Open Access
Add to BibSonomy
Abstract
We investigate a novel light source design for efficient collection of UV light from multiple LEDs. The design is based on a truncated cone with a large circular lid incorporating LEDs on one side, and a small circular exit aperture (diameter of 9 mm) on the other side. The exit aperture size allows a simple coupling with secondary optics of a microscope for hyperspectral imaging that requires a wide spectrum of frequencies of illuminating light. The light collection efficiency was optimized with respect to cone length and diameter of the LED lid. In all simulations, we use a highly UV-reflecting aluminum coating on the inner surfaces of the cone. The influence of the LED positions on the cone efficiency is determined by varying the LED distance from the central axis as well as the interLED distance. We found the maximum efficiency of the cone is realized for LEDs positioned at the center, and the shorter is the inter-LED distance, the better is the performance of the light source.
© 2017 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
A LED based excitation light source, which could be portable and produce sufficient incoherent light output over a broad range of light frequencies (UV and visible), is a key requirement in hyperspectral imaging (HSI) [1, 2]. The current state-of-the-art hyperspectral system uses 12 excitation bands ranging from 334 to 495 nm from a mercury (Hg) lamp [1–3]. The desired excitation wavelength of light is extracted from the spectrum of a mercury (Hg) lamp by aligning an excitation filter, microscope objective and the optical fiber, making the light illuminate the sample at the microscope stage. The output power at various excitation bands cannot be changed in a flexible, modular way. Compared to Hg lamps, LED light sources are more cost-effective, have longer lifespan, are durable, and have stable power output over time [4–6]. Therefore, LEDs have been widely applied in medicine and industry [5] as illumination sources. Despite these advantages, the application of UV-LEDs in HSI is limited, due to their relatively low powers. Therefore, the parameters of the light source have to be judiciously chosen in such a way that they minimize the optical losses of low power LEDs (UV), traversing through optical path.
Here, we present a theoretical design for a simple, inexpensive solution for the HSI light source, designed for UV-Vis LEDs that can replace the Hg lamp used for epifluorescence illumination without any microscope modifications. The design is based on a truncated cone geometry with parameters optimized using TracePro software.
We optimize the parameters of the cone, its length (L) and lid diameter (D). The coating of inner surfaces is chosen to be aluminum (good reflection properties in UV-Vis) and we keep the diameter of exit aperture, d, fixed at 9mm in order for the system to be highly compatible with standard microscope illumination. The D/L ratio of the optical system is optimized to determine the maximum efficiency, and LED positions (P), LED tilt (T) and interLED distance (IL) are simulated to obtain the optimized number of LEDs (N) at the lid of an optimized cone.
2. Material and methods
The simulated truncated cone consists of five surfaces: (1) slanting surfaces, inner and outer, (2) lid surfaces, inner and outer, and (3) exit surface. The geometry of the truncated cone is shown in Fig. 1.
Fig. 1 (A). Schematics of a truncated cone light source. (B). Truncated cone light source combined with the secondary optics of Hg lamp in an inverted microscope.
The ray model of a commercial LED “OSRAM TOP LEDs LB T676” has been used in the simulations in the form of cubes with size of 2.5 x 2.5 x 2.5 mm3 because the LEDs provide emissivity for a broad range of wavelengths from UV to visible starting from 380 nm to 620 nm. The performance of this optical system has been assessed with respect to its efficiency (normalized flux), E, defined as a ratio of output flux α to input flux β:
Aluminum is chosen as coating of inner surfaces and the lid of the source. The selection is due to its high reflectivity in the UV range in addition to good heat dissipation properties [7, 8]. Light from the LEDs is represented in the simulations by light rays at 392 nm. This wavelength is selected because it lies in the middle of deep UV to visible spectrum. The results of our simulations are generally valid for the whole range of wavelengths explored here, because the reflectance of Al is adequately similar (above 90%) for UV-Vis wavelengths (250-550 nm). The variation in reflectance with angle is also consistent (above 90%) at majority of the angles, see Fig. 2. However, to observe the effect of incidence angle (0 to 90 degree), the reflectance of aluminum based mirror surface has been varied by using the Fresnel Eqs. In this way, this design accommodates different wavelengths (UV to visible range).
Aluminum (Al) coating with a reflectivity of 92% is used in simulations which is basically a standard mirror with a surface roughness modelled with ABg scatter. So, in this study, the cone design has a certain surface roughness on all inner sides of the cone. Being theoretical design, the simulations require only the reflectance property of Aluminum that features the concept of coating thickness. However, for practical reasons, the thickness of coating is designed, depending upon the wavelength of incident light.
The impact of surface roughness for Al based standard mirror coating appears in the form of absorbed flux and lost flux observed during simulation. Practically, Al coating surfaces corrode with passage of time. However, our theoretical design requires consistency in reflectivity of Al mirror at 92% to maintain the quality of results. Therefore, a corrosion protection layer, not affecting the quality of Al surface is required during the fabrication of cone to increase the life of optical system.
A detector having a radius of 4.5 mm, thickness of 1 mm has been simulated to be a perfect absorber with 100% absorbance, zero reflectance or transmission and is positioned 1 mm away from the exit of cone during simulations. This detector simulates the size of exit aperture for the source.
We optimize the lid diameter and cone length with respect to maximum efficiency. To this aim, the cone lengths between 50 to 250 mm are considered and the lid radius has been varied between 5 to 255 mm for each cone length.
After optimization of D/L, the optimized value is used to determine the optimized LED position (P), interLED distance (IL), and the angle of LED (T). The X and Y positions of the LEDs are varied, and optical efficiency is assessed for each position. Thus, the cone performance for variable LED position is recorded.
The InterLED distance of LEDs located at the cone’s lid also affects source performance; therefore, the interLED distance of 3, 4 and 5 mm is simulated for 13 LEDs using a total of 10,000 rays per LED. In this test, an odd number of LEDs are placed symmetrically on the lid of cone, as illustrated in Fig. 3. We note that in the practical implementations of this light source, the LEDs can also be placed on the lid in compact concentric rings.
To optimize the angle of the LED surface with respect to the lid, the emitter surface of 2.5 x 2.5 mm2 LED is rotated stepwise from 0 to 90° at different positions on the lid, and the relevant trends in the cone efficiency are recorded. A total of 250,000 rays are traced in each simulation. Overall, the optical efficiency is found to vary with D/L, position, interLED distance and the angle of LEDs.
3. Results and discussion
The diameter to length (D/L) ratio of the cone is found to have a prominent effect on the response of our optical system as shown in Fig. 4.
Fig. 4 Optical efficiency has been plotted against diameter/Length (D/L) ratio of truncated cone for different cone lengths (CL) ranging from 50 mm to 250 mm. The figure indicates that for D/L = 1, the shortest length of cone produces the highest optical efficiency.
Major efficiency variations are observed for different diameters and lengths of the truncated cone. The efficiency of the cone decreases for longer cone lengths, as most of the rays lose their energy while making multiple reflections inside the cone. For example, Fig. 4 shows that at D/L = 1, only 28% and 82% rays are detected at the output of 250 mm and 50 mm long cones respectively. The optimized performance of 33.42% has been obtained for the light source within the range of investigated parameters achieved for the cone length and lid diameter of 50 mm, at D/L = 1.
For all cone lengths, two optical efficiency maxima are observed one at D/L = 1 and second at D/L = 2, ranging D/L from 0.4 to 2.2. The overall optical efficiency decreases after D/L = 1 and then rises again towards D/L = 2 for all cone lengths. However, the efficiency is lower at D/L = 2 than that for D/L = 1. For D/L = 2, the diameter of cone becomes twice of the length of cone, which indicates that almost equivalent number of rays reach the detector at D/L = 1 and 2, but due to larger surface area in case of D/L = 2, a relatively larger component of rays has been absorbed in the lid or slanting side of the cone, Fig. 5.
Fig. 5 The flux components of two LEDs, separated by interLED distance of 5mm and 20mm have been shown. The bar chart indicates that the most of rays are incident on slanting sides of cone and cone lid. In particular, for two LEDs, large interLED distance (20mm) causes more loss of rays (presented in dark green) as compared to small interLED distance (5mm).
Interestingly, the minimum output of each cone length depends upon the size of lid, Fig. 4. For shorter cones, the lid size is also small and vice versa, causing less absorption of rays in the lid and giving relatively high throughput of an optical light. It can be seen in Fig. 4 that the cone lengths between 50 to 75 mm have the lowest efficiency values for lid radius of 10 mm, whereas for longer cone of 200 mm, the lid radius increases to 35mm, producing minimum efficiency.
From Fig. 4, a 60mm cone length has the efficiency of 28% that is 5% less than that of 50 mm cone length, but this provides an advantage of more space on cone’s lid for fixing a higher number of LEDs. Practically, if one round LED of any wavelength occupies a space of 15 mm (176mm2) on the top of cone’s lid of 60 mm diameter (2826 mm2), then 16 LEDs can be accommodated on the lid. The diameter higher than 60 mm reduces the optical efficiency by more than 10%. Therefore, 60mm is selected for further simulations of the truncated cone.
The majority of flux is absorbed or lost either in the inner lid or slanting surface of the cone as shown in Fig. 5. However, the flux varies with the change in interLED distance between two LEDs; for example, if interLED distance is increased from 5mm to 20mm for a given length of cone (L = 60 mm), then relatively more light is lost and absorbed in the latter case, reducing the output, Fig. 5. So, LEDs have to be located as close as possible to the central axis of the cone in order to reduce the absorption and loss of light under the same conditions of roughness of Al-mirror coating.
The optical efficiency profile for 60 mm cone length and 60 mm lid diameter as a function of the LED position is shown in Fig. 6. As expected, the maximum efficiency is found at the central axis of the cone, and the efficiency decreases abruptly as the LED moves towards the boundary of the lid. The highest efficiency, 28% is found at the center of lid and the lowest efficiency, 5%, is observed 22 mm away from the center. This implies that to maximize the overall performance of a source with multiple LEDs, they should be placed as close as possible to the center of the lid (subject to geometrical constraints) with the low power LEDs to be positioned closer to the central axis, while high power LEDs can be installed further away.
Fig. 6 Variation of source efficiency in a cone as a function of distance of single LED from the center. Cone thickness is 1mm, L = 60 mm, D/2 = 30 mm, standard model conditions (Al-mirror coating, 250,000 ray tracing, d = 9mm) are maintained in this study.
To determine the achievable number of LEDs, different numbers and arrangements of LEDs are fixed on the lid and the corresponding effect on the performance of LED light source has been quantified, Fig. 7. The results show that the optical efficiency decreases as the gap between LEDs increases, and as the number of LEDs increases on the cone lid, Fig. 7. The data of Fig. 7 also reinforce the results of Fig. 5 in that, with increasing interLED distance, more rays are absorbed by the cone, reducing optical power at the exit.
Fig. 7 The LEDs number and interLED distance affect efficiency of cone. The efficiency decreases with increase in the number of LEDs and larger gaps between LEDs. Under standard model conditions (Al-mirror coating, d = 9mm), 10000 rays per LED are traced for 13 LEDs, cone thickness is 1mm, L = 60 mm, D/2 = 30 mm.
In our design, all LEDs are installed at the lid. From Fig. 6, the LEDs mounted within the 14 mm radius yield 7% efficiency of the source, and the area of the lid is sufficient to accommodate 14 LEDs each of 6.5mm side (42 mm2). It implies that reducing the size (and gap) per LED can increase the number of LEDs at the lid of cone.
The irradiance profile of the 392nm LED (1W flux, 250000 rays) at the exit of optimized cone varies with the position of LED on the lid of truncated cone. In Fig. 8(a), the irradiance is the highest, 20000 W/m2 at the center, and it falls steeply downwards to 1000 W/m2 within 4.5 mm on either side from the central axis. Figures 8(b), 8(d) show the change in output profile when LEDs are fixed on the opposite sides of X-axis at a distance of 20 mm. Similarly, Fig. 8(c) and 8(e) show output profiles on Y-axis for the LED at 20mm away from the center. The position of LEDs on the cone’s lid determines the irradiance profile.
Fig. 8 The irradiance profile of optical efficiency has been found shifting with change in the position of illuminating LEDs at the lid of an optimized cone, L = 60 mm, D/2 = 30 mm, Al mirror coating, 250,000 ray traces, d = 9mm, cone thickness = 1mm, 1 LED is positioned at the center (A) and 20mm off axis (-x, -y, + x, + y axes in figures B, C, D, E respectively).
Finally, the simulated output of LEDs is found to decrease with increase in LED angle; however, with increasing distance from the central axis, the efficiency decreases. The emitter surface in the LED (the light source) when rotated at an angle from 0 to 50° does not cause appreciable variation in the output of cone for any LED distance from the central axis of cone (5mm to 25mm). For example, the efficiency of the source remains the highest at 18% at small distance (5mm) from central axis. By contrast, the output remains always the lowest (around 5%) for large distance (25mm). The efficiency drops abruptly for angles in excess of 55 degree for a LED on the lid of optimized cone, and this effect has been observed to be more prominent at positions closer to central axis, for example at 5mm radius, Fig. 9. This allows us to conclude that to obtain high efficiency, low powered LEDs should be installed near the center of lid so that maximum power can be delivered straight through the cone, whereas high power LEDs can be installed further away from the center in our design.
Fig. 9 The optical output of the cone shows variation with the increase in tilt and distance of LED with respect to the central axis of truncated cone. For smaller LED tilts (0 to 50°) and distances from central axis (5mm), the optical output is more. Whereas, for the LED tilt higher than 50 degree the output starts decreasing for the same position of LED.
There are a large number of drawbacks associated with mercury, xenon and laser light sources which include heat production, limited lifespan, ozone formulation, excitation and emission filters and difficulty in modulation. Therefore, such light sources can be replaced with an LED based light source designed in our study, but the constraints and limitations of the design presented above need to be considered [4, 6]. These multiple LEDs in hyperspectral imaging lamps require highly reflective (aluminum) coating for the best optical performance from deep UV to IR range [7, 8]. Every LED provides its characteristic illumination which needs to be optimized at the output plane either keeping illumination pattern [9] or measuring optical efficiency as done in our study by optimizing the interLED distance and the number of LEDs.
The corrosion occurs due to oxidation of pure Aluminum. Commercially, UV enhanced Aluminum and protected Aluminum are available with the reflectivities of 90% and 95% respectively. They are respectively covered with MgF2 and SiO2 to avoid corrosion. The protective layer is transparent, and does not affect the reflectance of Aluminum layer adversely. Therefore, the effect of protective layer (MgF2 and SiO2) has not been included in our theoretical design. As the reflectivity of Al-based standard mirror set in our simulations is 92% therefore, any deterioration in reflectance of Al coating will lead to change the simulation results.
When various LEDs are combined then wavelength mixing, and illumination uniformity become challenging. In a previously published study, red, blue and green LEDs are illuminated in a light pipe with a diffuser, and it is found that the optimal length-width ratio 0.8<L/D<1.2 achieves a good optical efficiency and wavelength mixing. The ratio appears to be consistent with the optical output of our light source design [10, 11]. The non-uniform optical output originated from LEDs in the cone can be made uniform and more intense with the help of light pipe, lens if this system is to use as a standalone system [12, 13].
4. Conclusion
We optimized a truncated cone LED light source with a fixed size exit aperture with Al coating, and we found that it provides the highest optical efficiency for D/L = 1. The optical output of cone depends upon the position and number of LEDs, interLED distance, and the angle of LEDs installed at the lid of cone. The optical efficiency of 28.35% is obtained for an optimized cone having length and diameter each of 60mm. It has been found in this study that LEDs should be located close to each other over the cone’s lid and their tilt larger than 50 degree reduces the optical efficiency of cone. We determine that the irradiance profile of cone varies as a function of the LED position. The LED cone light source designed here can be used as a replacement for an Hg lamp or as a standalone excitation light source in hyperspectral imaging systems and other applications, where multiple wavelengths of light ranging from UV to visible are required.
Funding
Australian Research Council Center of Excellence Centre for Nanoscale Biophotonics (CE140100003); ARC DECRA fellowship (DE170100241); International Macquarie University Research Excellence Scholarship (iMQRES).
Acknowledgments
We acknowledge the contribution of Lambda Research Corporation for the provision of Tracepro software to support this research. We also acknowledge the support of Martin E. Gosnell, and Ayad G. Anwer, Department of Physics and Astronomy, Macquarie University, North Ryde, Sydney, 2109, NSW, Australia.
Disclosures
The authors declare that there are no conflicts of interest related to this article.
References and links
1. M. E. Gosnell, A. G. Anwer, J. C. Cassano, C. M. Sue, and E. M. Goldys, “Functional hyperspectral imaging captures subtle details of cell metabolism in olfactory neurosphere cells, disease-specific models of neurodegenerative disorders,” Biochimica et Biophysica Acta (BBA) - Molecular Cell Research 1863(1), 56–63 (2016).
2. M. E. Gosnell, A. G. Anwer, S. B. Mahbub, S. Menon Perinchery, D. W. Inglis, P. P. Adhikary, J. A. Jazayeri, M. A. Cahill, S. Saad, C. A. Pollock, M. L. Sutton-McDowall, J. G. Thompson, and E. M. Goldys, “Quantitative non-invasive cell characterisation and discrimination based on multispectral autofluorescence features,” 6 (1), 1–11 (2016).
3. T. E. Renkoski, B. Banerjee, L. R. Graves, N. S. Rial, S. A. H. Reid, V. L. Tsikitis, V. N. Nfonsam, P. Tiwari, H. Gavini, and U. Utzinger, “Ratio images and ultraviolet C excitation in autofluorescence imaging of neoplasms of the human colon,” J. Biomed. Opt. 18(1), 16005 (2013). [PubMed]
4. R. W. Cole and J. N. Turner, “Light-Emitting Diodes Are Better Illumination Sources for Biological Microscopy than Conventional Sources,” Microsc. Microanal. 14(3), 243–250 (2008). [PubMed]
5. B. Hohman, “LED light source: Major advance in fluorescence microscopy,” Biomed. Instrum. Technol. 41(6), 461–464 (2007). [PubMed]
6. I. Young, Y. Garini, H. Dietrich, W. van Oel, and G. Lung, “LEDs for fluorescence microscopy,” Proc. SPIE 5324, 208–215 (2004).
7. K. Balasubramanian, J. Hennessy, N. Raouf, S. Nikzad, M. Ayala, S. Shaklan, P. Scowen, J. Del Hoyo and M. Quijada, “Aluminum mirror coatings for UVOIR telescope optics including the far UV,”Proc. SPIE 9602, 96020I96021–96029 (2015).
8. I. Lindseth, A. Bardal, and R. Spooren, “Reflectance measurements of aluminium surfaces using integrating spheres,” Opt. Lasers Eng. 32(5), 419–435 (1999).
9. I. Moreno, M. Avendaño-Alejo, and R. I. Tzonchev, “Designing light-emitting diode arrays for uniform near-field irradiance,” Appl. Opt. 45(10), 2265–2272 (2006). [PubMed]
10. K. K. Li, S. Inatsugu, and S. Sillyman, “Design and optimization of tapered light pipes,” Proc. SPIE 5529, 48–57 (2004).
11. C.-C. Sun, I. Moreno, Y.-C. Lo, B.-C. Chiu, and W.-T. Chien, “Collimating lamp with well color mixing of red/green/blue LEDs,” Opt. Express 20(1S1), A75–A84 (2012). [PubMed]
12. F. J. Bolton, A. Bernat, S. L. Jacques and D. Levitz, “Development and testing of a homogenous multi-wavelength LED light source,”Proc. SPIE 10055, 100550R (2017).
13. C. Chen and X. Zhang, “Design of optical system for collimating the light of an LED uniformly,” J. Opt. Soc. Am. A 31(5), 1118–1125 (2014). [PubMed]
