State-of-the-art monitoring of fuel acidity

Available online at www.sciencedirect.com

Sensors and Actuators B 130 (2008) 871–881

State-of-the-art monitoring of fuel acidity Justyna Widera a,∗ , Bill L. Riehl b , Jay M. Johnson b , Douglas C. Hansen b a b

Adelphi University, Department of Chemistry, 1 South Avenue, Garden City, NY 11530, United States University of Dayton Research Institute (UDRI), 300 College Park, Dayton, OH 45469, United States Received 16 October 2007; accepted 25 October 2007 Available online 4 November 2007

Abstract Development of a novel iridium oxide (IrOx) based acidity sensor for off-line monitoring of fuel acidity is described. The sensor works in the potentiometric mode using an IrOx electrode as an indicating electrode and a Ag/AgCl or Ag/Ag2 O—reference electrode. The data show that the IrOx sensor responds to compounds present in fuel that have acid–base character. Using an off-line IrOx sensor, it is possible to determine the acidity of different fuels and discriminate between unstressed and thermally stressed fuels. It is possible to correlate the response of an IrOx sensor with the total acid numbers of different fuels. Experimental results also indicate that the low fuel conductance, the material used for sensor encapsulation, and/or the type of reference electrode may influence the response time of the IrOx sensor. Finally, the IrOx response has been demonstrated to be faster, better defined, more accurate and more reproducible than a glass electrode response for titrations of non-aqueous solutions. © 2007 Elsevier B.V. All rights reserved. Keywords: Iridium oxide; Sensor; Acidity; Non-aqueous solvents; Fuel; Total acid number

1. Introduction Some acids can be present in aviation turbine fuels due to naturally occurring organic acids, the presence of some fuel additives, acid treatment during the refining process, and/or degradation/oxidation products of the fuel formed during service and thermal stressing [1]. In aviation fuel, the constituents considered to have acidic characteristics include organic and inorganic acids, esters, phenolic compounds, lactones, resins, heavy metal salts, and additives, such as inhibitors and detergents. Significant acid contamination is not likely to be present because of the numerous quality control steps during the various stages of refining. However, trace amounts of acid can be present and are undesirable because of the consequent tendencies of the fuel to corrode metals and allow higher levels of dispersed water within the fuel [1]. The measurement of acidity is a difficult but important technique in the characterization of petroleum products. Information

∗

Corresponding author. Tel.: +1 516 877 4135; fax: +1 516 877 4485. E-mail addresses: [email protected] (J. Widera), [email protected] (J.M. Johnson). 0925-4005/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.snb.2007.10.056

about acid content in fuel or lubricants is crucial because it is an indicator of the quality of these products; an acidic reading may suggest, for example, that an aircraft fuel has undergone thermal oxidation or that a lubricant has completely lost the added antioxidants and needs to be replaced [2]. The ability to continuously monitor these systems would provide an early warning for failure of these fluids. Perhaps more important than “user-level” specification testing [3–5] is the fact that the fuels processing industry must deliver neutral products compatible with fuel systems worldwide. In this regard, the process industry must create a product and then perform acid number testing [6]. If in-line monitoring could be performed more quickly and accurately than the off-line testing, process changes could be affected more readily, reducing waste and reprocessing. Commercially available pH sensors are designed to conduct measurements in the aqueous phase. The measurement of acidity in organic solvent-based matrices like petroleum products is much more difficult due to their complexity and the fact that they are extremely non-conducting. Traditional electrochemical pH measurement technology is poorly suited for non-aqueous environments. Thus, non-aqueous pH measurements are meaningful only for monitoring the course of an acid–base titration or relative to some reference measurement made within the indi-

872

J. Widera et al. / Sensors and Actuators B 130 (2008) 871–881

vidual laboratory. Little, if any, confidence can be attached to absolute pH measurements in non-aqueous systems. Additionally, commercially available glass sensor electrodes are fragile and do not have the requisite stability, nor are they physically small enough to be easily incorporated into miniature reactors becoming more popular among research scientists [7–12]. The iridium oxide (IrOx) electrode has demonstrated superior long-term stability when compared to commercially available (glass electrode) and other experimental pH sensors (ion selective field effect transistors and other metal oxide electrode-based sensors) described in the literature [13–15]. Metal oxide electrodes (MOEs), in general, are inherently durable and suitable for miniaturization. However, based upon our empirical results, MOEs made by processes other than the thermal oxidation process are less stable (unpublished results). An IrOx pH sensor for aqueous applications has recently been developed and results suggest that IrOx sensors are superior to any other pH sensors on the market or in the published literature in terms of accuracy, long-term stability, sensitivity, reproducibility and response time [16]. This sensor is commercially available (SensIrOx, Inc.) and is fabricated using a proprietary process involving the thermal oxidation of iridium in a lithium carbonate melt. It has been demonstrated that lithium ions are inserted into the IrOx crystal matrix to form a new Li–IrOx compound. It is believed that it is the properties of this new compound that result in the improved performance of the sensor. As a result of their improved stability, the SensIrOx sensors do not require frequent calibration. They also can potentially be miniaturized and produced at a much lower cost. The SensIrOx sensor has shown excellent performance in aqueous solutions: the electrode exhibits excellent reversibility independent of the direction of the pH change or whether the pH is changed in small or large steps; it shows ideal Nernstian sensitivity (slope of 59.0 mV/pH), excellent reproducibility, and a response time on the order of seconds [16]. SensIrOx electrodes show good stability over a wide pH range, even at high temperatures [17], at high pressures [18], in aggressive environments, such as HF solutions [19], and fast response even in non-aqueous solutions [13]. Some preliminary work has also been done on using the SensIrOx electrode as a working electrode material for the development of acidity and basicity sensors for industrial lubricants [20]. These promising results motivated us to explore the use of the SensIrOx electrode for determining the acidity of aviation fuels. The measurement of acidity in certain organic-based liquids, such as petroleum products is a difficult application due to the complex and non-polar (non-conducting) matrix. Solving the “low conductance problem” is the primary focus in the development of an acidity sensor for the petroleum products application. Petroleum products exhibit high electrical resistance because of their non-polar nature. Non-polar solvents have low ionic mobility and conductance due to the fact that they have low dielectric constants, extended ion pairing, and multiple ionic associations. Making electrochemical measurements in non-polar solvents is further complicated by the fact that they tend not to be very good solvents for salts and the resulting high electrical resistance can cause severe signal distortion. The use of microelectrodes can largely mitigate the resistance problem as they can allow

effective electrochemical measurements in resistive media (e.g., in electrolyte-free liquids with low dielectric constants). However, for this work, since fabrication of the IrOx based indicator electrode in a micro-format was not considered a viable nearterm option, fuel samples were simply diluted into a more polar solvent for analysis. Currently, the measurement of Total Acid Number (TAN) in non-conducting fluids is performed by tedious, time consuming and solvent intensive methods (ASTM D 3242 [21] or D 664 [22]) based on titration. In these methods, the tested fuel sample is dissolved in a 1:1 mixture of toluene/isopropanol containing a small amount of water and titrated with standard alcoholic potassium hydroxide. In the D 3242 method, the end point is indicated by the color change of the added ␳-naphtholbenzein indicator solution. In the D 664 method, the inflection point during the titration is indicated potentiometrically using a glass indicating electrode and a calomel reference electrode. Clearly, a chemical sensor that could rapidly and reliably measure TAN in fuels and oils would be of tremendous use in specification testing, for monitoring/process control, as an R&D tool for additive development and for fuel thermal stability studies. An iridium oxide-based sensor (IrOx) may prove to be the best choice for these applications. In this paper, efforts to develop an IrOx-based acidity sensor suitable for work in low conducting media, such as aviation fuel are described. Studies using the IrOx sensing system, are presented which demonstrate the detection characteristics in non-aqueous solutions and the feasibility of the IrOx (SensIrOx) sensor as a fast, accurate, real-time acidity off-line sensor for the testing of fuels. The development of a novel type of reference electrode is also described, where a thin porous polymeric film of cellulose acetate/cellulose acetate butyrate (CA/CAB) is used to protect the reference element from the fouling effect of fuel samples and facilitate the communication between the reference and working electrode. 2. Experimental 2.1. Chemicals and materials All chemicals were obtained from Aldrich (Milwaukee, WI) and used without further purification. Water was deionized and purified using a Synergy Millipore water purification system. Fuel samples (Table 1) were obtained from the fuels branch of the Propulsion Directorate at Wright-Patterson Air Force Base. The fuel samples were identified using common Air Force nomenclature and reflect their content and properties that are proprietary. The IrOx electrodes were obtained from SensIrOx, Inc. (Columbus, OH). The manufacturing process for these electrodes has been described previously with a detailed compositional analysis of the iridium oxide thin film [16]. The fabrication process involves the thermal oxidation of an iridium wire in a lithium carbonate melt. Various materials were tested for sensor encapsulation. These included: Eccobond 55 epoxy (Emerson & Cuming, Billerica, MA), polyimide resin (Restek, Bellefonte, PA), glass powder 7556 (Corning, Corning, NY),

J. Widera et al. / Sensors and Actuators B 130 (2008) 871–881 Table 1 Response times (time-constant) of the epoxy encapsulated IrOx electrode Fuel type

Response time (time-constants) [min]

E @ 90% [mV]

3166n 2747n 3804n 3166s 2747s 3804s

116 22 46 110 70 36

28.7 23.7 31.3 108.0 64.4 58.1

Unstressed fuel Thermally stressed fuel

Average response

Standard deviation

27.9 76.8

3.9 27.2

n—unstressed fuel; s—thermally stressed fuel, fuel that was exposed to 140 ◦ C for 5 h. E90 response is defined as response at approximately three timeconstants.

ester cyanate RS-14A ML-2-100-1 (YLA Inc., Benicia, CA) and PTFE/FEP (polytetrafluoroethylene)/(fluorinated ethylenepropylene) dual shrink tubing (Small Parts, Inc., Miami Lakes, FL). All of the parts necessary to build the reference electrode were purchased as follows: silver wire 0.25 mm dia; 99.9985% purity Premion® (Alfa Aesar, Ward Hill, MA); flexible Teflon tube with ceramic or fiber junction frit (Cypress Systems, Inc., Lawrence, KS). The cellulose acetate/cellulose butyrate (CA/CAB) thin film that was used for coating the Ag/AgCl reference electrode was prepared according to a previously described procedure [23]. 2.2. Methods and measurements The experimental system consists of an electrochemical cell and the electrochemical workstation Model 650B from CH Instruments (Austin, TX). The hardware is controlled and data acquisition achieved using an external PC under a Windows environment. The sensing probe is potentiometric and consists of indicating electrode (IE) and a reference electrode (RE). An IrOx electrode is used as the indicating electrode. The IrOx was removed from a small 1 mm section at the end of the Ir wire, and a gold lead attached. In order to protect this contact area and provide both a solvent tight seal and electrical insulation, various encapsulating materials were evaluated, including epoxy resin, polyimide resin, glass powder, ester cyanate and PTFE/FEP dual shrink tubing. The IrOx sensor was tested in various solvents and in various combinations of solvents and supporting electrolytes. The test solution was stirred during the measurements. Initial measurements were made in acetonitrile (10 mL) and after establishing a stable background for the IrOx electrode 1.0 mL of 0.1 M phenol solution in acetonitrile was added in order to check the sensor response to the presence of a compound commonly present in fuel [24]. After exposing the sensing probe to the solution containing the analyte of interest, the signal was observed as a change in potential relative to the reference electrode. The IrOx electrode potential versus time was monitored until a stable plateau on the potential–time curve

873

was reached. The time at which the signal reached 90% of maximum potential was used as a criterion to determine the response time of the sensor. The response of the IrOx sensor to the fuel was initially investigated using two kinds of samples, A and B, each containing 10 mL of acetonitrile and 1 mL of fuel type 2747, (n) unstressed or (s) thermally stressed, respectively. The measurement was repeated three times on each sample by dipping the IrOx sensor in each sample for 1 h. The sensor was immersed in pure acetonitrile solution between consecutive runs to clean it. Long-term measurements were also performed by exposing IrOx probes for approximately 8 h to the fuel:acetonitrile (1:10) mixtures containing the following fuel samples: 2747, 3166 and 3804 unstressed and thermally stressed fuels. In the next series of tests, isopropanol was used as a fuel diluting solvent. Repeatability of the IrOx sensor exposed to the fuel (3686n)—solvent (isopropanol) mixture (1:10) was checked by running this test three times. The response of the IrOx sensor to fuels of various acidity was also checked. The following fuels were tested: 3804n, 3686n, 3804n with LT (low temperature additive) 4493 at 4 g/L, 3804 with LT4596 at 4g/L, 3804n with LT4708 at 4 g/L and 3804n with LT4708 at 5 g/L by exposing the IrOx sensing probe to the solution containing 10 mL of isopropanol, 1 g/L of tetrabutylammonium tetrafluoroborrate (TBATFB) and 1 mL of each type of fuel, respectively. The same experiment was repeated as mentioned above, adding various amounts (2, 1, 0.5 and 0.25 mL) of fuel (3804 with LT4493 at 4000 mg/L) to the tested mixture, but using a different reference electrode. In this case, a Ag/AgCl electrode protected by CA/CAB was used. The change in electrode potential was correlated with the TAN of the fuel solutions tested or with corrected volume of the same fuel added to the background, diluting solvent. The TAN’s for unstressed fuels were obtained from WrightPatterson Air Force Base internal reports and databases. The TAN’s for the stressed fuels are not routinely performed and therefore the titration of these fuels following the ASTM 3242 protocol [21] was performed. The 3242 ASTM method uses a visual detection end point indicated by the color change of the dye added to the titrated solution. The following stressed fuels were tested: 3166s, 3804s and 2747s. The same ASTM method was used to test stressed fuels containing low temperature additives (Table 1). The potentiometric titration experiment was performed using an automatic titration system (702SM Titrino, (Metrohm, Switzerland) following the ASTM 664 protocol [22]. A fuel sample (3804 unstressed fuel containing 4000 mg/L of 4708 low temperature additive) was dissolved in the mixture containing isopropanol, toluene and DI water. In the titration experiment, either the combined pH glass electrode or the IrOx electrode were used as a potentiometric end point titration detector. An internal Ag/AgCl electrode within a combination pH glass electrode was used as the reference electrode and the solution was stirred during the titration. The IrOx electrode potential versus time (or added volume of titrant) was monitored and an inflection point on the potential–time curve was used as the end point criteria for these measurements.

874

J. Widera et al. / Sensors and Actuators B 130 (2008) 871–881

2.3. Reference electrode Reference electrodes, of various designs, for the non-aqueous application were assembled and tested. All the potentials in the present work are referenced versus either Ag/AgCl or Ag/Ag2 O reference electrodes. To prepare the Ag/AgCl reference electrode, Ag wires (0.25 mm o.d.) were chloridized by immersion in 1 M FeCl3 –0.1 M HCl for 24 h. For the Ag/Ag2 O reference electrode preparation, the Ag wire was electro-oxidized for 10 min in 0.1 M aqueous solution of acetate buffer at potential +1 V versus a commercial Ag/AgCl reference electrode. The Ag/Ag2 O wire was placed in a Teflon tube terminated with a ceramic frit and containing isopropanol and TBATFB (1 g/L). A novel design of the reference electrode, which eliminated the necessity of tubing, frit and a liquid junction was tested. This particular design consists of a reference element (Ag/AgCl) protected by a thin, porous polymeric film of CA/CAB [23]. The CA/CAB film was prepared as follows: 4.552 g of CA were plasticized in 18.208 g of butyrolactone (1:4 ratio) and 1.138 g of CAB were plasticized in 4.553 g of butyrolactone (1:4 ratio). Both plasticized CA and CAB were combined together and mixed quickly to thoroughly incorporate the two components and to prevent vapor loss. An additional 7.500 g of butyrolactone were added to the plasticized polymer and mixed thoroughly. Vapor loss was minimized by capping the container with the prepared mixture. Then 34.050 g of nitromethane were added, mixed quickly and covered to minimize vapor loss. Finally 30.000 g of isopropanol were added and quickly mixed again to prevent nitromethane loss. The details associated with the application and drying of the CA/CAB film are proprietary. The chemicals necessary to prepare the CA/CAB thin film coating were bought from Eastman Chemical Company (Kingsport, TN). The reference element was dipped in the prepared polymeric solution and dried leaving a thin film of the CA/CAB polymer on the reference electrode surface. 3. Results and discussion 3.1. Sensor encapsulation All of the materials tested (epoxy resin, polyimide resin, glass powder and ester cyanate) were preliminarily found to be acceptable for electrode encapsulation purposes. The best performance was achieved using double-wall shrink/melt tubing. It is constructed of an outer layer of heat shrink PTFE (polytetrafluoroethylene) and an inner layer of heat melt FEP (fluorinated ethylene-propylene); the outer tubing shrinks while the inner layer melts when heat is applied. The shrink tubing allows for a tight fit around the encapsulated part while the near-solid encapsulation can withstand moisture and relatively severe mechanical stresses. 3.2. Reference electrode Ag/AgCl and Ag/Ag2 O reference electrodes proved to be the easiest to prepare and the most stable reference electrodes for non-aqueous applications. The reproducibility and long-term

stability of these reference electrodes were also tested and found to be acceptable. The responses across four electrodes of both types were within 20 mV of each other when all were immersed in acetonitrile or isopropanol with TBATFB. All further measurements were done using these reference electrodes. A novel reference electrode, which eliminated the necessity of tubing, frit and liquid junction was also tested. This particular design consists of a reference element (Ag/AgCl) protected by a thin, porous polymeric film of CA/CAB. Since this reference electrode doesn’t have a frit, there is no associated substantial IR drop that causes distortion of measurements in non-aqueous solutions. In this way the communication between the reference and the indicating electrodes is improved. Additionally, the reference electrode is apparently protected from the fouling influence of fuel by the thin film of CA/CAB. Future work will involve more in-depth studies to confirm and better quantify the ability of the CA/CAB film to protect the reference electrode and allow accurate TAN measurements in diluted fuels. 3.3. Acidity testing of non-aqueous solvents Several criteria were taken into account when choosing the proper solvent for our experiments: (a) it had to be miscible with fuel, (b) acidic species present in the fuel had to be soluble in it, and (c) it had to be a “differentiating” solvent. The term “differentiating” means that it is a solvent that can differentiate the strengths of acids (or bases) that are equivalent in water. The solvation of H+ differs from one solvent to another and, even in a solution of 1 M hydrogen ion, the activity of H+ differs drastically by solvent. In order to compare the acid–base properties in different solvents, it is convenient to define a theoretical pH scale that is common to various solvents. The pH scale in water is used as a reference to define the theoretical pH windows in various solvents in such a common pH scale [25]. If the solvent is a weaker base than water, the pH window expands below pH 0 (more acidic than water). On the other hand, if the solvent is of weaker acid than water, the pH window expands above pH 14 (more basic than water). These expanded pH windows open up various chemical possibilities: (1) In solvents with an expanded pH region below pH 0, the solvated protons (SH2 + ) have a very strong acidity; some acids, which behave as strong acids in water, tend to behave as weak acids of varying strengths. Thus, they can be determined separately by titration. Moreover, in such a solvent, some bases, which are too weak to titrate in water, can be titrated and their strengths can be determined. (2) In solvents with an expanded pH region above pH 14, the lyate ion (S− ) has very strong basicity and some bases, which behave as strong bases in water, tend to behave as weak bases of different strengths. They can be determined separately by titration. Moreover, in such a solvent, some acids, which are too weak to titrate in water, can be titrated and their strengths can be determined. Acetonitrile as well as isopropanol belong to the group of “differentiating” solvents and have expanded pH ranges

J. Widera et al. / Sensors and Actuators B 130 (2008) 871–881

compared to water solution [25]. According to the above consideration, both acetonitrile and isopropanol seemed to be good choices for a solvent. Initially, acetonitrile was used as a solvent into which all the fuel samples studied were diluted to overcome the conductivity limitations of the jet fuel. However, acetonitrile is only partially miscible with fuel and does not form a true solution; in subsequent measurements isopropanol was used to improve solubility of the jet fuel. The mixing ratio of all of the aliquots was 1:10 (fuel:solvent). All of the IrOx electrodes tested displayed good repeatability. The potential difference across three different IrOx sensors was approximately 15 mV (data not shown). Exposure of the IrOx sensor to acetonitrile solution containing 0.1 M phenol resulted in a potential shift of approximately +13 mV (data not shown). This suggests that the IrOx sensor responds to compounds having weak acid–base character (like phenol). Phenols are a representative group of compounds known to be present in fuel and play an important role in the thermal stability of fuel [24]. 3.4. TAN fuel testing Jet fuel is a complex mixture containing hundreds of organic species including alkanes, naphthenes and aromatics. Some other polar heteroatomic species, such as phenols, sulfur, and nitrogen species are present in extremely low concentrations (1%); however, many of the important properties of fuel result from their presence. These properties include thermal stability (the ability to resist the formation of surface deposits upon thermal exposure), lubricity, oxidizability, storage ability, and electrical conductivity [5,26,27]. Substantial chemical changes occur in the jet fuel upon thermal exposure in the presence of dissolved oxygen. These changes are not well understood and include the formation of surface and bulk deposits, peroxides, acids, as well as other oxygenates, such as aldehydes, ketones, and alcohols [28]. TAN values are routinely measured on purchased fuels to verify that thermal degradation has not occurred and the fuel is of otherwise high quality. Some amount of thermal degradation also normally occurs in aircraft fuel systems [29–31]. Testing unstressed as well as thermally stressed fuels provides the ability to follow the changes in fuel composition and properties. Information about formation and/or removal of the species during thermal oxidative exposure is invaluable in obtaining a better understanding of the autooxidative mechanism which results in the formation of insolubles and deposits. The results of the TAN fuel testing measurements are shown in Fig. 1. The signal increase is generated by the presence of acidic species in the samples, interacting with the IrOx sensor. The IrOx sensor shows repeatable results for replicate measurements in both unstressed (A) and stressed (B) fuels, respectively. The potential changes recorded after 1 h of measurement were similar as follows: A1 = 0.0241 V, A2 = 0.0240 V, A3 = 0.0238 V and B1 = 0.0411 V, B2 = 0.0417 V and B3 = 0.0412 V, respectively. These measurements also allow unstressed and stressed fuels to be differentiated. The potential shift for the sample containing stressed fuel is larger, consistent with the fact that

875

Fig. 1. IrOx sensor potentiometric response to two kinds of samples: A and B each containing 10 mL of acetonitrile and 1 mL of fuel type 2747 unstressed or stressed, respectively. A1, A2, A3, B1, B2, B3—consecutive runs on samples A and B, respectively. The IrOx electrode was encapsulated in epoxy resin. Reference electrode: Ag/AgCl electrode sealed in Teflon tubing terminated with a ceramic frit. The internal compartment contained acetonitrile and LiCl (1 g/L).

stressed fuel contains more acidic species created during the exposure of fuel to high temperatures. However, the background continued to rise throughout this experiment, probably because the sensor had not been given enough time to equilibrate between solution changes. In order to verify this, measurements of IrOx sensor response time were performed over a longer time frame in six different fuel cocktails: 2747, 3166 and 3804 both unstressed and stressed, respectively. (Fig. 2) These results suggest once again that using the IrOx sensor for off-line analysis, it is possible to measure the acidity of different fuels and distinguish between unstressed and stressed fuels. The unstressed fuels show very similar potential responses (Table 1). This is expected, because the TANs for these fuels as determined by the ASTM method were all very similar (Table 2). On the other hand, the stressed fuels would be expected to give higher responses, consistent Table 2 Total acid numbers (TAN) of various fuels determined by 3242 ASTM titration with visual detection end point Fuel type

TAN

2747n 3166n 3804n 3686n 2747s 3166s 3804s 3804n + LT 4493@4000 mg/L 3804n + LT 4596@4000 mg/L 3804n + LT 4708@4000 mg/L 3804n + LT 4708@5000 mg/L

0.001 0.002 0.006 0.04 0.0604 0.1010 0.0421 0.2488 0.1147 0.1729 0.2111

n—unstressed fuel, s—thermally stressed fuel, fuel that was exposed to 140 ◦ C for 5 h, LT—low temperature additive.

876

J. Widera et al. / Sensors and Actuators B 130 (2008) 871–881 Table 3 Typical response times:time-constants (s) Response times–time-constant (s) Aqueous

Epoxy Shrink tube Bare

5 5 1

Non-aqueous (fuel) CA/CAB–Ag/AgCl

Single junction

3600 120 20

3600 300 120

Non-aqueous test solution consisted of isopropanol solution containing TBATFB (1 g/L) and 1 g/L of 3804 stressed fuel. The single junction reference electrode was a Ag/Ag2 O reference electrode which was sealed in Teflon tubing terminated with a ceramic frit. The internal compartment contained isopropanol and TBATFB (1 g/L).

Fig. 2. Representative potential vs. time response of an epoxy encapsulated IrOx electrode in the presence of different fuels: 3166, 2747 and 3804 unstressed and stressed fuels (n—unstressed fuel; s—stressed fuel). The IrOx electrode was encapsulated in epoxy resin. Reference electrode: Ag/AgCl electrode sealed in Teflon tubing terminated with a ceramic frit. The internal compartment contained acetonitrile and LiCl (1 g/L).

with their higher TANs due to the increased concentration of acidic species caused by the stressing process and indeed the potential changes for the stressed fuel samples are significantly higher than the potential changes for the same type unstressed fuels (Table 1). The largest, most distinctive response from the fuels was observed for the 3166s thermally degraded fuel sample. This suggests that the 3166s fuel sample contains the largest amount of acidic species. The approximate time-constants for sensor response are presented in Table 1 and range from about 22 to 110 min. It is important to mention that all experiments to this point were done with the IrOx sensor encapsulated in epoxy resin. It is suspected that the long response time of the IrOx sensor in nonaqueous solution may not be inherent to the sensor, but rather somehow associated with the encapsulation material or the fact that fuel is not completely miscible with acetonitrile. In order to test this the time-constants (aqueous and non-aqueous) for sensor response were determined on bare unencapsulated sensors and sensors encapsulated with epoxy or double-walled shrink tubing. Sensors were tested in both aqueous buffers and isopropanol containing the supporting electrolyte TBATFB. Either aqueous acid/base solutions or fuel was added to shift the pH or TAN, respectively. Table 3 shows the results of this study. The shortest response time (1 s time-constant), was measured for the bare sensor in aqueous solution. Encapsulation of the IrOx sensor using either epoxy resin or shrink tubing significantly slows the response time of the sensor in both aqueous and non-aqueous solutions. All of the sensor types mentioned above responded much faster in aqueous solution than in non-aqueous solutions. This is expected because of the lower conductance of the non-aqueous

solutions and the potential complications associated with the fact that the sensor response may depend on or involve a hydrated iridium oxide layer. Other related and important factors include the specific design of the reference electrode and the indicating electrode encapsulation material. Specifically, shrink tubing appears to be a much better encapsulation material then epoxy, especially for the application in organic solvents/jet fuel. The shrink tubing encapsulated iridium oxide electrode in conjunction with a single junction Ag/AgCl reference electrode results in a response time-constant of 300 s and in combination with the CA/CAB–Ag/AgCl reference electrode a response timeconstant of only 120 s is obtained. In non-aqueous environments the communication between the IrOx indicating electrode and Ag/AgCl electrode can be severely compromised which contributes to very long response time values. This is seen especially in the case of using a common single junction reference electrode. When a simpler quasireference electrode is used, i.e. a CA/CAB coated Ag/AgCl electrode, a significant decrease of the response time values are observed (Table 3). This can be explained by comparing the conductance measured across the two types of reference electrodes (Table 4). The conductance across the single junction frit is very low, only 0.01 ␮, while the conductance across the CA/CAB coated Ag/AgCl electrode is at least two orders of magnitude larger (2.9 ␮). Thus, because of its high conductance the CA/CAB coated Ag/AgCl electrode may serve better for non-aqueous applications than the regular single junction Ag/AgCl reference electrode. However, the suitability of the CA/CAB coated Ag/AgCl quasi-reference electrode for general application in diluted fuel measurements needs to be verified further. Regardless of the type of reference electrode used for Table 4 Representative conductance values measured between Pt disc electrodes and between a Pt disc electrode and the two reference electrodes in isopropanol solution containing 1 g/L of electrolyte TBATFB Electrode 1–Electrode 2 system

Conductance (␮)

Pt–Pt Pt–CA/CAB/Ag/AgCl Pt–Ag/Ag2 O single junction

3.4 2.9 0.01

In all cases, 1 mm diameter Pt disc electrodes (BAS) were used. Measurements were done using a YSI Model 35 conductance meter.

J. Widera et al. / Sensors and Actuators B 130 (2008) 871–881

the measurements, shorter response times are observed for bare IrOx sensors than for encapsulated sensors. Thus, the shrink tubing encapsulated IrOx as an indicating electrode and the CA/CAB–Ag/AgCl as reference electrode, appear to be the best set of electrodes for obtaining rapid response times in fuel acidity measurements. All further experiments were run using an IrOx sensor encapsulated in shrink tubing and isopropanol as a diluent. For the next set of experiments, isopropanol was used as the diluent and the sensor probe consisted of the shrink tubing encapsulated IrOx (IE) and Ag/Ag2 O (RE), which was sealed within a short length of Teflon tubing terminated with a ceramic frit. The internal compartment of the RE salt bridge contained isopropanol and TBATFB (1000 mg/L). (A single junction RE electrode was used here as opposed to the CA/CAB coated Ag/AgCl RE to ensure that any contaminants that might affect the performance of the reference electrode were excluded.) The response of this sensor to the fuel–solvent mixture is reproducible within 3 mV (Fig. 3). The sensor response for the third run was calculated from the asymptote. Subsequently, the response of the IrOx sensor to fuels of various acidity was checked (Fig. 4) using the same experimental set-up and protocol as that used for the data displayed in Fig. 3. The potential shift recorded with the IrOx sensor shows good, linear correlation with the TAN’s of the fuels studied (r2 = 0.9840) (Fig. 5). Similar behavior was observed when various amounts of the same fuel (3804 with LT4493 at 4 g/L) were added to the tested mixture. Representative measurements showing the response of IrOx versus Ag/AgCl dip coated with CA/CAB to different amounts of the same fuel type added to the mixture (isopropanol and TBATFB) are presented in Fig. 6, where

Fig. 3. Reproducibility studies on the IrOx electrode response to the fuel:solvent mixture (1:10), where the fuel was 3686 unstressed and the solvent was isopropanol containing 1 g/L of TBATFB supporting electrolyte. All potentials are measured vs. a Ag/Ag2 O reference electrode, which was sealed in Teflon tubing terminated with a ceramic frit. The internal compartment contained isopropanol and TBATFB (1 g/L). The IrOx sensor was encapsulated in shrink tubing.

877

Fig. 4. Response of the IrOx sensor to fuels of various acidities. Fuel—various acidity fuels. Solvent—isopropanol + TBATFB (1 g/L); fuel:solvent = 1:10. All potentials are measured vs. a Ag/Ag2 O reference electrode which was sealed in Teflon tubing terminated with a ceramic frit. The internal compartment contained isopropanol and TBATFB (1 g/L). The IrOx sensor was encapsulated in shrink tubing.

the volume of the added fuel was corrected for the dilution. A linear relationship between the potential shift and the volume of fuel added was also observed (r2 = 0.9964) (Fig. 7). The response time of the sensor was much quicker in these measurements compared to the previous experiment (Fig. 4). In the first experiment (Fig. 4), time-constant ≈5 min, while in the latter experiment (Fig. 6) with the novel reference electrode timeconstant ≈1.5 min clearly demonstrating the advantage of using this particular reference electrode design.

Fig. 5. Calibration curve for IrOx sensor response vs. fuel-TAN. Response data from Fig. 4.

878

J. Widera et al. / Sensors and Actuators B 130 (2008) 871–881

Fig. 6. Response of the IrOx electrode vs. Ag/AgCl coated with CA/CAB to different amounts of fuel added to the solvent mixture (isopropanol with 1 g/L TBATFB). Fuel—3804 with low temperature additive 4493 at 4 g/L TAN = 0.2488. The IrOx sensor was encapsulated in shrink tubing.

3.5. Titration experiment IrOx sensors can be also used instead of a glass electrode as a potentiometric end point titration detector in the ASTM 664 method [22]. For comparison, we present the results of TAN determined for the same fuel (3804n fuel containing low temperature additive 4708 at 4000mg/L) by three independent methods: TAN = 0.1719 determined by the ASTM 3242 titration method with visual end point [21], TAN = 0.1721 determined by the ASTM 664 potentiometric titration method using a glass electrode [22] and TAN = 0.1714 determined by a modified ASTM 664 method—potentiometric titration using the IrOx sensor. All of these methods give comparable results demonstrating that the

Fig. 7. Calibration curve of IrOx sensor response vs. corrected volume of the added fuel. Response data from Fig. 6.

IrOx electrode can be reliably used instead of a glass electrode in the potentiometric titration of fuel samples. The graphs comparing the response of the IrOx sensor and the glass electrode during the potentiometric titration performed following the ASTM 664 protocol are shown in Fig. 8. The titration was done at two different rates: 0.05 and 0.5 mL/min. As shown in Fig. 8, in both of the plots: (A)—potential versus volume and (B)—the first derivative plot, the IrOx sensor (dotted–dashed line) performed better than the glass electrode (solid line). The IrOx response is faster, with better peak definition (Fig. 8B), more accurate and more repeatable than the glass electrode response. The apparent end point value determined using the glass electrode depends on the speed of titrant addition. Table 5 summarizes the results from Fig. 8A and B by comparing the calculated TAN’s for the same

Fig. 8. Comparison of IrOx sensor vs. glass electrode response during the titration process. All plots of potential vs. volume were normalized. The IrOx sensor was encapsulated in shrink tubing. Reference electrode: the internal Ag/AgCl electrode within the combination glass pH electrode.

J. Widera et al. / Sensors and Actuators B 130 (2008) 871–881 Table 5 TAN’s calculated from the data presented in Fig. 8A and B Indicator electrode

IrOx Glass pH

Titrant addition rate (mL/min) 0.5

0.05

0.1714 0.2131

0.1714 0.1721

fuel sample recorded with use of either the glass electrode or the IrOx electrode. These results show the superiority of performance of the IrOx electrode over the commercial glass electrode. Two different values for the end point at two different speeds of titration were obtained when the glass electrode was used while only one accurate, reproducible value of titration volume was obtained using the IrOx electrode using the same experimental conditions (Table 5). The same TAN value (0.1714) was obtained at both titration rates: 0.5 and 0.05 mL/min using IrOx as an indicating electrode. In the case of the commercial glass pH electrode, two different TAN values were obtained, 0.1721 and 0.2141, at the titration rates of 0.05 and 0.5 mL/min, respectively. These results demonstrate that the response of the IrOx electrode is fast and independent of the titration rate (Fig. 8B). The discrepancies caused by the slow response of the glass electrode indicate that it has limited utility for work in non-aqueous solutions. The slow response of the glass electrode has been attributed to the presence of a hygroscopic surface layer on the glass bulb, where the equilibrium of protons between the inside and outside of the glass electrode is established [32]. This is not an issue for the IrOx electrode, which does not have such a layer. Fig. 9 presents comparative graphs of the responses of the shrink tubing encapsulated IrOx indicating electrode with a commercially available glass pH electrode (ACCUMET) in isopropanol solution containing TBATFB (1 g/L) and 1 g/L of 3804 stressed fuel. The potential of both electrodes was measured versus the CA/CAB covered external Ag/AgCl reference electrode.

879

Clearly, the performance of the shrink tubing encapsulated IrOx electrode is much better than the response of the commercially available pH electrode. The superiority of the IrOx electrode can be seen in all aspects, the response time is faster, the drift is negligible and the noise much lower than in the case of the glass pH electrode. The actual pH sensing mechanism of iridium oxide-based electrodes has not been confirmed yet, however several possible models have been proposed [13,16,33–41]. The potentiometric response of the IrOx film to pH is related to film composition and possibly morphology [16,41,42]. The general reaction for the pH sensitivity of iridium oxide electrodes, representing the transition between two oxidation states of an electroactive couple (Ir(III) oxide and Ir(IV) oxide) can be written as follows: Ir(IV)oxide + xH+ + ne−  Ir(III)oxide + yH2 O

(1)

where n, x and y are stoichiometric coefficients and their values vary with the oxide preparation method and are usually not integers [33,43,44]. The same mechanism is believed to be involved in both the aqueous (pH) and non-aqueous response (TAN) of the sensor [13]. The composition of the iridium oxide film that is predetermined by the specific oxide formation process determines the mechanism of response. Anhydrous iridium oxide layers (IrO2 ) created by sputtering or thermal oxidation respond to pH changes according to the following mechanism [34–37]: IrO2 + H+ + e−  IrO · OH

(2)

or 2IrO2 + 2H+ + 2e−  Ir2 O3 + H2 O

(3)

The pH response mechanism of the hydrated, electrochemically grown iridium oxide layer can be most likely described by the following reaction [37]: 2[IrO2 (OH)2 · 2H2 O]2− + 3H+ +2e−  Ir2 O3 (OH)3 · 3H2 O]3− + 3H2 O

(4)

Buck and co-worker [38] proposed a very general model for the potential–pH response of metal oxide electrodes (MOE’s) called “oxygen intercalation” and is represented by the following equilibrium: MOx + 2δH+ + 2δe−  MOx−δ + δH2 O

Fig. 9. Comparison of the responses of a shrink tubing encapsulated IrOx indicating electrode with a commercially available glass pH electrode (ACCUMET) in isopropanol solution containing TBATFB (1 g/L) and 1 g/L of 3804 stressed fuel. Fuel was added at the point indicated by the arrows. The potential of both electrodes was measured vs. a CA/CAB covered external Ag/AgCl reference electrode.

(6)

where MOx is a higher oxidation state of a metal oxide and MOx−δ is a lower oxidation state of a metal oxide. Additionally, the nature of the iridium oxide film and consequently its electrochemical behavior can be considerably influenced by the insertion of lithium ions and the hydration of the oxide. Both processes, lithiation and hydration of the oxide film, make an important contribution to both the chemical and response stability of the IrOx pH electrode used in these studies. According to the “oxygen intercalation” mechanism, the equilibrium of the Li–iridium oxide couple in solution can be expressed by the following equation [16]: Lix IrOy + 2δH+ + 2δe−  Lix IrOy−δ + δH2 O

(7)

880

J. Widera et al. / Sensors and Actuators B 130 (2008) 871–881

where Lix IrOy is a higher oxidation state and Lix IrOy−δ is an oxygen deficient phase with a lower oxidation state. Buck and co-worker [38] have described in detail the requirements of a pH sensor. Although their discussion concerned the application of the pH sensor in aqueous solutions, their conclusions are valid for a pH sensor in non-aqueous media. According to Buck and co-worker, the surface of an ideal electrochemical ion-exchange pH sensor together with the external solution should provide a surface of nearly uniform composition that permits the rapid exchange of protons and the film or bulk phase should act as an ionic semiconductor without electronic conductivity. The reactive surface layer should reach equilibrium rapidly, and ion-exchange by other ions should not be energetically feasible. Also, it should be free from redox interferences. 4. Conclusions A novel IrOx electrode was used to develop an off-line sensor for monitoring the acidity of fuel. The sensor is accurate, stable, robust, has direct reading capability and requires only a very small volume of sample. The data obtained show that the IrOx sensor responds to compounds having acid–base character, such as phenol. This IrOx sensor is capable of determining the acidity of different fuels and can distinguish between unstressed and thermally stressed fuels. The potentiometric response of the IrOx sensor shows excellent linear correlation with the TANs of the fuels studied. During the IrOx sensor development process we identified and addressed a number of sources of slow sensor response, among them: poor fuel solubility, low conductance of the solution, nature of the sensor encapsulation material and design of the reference electrode. Isopropanol was used instead of acetonitrile as a solvent because fuel components were found to be more soluble in acetonitrile. Shrink tubing was used instead of epoxy resin in order to minimize fuel uptake by the encapsulation material, which contributed to slow sensor response. And finally, the reference electrode design was modified in order to obtain a faster response. In particular, the traditional frit was replaced with a special thin film coating made of CA/CAB. This reference electrode design allows faster response due to the higher conductance across the CA/CAB film versus the frit of the single junction design. However, the general suitability of this quasi-reference electrode for fuel based measurements needs to be further verified. All of these factors helped us to greatly improve the speed of response of the IrOx sensor. All of the above makes this IrOx based sensor a very useful tool for applications, such as specification testing, thermal stability monitoring, and as an R&D tool for additive development and for process control applications. Future efforts will focus on experiments aimed on designing an in-line sensor for direct measurements in undiluted fuel. This will involve explorations of the effect of the acid on the IrOx sensor over a longer testing period (both consecutive testing and several tests over an extended period of time) to predict the lifetime and long-term quality of the sensor.

Acknowledgements The authors are grateful to SenIrOx for providing the IrOx electrodes for these studies. This work has been supported through funding from the Air Force Research Laboratory, PRTG (Mr. William E. Harrison III). References [1] Aviation Fuels, Technical Review 2000, Chevron Products Company. [2] J.P. Allinson (Ed.), Criteria for Quality of Petroleum Products, 1973. [3] R.L. Johnson, J.C. Fitch, Handbook of Lubrication and Tribology, second ed., vol. 1, 2006, pp. 35/1–35/11. [4] CRC Report No. 635, Handbook of Aviation Fuel Properties, 2004 third ed., Coordinating Research Council, Inc. [5] Aviation Fuels, Technical Review 2004, Chevron Products Company. [6] Final Report of (PQIS) Petroleum Quality Information System, Defense Energy Support Center Product Technology and Standardization Division (DESC-BP), 2004. [7] Y.J. Kim, Y.C. Lee, B.K. Sohn, C.S. Kim, J.H. Lee, Electron. Lett. 39 (21) (2003) 1515–1516. [8] I. Walther, B. Van der Schoot, M. Boillat, A. Cogoli, Enzyme Microb. Technol. 27 (10) (2001) 778–783. [9] K. Fawcus, V.J. Gorton, M.L. Lucas, G.T.A. Mcewan, Comparative biochemistry and physiology, A: Physiol. 118A (2) (1997) 291–295. [10] K.H. Gan, W.P. Geus, C.B. Lamers, H.G. Heijerman, Dig. Dis. Sci. 42 (11) (1997) 2304–2309. [11] P. Lingstrom, T. Imfeld, D. Birkhed, J. Dent. Res. 72 (5) (1993) 865– 870. [12] S.v. Rithalia, A.J. Hewer, J. Tinker, P. Herbert, J. Biomed. Eng. 2 (2) (1980) 126–128. [13] K. Izutsu, H. Yamamoto, Anal. Sci. 12 (1996) 905–909. [14] H. Galster, pH Measurements—Fundamentals, Methods, Applications, Instruments, VCH Publishers, New York, 1991. [15] C. Liu, B. Bocchicchio, P. Overmeyer, M. Neuman, Science 207 (1980) 188–189. [16] S. Yao, S.M. Wang, M. Madou, J. Electrochem. Soc. 148 (4) (2001) H29–H36. [17] J.V. Dobson, P.R. Snodin, H.R. Thirsk, Electrochim. Acta 21 (1976) 527. [18] T. Katsube, I. Lauks, J.N. Zemel, Sens. Actuators 2 (1982) 399. [19] I. Lauks, M.F. Yuen, T. Dietz, Sens. Actuators 4 (1983) 375. [20] M.F. Smiechowski, V.F. Lvovich, Sens. Actuators 96 (2003) 261. [21] Annual Book of ASTM Standards, American Society for Testing and Materials (ASTM), 2001, D3242. [22] Annual Book of ASTM Standards, American Society for Testing and Materials (ASTM), 2001, D664. [23] J.M Johnson, J.L. Huntington, US Patent 5,766,839, (June 16, 1998). [24] L.M. Balster, S. Zabarnick, R.C. Striebich, Petroleum Chemistry Division Preprints 47(3) (2002) 161–164. [25] K. Izutsu, Electrochemistry in Nonaqueous Solutions, Wiley-VCH Verlag GmbH, Weinheim, 2002, 59–83. [26] N.W. Lundine, Oil Gas J. 65 (38) (1967) 95, 90. [27] M.W. Badger, S. Zabarnick, G.R. Wilson, Preprints—American Chemical Society, Division Petroleum Chem. 47 (3) (2002) 170–173. [28] N.J. Kuprowicz, S. Zabarnick, Z.J. West, J.S. Ervin, Energy Fuels 21 (2) (2007) 530–544. [29] S.P. Heneghan, S. Zabarnick, D.L. Ballal, W.E. III Harrison, Proceedings of the Intersociety Energy Conversion Engineering Conference 31st, 1996, pp. 63–68. [30] B. Grinstead, S. Zabarnick, Energy Fuels 13 (3) (1999) 756–760. [31] J.S. Ervin, S. Zabarnick, T.F. Williams, J. Energy Resour. Technol. 122 (4) (2000) 229–238. [32] R.G. Bates, Determination of pH. Theory and practice, John Wiley & Sons, Inc., New York, 1964, pp. 289–338. [33] R.K. Jaworski, J.A. Cox, J. Electroanal. Chem. 325 (1992) 111. [34] P. VanHoudt, Z. Lewandowski, Biotech. Bioeng. 40 (1992) 601–608. [35] M.L. Hitchman, S. Ramanathan, Analyst 113 (1988) 35–39.

J. Widera et al. / Sensors and Actuators B 130 (2008) 871–881 [36] P.J. Kinlen, J.E. Heider, D.E. Hubbard, Sens. Actuators B 22 (1994) 13. [37] L.D. Burke, J.K. Mulcahy, D.P. Whelan, J. Electroanal. Chem. 163 (1984) 117–128. [38] A. Fog, R.P. Buck, Sens. Actuators 5 (1984) 137–146. [39] J. Hendrikse, W. Olthuis, P. Bergveld, Sens. Actuators B 53 (1998) 97–103. [40] W. Olthuis, M.A.M. Robben, P. Bergveld, M. Bos, W.E. Van Der Linden, Sens. Actuators B 2 (1990) 247–256. [41] D-c. Chen, J-s. Zhen, C-y. Fu, Trans. Nonferrous Met. Soc. China 13 (6) (2003) 1459–1464. [42] M. Wang, S. Yao, Electroanalysis 15 (20) (2003) 1606–1615. [43] J.E. Baur, T.W. Spaine, J. Electroanal. Chem. 443 (1998) 208. [44] S. Glab, A. Hulanicki, G. Edwall, F. Ingman, Crit. Rev. Anal. Chem. 21 (1989) 29.

Biographies Justyna Widera received her M.Sc. degree in Chemistry from University of Lodz (Poland) in 1995 and her Ph.D. degree in Chemistry from University of Warsaw (Poland) in 2000. For the past 3 years, she has worked very closely with the Air Force Research Laboratory, at Wright-Patterson Air Force Base as a Member of the Fuel Branch Research Team being contracted through ISSI and University of Dayton Research Institute. Her research was focused on development of electrochemical and optical-based sensors and nanoscale fuel additives. She is presently working as an assistant professor at Department of Chemistry, Adelphi University. Her research interests include electroanalytical chemistry, nanotechnology and sensors. Bill L. Riehl received his B.S. degree in Chemical Engineering in 2003 from the University of Dayton and his M.S. in Materials Engineering in 2005 from the

881

University of Dayton. Currently, he is finishing his Ph.D. in Materials Engineering at the University of Dayton. Bill currently holds the Chief Operating Officer position at Riehl Engineering, Ltd., a chemical and materials engineering consulting and design firm. His research interests include chemical and biological sensing, corrosion engineering, electrochemistry, mass transport, thermodynamics and carbon nanotechnology. Bill also holds a Chair at Phoenix NanoSystems LLC, a start up nanotechnology position, as well as holding a professional engineers license in multiple states where he practices chemical, materials and structural engineering. Jay M. Johnson received his M.Sc. degree in Chemistry from Wright State University in 1975 and his Ph.D. degree in Chemistry from the University of Cincinnati in 1981. He spent 27 years working at YSI, Inc. where he held a number of positions involving chemical sensor development and R&D management. He is presently working as a Senior Research Scientist at the University of Dayton Research Institute, Division of Materials Engineering. He is also an Assistant Professor at the School of Chemical and Materials Engineering at the University of Dayton. His research interests include nanotechnology, nanobiotechnology, biomaterials, electroanalytical chemistry, chemical sensors and biosensors. Douglas C. Hansen received his Ph.D. degree from the University of Delaware in 1993. He is currently a Senior Research Scientist and Leader of the Materials Degradation & Electrochemical Engineering Group at the University of Dayton Research Institute (UDRI) Materials Engineering Division. Doug has over 15 years experience in the field of corrosion electrochemistry, and has more than 7 years industrial experience in electrochemical instrumentation design and development. His areas of expertise include surface analytical techniques, such as SEM, FTIR, Scanning Vibrating Probe (SVP) and Scanning Kelvin Probe (SKP) techniques, Localized Electrochemical Impedance Spectroscopy (LEIS) and electrochemical measurement (AC/DC) techniques.