal of Non-Crystalline Solids, 95-96, 655-661. HASHIM, S., AL-AHBABI, S., BRADLEY, D. A., WEBB, M., JEYNES, C., RAMLI, A. T. & WAGIRAN, H. 2009. The thermoluminescence response of doped SiO2 optical fibres subjected to photon and electron irradiations. Appl Radiat Isot, 67, 423-7. HIBINO, Y. & HANAFUSA, H. 1986. Defect structure and formation mechanism of drawing-induced absorption at 630 nm in silica optical fibers. Journal of Applied Physics, 60, 1797-1801. HIBINO, Y., HANAWA, F. & HORIGUCHI, M. 1989. Drawing-induced residual stress effects on optical characteristics in pure-silica-core single-mode fibers. Journal of Applied Physics, 65, 30-34. LEE, J.-W., SIGEL JR, G. H. & LI, J. 1998. Processing-induced defects in optical waveguide materials. Journal of NonCrystalline Solids, 239, 57-65. LU, P., BAO, X., BROWN, K. & KULKARNI, N. 2000. Gamma-induced attenuation in normal single-mode and multimode, Ge-doped and P-doped optical fibers: A fiber optic dosimeter for low dose levels. Canadian Journal of Physics, 78, 89-97. MAHDIRAJI, G. A. & ADIKAN, F. R. M. 2014. Optical fiber for highly sensitive dosimeter. MAHDIRAJI, G. A., ADIKAN, F. R. M. & BRADLEY, D. A. 2015a. Collapsed optical fiber: A novel method for improving thermoluminescence response of optical fiber. Journal of Luminescence, 161, 442-447. MAHDIRAJI, G. A., DERMOSESIAN, E., GHOMEISHI, M. & MAHAMD ADIKAN, F. R. 2016. Photonic Sensors: Glass Optical Fibers as Dosimeters. Reference Module in Materials Science and Materials Engineering. Elsevier. MAHDIRAJI, G. A., GHOMEISHI, M., DERMOSESIAN, E., HASHIM, S., UNG, N. M., ADIKAN, F. R. M. & BRADLEY, D. A. 2015b. Optical fiber based dosimeter sensor: Beyond TLD-100 limits. Sensors and Actuators A: Physical, 222, 48-57. MAHDIRAJI, G. M., DERMOSESIAN, E., SAFARI, M., MAHAMD ADIKAN, F. R. & BRADLEY, D. A. 2015c. Collapsed-hole Ge-doped Photonic Crystal Fiber as a Diagnostic Radiation Dosimeter. Lightwave Technology, Journal of, PP, 1-1.

MORADI, F., MAHDIRAJI, G. A., DERMOSESIAN, E., KHANDAKER, M. U., UNG, N. M., MAHAMD ADIKAN, F. R. & AMIN, Y. M. 2017a. Influence of dose history on thermoluminescence response of Ge-doped silica optical fibre dosimeters. Radiation Physics and Chemistry, 134, 62-70. MORADI, F., UNG, N. M., MAHDIRAJI, G. A., KHANDAKER, M. U., ENTEZAM, A., SEE, M. H., TAIB, N. A., AMIN, Y. M. & BRADLEY, D. A. 2017b. Angular dependence of optical fibre thermoluminescent dosimeters irradiated using kilo- and megavoltage X-rays. Radiation Physics and Chemistry, 135, 4-10. PAUL, M. C., BOHRA, D., DHAR, A., SEN, R., BHATNAGAR, P. K. & DASGUPTA, K. 2009. Radiation response behavior of high phosphorous doped step-index multimode optical fibers under low dose gamma irradiation. Journal of Non-Crystalline Solids, 355, 1496-1507. POLF, J. C., YUKIHARA, E. G., AKSELROD, M. S. & MCKEEVER, S. W. S. 2004. Real-time luminescence from Al2O3 fiber dosimeters. Radiation Measurements, 38, 227-240. SAKAGUCHI, S., ITOH, H., HANAWA, F. & KIMURA, T. 1985. Drawing-induced 1.53-μm wavelength optical loss in single-mode fibers drawn at high speeds. Applied Physics Letters, 47, 344-346. SANI, S. F. A., GRIME, G. W., PALITSIN, V., MAHDIRAJI, G. A., RASHID, H. A. A., MAAH, M. J. & BRADLEY, D. A. 2014. Micro-PIXE analysis of doped SiO2 fibers intended as TL dosimeters for radiation measurements. X-Ray Spectrometry, 1-8. WAHIB, N., ZULKEPELY, N. N., MAT NAWI, S. N., AMIN, Y. M., LING, Y. S., ABDUL SANI, S. F., MAAH, M. J. & BRADLEY, D. A. 2015. Gamma irradiated thermoluminescence response of Ge-doped SiO2 fibre. Applied Radiation and Isotopes, 105, 158-162. YAAKOB, N. H., WAGIRAN, H., HOSSAIN, M. I., RAMLI, A. T., BRADLEY, D. A., HASHIM, S. & ALI, H. 2011. Thermoluminescence Response of Ge- and Al-Doped Optical Fibers Subjected to Low-Dose Electron Irradiation. Journal of Nuclear Science and Technology, 48, 1115-1117. ZAHAIMI, N. A., OOI ABDULLAH, M. H. R., ZIN, H., ABDUL RAHMAN, A. L., HASHIM, S., SARIPAN, M. I., PAUL, M. C., BRADLEY, D. A. & ABDUL RAHMAN, A. T. 2014. Dopant concentration and thermoluminescence (TL) properties of tailor-made Ge-doped SiO2 fibres. Radiation Physics and Chemistry, 104, 297-301.

Highlights

Influence of optical fiber diameter on TL yield is demonstrated in detail Effect of different fiber sizes with similar and various core-to-cladding ratio is presented It is experimentally shown that the main TL signal generated from fiber core It is shown that by doubling fiber diameter, the TL yield will be quadruple before normalization