Technical Note pubs.acs.org/ac
Substrate-Integrated Hollow Waveguides: A New Level of Integration in Mid-Infrared Gas Sensing Andreas Wilk,† J. Chance Carter,‡ Michael Chrisp,‡ Anastacia M. Manuel,‡ Paul Mirkarimi,‡ Jennifer B. Alameda,‡ and Boris Mizaikoff*,† †
Institute of Analytical and Bioanalytical Chemistry (IABC), University of Ulm, Albert-Einstein-Allee 11, 89081 Ulm, Germany Lawrence Livermore National Laboratory (LLNL), 7000 East Avenue, Livermore California 94550, United States
‡
S Supporting Information *
ABSTRACT: A new generation of hollow waveguide (HWG) gas cells of unprecedented compact dimensions facilitating low sample volumes suitable for broad- and narrow-band mid-infrared (MIR; 2.5−20 μm) sensing applications is reported: the substrate-integrated hollow waveguide (iHWG). iHWGs are layered structures providing light guiding channels integrated into a solid-state substrate material, which are competitive if not superior in performance to conventional leaky-mode fiber optic silica HWGs having similar optical pathlengths. In particular, the provided flexibility in device and optical design and the wide variety of manufacturing strategies, substrate materials, access to the optical channel, and optical coating options highlight the advantages of iHWGs in terms of robustness, compactness, and costeffectiveness. Finally, the unmatched modularity of this novel waveguide approach facilitates tailoring iHWGs to almost any kind of gas sensor technology providing adaptability to the specific demands of a wide range of sensing scenarios. Device fabrication is demonstrated for the example of a yin-yang-shaped gold-coated iHWG fabricated within an aluminum substrate with a footprint of only 75 mm × 50 mm × 12 mm (L × W × H), yet providing a nominal optical absorption path length of more than 22 cm. The analytical utility of this device for advanced MIR gas sensing applications is demonstrated for the gaseous constituents butane, carbon dioxide, cyclopropane, isobutylene, and methane.
T
hollow core of the HWG confines a very small probed volume (few milliliters or less) within a well-defined optical path length (OPL). This high volumetric optical efficiency leads to a substantial increase in sensitivity compared to conventional open path or multipass measurements and, hence, renders the sensor response time significantly smaller compared to conventional multipass (e.g., Herriott or White) cells usually requiring the exchange of volumes exceeding 500 mL. It has been demonstrated that limits of detection in the ppmv−ppbv concentration range can be maintained for volatile organic compounds (VOCs) including benzene, toluene, o-/m-/pxylene, ethyl chloride, ethylene, CO, NO, sidestream tobacco smoke constituents, and several other volatiles using HWGs coupled to either FT-IR spectrometers or QCLs.5,9,24−48 However, as shown herein, conventional HWGs are not well suited for integration into compact sensing devices because of two main characteristics, i.e., waveguide length and flexibility. High-sensitivity MIR (and likewise Raman) gas sensing applications involving HWGs require increased pathlengths for ensuring trace level detection of relevant constituents. Conventional HWG devices usually rely on tubes drawn from glass or
oday, a substantial number of applications in trace gas detection are excellently served by direct sensing in the mid-infrared (2.5−20 μm; MIR) spectral range. Advances toward hand-held modern MIR sensing techniques largely capitalize on the unique properties of quantum cascade lasers (QCLs) serving as MIR light sources toward lab-on-chip implementations of MIR sensors.1−13 Much less attention has focused on the development of appropriate modular waveguide technology that may be smartly combined with broad- and narrowband MIR light sources for advanced gas sensing applications. Hollow waveguides (HWGs) are essentially hollow core light-pipes initially designed as a conduit for delivering highpeak power laser light for industrial and surgical applications.14−20 Those applications relied on HWG length (i.e., frequently meters), flexibility, and a high damage threshold. Only during the three past decades has the use of HWGs been successfully extended to include spectroscopic and sensing applications. Hyphenation of capillary gas chromatography and Fourier transform infrared (FT-IR) spectrometry was enabled by the development of light-pipe based interfaces (i.e., goldcoated glass tubes) already in the 1980s.21 Dielectrically coated HWGs may simultaneously serve as a waveguide propagating IR radiation and/or Raman excitation radiation and Raman scattering22,23 and as an efficient miniaturized gas cell. The © 2013 American Chemical Society
Received: July 31, 2013 Accepted: September 23, 2013 Published: September 23, 2013 11205
dx.doi.org/10.1021/ac402391m | Anal. Chem. 2013, 85, 11205−11210
Analytical Chemistry
Technical Note
mill. The inner sidewall surfaces of both parts, which ultimately form the sidewalls of the iHWG open channel upon assembly, were polished using a variety of commercially available hand polishing tools. The final result is shown in Figure 1a. The sidewall surface roughness was characterized using atomic force microscopy (AFM). As reflection losses follow (1-reflectivity)reflections, thin gold films (100−200 nm) were deposited onto the aluminum waveguides using a magnetron sputter deposition system described in detail elsewhere.54 The base pressure of the system was approximately 1.5 × 10−7 Torr prior to deposition. The sputter gas was 99.999% pure argon, and the pressure
silica with diameters usually exceeding 1 mm, which cannot be sharply bent or coiled. Consequently, for conventional IRHWG sensors the required length of HWG largely governs the operational footprint of the entire sensor, thereby limiting their practical use for applications requiring a compact device form factor. Moreover, the overall physical size of the HWG renders it susceptible to unwanted mechanical vibrations or temperature fluctuations. Therefore, conventional HWGs must be mechanically supported along their physical length. Also, such long tubular devices, especially if drawn from glass, are fragile, requiring very careful handling or additional protection from mechanical damage. Hence, while conventional HWGs have fundamentally enabled many gas sensing applications, they are not adequate solutions in applications where small dimensions and (long-term) robustness are required. Coiling smaller diameter waveguides has been demonstrated to reduce the overall form factor, however, at the expense of sensitivity and robustness.49−52 Moreover, there are, to the best of our knowledge, no long-term studies reported that demonstrate how the internal mechanical bending stresses imposed onto structural materials and optical coatings in a coiled conventional HWG affect the overall lifetime of the HWG. Tubular devices depend on wet-chemistry for applying internal coatings, which significantly limits the coating options. Besides being expensive, the lack of mechanical robustness and bulkiness render the potential for integration of conventional HWGs with sensor or peripheral components the main limitation for more widespread adoption of such transducers. In order to resolve the main shortcomings of conventional HWGs, we propose a new class of waveguides based on a modular concept.53 These so-called substrate-integrated hollow waveguides (iHWGs), regardless of their complexity, are similarly based on a layered structure with the light guiding channels integrated into a rigid solid-state substrate material. By stacking a base plate, a waveguide layer (which may optionally be incorporated into the base plate), and a top plate held in position via screws, clamping devices, an adhesive, or any other method of temporary or permanent bonding, the formation of enclosed cross-sectional waveguide profiles is achieved. In the present study we reduce to practice the outlined general iHWG concept to demonstrate the first functional iHWG prototype applicable for advanced MIR gas sensing.
■
EXPERIMENTAL SECTION Figure S-1 (Supporting Information) shows the scheme of the experimentally realized iHWG prototype with yin-yanggeometry integrated into a 7.5 cm × 5.0 cm aluminum substrate along with the modeled propagation path of the center ray (22.45 cm OPL, 14 reflections). This iHWG comprises four individual components that upon assembly establish an enclosed waveguide structure, as illustrated in the Supporting Information (Figure S-2a−d). The iHWG components were fabricated from commercially available AlMg3 alloy. The assembled iHWG had dimensions of 75 mm × 50 mm × 12 mm (L × W × H) with an open channel cross-section of 2.1 mm × 2.0 mm, which establishes the optical pathway. The height of the iHWG top- and base plates and internal waveguide sections were 5 mm and 2 mm, respectively. Using commercially available diamond polishing suspensions, the top- and base substrate surfaces were polished to a mirrorlike finish. The internal waveguide sections were initially fabricated as 75 mm × 50 mm × 2 mm (L × W × H) bulk substrates and subsequently shaped using a solid carbide end
Figure 1. (a) Aluminum yin-yang-iHWG base plate with assembled waveguide layer (after polishing). (b) Top- and bottom view of goldcoated top plate with gas inlet/outlet connectors. (c) Schematic of the experimental iHWG sensor setup with (1) Bruker IRcube FT-IR spectrometer, (2) planar gold-coated mirror, (3) gold-coated off-axis parabolic mirror (OAPM), (4) iHWG assembly, and (5) MCT detector. The IR beam propagates from 1 to 5. 11206
dx.doi.org/10.1021/ac402391m | Anal. Chem. 2013, 85, 11205−11210
Analytical Chemistry
Technical Note
of surface contacts (reflection losses), sharply meandered (serpentine) geometries (Figure S-4, Supporting Information) are favored instead. Supporting this argument, the power throughput through the yin-yang iHWG was analyzed using a low power thermopile detector (Gentec-EO XLP12, GENTEC-EO USA, Lake Oswego, OR) resulting in an absolute power transmission of T = 6.85 ± 0.11%. Assuming that a ray needs a minimum number of 14 reflections in order to pass this waveguide structure (see Figure S-1 in the Supporting Information), this translates into a minimum surface reflectance of R = 82.6% (analyzed without BaF2 windows). Moreover, as the geometric channel length is approximately 21.8 cm, the attenuation coefficient a may be calculated at a = 0.053/cm. Sensor Setup. Referring to Figure 1c, the described optical sensor setup used direct optical coupling of the iHWG output to the MCT detector without coupling optics. Remarkably, there is a roughly 5 mm gap between the outcoupling port of the iHWG (i.e., the focal point) and the ZnSe window of the detector aligned at the optical axis. Using a large cross-section (4 mm2) MCT element (FTIR-16-2.00 MSL-12, InfraRed Associates Inc., Stuart, FL) facilitated obtaining 2.90 × 104 ADC (analog-to-digital converter) counts, thereby rendering additional refocusing optics optional. Hence, using such a detector in combination with appropriate refocusing optics provides sufficient potential to further increase the achievable signal-to-noise ratio (SNR) when using iHWGs with substantially extended OPLs, yet maintaining the same small sensor form factor. Analytical Performance. First qualitative spectra on the sensor performance were acquired during the curing period of the iHWG silicone sealing. During the polymerization process, the silicone rubber outgasses acetic acid. This process was used to study gaseous acetic acid as an in situ produced analyte. The measurements shown in Figure 2a were performed several hours after the silicone was applied. At 20 s intervals, MIR spectra were collected (50 averaged scans, 40 kHz sampling rate, 4 cm−1 spectral resolution) acquiring a total of 100 spectra. During that data acquisition period, the iHWG was discontinuously flushed with nitrogen. Whenever the nitrogen flow was stopped, the rapidly rising concentration of acetic acid within the hollow waveguide was readily detectable. Figure S-5 (Supporting Information) shows the IR spectrum acquired at 1200 s into this measurement series, thereby proving that the detected analyte was indeed acetic acid. Each spectrum was then integrated in the 1600−1900 cm−1 range (compare Figure S-5 in the Supporting Information). The obtained peak areas correspond to the data points plotted in Figure 2a with time demonstrating the excellent repeatability of such continuous measurements with the newly developed iHWG MIR sensor system. Several days thereafter, i.e., when the curing of the silicone was completed and no acetic acid was detectable after a 12 h idle period without N2 flushing, spectra of several gas phase constituents serving as model analytes were acquired. For that purpose, certified gas standards of isobutylene, methane, cyclopropane, carbon dioxide, and butane (nominal concentrations, 1% in N2; Westfalen AG, Weißenhorn, Germany) were diluted with nitrogen to 5000 ppmv using a custom-made mass flow-controlled gas mixing prototype developed by IABC and LLNL and delivered into the iHWG (200 sccm nominal flow rate) via the Luer lock connections. Unless stated otherwise, all MIR spectra were collected using the same data acquisition
during deposition was set at 2 mTorr. A thin (approximately 20 nm) chromium layer was deposited between the aluminum substrate and the reflective gold layer for enhancing the adhesion. The calibration of the deposition rate was performed by depositing individual gold and chromium films; thereafter, the obtained films were analyzed via X-ray diffraction (XRD) using a PANalytical Pro MRD system. To facilitate the introduction of gases into the fully assembled iHWG, access ports were fabricated into the top plate of the iHWG (holes with d = 0.5 mm), and M5 threaded brass nuts were glued concentric with the sampling ports (Thorlabs 2K epoxy adhesive for structural bonding), as shown in Figure 1b. Subsequently, female V2A steel Luer lock connectors were attached (EHS Medizintechnik GmbH, Leinfelden-Echterdingen, Germany) and sealed with Teflon sealing tape. Luer lock valve interfaces were used to enable gas/ vapor delivery into the iHWG waveguide channel. A clamping device (EN AW-7065 T651 aluminum alloy) similar to a hand screw type clamp was developed as an alternative to using adhesives for securing and maintaining the alignment of the assembled waveguide components. Once the iHWG was secured in the clamp, BaF2 windows (OEC GmbH, Zusmarshausen, Germany) were attached using a roomtemperature acetoxy cure silicone at both iHWG optical channel ports (see Figure S-2d in the Supporting Information) used for radiation in- and outcoupling. The same silicone rubber was applied to the perimeter of the assembled and clamped iHWG. The combination of silicone sealant and BaF2 windows provided for a sufficiently airtight iHWG facilitating FT-IR gas studies. Figure 1c illustrates the final iHWG sensor setup. The collimated IR beam from an FT-IR spectrometer (IRcube, Bruker Optics, Billerica, MA) was focused via two gold-coated mirrors (Janos Technology LLC, Keene, NH), i.e., a planar and an off-axis parabolic mirror (OAPM; effective focal length = 50.8 mm), onto one of two optical ports of the iHWG. Light passes through the waveguide structure containing the gas/ vapor sample and propagates through the distal optical port onto an external liquid nitrogen cooled mercury-cadmiumtelluride (MCT) detector interfaced with the FT-IR spectrometer.
■
RESULTS AND DISCUSSION Waveguide Surface Quality and Geometry. AFM studies (Supporting Information, Figure S-3) revealed that the polishing techniques discussed herein yielded a combined surface roughness/waviness of RMS = (44 ± 11) nm. As the MIR radiation utilized in the present study is approximately 3− 12 μm in wavelength, the surface quality of the waveguide channels is sufficient according to the λ/10th criterion, thereby rendering the inside waveguide walls flat for MIR radiation propagation. A highly polished surface is desirable, as light scattering is among the loss mechanisms that may occur as photons are reflected by the channel surfaces. Ray trace modeling has shown that using a spiral-type waveguide channel geometry where the bending radius is large compared to the cross-sectional diameter (applies to, e.g., the yin-yang type shown here) is advisable whenever noncoherent, noncollimated radiation sources are coupled into the iHWG; spiral type geometries eliminate back reflections and lossy resonant type ray paths over a wide range of light incoupling angles (f/ numbers). For transmitting highly collimated sources (e.g., a laser) through the iHWG while minimizing the overall number 11207
dx.doi.org/10.1021/ac402391m | Anal. Chem. 2013, 85, 11205−11210
Analytical Chemistry
Technical Note
sensor at a concentration of 5000 ppmv. At these measurement conditions, an excellent SNR of up to 500:1 was obtained for all analytes. Figure S-6b in the Supporting Information also shows butane at a trace level concentration of 20 ppmv (1000 averaged scans). Given that a 22.5 cm long and straight conventional HWG would be required to yield a comparable SNR, the first prototype of the iHWG MIR sensor already demonstrates truly competitive dimensions and performance. Figure 2b,c presents mixture spectrum of all model analytes with varying individual concentrations ranging from 400 to 2000 ppmv. Furthermore, the mixture spectrum was decomposed for revealing the individual contributions of each analyte to the sum signal. Thereby, the spectral features that are characteristic for each gas are readily discernible for demonstrating the multiconstituent capabilities of the outlined sensor. In addition, the univariate limits of detection (LODs) derived from individual analytes without interfering agents of the current configuration (i.e., method detection limits) were calculated following Inczédy et al.,55 i.e., under the assumptions (i) that the blank variance (noise) is Gaussian distributed (corresponding to a random background), (ii) that the variance is constant, and (iii) that both probabilities of false positives and false negatives are equal to 0.05. All LODs with the exception of CO2 were calculated between 6 and 11 ppmv. The LOD for CO2 was calculated at 21 ppmv, which is due to the fact that statistic ambient air CO2 content fluctuations add to the noise level of the blanks within the CO2 absorption band. However, if the sensor is sealed against the ambient environment, it is expected that this value will be at the level of the other LODs. Using a refined sampling method (e.g., increasing the number of averaged scans by a factor of 10 resulting in 1000 averaged scans, thereby decreasing the LOD by a factor of 100.5 to approximately one-third), the achievable LOD may be further reduced to, e.g.,
