Open Access
Add to BibSonomyAbstract
We investigate the potential of microstructured optical fibers (MOFs) for highly sensitive absorption and fluorescence measurements by infiltrating a dye solution in the holey structure. Generally in a MOF only the evanescent part of the electromagnetic field penetrates into the sample material, providing a weak light-matter interaction. We compare such a MOF with a selectively filled hollow core photonic crystal fiber (HCPCF), in which most of the field energy propagates in the sample material. We show that dye concentrations down to 1×10-10 M can be detected in a HCPCF using only nanoliter sample volumes. Our experiments proof that HCPCFs are well suited for demanding sensing applications, outperforming existing fiber tools that rely on evanescent sensing.
©2007 Optical Society of America
1. Introduction
Microstructured optical fibers (MOFs) have received considerable attention as promising tools for a variety of applications. Due to their holey structure and the resulting specific optical properties, the use of MOFs has rapidly increased in the field of telecommunication [1], nonlinear optics [2] and, in particular chemical and biological sensing [3, 4]. The holey structure allows to infiltrate liquids or gases into the MOF which interact with the guided light. Usually, this interaction with the sample material is evanescent for MOFs, since the light is index-guided within the higher dielectric and only a small part of the field energy penetrates into the sample volume [3, 5, 6, 7, 8]. Among the various kinds of MOFs hollow core photonic crystal fibers (HCPCFs) [9, 10, 11] are especially promising for sensing applications due to the confinement of the light in the hollow core. This core is surrounded by smaller cladding holes and therefore the guiding mechanism is based on the photonic bandgap effect. By selectively injecting liquids into the hollow core [12, 13] the guiding mechanism is changed to index guiding within the liquid [14]. In this way a very strong interaction between light and matter can be achieved as most of the field energy propagates in the sample material.
There have been several reports on the application of MOFs as sensing devices for gases [15, 16] and liquids [14, 17, 18], applying either absorption [6, 7, 8] or fluorescence measurements [19, 20, 21, 22]. However, a comprehensive study for both types of measurements on these fibers is still missing, especially with regard to the achievable detection sensitivity. For the design of efficient sensing tools it is required to evaluate the sensitivity of solid core MOFs (SCMOFs), which rely on evanescent sensing, and HCPCFs in which almost the entire field intensity interacts with the sample material. Only then the question can be addressed whether the advantages of HCPCFs justify their application instead of the usually much more cost effective SCMOFs.
In this paper we present in detail absorption and fluorescence measurements on dye solutions in SCMOFs and HCPCFs using the same experimental setup, thus enabling a direct comparison of both types of fibers. The interaction coefficient of the light and the sample is determined experimentally and the achievable detection limit is evaluated for each fiber.
2. Fiber geometry and numerical simulation
For our experiments we use a silica SCMOF with a Ge doped, microstructured core (provided by the Institute for Photonic Technology (IPHT) Jena, Germany) as well as a commercially available silica HCPCF (Crystal Fibre HC-532-01), which are shown schematically in Fig. 1(a) and Fig. 1(b), respectively. The SCMOF has a pitch of Λ = 6.3 μm and the diameter of the cladding holes is d clad = 3.7 μm. The refractive index of the core, which has an outer diameter of 19 μm, is increased to n core = 1.46 by doping the silica (n si = 1.45) with Ge. Furthermore seven holes with reduced diameter d core = 1.6 μm are incorporated into the core to enhance the interaction with the sample material. Because of a higher air filling fraction in the cladding structure and a higher refractive index n core compared to n si the effective refractive index of the cladding n clad eff is smaller than that of the core n core eff. The light is therefore guided in the core due to total internal reflection (TIR). When a sample liquid is injected into the holes of the fiber this index guiding persists as long as n sample < n si. With regard to sensing applications the most important factor which determines the detection sensitivity is the interaction coefficient γ, i.e. the fraction of the electric field intensity that interacts with the sample material. To quantify g we calculate the field distribution of the guided modes using a finite element method. The simulated electric field intensity distribution (at wavelength λ = 530 nm) of the fundamental mode for the completely filled SCMOF is shown in Fig. 1(c). Over 99% of the electric field intensity is guided in the Ge doped silica due to its higher refractive index and only the small evanescent part of the field penetrates into the surrounding holes. Thereby the penetration depth, and consequently γ, increases with increasing n sample, as can be seen in Table 1. In our experiments we therefore use ethylene glycol with n sample = 1.43 as solvent for the infiltration. The interaction coefficient γ depends further on the wavelength λ of the light [14] and the number of guided modes which are excited. For the wavelength region around λ = 530 nm we estimate γ to be in the range of 0.3% to 1.0%.
Fig. 1. (color online) (a) Schematic representation of the completely filled SCMOF and (b) the selectively filled HCPCF. (c) Corresponding simulated electric field intensity distribution of the fundamental eigenmode at a wavelength of λ = 530 nm in the completely filled SCMOF and (d) in the selectively filled HCPCF. Ethylene glycol is assumed as sample material in both cases.
The HCPCF is designed for a central wavelength of 510 nm. The fiber parameters are Λ = 1.6 μm, d clad = 1.45 μm and d core = 5.3 μm. As the refractive index of the air filled hollow core is smaller than that of the cladding structure, which has an air filling fraction of about 90% and an effective refractive index n clad eff = 1.15, light can only be guided in the core in a limited wavelength range due to the photonic bandgap effect. By filling the entire holey structure of the HCPCF with a liquid (n sample < n si) the bandgap shifts to a different spectral region. Although it is possible to design a HCPCF that exhibits a bandgap in the desired wavelength range when completely filled with a liquid, this wavelength window will be relatively narrow and very sensitive to n sample. However, by selectively injecting the sample material into the hollow core n core eff becomes larger than n clad eff and the light is index guided in the selectively filled core [14]. Hence almost the entire field intensity interacts with the sample. This guiding mechanism works for the entire visible and near infrared wavelength region and is not limited to a specific solvent. In Fig. 1(d) the simulated electric field intensity distribution (at λ = 530 nm) of the fundamental mode for the selectively filled HCPCF is shown. Again, ethylene glycol is used as solvent. Contrary to the SCMOF the relative influence of n sample on the interaction coefficient is negligible (see Table 1) and γ yields approximately 92% to 99% for λ = 530 nm, depending on the different modes which are excited. It should be noted that above λ = 450 nm no index guided modes can exist in the cladding of the HCPCF due to the extremely small thickness (≈ 150 nm) of the web structure. This suppresses unwanted light propagation in the cladding, which would disturb any absorption measurements.
Table 1. Interaction coefficient γ for the fundamental mode λ = 530 nm in the completely filled SCMOF and the selectively filled HCPCF for different infiltrated solvents.
Table 2. Internal numerical aperture NA for the fundamental mode at λ = 530 nm in the completely filled SCMOF and the selectively filled HCPCF for different infiltrated solvents.
In absorption measurements the interaction coefficient γ directly determines the absorbance α=γLcε, where L is the fiber length, c is the concentration of the absorber in the solution, and ε is the molar extinction coefficient. According to Lambert-Beers law the incoming intensity I 0 and the transmitted intensity I are related by
Thus γ is an important measure to quantify the detection sensitivity of fiber based sensing devices. In fluorescence measurements emitted light is detected via guided modes of the fiber. In this case the interaction coefficient not only determines the absorption of the excitation light but also the coupling strength of the emitters to the guided modes. The detection efficiency of the fluorescence further depends on the acceptance angle of the radiation, which is limited by the critical angle of TIR. The latter can be described in terms of the internal numerical aperture
which is here a measure for the fraction of radiation that is emitted into guided modes. n core eff and n cladding eff are determined by calculating the effective refractive index of the fundamental core mode and the lowest (lossy) cladding mode, respectively. As seen in Table 2, NA depends on the solvent used for infiltration in both types of fibers, being considerably larger for the HCPCF (≈ 0.59 for ethylene glycol) than for the SCMOF (≈ 0.14 for ethylene glycol) due to the higher index contrast between core and cladding. Using ethylene glycol as solvent and assuming an isotropic emission this corresponds to a fraction of 9.6% and 0.5% of the radiated light which is collected, respectively. Thus, the detection sensitivity in fluorescence measurements can be optimized by using a selectively filled HCPCF and sample materials with a high refractive index, thereby achieving a simultaneous maximization of the interaction coefficient γ and the NA of the MOF.
Fig. 2. (color online) (a) Optical microscope image of the end facet of a HCPCF after the cladding holes have been sealed by a fusion splicer. The arrow indicates the point where only the hollow core is still open, enabling a selective infiltration of liquids. (b) Image of the white light transmission through the selectively filled HCPCF. Here, two adjacent cladding holes are filled in addition to the core, clearly demonstrating the index guiding. (c) Part of the experimental setup. R6G solved in ethylene glycol is selectively injected into the HCPCF and exited by an Ar+ laser at λ = 514 nm.
3. Experimental setup
In order to verify the theoretical predictions we perform absorption and fluorescence measurements on the SCMOF and the HCPCF which have been simulated. The fiber length L is approximately 10 cm in all measurements. The MOFs are filled with a solution of rhodamine 6G (R6G). As solvent ethylene glycol is used due to its high refractive index and the very low evaporation rate. For the selective infiltration of the central hole of the HCPCF a fusion splicer technique is used to selectively seal the cladding holes at both end facets [12, 23] while the central hole remains open (see Fig. 2(a)). This technique is based on the arc discharge between two electrodes. To avoid a collapse of the holes which shall remain open it is important that these holes are located in the center of the MOF and are larger than the surrounding cladding holes. Therefore this technique cannot be applied to selectively fill the core of the SCMOF as d core < d clad.
According to Poiseuille’s law it takes 53 min to infiltrate the central hole (d core = 5.3 μm) of a 10 cm long HCPCF with ethylene glycol by using only capillary forces. We observe that in this case the infiltrated solution contains a lot of microbubbles and the detection efficiency decreases due to strong scattering of the guided light. We therefore pump the R6G solution into the fibers using moderate pressure (≈ 2 bar) exerted by a syringe in which the needle is replaced by the fiber. In this way the occurrence of microbubbles is largely reduced and the filling time decreases significantly to about 10 min. The result of such an infiltration procedure is shown in Fig. 2(b). The applied technique allows us to refill a MOF by simply changing the sample material in the syringe. Thus it is possible to rinse a MOF with a pure solvent after one measurement and subsequently re-use it for further measurements. An investigation of the fluorescence emission of remaining emitters in a selectively filled HCPCF verified that a fiber filled with a dye solution of the initial concentration c = 10-6 M can be rinsed and only a dye concentration lower than 10-8 M remains in the fiber. We also compared fluorescence spectra of freshly filled MOFs with those of refilled ones and did not detect significant differences. Consequently the refilling technique enables a complete series of measurements on the same fiber as long as the desired detection limit is higher than c = 10-8 M. It may be further decreased by using more efficient chemicals (instead of pure solvent) for rinsing.
Figure 2(c) shows a part of the experimental setup which can be simultaneously used for absorption and fluorescence measurements. Light is coupled into the fiber and collected in forward direction by 20×/ 0.4NA objectives, spectrally dispersed by a spectrograph (Acton SpectraPro 2550i), and detected with a CCD camera (Roper Scientific), using integration times of 1 s. For absorption measurements a white LED is used as light source. In this case the MOFs are first filled with pure solvent for a reference transmission spectrum and are refilled afterwards with the dye solution. For fluorescence measurements the fluorophore is exited at 514 nm by an Ar+ laser with an excitation power of 2 μW.
4. Experimental results
In the following the experimental results for the absorption and fluorescence measurements are presented. From the acquired absorption spectra the interaction coefficient γ of the guided light with the sample material in the fiber can be determined by using Eq. (1). Since measurements on R6G solved in ethanol and ethylene glycol show very similar absorption behavior we assume that the molar extinction coefficient e is the same for ethylene glycol as for ethanol, which is well documented ε = 116000 L mol-1 cm-1 at λ = 530 nm. Once γ is known the achievable detection sensitivity for a specific type of MOF can be estimated. As an appropriate measure we introduce the minimum concentration
which can be detected. The constant [ln(I 0/I)]min depends on the accuracy of the measurement. With our setup we can reliably detect differences in transmission spectra of I 0/I = 1.05, which is limited by uncertainties in the reference spectrum. As a further measure for the detection sensitivity we use the minimum number of molecules N min = cV needed for acquiring a suitable spectrum. Here, the sample volume V is 58 nL for a completely filled 10 cm SCMOF and only 2.2 nL for a selectively filled HCPCF of the same length.
Fig. 3. (color online) (a) Absorption and fluorescence spectrum of R6G solved in ethylene glycol and completely infiltrated into the SCMOF (black curves). The dye concentration is 5×10-5 M and 5×10-6 M, respectively. (b) Corresponding absorption and fluorescence spectrum of the selectively infiltrated HCPCF (black curves) for concentrations of c = 5×10-7 M and c = 1×10-9 M, respectively. The red curves in (a) and (b) indicate free space measurements of R6G (c = 1.7×10-6 M). (c) and (d) Averaged absorption results for different concentrations of R6G in the completely filled SCMOF and the selectively infiltrated HCPCF, respectively. The slope of the graphs represents the interaction coefficient γ. (e) and (f) Averaged fluorescence intensity for different concentrations of R6G in the completely filled SCMOF and the selectively infiltrated HCPCF, respectively.
In Fig. 3(a) absorption and fluorescence spectra of the SCMOF are shown. Corresponding free space measurements are also displayed for comparison. The interaction coefficient γ is determined from the absorption spectra by plotting ln(I 0/I)/(Lε) over c and fitting a linear regression curve to the data. The result for several SCMOFs filled with different concentrations of R6G is presented in Fig. 3(c). Note that each data point represents the mean value of five measurements performed at the same concentration, as individual measurements can produce slightly differing results due to different coupling conditions. We believe that these uncertainties in the fiber coupling, which are most critical at short wavelengths, are also responsible for the observed discrepancies at the short wavelength side of the fiber absorption spectra (compared to the free space measurement, see Fig. 3(a) and Fig. 3(b)). Additionally, there might also be modifications of the spectra due to surface interaction of the molecules with the inner fiber walls. After fitting the data we obtain an interaction coefficient of γ = (1.08 ± 0.07)%, which is somewhat larger than the simulated range of values (0.3% to 1.0%, depending on the excited modes). This may be attributed to cladding modes, which can also interact with the sample due to the complete filling of the fiber and which have not been taken into account in the simulation. Furthermore, the dye molecules might get attached to the walls of the holes, thereby locally increasing the concentration where the evanescent field is largest. The lowest dye concentration we were able to detect in absorption is c (10 cm) min = 2 × 10-6 M.
The sensitivity of absorption measurements is mainly limited by the accuracy of the reference spectra, which depend on the exact coupling conditions of the fiber. On the contrary, in fluorescence measurements (see Fig. 3(a) and Fig. 3(b)) there is no need for any reference. We therefore expect an improvement in the detection sensitivity for fluorescence measurements, as they are only limited by the background signal from the fiber and the sensitivity of the detector. In the SCMOF we were able to acquire fluorescence spectra of dye concentrations down to 1 × 10-6 M (see Fig. 3(e)), which corresponds to N min = 3.5 × 1010 molecules in the entire fiber. Obviously, there is no significant improvement in the sensing performance of fluorescence measurements in comparison to absorption measurements for this type of fiber, which we attribute to the weak interaction of the electric field intensity with the sample material and the low collection efficiency due to the small NA of the fiber. It should be noted that the interaction coefficient γ of the SCMOF can be increased by reducing the size of the core and/or by increasing the d/Λ ratio of the core holes (but still keeping n core eff > n cladding eff) [8]. However, this will not improve the collection efficiency, since in general NA will decrease.
In Fig. 3(b) absorption and fluorescence spectra of the selectively filled HCPCF and corresponding free space measurements are shown. The interaction coefficient γ is determined in the same way as for the SCMOF (see Fig. 3(d)) and yields a value of γ = (86 ± 10)%, which is in good agreement with the numerical simulation (92% to 99%). The lowest dye concentration we were able to detect in absorption is c (10 cm) min = 5 × 10-8 M. The selectively filled HCPCF exhibits therefore a 40 times higher detection sensitivity in absorption measurements than the investigated SCMOF, which can mainly be attributed to the larger interaction coefficient γ.
Comparing these results to other evanescent sensing techniques, where a maximum interaction coefficient of γ = 5.2% [6] and γ = 6.5% [7] has been reported, the selective infiltration of a HCPCF enhances the detection sensitivity by about an order of magnitude, while at the same time the sample volume is considerably reduced. Studies on selectively coated HCPCF [19] also yield an improved sensing performance (with γ ≈ 11%), although not as high as reported here due to the spatial location of the sample in the weaker part of the electromagnetic field.
With regard to fluorescence measurements the impact of using a HCPCF on the achievable detection limit is even more pronounced than for absorption measurements. We were able to acquire fluorescence spectra of dye concentrations down to 1 × 10-10 M in a selectively filled HCPCF (see Fig. 3(f)), which corresponds to N min = 1.2 × 105 molecules in the fiber. Compared to the minimum concentration which was achieved with the SCMOF (1 × 10-6 M, see Fig. 3(e)) the detection sensitivity is thus enhanced by four orders of magnitude. This dramatic improvement can once more be attributed to the larger interaction coefficient, which means that nearly all emitters are excited by the laser light and the emission is coupled very efficiently to the guided modes of the HCPCF. Additionally, the collection efficiency for the fluorescence emission is about 19 times higher in the HCPCF than in the SCMOF due to the larger NA.
Currently the detection sensitivity is limited by the background signal from the fiber, which becomes significant for dye concentrations below 1 × 10-9 M. We were not able to identify the origin of this background (which also occurs in unprocessed HCPCFs), since pure silica should not exhibit such features at power levels as low as used here. Maybe some impurities are incorporated into the fiber during the fabrication process. However, by identifying and possibly eliminating the source of this background signal the detection limit might be pushed to even lower concentrations. Another promising approach to improve the detection limit relies on surface enhanced Raman scattering (SERS) inside HCPCFs, as has been demonstrated recently [24, 25]. The light-matter interaction cross section for SERS is higher than that for absorption/emission processes and would in principle allow a further boost of the detection efficiency. Most importantly, it can also be applied to non-fluorescent samples, opening a much broader field of applications.
5. Conclusions
In conclusion we have studied the performance of a SCMOF and a HCPCF as liquid sensing devices. By selectively filling the hollow core of the HCPCF with a dye solution an interaction coefficient of the guided light with the sample material of nearly 100% is achieved, while it yields a value of only 1% for the completely filled SCMOF. We have demonstrated that dye concentrations down to 5×10-8 M and 1×10-10 M can be detected with a HCPCF in absorption and fluorescence measurements, respectively. The required sample volume is only a few nanoliters (for a fiber length of 10 cm), which corresponds to emitter quantities of 110 amol and 0.2 amol, respectively. Our studies clearly show that selectively infiltrated HCPCFs outperform existing evanescent sensing devices. Especially with regard to fluorescence sensing the detection limit is improved by four orders of magnitude compared to the investigated SCMOF. We thus believe that, although SCMOFs are usually much more cost effective and easier to fabricate, the clear advantages of HCPCFs make them ideal tools for the most demanding sensing applications where high sensitivities are needed. The strong confinement of the sample material and the probe light into small volumes (typically few nanoliters) provides very large surface-to-volume ratios, which are highly advantageous for biochemical sensing. By functionalization of the inner core walls prior to the infiltration various chemical reactions or biological processes might be monitored very efficiently. Furthermore, due to the small sample volume needed for sensing, an integration into optofluidic devices can be managed [7].
Acknowledgments
We would like to thank Jens Kobelke from the IPHT Jena for providing the Ge doped SCMOF. This project was funded by the EFRE program of the European Union.
References and links
01. M.D. Nielsen et al., “Low-loss photonic crystal fibers for transmission systems and their dispersion properties,” Opt. Express 12, 1372–1376 (2004). [CrossRef] [PubMed]
02. V. V. Ravi Kanth Kumar et al., “Extruded soft glass photonic crystal fiber for ultrabroad supercontinuum generation,” Opt. Express 10, 1520–1525 (2002).
03. J. B. Jensen et al., “Selective detection of antibodies in microstructured polymer optical fibers,” Opt. Express 13, 5883–5889 (2005). [CrossRef] [PubMed]
04. T. M. Monro et al., “Sensing with microstructured optical fibres,” Meas. Sci. Technol. 12, 854–858 (2001). [CrossRef]
05. B. J. Eggleton et al., “Microstructured optical fiber devices,” Opt. Express 9, 698–713 (2001). [CrossRef] [PubMed]
06. J. B. Jensen et al., “Photonic crystal fiber based evanescent-wave sensor for detection of biomolecules in aqueous solutions,” Opt. Lett. 29, 1974–1976 (2004). [CrossRef] [PubMed]
07. L. Rindorf et al., “Towards biochips using microstructured optical fiber sensors,” Anal. Bioanal. Chem. 385, 1370–1375 (2006). [CrossRef] [PubMed]
08. C. M. B. Cordeiro et al., “Microstructured-core optical fibre for evanescent sensing applications,” Opt. Express 14, 13056–13066 (2006). [CrossRef] [PubMed]
09. R. F. Creganet et al., “Single-Mode Photonic Band Gap Guidance of Light in Air,” Science 285, 1537–1539 (1999). [CrossRef]
10. J. C. Knight, “Photonic crystal fibers,” Nature 424, 847–851 (2003). [CrossRef] [PubMed]
11. P. Russell, “Photonic crystal fibers,” Science 299, 358–362 (2003). [CrossRef] [PubMed]
12. S. Yiou et al., “Stimulated Raman scattering in an ethanol core microstructured optical fiber,” Opt. Express 13, 4786–4791 (2005). [CrossRef] [PubMed]
13. K. Nielsen et al., “Selective filling of photonic crystal fibres,” J. Opt. A: Pure Appl. Opt. 7, L13–L20 (2005). [CrossRef]
14. J. M. Fini, “Microstructure fibres for optical sensing in gases and liquids,” Meas. Sci. Technol. 15, 1120–1128 (2004). [CrossRef]
15. F. Benabid et al., “Compact, stable and efficient all-fibre gas cells using hollow-core photonic crystal fibres,” Nature 434, 488–491 (2005). [CrossRef] [PubMed]
16. T. Ritari et al., “Gas sensing using air-guiding photonic bandgap fibers,” Opt. Express 12, 4080–4087 (2004). [CrossRef] [PubMed]
17. M. Lelek et al., “High sensitivity autocorrelator based on a fluorescent liquid core fiber,” Appl. Phys. Lett. 89, 061117 (2006). [CrossRef]
18. L. Rindorf et al., “Photonic crystal fiber long-period gratings for biochemical sensing,” Opt. Express 14, 8224–8231 (2006). [CrossRef] [PubMed]
19. S. Smolka et al., “Selectively coated photonic crystal fiber for highly sensitive fluorescence detection,” Appl. Phys. Lett. 90, 111101 (2007). [CrossRef]
20. S. O. Konorov et al., “Photonic-crystal fiber as a multifunctional optical sensor and sample collector,” Opt. Express 13, 3454–3459 (2005). [CrossRef] [PubMed]
21. C. M. B. Cordeiro et al., “Lateral access to the holes of photonic crystal fibers - selective filling and sensing applications,” Opt. Express 14, 8403–8512 (2006). [CrossRef] [PubMed]
22. S. J. Myers et al., “Manipulation of spontaneous emission in a tapered photonic crystal fibre,” Opt. Express 14, 12439–12444 (2006). [CrossRef] [PubMed]
23. L. Xiao et al., “Fabrication of selective injection microstructured optical fibers with a conventional fusion splicer,” Opt. Express 13, 9014–9022 (2005). [CrossRef] [PubMed]
24. H. Yan et al., “Hollow core photonic crystal fiber surface-enhanced Raman probe,” Appl. Phys. Lett. 89, 204101 (2006). [CrossRef]
25. Y. Zhang et al., “Liquid core photonic crystal fiber sensor based on surface enhanced Raman scattering,” Appl. Phys. Lett. 90, 193504 (2007). [CrossRef]
