Article pubs.acs.org/ac
Microgravimetric Thermodynamic Modeling for Optimization of Chemical Sensing Nanomaterials Pengcheng Xu, Haitao Yu, Shuanbao Guo, and Xinxin Li* State Key Lab of Transducer Technology, Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, 865 Changning Road, Shanghai 200050, China S Supporting Information *
ABSTRACT: On the basis of microgravimetric sensing data, an analytical modeling method is proposed for comprehensive evaluation and optimization of gas sensing or adsorbing related functional materials. Resonant microcantilever is loaded with the material to be evaluated for a gravimetric sensing experiment. With sensing isotherm curves obtained at different temperatures, key thermodynamic and kinetic parameters of the material, such as enthalpy ΔH°, Gibbs free energy, adsorption rate constant Ka, and coverage θ, etc., can be quantitatively extracted for optimal selection and design. On the basis of the gravimetric experiment, the modeling method is used on three sorts of trimethylamine sensing nanomaterials of mesoporous silica nanoparticles (MSNs). The COOH-functionalized material is clearly identified as the best sensing material among the three similar ones, thereby validating high accuracy of the proposed model. Broad applicability of the modeling method to other sensing materials and/or target gases is also experimentally confirmed, where sensing properties of a functionalized hyper-branched polymer to organophorous simulant of dimethyl methylphosphonate (DMMP) are still evaluated well. In addition to sensing materials, the gravimetric experiment-based modeling method can be expanded to other functional materials like moisture absorbents or detoxification agents. Water adsorbing experiment on KIT-5 mesoporous-silica is modeled, with the low −ΔH° value (i.e., low adsorption heat) result, indicating that the KIT-5 is a good adsorbent to humidity. Alternatively, the modeled high −ΔH° value (i.e., high reaction heat) shows promising usage of SBA-15 mesoporous-silica as detoxification material to hazardous organophorous chemicals. Therefore, the analytical modeling technology can be used for developing and evaluating new adsorbing materials for gas sensing, fixing, and detoxification applications.
G
underdeveloped gas sensor cannot be quantitatively determined until the device is fabricated and tested. Only after the performance is experimentally obtained, will the people be able to judge whether the sensor meets the requirements or not. From this point, development of the precise model for gas sensor performance is highly demanded. Being not so simple as physical microsensors like the capacitive accelerometers,6 which only possess one transduction interface from acceleration-induced displacement to capacitance change, a chemical sensor normally contains two transduction interfaces.7,8 Besides the similar interface to that of the accelerometer for transduction from physical effect to electric signal, another molecule-recognition interface exists for translating specific molecules adsorption into a physical sensing phenomenon like heating, resistance change, mass-induced
as molecule adsorption with advanced materials is very useful for sensing, fixing, and retrieving applications. With gas sensing as an example, micro/nano gas sensors have been long-time researched for on-site quick detection of chemical vapors such as volatile organic compounds (VOCs).1−4 Frankly speaking, industrialization of such chemical sensors has not been as successful as their counterpart of physical microsensors like pressure sensors and inertial sensors, etc. For those industrialized physical sensors, the well-established sensor modeling technology, based on theoretical analysis and software simulation, can precisely guide optimal design of sensor performance. Key sensing parameters, such as sensitivity and signal linearity, can be quantitatively determined at the design stage,5 and the designed parameters can be verified by testing data of the fabricated sensors. Comparatively, many underdeveloped chemical sensors have not reached such a modeling-design-fabrication-test-verification close-looped level. The lack of quantitative parameter modeling is a big issue for such chemical sensors. In many cases, the sensitivity of an © 2014 American Chemical Society
Received: October 29, 2013 Accepted: March 28, 2014 Published: March 28, 2014 4178
dx.doi.org/10.1021/ac403498x | Anal. Chem. 2014, 86, 4178−4187
Analytical Chemistry
Article
frequency-shift, or optical illumination. To realize the molecular recognition, gas sensing material is indispensable. The performance of a gas sensor strongly depends on the utilized sensing material. With the rapid development of nanotechnologies, nowadays a great number of nanomaterials have been explored for gas sensing.7,9−13 Gas sensing nanomaterials normally feature a high specific surface area for enhanced adsorption and improved limit of detection (LOD) for measurement of tracelevel VOCs. Now that the specific adsorption of the targeted gas molecules on the sensing material determines the sensing performance at the molecule-recognition interface, the material needs to be comprehensively designed and optimized. For repeatable detection, the sensing material needs to exhibit well in terms of both the molecule adsorption (during sensing) and desorption (for postsensing recovery). Hence, the molecular interaction cannot form a covalent chemical bond (i.e., pure chemisorption), since such type of strong interaction does not facilitate fast recovery after sensing and repeated detection. Normally, hydrogen-bonding interaction is preferred for reversible chemical detection. In the other extreme, pure physisorption is also not preferred due to the lack of specificity to the gas.14 Unlike the gas sensing, other applications such as moisture adsorbing or VOC detoxification reaction requires a different adsorption type of pure physisorption or substitution reaction, respectively. In general, for optimal and comprehensive design of functional materials to fulfill concrete gasadsorbing requirements, it is really crucial to build a comprehensive model where all the key criteria are taken into account.
Scheme 1. Modeling Route for Thermodynamic and Kinetic Parameters of Gas Adsorbing/Sensing Functional Materiala
■
CONCEPT Known as the central importance of sensing or other adsorbingfunctional materials, the thermodynamic parameters of enthalpy (ΔH°), entropy (ΔS°), and Gibbs free energy (ΔG°) have been widely utilized as staple criteria for predicting reaction path and distinguishing interaction mechanism between the materials and gas molecules.15,16 Starting from the thermodynamic properties, key kinetic parameters of the material, such as rate constant Ka (or Kd) that is important concern for high-speed adsorbing (or desorbing) response, are expected to be further obtained and evaluated. Nowadays the available methods to measure the thermodynamic data are based on either bulky instruments or density functional theory calculation that are generally high cost, time-consuming, and with limited application scope.17−20 For gas sensors, a simple and low-cost microplatform is highly required to conveniently real-time evaluate gas sensing materials. Gravimetric resonant microcantilever can in situ detect the adsorbed molecule quantity by online recording the added mass with frequencyshift signal. This inexpensive tiny tool can be used to characterize the material in very small volume. Such a microcantilever can easily be put into a temperature-controlled chamber to obtain an isotherm curve. Having been welldeveloped as gravimetric bio/chemical sensors,12,21−24 the resonant microcantilevers are herein utilized to model thermodynamic and kinetic parameters of the materials. With pictogram-level mass resolution in atmosphere, the resonant microcantilever is used to real-time detect the quantity of the adsorbed/desorbed molecules (with various partial pressures) onto/from the material. The measured data can be used to further explore the thermodynamic and kinetic properties for sensing-material comprehensive modeling. The modeling procedure is outlined in Scheme 1 and described as follows.
a
Isotherm curves are extracted from a gravimetric gas adsorbing experiment at different temperatures by using a resonant cantilever sensor. Then, calculation is performed for obtaining (at the left branch) enthalpy (ΔH°) and (at the right branch) adsorbing site number (N), coverage (θ), and equilibrium constant (K). Gibbs free energy and entropy are finally derived.
Temperature-Varying Experiment for Isotherms and ΔH°. The added gravimetric at a constant temperature (i.e., the frequency response signal of the resonant cantilever caused by the adsorbed mass of the gas molecules) can be detected and used to plot an isotherm against gas pressures (p). Herein, p is linearly associated with the gas concentration. Together with the embedded resonant microcantilever, the testing chamber is immersed into a temperature-controllable water bath. In this way, the temperature for gas adsorbing/sensing experiments can be adjusted. At a certain temperature, one sensing response curve can be obtained with regard to various gas concentrations. By changing the temperature of the water bath, a series of isotherm curves under different temperatures can be obtained. It is noted that the silicon cantilever itself features an ultra low temperature coefficient of resonance frequency. Originated from the merely −60 ppm/°C thermal drift of silicon Young’s modulus, the frequency temperature-coefficient is only −30 ppm/°C that has negligible influence on the accuracy of the data obtained from the temperature-varying experiment. Under different temperatures, identical frequency shift of the sensing signal from the same sensor indicates equivalent 4179
dx.doi.org/10.1021/ac403498x | Anal. Chem. 2014, 86, 4178−4187
Analytical Chemistry
Article
of the ≡Si−OH is too weak or not for firmly capturing the alkaline TMA. We also consider an alternative way to premodify the surface of MSNs. In this study, −COOH groups with moderate acidity and −SO3H with strong acidity are functionalized onto the mesoporous surface, respectively. The modified more chemically stable organic groups are expected to desirably tune the acidity of the MSN surface. The modeling method has been experimentally verified, being usable for various types of sensing material (e.g., hyperbranched polymer) to different gases (e.g., organophosphorus vapor). In addition to gas detection, the modeling can be expanded to evaluation of other functional materials, like moisture absorbents or detoxification agents. Pure physisorption of KIT-5 mesoporous-silica to H2O molecules is convinced by the modeling method that the material is a good moisture absorbent agent. Still based on the modeling method, the detoxification function of the material of SBA-15 mesoporoussilica to hazardous organophorous simulant of dimethyl methylphosphonate (DMMP) vapor is proved, where the extracted high −ΔH° parameter and the GC/MS results identify the substitution reaction of −OH with −O−CH3.
quantities of the molecules adsorbed on the sensing materials (i.e., identical fractional coverage θ). On the basis of at least two isotherms obtained from the experiments at two temperatures, the ΔH° value can be calculated by using the Clausius−Clapeyron equation. In general, the value of −ΔH° lower than 40 kJ mol−1 indicates physisorption, while the −ΔH° value of larger than 80 kJ mol−1 should be from strong chemisorption.25 As for design of gas sensing material, the ideal value of −ΔH° would be slightly larger than 40 kJ mol−1. Featuring weak chemisorption, such a moderate ΔH° value provides the material with not only desired specificity to the gas molecules but also good reliability for repeatable sensing. As for other applications like developing a moisture-absorbent agent, physisorption with low −ΔH° value is preferred, since it allowed absorbent material for multiple recycling usages. Alternatively, chemisorption with a high −ΔH° value is obviously favorable for gas fixing or chemical detoxification. Equilibrium Constant (K). From the Langmuir equation, we use one gravimetric sensing curve to lineally plot p/V versus p. The intercept of the plot is used to calculate adsorption/ desorption equilibrium constant K.25 Using fundamental thermodynamic equations, ΔG° and ΔS° can be calculated from the known K and ΔH°. The obtained thermodynamic parameters can be used to evaluate the material. Fractional Coverage (θ). With K and partial pressure p, θ can be calculated. By combining the sensing response and the mass sensitivity of the cantilever, the adsorbed molecule number (n) is obtained. With θ and n, the total sites number for adsorption N can be further obtained. The obtained parameters can be used to evaluate the adsorption capability of the material to the targeted gas. Adsorption/Desorption Rate Constant Ka/Kd. It is reasonable to assume that desorption is very weak at the initial stage of gas adsorption. The adsorption rate constant Ka can be calculated approximately according to the tested adsorption velocity (i.e., the slope of the frequency-shift) at the initial stage. With known K and Ka, Kd can be calculated. With the parameters of Ka and Kd obtained, the kinetic performance of the functional material, such as adsorbing/desorbing speed, can be considered overall. By now, it is clear that, with the resonant cantilever as gravimetric experiment tool, various thermodynamic and kinetic parameters of a gas adsorbing material can be precisely quantified. The parameters are very useful for comprehensive evaluation and optimal selection of gas adsorbing functional material. In order to validate the proposed modeling, we have herein used the gravimetric experiment-based thermodynamic parameter modeling method to identify the best trimethylamine (TMA, a malodorous VOC) sensing material from three sorts of functionalized mesoporous silica nanoparticles (MSNs). With this example, we hope to validate the accuracy of the modeling method. Three similar nanomaterials of the −OH group covered unmodified MSNs, −COOH-functionalized MSNs, and −SO3H-functionalized MSNs are prepared as potential candidates to be selected for detection of trace-level TMA. All three nanomaterials feature high specific surface area and nanoscale particle size of less than 100 nm. It is known that there are already numerous silanol groups (≡Si−OH, played as Brønsted acid sites) at the surface.26,27 Therefore, unmodified MSNs are capable of adsorbing basic gases like amine. However, the silanol groups of the unmodified MSNs are reactive and unstable. It needs to be known whether the acidity
■
EXPERIMENTAL SECTION Material Preparation and Microcantilever Fabrication. The mesoporous silica nanoparticle (MSN) sensing materials are prepared using a modified “Stöber” protocol.28−30 The fluorinated phenol functionalized hyper-branched polymer is synthesized with a modified method, as described in the previous report.31,32 The KIT-5 and SBA-15 mesoporous silica materials are both prepared by using hydrothermal methods.33,34 The technical details and sample characterization results are provided in the Supporting Information. Sensing Material Loaded on Cantilever. The mesoporous silica (including MSNs, KIT-5, and SBA-15) material (about 0.01 g) is added into 1 mL of deionized water (under ultrasonic) to form a crude suspension. About 0.1 μL of suspension is loaded onto the top-surface of the cantilever end region by using a commercial micromanipulator (Eppendorf made, model: PatchMan NP2). The process control is aided by inspection under microscopy (Leica-made, model: DM4000). Then, the microcantilever together with the material is dried in an oven at 333 K (60 °C) for about 2 h. This method is also suitable for loading the hyper-branched polymer onto the cantilever, while the water used for mesoporous silica dispersion is replaced by some organic solvent of tetrahydrofuran (THF). Vapor Generator. Either TMA or DMMP vapor with desired concentration is prepared from a commercially available standard vapor generator (model: Molecular Analysis series 8000S, Taiwan), which is equipped with a temperature programmable oven. For generation of the TMA vapor, a TMA permeation tube (Kin-Tec made, La Marque, Texas), with the permeation rate calibrated by weight loss speed of 242 ng/min, is inserted into the oven. Highly pure N2 is used as the carrier, where the flow rate is set as 1000 ± 1 sccm (standard cubic centimeter per minute). After stabilization at 303 K for 3 days, the N2-diluted TMA vapor with constant concentration is generated. With the use of another mass flow controller, the TMA vapor can be further diluted by pure N2 down to the desired concentration and, then, is introduced to a testing chamber, where the cantilever is inside. The DMMP vapor is obtained by using the same vapor generator. The permeation rate of the commercially available 4180
dx.doi.org/10.1021/ac403498x | Anal. Chem. 2014, 86, 4178−4187
Analytical Chemistry
Article
DMMP permeation tube (Kin-Tec made, La Marque, Texas) is calibrated as 454 ng/min, and the work temperature is set as 373 K. For generating water vapor, highly pure N2 is introduced to a water bubbler (containing 50 mL of water by accurate mass testing) and used as a carrier to blow and vaporize the water, where the flow rate is set as a fixed value in the range of 100− 1000 sccm. After 3 days, the mass of the water bubbler is weighed. In accordance with the mass loss of the water and the N2 flowing rate, the concentration of H2O vapor can be calibrated. Characterizations. High-resolution scanning electron microscopy (HR-SEM) and transmission electron microscopy (TEM) images of the samples are taken using an FEI Magellan 400 XHR ultrahigh resolution cold field emission scanning electron microscope (operated at 15, 20, and 30 kV) and a JEOL-2010F TEM apparatus. The nitrogen sorption isotherm is measured at 77 K by using a Micromeritics ASAP 2020 M system. Specific surface area and pore size distribution are calculated using Brunauer−Emmett−Teller (BET) and Barrett−Joyner−Halenda (BJH) methods. The Fourier Transform-infrared spectrum (FT-IR) experiment is performed with a Bruker Vertex 70v infrared spectrometer (under the vacuum,
