A mid-infrared lab-on-a-chip for dynamic reaction monitoring

Abstract

Mid-infrared spectroscopy is a sensitive and selective technique for probing molecules in the gas or liquid phase. Investigating chemical reactions in bio-medical applications such as drug production is recently gaining particular interest. However, monitoring dynamic processes in liquids is commonly limited to bulky systems and thus requires time-consuming offline analytics. In this work, we show a next-generation, fully-integrated and robust chip-scale sensor for online measurements of molecule dynamics in a liquid solution. Our fingertip-sized device utilizes quantum cascade technology, combining the emitter, sensing section and detector on a single chip. This enables real-time measurements probing only microliter amounts of analyte in an in situ configuration. We demonstrate time-resolved device operation by analyzing temperature-induced conformational changes of the model protein bovine serum albumin in heavy water. Quantitative measurements reveal excellent performance characteristics in terms of sensor linearity, wide coverage of concentrations, extending from 0.075 mg ml^-1 to 92 mg ml^-1 and a 55-times higher absorbance than state-of-the-art bulky and offline reference systems.

Fig. 1. FTIR spectrum of bovine serum albumin (BSA) & schematic of the QCLD device.

a Attenuated total reflection Fourier-transform infrared spectrometer (ATR-FTIR) reference measurement of the thermal denaturation process of BSA, analyzed in the range of the amide I${}^{\prime }$ band between 50 ^∘C (blue) and 90^∘C (red). The temperature-induced transition from α-helix (1651 cm⁻¹, blue) to β-sheet (1615 cm⁻¹, red) is indicated. b On-chip sensor concept including indicated plasmonic mode. Emitter (QCL, 10 μm wide) and detector (QCD, 15 μm wide) are connected through a 48 μm long tapered SiN-based plasmonic waveguide. The whole sensor is submerged into the sample solution (D₂O + BSA), which is shown by the blue transparent layer on the chip. The gold layer (plasmonic waveguide and electrical contacts) is indicated in gold color, the SiN passivation and dielectric loading layer are shown in brown and the InP substrate is indicated in dark gray.

Fig. 2. FEM-based simulations at 1620 cm⁻¹ (=6.17 μm).

a The mode cross-section of the SiN on Au DLSPP waveguide (n_SiN = 1.79, dimensions in inset: d_Au: gold thickness, d_SiN: SiN thickness and w_SiN: width SiN slab), b 2D topview simulation along the tapered 48 μm DLSPP waveguide between QCL and QCD in air (left) and D₂O (right) as well as longitudinal cross-section profile along the 48 μm DLSPP waveguide conducted in: c air and d D₂O. The white line in a, c, and d represents the Au plasmonic layer.

Fig. 3. Concentration measurement setup.

A peristaltic pump continuously pumps the stock solution (50 ml BSA in D₂O at 150 mg ml⁻¹) to the measurement beaker, initally filled with pure D₂O. The QCLD-sensor is directly submerged in this second beaker to monitor the concentration changes.

Fig. 4. Absorbance vs concentration measurements at 1597 cm⁻¹.

Results in absorbance units (AU) of the QCLD sensor for BSA in D₂O with (red stars) and without (blue squares) 18-mV crosstalk correction (left scale) and in comparison to the single-reflection ATR-FTIR system (violet circles, right scale). The right scale is divided by a factor of 10 as compared to the left one for better visibility of the ATR-FTIR signal.

Fig. 5. Denaturation measurement setup.

We use a custom-made 60-μl cell for this experiment. The stock solution (35 ml BSA in D₂O at 20, 40, and 60 mg ml⁻¹) is constantly heated from room temperature to 90 ^∘C, while being continuously pumped through a beaker with cooling liquid at 20^∘C and into the cell containing the QCLD-sensor. After (continuous) measurement it is pumped out and disposed of.

Fig. 6. Denaturation measurement results at 1620 cm⁻¹.

Investigation of three different concentrations of BSA: 20 (red), 40 (blue), and 60 mg ml⁻¹ (violet) and extraction of the transition temperature x0. a Absolute measurement values (circles) and sigmoidal Boltzmann-Fit curves (solid lines) in absorbance units (AU). b Comparison of the (individually normalized) Boltzmann-fit curves, showing the temperature- and concentration-dependence of the sigmoidal-shaped absorbance curves.

Fig. 7. Submersion experiment in native biophysical conditions, i.e., water.

Detector Signal when operating the QCLD sensor in DI-H₂O for about 1 min and increasing the temperature by about 0.1 ^∘C.