An evanescent field sensor utilizing layered polymeric-inorganic composite waveguide configuration was developed in this work. The composite waveguide structure consists of a UV-imprint patterned polymer inverted rib waveguide with a Ta2O5 thin film sputter-deposited on top of the low refractive index polymer layers. The results suggest that the polymer based sensor can achieve a detection limit of 3 × 10−7 RIU for refractive index sensing and corresponding limit of about 100 fg/mm2 for molecular adsorption detection. Besides enhancing the sensitivity significantly, the inorganic coating on the polymer layer was found to block water absorption effectively into the waveguide resulting in a stabilized sensor operation. The ability to use the developed sensor in specific molecular detection was confirmed by investigating antibody – antigen binding reactions. The results of this work demonstrate that high performance sensing capability can be obtained with the developed composite waveguide sensor.
© 2012 OSA
Optical planar waveguide sensors utilizing evanescent field probing scheme have drawn great interests in biosensing due to its high sensitivity and integration rate for label-free multi-analyte detection. Typical planar waveguide sensors are fabricated with inorganic dielectrics, such as, SOI , Si3N4 , III-V group alloys , and Ta2O5  with claddings of SiO2. To reduce sensor manufacture costs alternative materials and sensor configurations suitable for low-cost fabrication process have been investigated. Polymers possessing good optical properties, versatile process-abilities including molding and spin coating techniques can therefore be compatible for low-cost production of optical sensors. Several polymer-based planar waveguide biosensors have been developed recently, such as, ring resonators [5, 6], interferometers [7, 8], and grating couplers [9, 10]. Compared to inorganic materials, the refractive index (RI) of polymers differs only a little leading to a low RI contrast between the core and cladding in a waveguide. Consequently, this limits available configurations of polymer planar waveguide sensors. For example, field localization and high quality resonators are generally difficult to realize. According to our earlier studies on layered inorganic – polymeric composite waveguides , the optical field distribution can be manipulated by depositing a thin layer of Ta2O5 on top of polymer layers. With the help of the high index (n = 2.1) thin film, the guided mode field can be pushed up towards the surface resulting in an enhanced optical field probing at the waveguide surface, where binding interactions of biomolecules take place.
The motivation of this work was to develop a low cost biosensor yet having sensitivity comparable to inorganic based integrated optical (IO) biosensors. With the proposed inverted rib waveguide structure  a simple process consisting of room-temperature UV-imprinting, spin-coating, and sputtering of Ta2O5 high index layer is utilized in the sensor fabrication. The influence of water absorption into the polymer is usually neglected in reports on polymer-based waveguide sensors, though the phenomenon of water molecular penetration and absorption exists widely in polymeric materials . In this paper, the effect of water penetration in polymer waveguides is examined and the water blocking effect of the thin Ta2O5 layer is investigated. Finally, the behavior of the developed polymeric-inorganic waveguide sensors in bio-measurements is evaluated by investigating the specific binding of CRP (c-reactive protein) antibodies and antigens.
2. Waveguide sensor fabrication and characterization
Interferometers are considered as one of the most sensitive devices  for optical sensing due to the background compensation between the sensing and reference arm against environmental disturbances, such as temperature fluctuations. Figure 1(a) shows a schematic image of the Young interferometer (YI) biosensor configuration utilized in this work. The Young interferometer was chosen as the transducer platform due to its linear correlation between the spatial change of the interference pattern and the induced phase change in the waveguide .
The materials used for sensor fabrication consist of the UV-curable Ormocer series of hybrid polymers (Micro resist technology GmbH). Ormocomp, Ormocore and Ormoclad having refractive index of 1.520, 1.553 and 1.536 at wavelength of 633 nm are used for the waveguide’s under cladding, core and upper cladding respectively. The details of the waveguide fabrication are described in Ref. 12. Fabrication of the polymer inverted-rib waveguide is an etch-free process. Except the final deposition of the high index layer, the whole fabrication process incorporates polymeric materials only.
The waveguide structure was built on a silicon wafer offering mechanical support. Fabrication of the polymer waveguide sensor consists of two main steps: patterning the waveguide under-cladding and filling the patterned grooves with the core material. UV- imprinting used for the rib/groove patterning does not require high pressure during the imprinting process and it is performed at room temperature. When the core material diluted in 1-methoxy-2-propyl acetate is spin coated with a speed of 3000 rpm, a residual layer having a thickness around 400 nm is left above the patterned under cladding. A cross sectional SEM image of the fabricated inverted rib waveguide located inside the sensing window is shown in Fig. 1(b). The groove imprinted in Ormocomp under cladding is filled with Ormocore core material. Ta2O5 coating can be seen as a brighter thin layer on top of the polymer layers. The shown waveguide has a dimension of about 2 µm (width) × 1.6 µm (height) including the slab residual core layer. To create a sensing channel on the other waveguide arm, a thick layer (about 20 µm) of Ormoclad was spin coated above the waveguide structure and a sensing window with a length of 10 mm was formed photo-lithographically. For the under cladding layer Ormocomp material was chosen because of its better compatibility with the diluted Ormocore core layer. For the upper cladding Ormoclad was used due to lower refractive index contrast with the core material resulting in a single-mode operation. Finally, a thin layer of Ta2O5 was reactively sputtered using a 4N pure Ta metal target (Helios XL Pro, Leybold Optics) on top of the fabricated polymer waveguide.
2.2 Waveguide and sensor characterization
In the fabricated integrated YI, the distance between the sensing and referencing waveguide is 50 µm. Part of the YI sensor is shown in the top view microscope image in Fig. 2(a) . According to the simulations (Fimmwave, Photon Design Inc.) at 633 nm, the sensing waveguide supports only one TM propagation mode in the sensing window, where DI water serves as its upper cladding with the refractive index of 1.33. The waveguide also operates in single mode in the areas with Ormoclad over cladding layer. The coupling efficiency Γ2 between the two regions in the sensing waveguide is defined as an overlap integral  according toFig. 2(a) at the indicated locations in the sensing waveguide.
By using the shadow masking technique in the Ta2O5 sputtering process, an adiabatic region with gradually increasing Ta2O5 thickness was formed between the Ormoclad covered region and the Ta2O5 coated region. Without the adiabatic transition the coupling efficiency between the two regions was calculated to be 60%. With the adiabatic transition starting from 20 nm of Ta2O5 coating at the front edge of the sensing window, the obtained coupling efficiency was approximately 90%. Assuming that the power loss occurs at the front and the back edge of the sensing window, the overall transmission efficiency was estimated to be around 81%. As evaluated in , the intrinsic material loss of a sputtered Ta2O5 thin film is about 1 dB/cm. The propagation loss in a composite polymeric waveguide with 100 nm thick Ta2O5 layer was estimated to be about 1.6 dB/cm . These attenuation values are small enough to have a good interference fringe visibility at the output of the sensor chip facet. The generated fringe pattern from the referencing waveguide and sensing waveguide with adiabatic transitions is shown in Fig. 2(b). The clear visibility of the interferogram gives a good resolution when analyzing the fringe pattern phase shifts by two-dimensional fast Fourier transform.
2.3 Theoretical sensitivity evaluation
The IO waveguide biosensors respond to the change of the bulk refractive index of the liquid sample and the surface adsorption of biomolecules. The bulk RI change of the liquid sample is seen as a change in the refractive index of the upper cladding nc. The adsorption of the biomolecules is translated as a thickness change of the adlayer tad growing above the sensing waveguide. The induced effective RI change Δneff of the sensing waveguide can be described by Eq. (2)Fig. 1(b), the theoretical sensitivities of ∂neff/∂nc and ∂neff/∂tad were examined using FMM method. For the calculation of ∂neff/∂tad, nad = 1.45 was used for the RI of adsorbed biomolecule layer. With increasing coating thickness, the optical field distribution of the propagating mode is pushed up towards the waveguide surface resulting in improved sensitivity. As can be seen in Fig. 3(a) and 3(b), the curves for homogeneous refractive index sensing and surface adsorption sensing resemble each other though there is a slight Ta2O5 thickness difference for the maximum sensitivity. For TE-polarization state, sensitivity increases with increasing Ta2O5 thickness upto about 60 nm for homogenous refractive index sensing and about 70 nm for adsorption sensing after which the further increase in high-index layer thickness does not improve sensitivity. Corresponding thicknesses for TM-polarization state are 120 nm and 140 nm, respectively.
The enhancement in sensitivity is associated with the decreased confinement inside the inverted guiding ridge. Therefore, the composite waveguide structure becomes leaky in the lateral direction, when a much thicker high-index coating is deposited on the polymer layers. To optimize the sensitivity and mode confinement in the sensing region, 80 nm coating and TM-polarization was chosen. According to the simulations the obtained sensitivity of 80 nm Ta2O5 coating is 0.12 and 2.1 × 10−4 nm−1 for homogeneous RI and surface molecular adsorption sensing respectively. The sensitivities are enhanced over 40 times when comparing to the polymer waveguide sensor without high index coating while still having 17% power confinement inside the inverted ridge resulting in proper waveguide operation.
3. Experiment and results
A He-Ne (632.8 nm) laser was used as a light source in the measurements. TM-polarization state was adjusted by rotating a polarizer in front of the laser. Microscope objectives were used for coupling light into and out from the sensing chip. The liquid analytes were pumped through a flow cell attached above the sensing element by a continuous flow syringe pump (OPAM Instruments). A CMOS camera (PixeLink) with a resolution of 1280 × 1024 recorded fringe pattern images at a rate of 1 Hz. For each image two dimensional fast Fourier transform was applied to extract the phase from the fringe pattern.
3.1 Effect of water absorption
When subjected to water absorption the refractive index and volume of a polymer may change . To examine the water absorption in Ormocer waveguide two YI waveguide sensors, one with and the other without Ta2O5 coating, were used. The phase responses when injecting water into the sensing window are shown in Fig. 4(a) and 4(b).
In the beginning, when water was just introduced into the sensing window, an abrupt change in the phase response was triggered by an immediate change of the refractive index from the value of air (n = 1) to the value of water (n = 1.33). The swift change resulted in big spikes where a gray area is masked over the sensorgram. Inside these regions the phase responses were not able to be defined explicitly. At the same time, water started to be absorbed in the polymer waveguide sensor without Ta2O5 coating resulting in a continuous change in the phase response as shown in Fig. 4(a). The phase converged gradually when the diffusion of the water molecules reached equilibrium. When the polymer waveguides were coated with Ta2O5, the phase response stabilized in a couple of minutes right after the water completely filled the sensing window as shown in Fig. 4(b). These results indicate that 80 nm Ta2O5 coating can effectively block the penetration of water into the underneath polymer waveguide. The refractive index change of the used polymers can be estimated by assuming that the water absorption into the polymer waveguide dominates the refractive index change and the amount of change is the same in the core and cladding. Figure 4(c) shows the calculated phase change Δφ when the refractive index of the core and cladding was varied between 10−6 and 10−3 RIU. The red dot indicates Δφ of 14 rad corresponding to the RI change of 1.4 × 10−4 in the core and cladding. Taking into account the ambiguity in the measurement transient as illustrated with the grey area in Fig. 4(a), the actual RI change in the waveguide core and cladding could be even larger than 10−4. The obtained value of refractive index change of Ormocer is comparable to that of the silicone resin and epoxy resin 10−4  and possibly smaller than the polymide  and PMMA 10−3  at 100% relative humidity.
3.2 Homogeneous sensing
To evaluate the response against the change of the bulk refractive index, DI-water with varying glucose concentrations (0.005%, 0.02%, 0.04%, 0.05%, 0.1%, 0.2%, 0.6%, 1% w/v) was applied to the sensing surface. The refractive index change of glucose solution with respect to DI-water was calculated based on the empirical equation Fig. 5 . The effective RI change of the sensing waveguide can be extracted from the phase shift of the interference pattern according to the equation ∆neff = ∆φ∙λ/2πL, where ∆φ is the phase shift of the fringe pattern, λ is the wavelength in vacuum and L is the interaction length.
Figure 6 shows the obtained ∆neff values versus ∆nc for all eight glucose solutions. The slope of the linear fit indicates that the actual sensitivity ∂neff/∂nc equals to 0.11, which is in a good agreement with the simulated sensitivity of 0.12. The noise level, defined as the standard deviation of the baseline, was in the order of 0.003 rad for a 5 minutes measurement. With this value the potential detection limit can be derived to be about 3 × 10−7 RIU for the bulk index changes.
Compared to some of the polymer based planar waveguide sensors reported previously [5, 6, 8 ,9] the composite waveguide polymeric-inorganic sensor has a detection limit of one order of magnitude better. The obtained detection limit is also comparable or better than some of the inorganic dielectric based biosensors reported recently where the detection limit lies in the range of 10−5~10−7 RIU [1, 2, 22–24] for homogeneous bulk refractive index sensing.
3.3 Surface sensing of protein adsorption
The performance of the developed sensor in biosensing applications was examined by detecting specific molecular binding events between c-reactive protein (CRP antigens) and monoclonal anti-human C-reactive protein (CRP antibodies). Their phase responses are plotted in Fig. 7(a) and corresponding biomolecule bindings are illustrated in Fig. 7(b).
Phosphate buffered saline (PBS, pH 7.4) solution was used as the buffer solution throughout the surface sensing experiments. First, the sensing window was filled with PBS buffer after which 0.5 mg/ml of CPR antibody was pumped continuously through the sensing window with the speed of 16.7 µl/min for 1 hour. CRP antibodies flowing close to the sensing surface were adsorbed nonspecifically to the waveguide surface. Subsequent to a PBS wash, 2 mg/ml bovine serum albumin (BSA) solution was used to block the sensing surface. When the surface was prepared, CRP antigen was pumped through the sensing window. Two different concentrations 2 µg/ml (16 nM) and 0.1 µg/ ml (0.8 nM) were used, each on a new sensor chip. In step 1 the slightly different phase responses during the attachment of the primary CRP antibodies might be due to differential adsorption of the biomolecules on the bare polymer waveguide surface. However, after BSA surface blocking procedure (step 3), the phase change of the two different measurements showed rather similar values. As shown in the zoomed-in curves of steps 5 and 6 in Fig. 7(a), the binding events caused a phase change after PBS wash of about 8.2 rad and 2.3 rad for 2 µg/ml and 0.1 µg/ ml CRP concentrations, respectively. The phase shifts might be partially suppressed due to a random orientation of nonspecifically adsorbed primary antibodies which prevents binding of the antigen to all antibodies on the surface. Using the theoretical sensitivity of ∂neff/∂tad = 2.1 × 10−4 nm−1, the growth of the adlayer was determined to be 0.44 nm and 0.12 nm corresponding to the phase changes. The obtained layer thickness does not indicate an exact molecular height but an average thickness taking into account the molecular packing density. The corresponding surface mass density M can be calculated with De Feijter’s formula25]. The calculated surface mass density is 280 pg/mm2 for 2µg/ml and 80 pg/mm2 for 0.1 µg/ml CRP antigens. Taking into account of the noise level of 0.003 rad, the achieved detection limit is estimated to be about 100 fg/mm2. The obtained measurement results are much less than a monolayer surface mass coverage of 1.48 ng/mm2 . It could be partially due to a random orientation of the CRP antibodies meaning that only part of the CRP antibodies has their binding sites facing out from the waveguide and are available to bind CRP antigens. Moreover, as shown in Fig. 7(a) with the blue and red curves, the signal levels were different for the two sensor chips under investigation after step 1 and step 2. These results indicate that the CRP antibodies in step 1 did not form a dense monolayer leading to a lower surface mass density of the bound CRP antigens. According to previous studies [27, 28], it is not easy to achieve a single but dense monolayer. The surface densities of biomolecules can vary in a broad range being strongly dependent on the surface and the surface modification protocol.
To assure the induced phase responses were not due to nonspecific binding of CRP on the waveguide surface, we performed a negative control test using mouse IgG instead of CRP-antibodies as a receptor. Due to a low binding affinity between the mouse IgG and CRP, no obvious phase response was detected (black curve in Fig. 7(a)). The specific biomolecular interaction was further confirmed by running the secondary CRP antibodies labeled with fluorescence markers through the sensing window. As seen in step 7 of Fig. 7(a), attachment of secondary antibody to primary antibody-antigen complex caused a big phase response (red line) whereas in the negative control experiment (black line) only a very small phase change was detected, possibly representing slight nonspecific binding. A CRP molecule is composed of five identical subunits forming an annular configuration with cyclic pentameric symmetry . Each subunit contains an epitope which is specifically recognized by the antibody. Thus a single CRP molecule can be bound by five different antibody molecules. When a CRP molecule was bound by a CRP antibody on the sensing surface, four available epitopes are left unoccupied. When applying a high concentration (13.2 µg/ml) of secondary antibodies, all the available epitopes were expected to become occupied producing a corresponding about four times increase in phase response as CRP antigens (115 kDa) and CRP antibodies (150 kDa) have a comparable molecular weight. Fluorescence microscopy images in Fig. 7(a) insert show attachment of fluorescent labeled secondary antibodies onto the CRP antibody activated surface but not onto the negative control.
In this paper a highly sensitive polymeric-inorganic waveguide sensor was developed. The emphasis was to demonstrate that high-performance sensor configuration is attainable with simplified and potentially low-cost fabrication methods. By depositing 80 nm Ta2O5 coating on top of the UV-imprinted polymer YI sensor the sensitivity was improved over 40 times. It was also demonstrated that the deposited thin inorganic layer prevented the water penetration into the polymer waveguides and therefore the sensing response was quickly stabilized after exposure to the aqueous analyte. The fabricated sensor was calibrated with glucose DI-solutions. The actual sensitivity 0.11 was found to have a good correspondence with the theoretical value. An estimated detection limit was about 3 × 10−7 RIU for homogeneous ambient sensing and 100 fg/mm2 for surface mass adsorption detection. The biosensing applicability was successfully demonstrated with the immunoassay of CRP.
This work was supported by Infotech Oulu Graduate School and Academy of Finland (Grant No.133814).
References and links
1. A. Densmore, D. X. Xu, S. Janz, P. Waldron, T. Mischki, G. Lopinski, A. Delâge, J. Lapointe, P. Cheben, B. Lamontagne, and J. H. Schmid, “Spiral-path high-sensitivity silicon photonic wire molecular sensor with temperature-independent response,” Opt. Lett. 33(6), 596–598 (2008). [CrossRef]
2. A. Ymeti, J. S. Kanger, R. Wijn, P. V. Lambeck, and J. Greve, “Development of a multichannel integrated interferometer immunosensor,” Sens. Actuators B Chem. 83(1-3), 1–7 (2002). [CrossRef]
3. D. R. Cassidy and G. H. Cross, “Picometer resolution wavelength tracking in the C -band using an InP–InGaAsP dual-slab interferometer,” IEEE Photon. Technol. Lett. 19(14), 1075–1077 (2007). [CrossRef]
4. K. Schmitt, B. Schirmer, C. Hoffmann, A. Brandenburg, and P. Meyrueis, “Interferometric biosensor based on planar optical waveguide sensor chips for label-free detection of surface bound bioreactions,” Biosens. Bioelectron. 22(11), 2591–2597 (2007). [CrossRef]
5. C.-Y. Chao, W. Fung, and L. J. Guo, “Polymer microring resonators for biochemical sensing applications,” IEEE J. Sel. Top. Quantum Electron. 12(1), 134–142 (2006). [CrossRef]
6. G.-D. Kim, G.-S. Son, H.-S. Lee, K.-D. Kim, and S.-S. Lee, “Integrated photonic glucose biosensor using a vertically coupled microring resonator in polymers,” Opt. Commun. 281(18), 4644–4647 (2008). [CrossRef]
7. R. Bruck, E. Melnik, P. Muellner, R. Hainberger, and M. Lämmerhofer, “Integrated polymer-based Mach-Zehnder interferometer label-free streptavidin biosensor compatible with injection molding,” Biosens. Bioelectron. 26(9), 3832–3837 (2011). [CrossRef]
8. M. Wang, S. Uusitalo, C. Liedert, J. Hiltunen, L. Hakalahti, and R. Myllylä, “Polymeric dual-slab waveguide interferometer for biochemical sensing applications,” Appl. Opt. 51(12), 1886–1893 (2012). [CrossRef]
9. M.-S. Kwon and S.-Y. Shin, “Refractive index sensitivity measurement of a long-period waveguide grating,” IEEE Photon. Technol. Lett. 17(9), 1923–1925 (2005). [CrossRef]
10. J.-W. Kim, K. J. Kim, J. A. Yi, and M.-C. Oh, “Polymer waveguide label-free biosensors with enhanced sensitivity by incorporating low-refractive-index polymers,” IEEE J. Sel. Top. Quantum Electron. 16(4), 973–980 (2010).
11. J. Hiltunen, S. Uusitalo, P. Karioja, S. Pearce, M. Charlton, M. Wang, J. Puustinen, and J. Lappalainen, “Manipulation of optical field distribution in layered composite polymeric-inorganic waveguides,” Appl. Phys. Lett. 98(11), 111113 (2011). [CrossRef]
12. M. Wang, J. Hiltunen, S. Uusitalo, J. Puustinen, J. Lappalainen, P. Karioja, and R. Myllylä, “Fabrication of optical inverted-rib waveguides using UV-imprinting,” Microelectron. Eng. 88(2), 175–178 (2011). [CrossRef]
13. J. Comyn, Polymer Permeability (Chapman & Hall, 1985).
14. W. Lukosz, “Principles and sensitivities of integrated optical and surface plasmon sensors for direct affinity sensing and immunosensing,” Biosens. Bioelectron. 6(3), 215–225 (1991). [CrossRef]
15. G. H. Cross, Y. Ren, and N. J. Freeman, “Young’s fringes from vertically integrated slab waveguides: applications to humidity sensing,” J. Appl. Phys. 86(11), 6483–6488 (1999). [CrossRef]
16. G. T. Reed and A. P. Knights, Silicon Photonics, (Wiley, 2004).
17. S. J. Pearce, M. D. B. Charlton, J. Hiltunen, J. Puustinen, J. Lappalainen, and J. S. Wilkinson, “Structural characteristics and optical properties of plasma assisted reactive magnetron sputtered dielectric thin films for planar waveguiding applications,” Surf. Coat. Tech. 206(23), 4930–4939 (2012). [CrossRef]
18. S. Bäumer, Handbook of Plastic Optics (Wiley-VCH, 2005).
19. T. Watanabe, N. Ooba, Y. Hida, and M. Hikita, “Influence of humidity on refractive index of polymers for optical waveguide and its temperature dependence,” Appl. Phys. Lett. 72(13), 1533–1535 (1998). [CrossRef]
20. T. Shioda, N. Takamatsu, K. Suzuki, and S. Shichijyo, “Influence of water sorption on refractive index of fluorinated polyimide,” Polymer (Guildf.) 44(1), 137–142 (2003). [CrossRef]
21. R. C. Weast, Handbook of Chemistry and Physics (Cleveland Ohio USA, 1974).
22. B. Sun, M. Y. Chen, Y. K. Zhang, J. C. Yang, J. Q. Yao, and H. X. Cui, “Microstructured-core photonic-crystal fiber for ultra-sensitive refractive index sensing,” Opt. Express 19(5), 4091–4100 (2011). [CrossRef]
23. K. Zinoviev, A. Gonzáez-Guerrero, C. Domíguez, and L. Lechuga, “Integrated bimodal waveguide interferometric biosensor for label-free analysis,” J. Lightwave Technol. 29(13), 1926–1930 (2011). [CrossRef]
24. X. Tu, J. Song, T. Y. Liow, M. K. Park, J. Q. Yiying, J. S. Kee, M. Yu, and G. Q. Lo, “Thermal independent silicon-nitride slot waveguide biosensor with high sensitivity,” Opt. Express 20(3), 2640–2648 (2012). [CrossRef]
26. S. Lin, C.-K. Lee, Y.-M. Wang, L.-S. Huang, Y.-H. Lin, S.-Y. Lee, B.-C. Sheu, and S.-M. Hsu, “Measurement of dimensions of pentagonal doughnut-shaped C-reactive protein using an atomic force microscope and a dual polarisation interferometric biosensor,” Biosens. Bioelectron. 22(2), 323–327 (2006). [CrossRef]
27. D. X. Xu, A. Densmore, A. Delâge, P. Waldron, R. McKinnon, S. Janz, J. Lapointe, G. Lopinski, T. Mischki, E. Post, P. Cheben, and J. H. Schmid, “Folded cavity SOI microring sensors for high sensitivity and real time measurement of biomolecular binding,” Opt. Express 16(19), 15137–15148 (2008). [CrossRef]
28. E. Melnik, R. Bruck, R. Hainberger, and M. Lämmerhofer, “Multi-step surface functionalization of polyimide based evanescent wave photonic biosensors and application for DNA hybridization by Mach-Zehnder interferometer,” Anal. Chim. Acta 699(2), 206–215 (2011). [CrossRef]
29. A. K. Shrive, G. M. T. Cheetham, D. Holden, D. A. A. Myles, W. G. Turnell, J. E. Volanakis, M. B. Pepys, A. C. Bloomer, and T. J. Greenhough, “Three dimensional structure of human C-reactive protein,” Nat. Struct. Biol. 3(4), 346–354 (1996). [CrossRef]