sensors Article

Characterization of Industrial Coolant Fluids and Continuous Ageing Monitoring by Wireless Node—Enabled Fiber Optic Sensors Alexandros El Sachat 1,2 , Anastasia Meristoudi 1 , Christos Markos 1,3 , Andreas Sakellariou 4 , Aggelos Papadopoulos 5 , Serafim Katsikas 4 and Christos Riziotis 1, * 1

2 3 4 5

*

National Hellenic Research Foundation, Theoretical and Physical Chemistry Institute, Photonics for Nanoapplications Laboratory, Athens 11635, Greece; [email protected] (A.E.S.); [email protected] (A.M.); [email protected] (C.M.) Catalan Institute of Nanoscience and Nanotechnology (ICN2), CSIC and the Barcelona Institute of Science and Technology, 08193 Barcelona, Spain Technical University of Denmark, Department of Photonics Engineering, DTU Fotonik, Kgs. Lyngby 2800, Denmark PRISMA Electronics S.A., Research & Development Department, Industrial area of Alexandroupolis, Alexandroupolis 68132, Greece; [email protected] (A.S.); [email protected] (S.K.) Kleemann S.A., Kleemann Group of Companies, Kilkis Industrial Area, Kilkis 6100, Greece; [email protected] Correspondence: [email protected]; Tel.: +30-210-7273887; Fax: +30-210-7273794

Academic Editor: Elfed Lewis Received: 7 December 2016; Accepted: 9 March 2017; Published: 11 March 2017

Abstract: Environmentally robust chemical sensors for monitoring industrial processes or infrastructures are lately becoming important devices in industry. Low complexity and wireless enabled characteristics can offer the required flexibility for sensor deployment in adaptable sensing networks for continuous monitoring and management of industrial assets. Here are presented the design, development and operation of a class of low cost photonic sensors for monitoring the ageing process and the operational characteristics of coolant fluids used in an industrial heavy machinery infrastructure. The chemical, physical and spectroscopic characteristics of specific industrial-grade coolant fluids were analyzed along their entire life cycle range, and proper parameters for their efficient monitoring were identified. Based on multimode polymer or silica optical fibers, wide range (3–11) pH sensors were developed by employing sol-gel derived pH sensitive coatings. The performances of the developed sensors were characterized and compared, towards their coolants’ ageing monitoring capability, proving their efficiency in such a demanding application scenario and harsh industrial environment. The operating characteristics of this type of sensors allowed their integration in an autonomous wireless sensing node, thus enabling the future use of the demonstrated platform in wireless sensor networks for a variety of industrial and environmental monitoring applications. Keywords: ageing monitoring; photonic sensors; industrial coolants; sol-gel; optical fibers; predictive maintenance; wireless sensors

1. Introduction There is a growing interest on the development of new techniques for efficient infrastructure monitoring and assets maintenance management in different industrial sectors, further to support calendar-based maintenance which is now a widely used industrial standard procedure. Predictive maintenance [1,2] suggests a viable solution, recently emerged, which is based on the continuous Sensors 2017, 17, 568; doi:10.3390/s17030568

www.mdpi.com/journal/sensors

Sensors 2017, 17, 568

2 of 20

monitoring and assessment of the real conditions and the extraction of results regarding the optimum time plan and actions for maintenance [2,3]. Implementation of such predictive maintenance systems requires efficient and reliable sensing devices that could be interconnected in a network of sensors collecting and handling the information and feeding this to a decision making system for appropriate maintenance actions [2]. Photonics technology is currently a quite favorable solution for the development of environmentally robust sensors [2] with operational safe characteristics, even in demanding explosive or flammable environments [4] and which also exhibit electromagnetic interference immunity (EMI), crucial to a number of industrial applications. Certain photonics technologies [5] mainly focused on large core optical fibers, have demonstrated some unique characteristics such as low cost, easy handling, compatibility with inexpensive transceivers, and also compatibility for easy and rapid customization [6,7]. Combining such low cost fibers functionalized with sensitive materials [8,9] is a realistic way for the development of reliable yet affordable sensing heads that could be employed as reusable or disposable sensors. Further to the above requirements there are also certain operational requirements regarding the interrogation complexity, circuitry and transceiver complexity that should be kept at a minimum level of cost. To avoid complex interrogation systems operating in the frequency domain, for example, by employing components with Bragg gratings [10], amplitude interrogation is proposed as an effective balance between accuracy and complexity. Further, overcoming the need of expensive stabilized laser sources, the use of non-coherent LED sources and inexpensive photodetectors provide solutions for sensor integration in low power consumption autonomous sensing nodes. The viability of the specific sensing platform in wireless-based systems for predictive maintenance and management is demonstrated here based on an indicative and rather challenging application of coolants fluids’ ageing monitoring [11] in a harsh industrial environment. Such autonomous and embedded sensors equipped with wireless interconnectivity capabilities [12–14] are expected to play a key role in the development of future Internet of Things (IoT) applications [15,16]. More specifically, the demonstration case in this work is focused on monitoring the ageing process of coolant fluids [17,18] in industrial metal processing of metallic cylinders (pistons) used in hydraulic elevators in the industrial collaborator Kleemann S.A. (Kilkis, Greece). Different machines for grinding (BOSSI s.r.l, Milano, Italy) and for polishing with custom made Honnen machine (Kleemann S.A., Kilkis, Greece) are employed for metal processing to strict manufacturing requirements (Figure 1). For the dissipation of the heat and the maintenance of constant lubricating action the use of coolant fluids is essential [19]. Depending on the coolant type there are specific conditions for their proper operation in order to provide the desired cooling and lubricating actions together with anticorrosion and antibacterial protection [18]. Failure to maintain proper cooling action can lead to significant deterioration of metal processing quality, leading in turn to damaged metallic parts or broken machine tools, with the corresponding time delays in the production line and the associated often very serious economic costs. Additionally, during the metal grinding process, the change of coolant’s characteristics can lead to development of bacterial or microbial biohazardous load [20] where at a certain level of ageing becomes non-usable, due also to serious intolerance by the personnel and other associated health safety issues [21]. Despite the fact that there are no systematic studies in the literature for quantifying the coolants’ ageing process or providing continuous monitoring strategies, there is in contrast a large number of studies on the toxicity and health effects of coolants associated with skin irritation and respiratory effects to even more severe pulmonary complications [21,22]. Therefore the continuous monitoring of coolants is very important not only for guaranteeing end-manufacturing specifications but also for meeting the strict environmental and health issues regulations. Furthermore at the end of a coolants life-cycle when it becomes unusable there arises in coolants’ management process the issue of their proper disposal and replacement. This is also directly related to environmental and economic issues associated with optimized fluid waste handling policies [23] and efficient storage and logistic processes. Additionally the storage tanks of pristine coolants need also to be monitored as they are highly susceptible to deterioration due to potential development of bacterial load under

Sensors 2017, 17, 568

3 of 20

certain environmental storage conditions of temperature, humidity or atmospheric pollutants [22]. Sensors 2017, 17, 568 3 of 19 The presence of different metal processing machines in the same large scale infrastructure together with remote sites of storage wastesites coolant tanks create the need of a coordinated infrastructure together withand remote of storage and waste coolant tanks create monitoring the need ofbya wireless sensor networks for providing the overall combined condition of coolants and allowing the coordinated monitoring by wireless sensor networks for providing the overall combined condition efficient management by a decision making system. of coolants and allowing the efficient management by a decision making system.

Figure 1. A Honnen metal polishing machine (the inset demonstrates a typical required quality of the Figure 1. A Honnen metal polishing machine (the inset demonstrates a typical required quality of the produced metal surfaces with special line features on it.) Detail of the polishing head with the produced metal surfaces with special line features on it.) Detail of the polishing head with the presence presence of coolant flow. Coolant’s exit from the grinding machine and detail of coolants’ initial of coolant flow. Coolant’s exit from the grinding machine and detail of coolants’ initial filtering and filtering and transient temporary storage at the tanks. transient temporary storage at the tanks.

In terms now of the actual sensing strategy the key point is to identify if possible a simple In terms now of the actual sensing strategy the key point is to identify if possible a simple parameter approach in order to monitor the condition of such chemically complex multi-component parameter approach in order to monitor the condition of such chemically complex multi-component coolant fluids. As will be shown later the acidity (pH) can be related very accurately to the condition coolant fluids. As will be shown later the acidity (pH) can be related very accurately to the condition of the coolant. However the operation of a sensor in direct contact with coolant fluids has not been of the coolant. However the operation of a sensor in direct contact with coolant fluids has not been systematically evaluated before and there is no prior information on the actual response of similar systematically evaluated before and there is no prior information on the actual response of similar sensors. The behavior of such sensors is expected to be much different in real coolant fluids as sensors. The behavior of such sensors is expected to be much different in real coolant fluids as compared to standard solutions of controllable pH. Different large core optical fiber sensors were compared to standard solutions of controllable pH. Different large core optical fiber sensors were characterized in coolants monitoring and successful implementations were identified here as robust characterized in coolants monitoring and successful implementations were identified here as robust enough in terms of lifetime and repeatability. enough in terms of lifetime and repeatability. Preliminary results of coolants characterization and sensors behavior were previously reported Preliminary results of coolants characterization and sensors behavior were previously [11], while here a thorough analysis leading to the optimization of the sensors response that enabled reported [11], while here a thorough analysis leading to the optimization of the sensors response also their integration in low power driving circuits boards is presented. By properly selecting low that enabled also their integration in low power driving circuits boards is presented. By properly power optical transceivers it was possible the integration of fiber optic sensor in an autonomous selecting low power optical transceivers it was possible the integration of fiber optic sensor in an wireless sensing node in a ZigBee wireless sensor network. This functionality can enable the efficient autonomous wireless sensing node in a ZigBee wireless sensor network. This functionality can enable monitoring and management of the entire cycle of coolants’ lifetime, from storage, to operation and the efficient monitoring and management of the entire cycle of coolants’ lifetime, from storage, to finally to the disposal phase, in a large scale industrial infrastructure. operation and finally to the disposal phase, in a large scale industrial infrastructure. The paper is organized in the following way: first the detailed chemical, physical and The paper is organized in the following way: first the detailed chemical, physical and spectroscopic spectroscopic characterization of the coolants aging process is presented in order to reveal the characterization of the coolants aging presented in to reveal the ageing mechanism ageing mechanism and identify the process suitableisparameter fororder efficient monitoring, thus allowing and the identify the suitable parameter for efficient monitoring, thus allowing the design of appropriate sensing design of appropriate sensing components. Then is presented the fiber optic sensors development by components. Then is presented the fiber optic sensors development by specially designed coating of specially designed coating of sol-gel derived pH sensitive materials, followed by the characterization sol-gel derived pH sensitive materials, followed by the characterization sensors’ of sensors’ performance in standard ideal solutions and in real coolant of fluids for aperformance large range in of standard ideal solutions and in real coolant fluids for a large range of ageing grade and ageing grade and associated pH index. Finally the circuitry integration of the sensors in associated a wireless pH index. Finally the circuitry integration of the sensors in a wireless sensing node is discussed sensing node is discussed and demonstrated. and demonstrated.

Sensors 2017, 17, 568 Sensors 2017, 2017, 17, 17, 568 568 Sensors Sensors 2017, 17, 568

4 of 19 of 19 19 44 of 4 of 20

2. Materials and Methods: Coolants Characterization and Sensing Approach 2. Materials Materials and and Methods: Methods: Coolants Coolants Characterization Characterization and and Sensing Sensing Approach Approach 2. 2. Materials and Methods: Coolants Characterization and Sensing Approach 2.1. Coolant Fluids’ Study and Characterization 2.1. Coolant Coolant Fluids’ Fluids’ Study Study and and Characterization Characterization 2.1. 2.1. Coolant Fluids’ Study and Characterization 2.1.1. Industrial Coolant and Lubricant Fluids 2.1.1. Industrial Industrial Coolant Coolant and and Lubricant Lubricant Fluids Fluids 2.1.1. 2.1.1. Industrial Coolant and Lubricant Fluids The study has been based on two representative and commercially available coolant fluids The study study has has been been based based on on two two representative representative and and commercially commercially available available coolant coolant fluids fluids The The study has been based on two representative and commercially available coolant fluids widely widely used in industry. Specifically, those coolants were employed in metal grinding and polishing widely used in industry. Specifically, those coolants were employed in metal grinding and polishing widely used in industry. Specifically, those coolants were employed in metal grinding and polishing used in industry. Specifically, those coolants were partner employed in metal grinding and polishing machinery machinery in the infrastructure of the industrial Kleemann S.A., providing thus access to machinery in the the infrastructure infrastructure of the the industrial partner Kleemann S.A., providing providing thus access access to machinery in of industrial partner Kleemann S.A., thus to in the infrastructure of the industrial partner Kleemann S.A., providing thus access to real working real working conditions data and allowing the detailed characterization of coolants aging process. real working conditions data and allowing the detailed characterization of coolants aging process. real working conditions data and allowing the detailed characterization of coolants aging process. conditions datacoolant and allowing detailed characterization of coolants aging process. two specific The two specific liquids used are Syntilo 81 BF by Castrol [24] and Multan 61-3 DF by Henkel The two specific specific coolant liquidsthe used are Syntilo Syntilo 81 BF BF by by Castrol Castrol [24] and and Multan 61-3 The DF by by Henkel The two coolant liquids used are 81 [24] Multan 61-3 DF Henkel coolant liquids used are Syntilo 81 BF by Castrol [24] and Multan 61-3 DF by Henkel [25]. [25]. [25]. [25]. Syntilo ayellow yellowcolored coloredcoolant coolantliquid liquidthat thatpresents presentsaa ahigh highlevel levelof viscosity.It Syntilo 81 BF is viscosity. is Syntilo 8181 BFBF is is yellow colored coolant liquid that presents high level ofofviscosity. viscosity. It It is is Syntilo 81 BF is aaa yellow colored coolant liquid that presents a high level of It is mainly used heavy industrial machinery 8% diluted form. The main ingredient Syntilo mainly used in heavy industrial machinery in an 8% diluted form. The main ingredient of Syntilo mainly used inin heavy industrial machinery inin anan 8% diluted form. The main ingredient ofof Syntilo mainly used in heavy industrial machinery in an 8% diluted form. The main ingredient of Syntilo 81BF is triethanolamine (35%–40%) in addition to several other ingredients such as ethanolamine 81BF is triethanolamine (35%–40%) in addition to several other ingredients such as ethanolamine 81BF is is triethanolamine triethanolamine (35%–40%) (35%–40%) in in addition addition to to several several other other ingredients ingredients such such as as ethanolamine ethanolamine 81BF (1%–5%),3-iodo-2-propynylbutyl 3-iodo-2-propynylbutyl carbamate (0.1%–1%) aswell wellas aspoly(quaternary poly(quaternaryammonium ammonium (1%–5%), carbamate (0.1%–1%) as (1%–5%), 3-iodo-2-propynylbutyl carbamate (0.1%–1%) as well as poly(quaternary ammonium (1%–5%), 3-iodo-2-propynylbutyl carbamate (0.1%–1%) as well as poly(quaternary ammonium chloride) (0.1%–0.25%). Table 1 summarizes the chemical composition Syntilo coolant liquid chloride) (0.1%–0.25%). Table summarizes the chemical composition of Syntilo 81 BF coolant liquid chloride) (0.1%–0.25%). Table summarizes the chemical composition ofof Syntilo 8181 BFBF coolant liquid chloride) (0.1%–0.25%). Table 111 summarizes the chemical composition of Syntilo 81 BF coolant liquid which is commercially available from Castrol. The main ingredients of Multan 61-3 DF by Henkel which is commercially available from Castrol. The main ingredients of Multan 61-3 DF by Henkel which is is commercially commercially available available from from Castrol. Castrol. The The main main ingredients ingredients of of Multan Multan 61-3 61-3 DF DF by by Henkel Henkel which (Table 1) are polyalkylene glycol esters; however, the exact amount (%) of this particular material (Table 1) are polyalkylene glycol esters; however, the exact amount (%) of this particular material is (Table 1) 1) are are polyalkylene polyalkylene glycol glycol esters; esters; however, however, the the exact exact amount amount (%) (%) of of this this particular particular material material is (Table is is not provided in the Henkel specifications. Similarly, this coolant liquid contains a small amount not provided in the Henkel specifications. Similarly, this coolant liquid contains a small amount not provided provided in in the the Henkel Henkel specifications. specifications. Similarly, Similarly, this this coolant coolant liquid liquid contains contains aa small small amount amount not (0.25%–0.3%) pyridine-2-thiol 1-oxide sodium salt, well several inorganicsalts. salts. (0.25%–0.3%) of pyridine-2-thiol 1-oxide sodium salt, as well as several inorganic (0.25%–0.3%) ofof pyridine-2-thiol 1-oxide sodium salt, asas well asas several inorganic salts. (0.25%–0.3%) of pyridine-2-thiol 1-oxide sodium salt, as well as several inorganic salts. Table 1. Chemical composition of industrial grade coolant fluids, Syntilo 81BF and Multan 61-3 DF. Table 1. Chemical composition of industrial grade coolant fluids, Syntilo 81BF and Multan 61-3 DF. Table 1. 1. Chemical Chemical composition composition of of industrial industrial grade grade coolant coolant fluids, fluids, Syntilo Syntilo 81BF 81BF and and Multan Multan 61-3 61-3 DF. DF. Table

Coolant CoolantCoolant Coolant

Material Material Material Material

Quantity (%) Quantity (%) Quantity (%) Quantity (%)

Polyalkylene glycol esters Polyalkylene esters Polyalkylene glycolglycol esters Polyalkylene glycol esters Multan 61-3 DF Multan 61-3 DF Multan 61-3 DF Multan 61-3 DF (Henkel) Pyridine-2-thiol 1-oxide (Henkel) Pyridine-2-thiol 1-oxide Pyridine-2-thiol 1-oxide Pyridine-2-thiol 1-oxide (Henkel) (Henkel) sodium salt sodium salt sodium salt sodium salt

Syntilo 81 BF Syntilo 81 81 BF BF Syntilo (Castrol) Syntilo 81 BF (Castrol) (Castrol) (Castrol)

N/A N/A N/A N/A 0.25–0.3 0.25–0.3 0.25–0.3 0.25–0.3

Inorganic salts InorganicInorganic salts salts Inorganic salts

N/A N/A N/A N/A

Triethanolamine Triethanolamine Triethanolamine Triethanolamine

35–40 35–40 35–40 35–40

Ethanolamine Ethanolamine Ethanolamine Ethanolamine

1–5 1–51–5 1–5

3-Iodo-2-propynyl 3-Iodo-2-propynyl 3-Iodo-2-propynyl 3-Iodo-2-propynyl butylcarbamate butylcarbamate butylcarbamate butylcarbamate Poly(quaternary Poly(quaternary Poly(quaternary ammonium Poly(quaternary chloride) ammonium chloride) ammonium chloride) chloride) ammonium

Chemical Structure Structure Chemical Chemical Structure Chemical Structure

N/A N/A N/A N/A

0.1–1 0.1–1 0.1–1 0.1–1 0.1–0.25 0.1–0.25 0.1–0.25 0.1–0.25

----

Deterioration of coolants in metal grinding applications [18,19] can have several implications in Deterioration coolants in metal grinding applications [18,19] several implications Deterioration of of coolants in metal metal grinding applications [18,19] can can havehave several implications in Deterioration of coolants in grinding applications [18,19] can have several implications in the effectiveness of their operation as well as in the environmental conditions [20–23], and more in the effectiveness of their operation as well as in the environmental conditions [20–23], and more the effectiveness effectiveness of of their their operation operation as as well well as as in in the the environmental environmental conditions conditions [20–23], [20–23], and and more more the specifically could result in the presence of rust on the metal parts, premature tool wear due to lack of specifically could result the presence rust the metal parts, premature tool wear due lack specifically could result inin the presence ofof rust onon the metal parts, premature tool wear due toto lack ofof specifically could result in the presence of rust on the metal parts, premature tool wear due to lack of lubricant in the coolant, excessive split emulsion that could induce particle filter blockages, as well lubricant in the coolant, excessive split emulsion that could induce particle blockages, as well lubricant in in the the coolant, coolant, excessive excessive split split emulsion emulsion that that could could induce induce particle particle filter filter blockages, blockages, as as well wellas lubricant as excessive skin sensitivity and foul odors due to emissions of sulfuric and hydrochloric acid. The foul odors due to emissions of sulfuric and and hydrochloric acid. acid. The ageing asexcessive excessiveskin skinsensitivity sensitivityand and foul odors due to emissions emissions of sulfuric sulfuric and hydrochloric acid. The as excessive skin sensitivity and foul odors due to of hydrochloric The ageing of the coolant can be attributed partially to the elevated operating temperature in the of the coolant can be attributed partially to the elevated operating temperature in the working parts ageing of the coolant can be attributed partially to the elevated operating temperature in theof ageing of the coolant can be attributed partially to the elevated operating temperature in the ◦ C) and also working parts of the machines (35–50 °C) and also to the metallic impurities trapped inside the the machines (35–50 to the metallic impurities trapped inside the liquids, after the long working parts parts of of the the machines machines (35–50 (35–50 °C) °C) and and also also to to the the metallic metallic impurities impurities trapped trapped inside inside the the working liquids, after the long friction with the metal pipes. Based on the temperature measurements frictionafter with the the metal pipes. Based themetal temperature measurements during the grinding liquids, after the long friction friction withon the metal pipes. Based Based on on the the performed temperature measurements liquids, long with the pipes. temperature measurements ◦C performed during the grinding process the temperature range was between 35 and 50 °C at the process the temperature range was between 35 and 50 at the friction/grinding area. coolant performed during during the the grinding grinding process process the the temperature temperature range range was was between between 35 35 and and 50 50 The °C at at the performed °C the ◦ C and afterwards friction/grinding area. The coolant fluid exits the machine at a temperature between 32 and 35 °C fluid exits the machine at a temperature between 32 and 35 is stored at room friction/grinding area. area. The The coolant coolant fluid fluid exits exits the the machine machine at at aa temperature temperature between between 32 32 and and 35 35 °C °C friction/grinding and afterwards is stored at room temperature at the adjacent tank (Figure 1). The temperature at the temperature at the adjacent tank (Figure 1). The temperature at the friction area is estimated to and afterwards is stored at room temperature at the adjacent tank (Figure 1). The temperature at the and afterwards is stored at room temperature at the adjacent tank (Figure 1). The temperature at thebe

Sensors 2017, 17, 568 Sensors 2017, 17, 568

5 of 20 5 of 19

◦ C. An friction is estimated to continuous be up to 60monitoring °C. An efficient and continuous monitoring of coolants up to 60 area efficient and of coolants acidity along their life cycle could acidity along their tool life for cycle could provide efficient tool for their and could alsothe to provide an efficient their operation andancould also to reveal usefuloperation trends that would allow reveal useful trends that would allowreplacement. the on-time predictive actions for coolant replacement. on-time predictive actions for coolant

2.1.2. 2.1.2. pH Characterization Characterization Coolants Coolants are are designed designed to to be be alkaline alkaline in in order order to to neutralize neutralize the the acidic acidic nature nature of of the the developed developed biological biological content content (e.g., (e.g., bacteria) bacteria) entering entering and and growing growing in in the the coolant coolant fluid fluidtank. tank. Empirically Empirically the the maintenance and the management of the coolant fluid tanks is addressed by manually measuring maintenance the management of the coolant fluid tanks is addressed by manually measuring its its acidity. Here a detailed study monitoring coolants was performed different stages acidity. Here a detailed study of of pHpH monitoring of of coolants was performed at at different stages of of aging ordertotoreveal revealthe thedependence dependenceofofpH pH with with coolants’ coolants’ aging. The coolants were sampled aging in in order sampled during collected in in bottles andand measured afterwards in the during the theoperation operationofofthe thegrinding grindingmachine, machine, collected bottles measured afterwards in lab the environment using using a high aresolution commercial electro-chemical pH meter (PIONEER, lab environment high resolution commercial electro-chemical pH meter Radiometer (PIONEER, Analytical SAS, Lyon, France). Figure 2 summarizes measuredthe values of pH for the Syntilo 81 the BF Radiometer Analytical SAS, Lyon, France). Figure 2the summarizes measured values of pH for by Castrol (black line) and Multan 61-3 DF by Henkel (red line). Considering the first coolant liquid Syntilo 81 BF by Castrol (black line) and Multan 61-3 DF by Henkel (red line). Considering the first by Castrol, it is by clearly observed that as the coolant’s increases, the corresponding coolant liquid Castrol, it is clearly observed thatuse as duration the coolant’s use duration increases, pH the index are decreasing monotonically. These variations range from 9.6 to 7.3 during a 29 days use corresponding pH index are decreasing monotonically. These variations range from 9.6 to 7.3 during period. Similar behaviorSimilar was observed the second coolant liquid, provided Henkel. However, a 29 days use period. behaviorin was observed in the second coolantby liquid, provided by the pH values of Multan 61-3 DF are compared BF. The range of the81second liquid’s Henkel. However, the pH values of lower Multan 61-3 DF to areSyntilo lower 81 compared to Syntilo BF. The range pH index is from 10.5 to overisthe 25 days period of use. of the second liquid’s pH6.6 index from 10.5 to 6.6 over the 25 days period of use.

Figure pH characterization of Castrol and Henkel at different degrees ageing Figure 2. pH2.characterization of Castrol and Henkel coolantcoolant fluids atfluids different degrees of ageingof(days used). (days used).

Due to specific operational requirements of the production line during its operation, fresh coolant Due to specific operational requirements of the production line during its operation, fresh was added when this was empirically judged by the personnel as necessary due to the anticipated coolant was added when this was empirically judged by the personnel as necessary due to the increased need of cooling and lubricating action. This is usually an empirical action based on the anticipated increased need of cooling and lubricating action. This is usually an empirical action visual inspection of metal pipes for grinding and depends on their surface quality characteristics based on the visual inspection of metal pipes for grinding and depends on their surface quality like roughness or cleanness, as this determines the required grinding strength. The addition of fresh characteristics like roughness or cleanness, as this determines the required grinding strength. The coolant in the tank took place as demonstrated in the days numbered as 5, 6, 8, 23 and 25 and indicated addition of fresh coolant in the tank took place as demonstrated in the days numbered as 5, 6, 8, 23 with the irregular pH values not following the monotonically decreasing trend. This set of data was and 25 and indicated with the irregular pH values not following the monotonically decreasing trend. intentionally included in the study, regardless of the associated interpretation problem caused by the This set of data was intentionally included in the study, regardless of the associated interpretation addition of the coolant that had to be done according to the current rules of the factory production line, problem caused by the addition of the coolant that had to be done according to the current rules of as it demonstrates characteristically a common current coolant management approach. This empirical the factory production line, as it demonstrates characteristically a common current coolant management approach. This empirical decision and followed management process often leads to an

Sensors 2017, 17, 568

6 of 20

Sensors 2017,followed 17, 568 6 ofthe 19 real decision and management process often leads to an over- or underestimation of situation with the obvious associated inefficiency of the maintenance strategy. over- or underestimation of the real situation with the obvious associated inefficiency of the During the operation maintenance strategy. of the grinding machines the coolants are recirculated and stored in tanks after passing through suitable microporous textilethe filters (Figure 1) in order tostored remove residues. During the operation of the grinding machines coolants are recirculated and in tanks The collected samples are filtered by this standard way but a further filtration step would be possible after passing through suitable microporous textile filters (Figure 1) in order to remove residues. The if required, insamples order toare assure thebyminimum level of but particle contaminants and residues. This would collected filtered this standard way a further filtration step would be possible if required, in order to assure level particleelement contaminants andpossibly residues.toThis would help the monitoring process of the minimum coolants by theofsensing leading longer lifetime the monitoring process of the coolants by the sensing element leading possibly to longer of thehelp sensors. lifetime of the sensors. Collected coolant samples were filtered by microporous paper-type filter while examining how Collected coolant samples were filtered by microporous paper-type filter while examining how this process affects the pH value of the samples. The comparison of the pH values of the samples this process affects the pH value of the samples. The comparison of the pH values of the samples before and after filtration is shown in Figure 3, where it can be noticed that after the filtration the aged before and after filtration is shown in Figure 3, where it can be noticed that after the filtration the coolants’ values by almost a full apH However, thethe overall pH trend aged pH coolants’ pHdecreased values decreased by almost fullunit. pH unit. However, overall pH trendofofthe theaged coolant remains the same, characterized by a monotonically decreasing trend. aged coolant remains the same, characterized by a monotonically decreasing trend.

Figure 3. pH measurement before(black) (black) and and after after (red) coolants’ filtration for the Figure 3. pH measurement before (red)the theadditional additional coolants’ filtration for the Henkel Multan 61-3DF. A clear reduction of the pH index for the filtered coolant is noticeable. Henkel Multan 61-3DF. A clear reduction of the pH index for the filtered coolant is noticeable.

2.1.3. Spectroscopic Characterization

2.1.3. Spectroscopic Characterization

In order to study the properties of the coolant fluids and also monitor their behavior during the

In order to study the properties of was the coolant fluids and also monitor their behavior aging period a full spectroscopic study performed employing spectroscopic techniques such during as the aging period a full and spectroscopic performed spectroscopic techniques UV-Vis absorption Attenuated study Total was Reflectance (ATR)employing spectroscopy. The spectroscopic of the coolant wereTotal monitored during(ATR) the aging process while employed in such characteristics as UV-Vis absorption and materials Attenuated Reflectance spectroscopy. The spectroscopic the metal grinding machinesmaterials over a typical period of 28 days. For the samples of aged coolants characteristics of the coolant were monitored during thestudy aging process while employed collected at the first dayover of operation afterwards every 7For days. shows the in thewere metal grinding machines a typicaland period of 28 days. theFigure study 4a samples of aged absorption spectrum (UV-Vis) from 300 to 850 nm wavelength of Castrol Syntilo 81BF, where a clear coolants were collected at the first day of operation and afterwards every 7 days. Figure 4a shows the shift of the entire absorption spectrum to higher values with respect to the days of use is observed. absorption spectrum (UV-Vis) from 300 to 850 nm wavelength of Castrol Syntilo 81BF, where a clear This change can be attributed to the increase of scattering phenomena due to the presence of metallic shift impurities of the entire higher values respect to the days4b of provides use is observed. in absorption the sample spectrum during thetoevolution of the with grinding process. Figure as This change can be attributed to the increase of scattering phenomena due to the presence of metallic reference the absorption spectrum of the unused pristine coolant sample showing that it remains impurities in the sample during the evolution of the conditions. grinding process. 4b provides as reference identical after 28 days regardless of the storage In orderFigure to further investigate the the absorption spectrum of the unused pristine coolant caused sampleby showing that it remains identical after material’s ageing process and operational degradation its use, infrared ATR spectroscopy overregardless the range 650 cm−1storage and 1800 cm−1 was applied. 4c presents the peak 28 days of the conditions. In orderFigure to further investigate the characteristics material’s ageing −1, it is clearly introduced by the group vibrations of the material ingredients. In particular, at 1550 cm process and operational degradation caused by its use, infrared ATR spectroscopy over the range observed the vibration of the C-H-N group due to the existence of amines in the solution. − 1 − 1 650 cm and 1800 cm was applied. Figure 4c presents the peak characteristics introduced by the group vibrations of the material ingredients. In particular, at 1550 cm−1 , it is clearly observed the vibration of the C-H-N group due to the existence of amines in the solution.

Sensors 2017, 17, 568

7 of 20

Sensors 2017, 17, 568

7 of 19

Sensors 2017, 17, 568

7 of 19

−1 and similarly at ~620 cm−1, the vibration of C–I vibration of NH at 1400 + 4+ occurs 1 and TheThe vibration of NH at 1400 cm−cm similarly at ~620 cm−1 , the vibration of C–I groups, 4 occurs + −1, the vibration of C–I Thedue vibration of NH4 occurs at 1400 cm−1 and similarly at ~620 cmother groups, to the presence of 3-iodo-2-propynylbutyl carbamate. All the vibrational peaks due to the presence of 3-iodo-2-propynylbutyl carbamate. All the other vibrational peaks appearing in groups, due to the presence of could 3-iodo-2-propynylbutyl All theintroduced other vibrational peaks appearing in the ATR spectrum possibly be causedcarbamate. by the impurities in the coolant the ATR spectrum could possibly be caused by the impurities introduced in the coolant liquid during appearing in the spectrum could possibly be caused by the impurities introduced therange coolant liquid during its ATR operation. Furthermore, a relative intensity variation of the peaks ininthe of 1 and its operation. Furthermore, a relative intensity variation ofofthe peaks in the range of 1030 cm−of −1 and liquidcm during its operation. Furthermore, a relative intensity variation thegroups peaks in the 1030 1100 cm−1 which is the vibrational region –OH and of C–O can berange clearly 1 which is the vibrational region of –OH and C–O groups can be clearly observed. This intensity 1100observed. cm−cm −1 and 1100 cm−1 which is the vibrational region of –OH and C–O groups can be clearly 1030 This intensity variation probably occurs due to protonation and deprotonation caused by variation probably occurs due to protonation and deprotonation caused bysame the presence ofcaused impurities observed. This of intensity variation occurs due toFollowing protonation and deprotonation by in the presence impurities in probably the liquid sample. the procedure all the the liquid sample. Following the same procedure all the aforementioned spectroscopic techniques the presence of impurities in the liquid sample. Following the same procedure all the aforementioned spectroscopic techniques are repeated in order to study the second coolant fluid are aforementioned spectroscopic techniques repeated in Multan order to61-3 study the second coolant fluid repeated in order to61-3 study the second coolantare fluid (Henkel DF). (Henkel Multan DF). (Henkel Multan 61-3 DF).

Figure 4. (a) Absorption spectrumofofCastrol Castrol Syntilo Syntilo 81 2828 days of use (b) (b) comparison Figure 4. (a) Absorption spectrum 81BF BFcoolant coolantover over days of use comparison Figure 4. (a) Absorption spectrum of Castrol Syntilo 81 BF coolant over 28 days of use (b) comparison of the pure material (black) and after 28 days (green) absorption spectra and (c) ATR spectrum of of of the pure material (black) and after 28 days (green) absorption spectra and (c) ATR spectrum of the pure material (black) and after 28 days (green) absorption spectra and (c) ATR spectrum of coolant before use (black), after 7 days (red) and 21 days (blue) of use. coolant before use (black), after 7 days (red) and 21 days (blue) of use. coolant before use (black), after 7 days (red) and 21 days (blue) of use.

Figure 5a shows the absorption spectrum (UV-Vis) from 300 to 800 nm, and Figure 5b shows the Figure 5a shows thethe absorption spectrum (UV-Vis) from 300toto800 800 nm, and Figure 5bcoolant shows Figure 5a shows absorption spectrum (UV-Vis) from 300 nm, and 5b shows the the ATR absorbance spectrum trend over 9 days. The irregular behavior is due to theFigure additional ATRliquid absorbance spectrum trend over 99days. The irregular behaviorisof isdue due to additional coolant ATR absorbance spectrum trend over days. The irregular behavior to thethe additional coolant added in the machine, thus changing the chemical composition the samples as described in liquid added in the machine, thuschanging changingthe the chemical chemical composition the samples as described in in liquid added in section. the machine, thus compositionofof the samples as described the previous the previous section. the previous section.

Figure 5. Spectroscopic characterization of coolant Henkel Multan 61-3 DF (a) absorption; and (b) ATR spectrum, for a 9 day ageing interval.

Sensors 2017, 17, 568

8 of 20

2.2. Fiber Optics Sensor Design and Development The previous analysis of the two coolant fluids’ ageing mechanism, has allowed the determination of their chemical and physical parameters in connection to their ageing mechanisms and the identification of suitable monitoring parameters. Due to the nature of the coolants and the apparent residues and strong presence of biological content such as bacteria grown during their operation, special care should be taken in order to either protect a rather expensive measuring device or employ a low cost and disposable sensing head that would be able to withstand for a reasonable and adequate time of operation in a reliable way. Given that the coolants fluids are actually a harsh environment for monitoring, the use of conventional pH meters suggests a problem with the lifetime of their employed electrodes as they are immersed on those liquids [26]. On the other hand, optical fiber-based pH sensors, compared with the corresponding electrical pH sensors, exhibit many advantages such as compact size, electrically passive operation, immunity to electromagnetic interference, capability of remote measurement and multiplexed detections [27,28]. Several fiber-optic pH sensors have been demonstrated to date by depositing a pH-sensitive coating onto the fiber surface such as hydrogels [29], cellulosic films [30], sol-gel derived glassy materials [31,32]. Recently new methods have been proposed which are based on layer-by-layer (LbL) electrostatic self-assembly (ESA) techniques [33,34] combined with the utilization also of special structures like nanostructure cavities [35], long period gratings LPG [36] or tilted fiber Bragg gratings [37] but with a limited pH range monitoring capability. However in order to provide a platform with the desired operational characteristics combined with low cost components, low power consumption, ease of handling, low cost processing, and environmental robustness in terms of electromagnetic interference EMI, the large core optical fibers [8,9] with appropriate sensitive overlayers offer a well-balanced solution. Therefore, polymer and silica optical fibers were considered and functionalized here using the fabrication flexibility of the sol-gel derived materials [38] by the appropriate incorporation of suitable pH indicators towards the development of pH sensitive overlayers. The experiments were performed with both polymer and silica fibers in order to demonstrate a direct comparison of their performance in terms of pH variation in standard solutions and in industrial coolant liquids respectively. Silica fiber of total diameter 630 µm was purchased from Thorlabs GmbH (Munich, Germany) and the polymer fiber used was an ESKA GH-4001P of 1 mm total diameter, manufactured by Mitsubishi-Rayon Co. (Tokyo, Japan). To retain the simplicity of the sensing device the sensor measuring principle was based on an amplitude interrogation system that is the simplest possible, but also adequate for its purpose in this demonstrated application. The operation of the proposed sensing device is based on evanescent wave absorption (EWA) phenomena [39,40]. These sensors are typically used to monitor the concentration of an absorbing material allowing the direct interaction with it. Evanescent waves in the cladding region were exploited in order to develop sensors based on evanescent wave absorption mechanism [41,42]. Evanescent field absorption occurs when the medium, which forms the cladding of the waveguide, absorbs the light at the wavelength being transmitted through the fiber. In our case the intermediate reagent which will interact optically with the coolants is the sol-gel based fiber’s overlayer. Three different pH indicators were used in order to extent the response to the required detection pH range of coolants. The aforementioned pH indicators were activated under specific pH conditions causing alterations in absorbance measurements at a specific wavelength, thus affecting the transmitted guided light. The quality and the efficient deposition of the sol-gel layer on the fiber surface, the proper functionalization of the fiber surface before the deposition and the proper selection of the pH indicators are critical parameters to the development of the fiber sensor. 2.2.1. Sol-Gel Overlayers Synthesis The chemical reagents and the materials used for the synthesis of the active sol-gel derived sensing material were tetraethylorthosilicate (TEOS) combined with cresol red, bromophenol blue and chorophenol red indicators which were all obtained from Aldrich Chemical Co (Milwaukee, WI, USA).

Sensors 2017, 17, 568 Sensors 2017, 2017, 17, 17, 568 568 Sensors

9 of 19 of 19 19 99 of

Sensors 2017, 17, 568reagents The chemical

9 of 20 and the materials used for the synthesis of the active sol-gel derived The chemical chemical reagents reagents and and the the materials materials used used for for the the synthesis synthesis of of the the active active sol-gel sol-gel derived derived The sensing material were tetraethylorthosilicate (TEOS) combined with cresol red, bromophenol blue sensing material material were were tetraethylorthosilicate tetraethylorthosilicate (TEOS) (TEOS) combined combined with with cresol cresol red, red, bromophenol bromophenol blue blue sensing and chorophenol red indicators which were all obtained from Aldrich Chemical Co (Milwaukee, WI, and chorophenol red indicators which were all obtained from Aldrich Chemical Co (Milwaukee, WI, order to adjust pH to the desired suitablefrom quantities of Chemical 0.1 mol/LCo hydrochloric acid (HCl) andIn chorophenol red the indicators which werevalue, all obtained Aldrich (Milwaukee, WI, USA). In order to adjust the pH to the desired value, suitable quantities of 0.1 mol/L hydrochloric USA). In order to adjust the pH to the desired value, suitable quantities of 0.1 mol/L hydrochloric and 0.1 mL/L sodium hydroxide (NaOH) solutions were used.The synthesis of the active material USA). In order to adjust the pH to the desired value, suitable quantities of 0.1 mol/L hydrochloricwas acid (HCl) and 0.1 mL/L sodium hydroxide (NaOH) solutions were used.The synthesis of the active acidconducted (HCl) and andat0.1 0.1 mL/L sodium sodium hydroxide (NaOH) solutions were used.The synthesis of theof active ambient conditions by mixing 3 mL ofsolutions TEOS with 1 mL deionized water, of 3 mL ethanol, acid (HCl) mL/L hydroxide (NaOH) were used.The synthesis the active material was conducted at ambient conditions by mixing 3 mL of TEOS with 1 mL deionized water, material was conducted at ambient ambient conditions by mixing mixing mL of4.2 TEOS with 11 mL mL deionized deionized water, alongwas withconducted 4.6 mg cresol red, 8.2conditions mg bromophenol blue andof mg chlorophenol red. Finally, a 3% material at by 33 mL TEOS with water, 33 mL of ethanol, along with 4.6 mg cresol red, 8.2 mg bromophenol blue and 4.2 mg chlorophenol mL of ethanol, along with 4.6 mg cresol red, 8.2 mg bromophenol blue and 4.2 mg chlorophenol solution of Triton X-100 was added to the material solution in order to act as a surface active reagent 3 mL of ethanol, along with 4.6 mg cresol red, 8.2 mg bromophenol blue and 4.2 mg chlorophenol red. Finally, a 3% solution of Triton X-100 was added to the material solution in order to act as a red.and Finally, 3% solution solution of Triton X-100 was was added to the the material solution in order to act act as let significantly reduceof theTriton micro-cracks in the formed sol-gel derived film. in The solution was red. Finally, aa 3% X-100 added to material solution order to as aa to surface active reagent and reduce the micro-cracks in the formed sol-gel derived film. ◦ C, significantly surface active reagent and significantly reduce the micro-cracksand in the the formed sol-gel derived film. stir for 60 min at 60and in order to allow thethe polymerization theformed network formation of the silica surface active reagent significantly reduce micro-cracks in sol-gel derived film. The solution was let to stir for 60 min at 60 °C, in order to allow the polymerization and the network The solution was let to stir for 60 min at 60 °C, in order to allow the polymerization and the network which incorporates the pHatindicators. Tableto2 allow showsthe thepolymerization characteristics of the pH indicators Thematrix, solution was let to stir for 60 min 60 °C, in order and the network formation of the silica matrix, which incorporates the pH indicators. Table shows the formation of the the silica matrix, matrix, which incorporates the pH pH indicators. indicators. Table Table 222 shows shows the the used in our experiments having which dynamic pH range (3.0–10.8). formation of silica incorporates the characteristics of the pH indicators used in our experiments having dynamic pH range (3.0–10.8). characteristics of of the the pH pH indicators indicators used used in in our our experiments experiments having having dynamic dynamic pH pH range range (3.0–10.8). (3.0–10.8). characteristics

Table 2. pH indicators used for the development of the sol-gel derived pH sensitive overlayer. Table 2. pH indicators used for the development of the sol-gel derived pH sensitive overlayer. Table 2. pH pH indicators indicators used used for for the the development development of of the the sol-gel sol-gel derived derived pH pH sensitive sensitive overlayer. overlayer. Table 2. Indicator Chemical Structure pH range Indicator Chemical Structure pH range Indicator Chemical Structure Structure pH range range Indicator Chemical pH

Bromophenol blue Bromophenol Bromophenol blue blue Bromophenol blue

3.0–4.6 3.0–4.6 3.0–4.6 3.0–4.6

Chlorophenol red Chlorophenol red Chlorophenol red Chlorophenol red

4.8–6.7 4.8–6.7 4.8–6.7 4.8–6.7

Cresol red Cresol red red Cresol Cresol red

7.0–10.8 7.0–10.8 7.0–10.8 7.0–10.8

2.2.2. Fiber Surface Preparation 2.2.2. Fiber Fiber Surface Surface Preparation Preparation 2.2.2. 2.2.2. Fiber Surface Preparation Two different type of fibers were considered and studied in this work, silica (SiO optical Two different different type type of of fibers fibers were were considered considered and and studied studied in in this this work, work, aaa silica silica (SiO (SiO222))) optical optical Two Two different typeequal of fibers were considered and studied in this work, aalso silicaa (SiO fiber 2 ) optical fiber with core diameter to 600 μm and total diameter 630 μm and polymer fiber fiber with core diameter equal to 600 μm and total diameter 630 μm and also a polymer fiber fiber with core diameter equal toµm 600and μmtotal anddiameter total diameter 630 μm and also afiber polymer fiber with core diameter equal to 600 630 µm and also a polymer (PMMA) with (PMMA) with core diameter equal to μm. Both fibers have removable cladding, 15 μm (PMMA) with with core core diameter diameter equal equal to to 980 980 μm. μm. Both Both fibers fibers have have aaa removable removable cladding, cladding, aaa 15 15 μm μm (PMMA) core diameter equal to 980 µm. Both980 fibers and havea a10removable cladding, a 15 µm thickness polymer thickness polymer cladding for the silica fiber, μm thickness fluorinated polymer cladding, thickness polymer cladding for the silica fiber, and a 10 μm thickness fluorinated polymer cladding, thickness polymer for the and a 10 μm thickness fluorinated polymer cladding, cladding for fiber. thecladding silica fiber, andsilica a 10 fiber, µm thickness fluorinated polymer cladding, forfibers the polymer for the polymer For the preparation of the fiber sensing region, the cladding of the was for the the polymer polymer fiber. fiber. For For the the preparation preparation of of the the fiber fiber sensing sensing region, region, the the cladding cladding of of the the fibers fibers was was for fiber. For the preparation of the fiber sensing region, the cladding of the fibers was initially removed initially removed using acetone. Thereafter, both the uncladded silica and polymer surfaces have initially removed removed using using acetone. acetone. Thereafter, Thereafter, both both the the uncladded uncladded silica silica and and polymer polymer surfaces surfaces have have initially using acetone. Thereafter, both 3the uncladded silica surfaces have been activated been activated by using aa 30% HNO solution, in order to and assistpolymer the bonding of the glassy sol-gel film by been activated by using 30% HNO 3 solution, in order to assist the bonding of the glassy sol-gel film been activated by usingsolution, a 30% HNO 3 solution, in order to assist the bonding of the glassy sol-gel film using a 30% HNO in order to assist the bonding of the glassy sol-gel film with the surface 3 with the surface of the fibers. Tetraethylorthosilicate (TEOS) was used as the main liquid precursor with the the surface surface of of the the fibers. fibers. Tetraethylorthosilicate Tetraethylorthosilicate (TEOS) (TEOS) was was used used as as the the main main liquid liquid precursor precursor with of the fibers. Tetraethylorthosilicate (TEOS) was used as the main liquid precursor for preparing the for preparing the pure silica film on the unclad fiber, whereas the silica glass gives refractive index for preparing preparing the the pure pure silica silica film film on on the the unclad unclad fiber, fiber, whereas whereas the the silica silica glass glass gives gives aaa refractive refractive index index for pure silica film on the unclad fiber, whereas the silica glassthe gives a refractive index less than that of less than that of the fiber core and consequently, allows total internal reflection guiding less than than that that of of the the fiber fiber core core and and consequently, consequently, allows allows the the total total internal internal reflection reflection guiding guiding less the fiber core and consequently, allows the total internal reflection guiding condition to be satisfied. condition to be satisfied. Generally, TEOS forms three-dimensional network of silicon dioxide condition to to be be satisfied. satisfied. Generally, Generally, TEOS TEOS forms forms aaa three-dimensional three-dimensional network network of of silicon silicon dioxide dioxide condition Generally, formsand a three-dimensional networkand of silicon dioxide can (SiObe through hydrolysis and 2 ) trapped (SiO 2) throughTEOS hydrolysis polymerization process pH indicators in structure (SiO 2) through hydrolysis and polymerization process and pH indicators can be trapped in structure (SiOpolymerization 2) through hydrolysis and polymerization process and pH indicators can be trapped in structure process and pH indicators candeposition be trappedtointhe structure without anysurface degradation or without any degradation or leaching. The material glass and polymer of the without any any degradation degradation or or leaching. leaching. The The material material deposition deposition to to the the glass glass and and polymer polymer surface surface of of the the without leaching. The material deposition to the glass and polymer surface of the fibers was achieved in a fibers was achieved in controllable manner by immersing the fibers (dip-coating method) into the fibers was was achieved achieved in in aaa controllable controllable manner manner by by immersing immersing the the fibers fibers (dip-coating (dip-coating method) method) into into the the fibers controllable manner by immersing the fibers (dip-coating method)After into the the sol-gel solutions for the a fixed sol-gel solutions for fixed amount of time with successive repeats. deposition/coating, sol-gel solutions solutions for for aaa fixed fixed amount amount of of time time with with successive successive repeats. repeats. After After the the deposition/coating, deposition/coating, the the sol-gel amount of time with successive repeats. After the deposition/coating, the fiber was left in a fume hood fiber was left in fume hood at atmospheric pressure and room temperature for ~20 days. Finally, fiber was was left left in in aaa fume fume hood hood at at atmospheric atmospheric pressure pressure and and room room temperature temperature for for ~20 ~20 days. days. Finally, Finally, ◦ fiber at atmospheric pressure and room temperature for ~20 days. Finally, the fibers were annealed at 60 C the fibers were annealed at 60 °C for 24 h to remove any remaining organic compounds and the fibers fibers were were annealed annealed at at 60 60 °C °C for for 24 24 h h to to remove remove any any remaining remaining organic organic compounds compounds and and the for 24 residues h to remove any remaining organic compounds and hydroxyl residues and were also washed hydroxyl were also washed with deionized water to remove the unbound indicators. hydroxyl residues residues and and were were also also washed washed with with deionized deionized water water to to remove remove the the unbound unbound indicators. indicators. hydroxyl with deionized and water to remove the unbound indicators. 2.2.3. Experimental Set-Up 2.2.3. Experimental Experimental Set-Up Set-Up 2.2.3.

Sensors 17, 568 Sensors 2017,2017, 17, 568

10 of 20 10 of 19

The characterization of the optical and electrical response of the developed fiber optic sensors 2.2.3. Experimental Set-Up

was performed by using a low-cost LED source operating at 650 nm with maximum output power of The characterization of the optical and electrical response of the developed fiber optic sensors 1 mW, (IF-E99, Industrial Fiber Optics, Tempe, AZ, USA). was performed by using a low-cost LED source operating at 650 nm with maximum output power of The light was coupled into the fibers and collected by an optical power meter (model 2832-C 1 mW, (IF-E99, Industrial Fiber Optics, Tempe, AZ, USA). Dual Channel, Newport Corporation, Irvine, CA, USA) connected to a PC. The measured optical The light was coupled into the fibers and collected by an optical power meter (model 2832-C Dual signal was continuously recordedIrvine, overCA, time using custom-built based on LabView. Channel, Newport Corporation, USA) connected to a PC. software The measured optical signal was The characterization of the electrical response was performed also on with a lowThe cost photodarlington continuously recorded over time using custom-built software based LabView. characterization detector (IF-D93, Industrial Optics) with integrated high gain and linear response. of the electrical responseFiber was performed also with a low costoptical photodarlington detector (IF-D93,Both Industrial Fiber Optics) with integrated high optical gain and linear response. Both the photodiode and the photodiode and phototransistor have integrated micro-lenses for efficient coupling to standard have fiber integrated micro-lenses efficient coupling to standard 1000 µm opticalwere 1000phototransistor μm core optical facilitating the for easy connection with the fibers. Thecore sensors fiber facilitating the easy connection with the fibers. The sensors were characterized both in standard characterized both in standard solutions with predetermined acidity where the pH indices were solutions with predetermined acidity where and the pH were achieved by adding HCl or NaOH achieved by adding HCl or NaOH solutions, theindices resulted value was also confirmed measuring it solutions, and the resulted value was also confirmed measuring it with a high resolution commercial with a high resolution commercial electro-chemical pH meter (PIONEER). Figure 6 shows the electro-chemical pH meter (PIONEER). Figure 6 shows the employed experimental set-up in order to employed experimental set-up in order to perform the optical measurements and characterization of perform the optical measurements and characterization of the fiber sensors. The active sensing region the fiber sensors. The active sensing region of the fibers (silica and polymer) is illustrated below. The of the fibers (silica and polymer) is illustrated below. The same experimental set-up was used for both samesilica experimental set-up used for both silica and polymer fiber sensor characterization. and polymer fiber was sensor characterization.

Figure 6. Schematic of the experimentalset setup up with with indicative from both silica andand polymer Figure 6. Schematic of the experimental indicativephotos photos from both silica polymer optical fiber sensing heads after the deposition of the sol-gel derived pH sensitive overlayer. optical fiber sensing heads after the deposition of the sol-gel derived pH sensitive overlayer.

specific LED source 650nm nmwas was selected selected taking thethe optimised wavelength The The specific LED source atat650 takinginto intoaccount account optimised wavelength for operation of the specific pH indicators mixture that has been previously identified to be for operation of the specific pH indicators mixture that has been previously identified toaround be around 620 nm [41]. The selected source was the best compromise given that at the time of sensors’ design 620 nm [41]. The selected source was the best compromise given that at the time of sensors’ design and development it was the most powerful commercially available compact LED source with low and development it was the most powerful commercially available compact LED source with low power consumption.

power consumption.

3. Results and Discussion

3. Results and Discussion

This section focuses on the performance characterization of the developed fiber optic sensors. Polymerand focuses silica fiber-based sensors werecharacterization tested initially in of standard pH solutions ordersensors. to This section on the performance the developed fiberinoptic validate their performance and provide reference response, andin afterwards the selected samples of to Polymerand silica fiber-based sensorsawere tested initially standardin pH solutions in order ◦ C. aged coolant fluids. The temperature of the liquid samples during the experiments was kept at 21 validate their performance and provide a reference response, and afterwards in the selected samples

of aged fluids. The temperature the liquid samples during the experiments was kept at 21 3.1. coolant Sensors’ Characterization in Standard of Solutions °C. Experiments have been carried out for the performance evaluation of the developed fiber optical sensors for both silica and polymer material platforms. The diagrams in Figure 7 shows

3.1. Sensors’ Characterization in Standard Solutions

Experiments have been carried out for the performance evaluation of the developed fiber optical sensors for both silica and polymer material platforms. The diagrams in Figure 7 shows the effect of solutions’ pH index variations on the transmitted optical power in the case of silica fiber. The variation of the pH levels of the under study solution achieved by adding HCl or NaOH

pH meter. As illustrated in Figure 7a, successive changes of solution’s pH index produces corresponding changes to the optical power of the optical fiber sensor with very good dynamic response and reversible characteristics. The electrical characteristics of silica fiber-based sensor’s response was also measured as a Sensors 568 phototransistor converted the current response into a voltage. The measured 11 of 20 resistor 2017, after17,the electric signal range was 2.09 V to 2.24 V (without using an amplifying circuit) for the entire monitored range, as pH shown Figure 7b.onThe exhibited anpower overallinoptical the effect pH of solutions’ indexinvariations the sensor transmitted optical the caseresponse of silica of about 5.2% in the (10.5–4.5) pH range. The repeatability of the electrical response was evaluated fiber. The variation of the pH levels of the under study solution achieved by adding HCl or NaOH as well, and is also shown in Figure 7b along a full during cycle of pH changes, demonstrating very solutions in different (random) time points respectively theaexperimental procedure. During the consistent results for such a wide pH range. The range of the electrical response in the whole sensor characterization the acidity of the solutions was also measured with an electrochemical pH measured index region was7a, equal to 160 mV. Thisofvoltage range evaluated as corresponding compatible with meter. AspH illustrated in Figure successive changes solution’s pH is index produces digital board circuitry systems that can allow the integration of sensors in wireless sensing and nodes changes to the optical power of the optical fiber sensor with very good dynamic response with appropriate interrogation systems. reversible characteristics.

Figure 7. (a) Measured optical response of the silica fiber sensor standard pH solutions following Figure 7. (a) Measured optical response of the silica fiber sensor standard pH solutions following dynamic changes responseofofthe thesensor sensorininsuccessive successive solution’s dynamic changesininsolutions’ solutions’acidity. acidity. The The dynamic dynamic response solution’s acidity changes electrically converted convertedresponse responseofofthe the silica optical acidity changesisisclearly clearlydemonstrated. demonstrated. (b) (b) The The electrically silica optical fiber sensor, as a function of the pH, exhibiting an almost linear and also reversible response with fiber sensor, as a function of the pH, exhibiting an almost linear and also reversible response with an an adequate 160 adequate 160mV mVrange. range.

Further to silica-based optical fiber sensors, sensor’s standard PMMA polymer opticalas fiber-based The electrical characteristics of silica fiber-based response was also measured a resistor sensors were also characterized. Following the same characterization method as above thesignal sensor after the phototransistor converted the current response into a voltage. The measured electric was evaluated and the results are presented in Figure 8a, showing the variations of the output range was 2.09 V to 2.24 V (without using an amplifying circuit) for the entire monitored pH range, as optical power for a7b. range solution pH indexes. As illustrated the behavior the in polymeric optical shown in Figure Theof sensor exhibited an overall optical response of aboutof5.2% the (10.5–4.5) fiber sensor seems similar to that of the silica fiber in the case of pH standard solutions exhibiting pH range. The repeatability of the electrical response was evaluated as well, and is also shown in monotonous linear responses. fiber sensor an results overallfor optical Figure 7b along a full cycle of The a pHpolymer changes,optical demonstrating veryshowed consistent such aresponse wide of pH about 7.5%The in the (10.5–4.5) pH rangeresponse while the electrical increased range. range of the electrical in corresponding the whole measured pH signal index region wasfrom equal2.75 to V to 160 2.87mV. V, as shown in Figure respectively. This voltage range is8a,b, evaluated as compatible with digital board circuitry systems that can allow the integration of sensors in wireless sensing nodes with appropriate interrogation systems. Further to silica-based optical fiber sensors, standard PMMA polymer optical fiber-based sensors were also characterized. Following the same characterization method as above the sensor was evaluated and the results are presented in Figure 8a, showing the variations of the output optical power for a range of solution pH indexes. As illustrated the behavior of the polymeric optical fiber sensor seems similar to that of the silica fiber in the case of pH standard solutions exhibiting monotonous linear responses. The polymer optical fiber sensor showed an overall optical response of about 7.5% in the (10.5–4.5) pH range while the corresponding electrical signal increased from 2.75 V to 2.87 V, as shown in Figure 8a,b, respectively. The range of the electric voltage for the whole region of the pH index variation is measured to be 120 mV, with very good repeatability.

optical power for a range of solution pH indexes. As illustrated the behavior of the polymeric optical fiber sensor seems similar to that of the silica fiber in the case of pH standard solutions exhibiting monotonous linear responses. The polymer optical fiber sensor showed an overall optical response of about 7.5% in the (10.5–4.5) pH range while the corresponding electrical signal increased from 2.75 V 2017, 568 in Figure 8a,b, respectively. 12 of 20 to Sensors 2.87 V, as 17, shown

Figure 8. (a) Measured optical response of the polymer optical fiber sensor standard pH solutions following closely dynamic changes in solutions’ acidity. The dynamic response of the sensors in a successive solution’s acidity changes is clearly demonstrated. (b) The electrically converted response of the polymer optical fiber sensor, as a function of the pH, exhibiting an almost linear and also reversible response with an adequate 120 mV range.

3.2. Sensors’ Response in Coolant Fluids The initial characterization of sensors in ideal standard solutions with predetermined pH values, was important in order to assess their capabilities and verify their effectiveness. However, the testing in ideal pristine solutions [18] is far from the real aged coolant fluids situation we are dealing with in this study, which have a quite complicated chemical composition [20,22]. More importantly coolant fluids develop during their aging progresses a quite heavy and complex biological load (e.g., bacteria) and incorporate also a number of residues and pollutants from the ferrous materials and the surrounding industrial environment present even after the filtering process. Therefore, the coolant fluid system could be characterized as a rather harsh environment for monitoring. This would definitely affect the measuring needs and also affect strongly the lifetime of the monitoring sensors. Indeed, although the first evaluation of the sensors’ performance showed similar results for polymer and silica optical fiber platforms, their behavior in coolant liquids is quite different as presented below. Figure 9 shows the optical response of the polymer fiber sensor in two different coolant samples of different aging (Week 1 and Week 5) of Syntilo 81BF, where the optical response appeared to be unstable and inconsistent after successive measurements. This could be attributed to the gradual deterioration of the overlayer’s quality in the presence of the attacking coolant fluid. In general, it is anticipated that due to the fact that the glassy nature of the sol-gel originated overlayer has different mechanical characteristics (e.g., expansion and thermal coefficients) compared to the polymer fiber surface would be more susceptible to cracks. This will result in partial delamination of the overlayer and the overall deterioration of the sensor. In contrast, the better deposition quality of the sol-gel derived material on the silica modified fiber substrate, led to the development of a stable sensitive layer strongly bonded to the silica substrate thus capable to withstand the presence of coolants and other impurities and allow its operation to determine the pH in this harsh environment. Figure 10a,b shows the different optical responses of the silica fiber sensor in the two different coolant fluids. The time response of the sensing heads is almost instantaneous. For the cases of monitoring the pH range (7.3–8.6) for the Syntilo and (6.4–10.5) for Multan samples, the total reduction in percent of the optical signal was calculated based on the experimental results and found equal to 34% and 38%, respectively. The pH sensor is capable of determining the critical points of coolant deterioration during the ageing process.

unstable and inconsistent after successive measurements. This could be attributed to the gradual deterioration of the overlayer’s quality in the presence of the attacking coolant fluid. In general, it is anticipated that due to the fact that the glassy nature of the sol-gel originated overlayer has different mechanical characteristics (e.g., expansion and thermal coefficients) compared to the polymer fiber surface would be more susceptible to cracks. This will result in partial delamination of the overlayer Sensors 2017, 17, 568 13 of 20 and the overall deterioration of the sensor.

Sensors 2017, 17, 568

13 of 19

substrate thus capable to withstand the presence of coolants and other impurities and allow its operation to determine the pH in this harsh environment. Figure 10a,b shows the different optical responses of the silica fiber sensor in the two different coolant fluids. The time response of the sensing heads is almost instantaneous. For the cases of Figure Measured response of of polymer optical fiber sensor in in two different samples of of coolant fluid Figure9.9. Measured response polymer optical fiber sensor two different samples coolant monitoring the pH range (7.3–8.6) for the Syntilo and (6.4–10.5) for Multan samples, the total Castrol Syntilo 81BF with two different aging degrees (Week 1 and Week 5), after successive fluid Castrol Syntilo 81BF with two different aging degrees (Week 1 and Week 5), after successive reduction in percent of the optical signal was calculated based on the experimental results and found measurements. the sensor’s behavior in real real coolants,due due probably to measurements. Theinconsistency inconsistencyofof thepH sensor’s in coolants, probably its its of equal to 34% and The 38%, respectively. The sensorbehavior is capable of determining the criticaltopoints deterioration response tothe attack coolant’s is is clearly demonstrated. deterioration ininresponse to attack by the the coolant’sspecific specificcomponents components clearly demonstrated. coolant deterioration during ageing process.

In contrast, the better deposition quality of the sol-gel derived material on the silica modified fiber substrate, led to the development of a stable sensitive layer strongly bonded to the silica

Figure 10. Measured optical response of silica optical fiber based sensor in both coolants (a) Castrol Figure 10. Measured optical response of silica optical fiber based sensor in both coolants (a) Castrol Syntilo and (b) Henkel Multan. Syntilo and (b) Henkel Multan.

Based on the availability of coolant fluids, another set of measurements was performed in order Basedthe on the availability coolant fluids,sensors another along set of measurements was performed in order to assess reversibility ofofthe employed a full cycle coolants’ use. Figure 11a to assess the reversibility of the employed sensors along a full cycle coolants’ use. Figure 11a presents presents the optical response together with the time response characteristics of the silica fiber sensor the optical response together the timeofresponse characteristics of the silica fiber sensor during of a during a full ageing cycle with experiment the Henkel Multan coolant. Following a number full ageing cycle experiment of the Henkel Multan coolant. Following a number of additional sensor additional sensor characterization measurements the electrical response was also derived for the characterization measurements the electrical response was also derived for the entire working pH entire working pH range of the Henkel coolant (Figure 11b). The specific pH range was chosen range the consideration Henkel coolantthe (Figure 11b).ageing The specific pH of range chosen taking consideration takingofinto different intervals the was coolant liquids, as into already shown in the different ageing intervals of the coolant liquids, as already shown in Figure 2. Based on this fact, Figure 2. Based on this fact, the determination and prediction of their full life cycle can be achieved the determination and prediction of their full life cycle can be achieved from the exhibited monotonic from the exhibited monotonic linear behavior. linear behavior. The experimental procedure for the sensor characterization lasted in the last demonstration example more than three hours and during this period the sensor was able to accurately monitor in real time the pH index of coolants. Similar behavior in time extended characterization was observed in other coolants and in standard pH solutions. This continuous contact of the sensing heads with the

Sensors 2017, 17, 568

14 of 20

coolants could have an impact to the expected lifetime of the sensor, but deterioration is not obvious in the demonstrated examples. Given the relatively slow change of the coolants’ characteristics during the real monitoring process, the coolants can be sampled and assessed for only a few seconds with a typical sampling period of 1 to 2 h, therefore the operational lifetime of the sensor could readily exceed one month which response is more than enough the based coolant is typically almost (a) fully replaced Figure of 10.operation Measured optical of silica opticalasfiber sensor in both coolants Castrol in three weeks. The outcome of this procedure demonstrates the efficient and reversible operation Syntilo and (b) Henkel Multan. of the proposed inexpensive sensors under harsh conditions, making them viable for industrial applications. for real applications of the was sensors are needed for Based on Of thecourse availability of coolant fluids,certain anotherimprovements set of measurements performed in order efficient thatofwould enable thesensors easy replacement of the sensing headuse. which is in any to assessstandardization the reversibility the employed along a full cycle coolants’ Figure 11a case of very cost. We alsotogether have towith notethe that dueresponse to the nature of aged coolants, thefiber developed presents the low optical response time characteristics of the silica sensor biological considerably, evenofunder normal laboratory storageFollowing conditions aeven without during a load full increases ageing cycle experiment the Henkel Multan coolant. number of any presence of metal processing. Therefore it was not possible to directly measure and compare with additional sensor characterization measurements the electrical response was also derived for the the same sensingpH heads theofpreviously coolants11b). as after few days/weeks their entire working range the Henkelexamined coolant (Figure Theaspecific pH range period was chosen composition had changed. the However we ageing have examined of the sensing inshown standard taking into consideration different intervalsthe of response the coolant liquids, as heads already in solutions after 5 and 10 days following their use for coolant characterization and confirmed a very Figure 2. Based on this fact, the determination and prediction of their full life cycle can be achieved good acceptable repeatability. from and the exhibited monotonic linear behavior.

Figure 11. 11. (a) Experimental evaluation the silica silica optical optical fiber-based fiber-based sensors’ sensors’ response response and and its its Figure (a) Experimental evaluation of of the exhibited reversibility during monitoring of the Henkel coolant ageing process. The fast time exhibited reversibility during monitoring of the Henkel coolant ageing process. The fast time response response to thepresence coolants’ispresence is also demonstrated. (b) Normalized electrical response of the to the coolants’ also demonstrated. (b) Normalized electrical response of the silica optical silica optical fiber sensor as a function of the pH index in the Henkel coolant fluid, for the entire fiber sensor as a function of the pH index in the Henkel coolant fluid, for the entire pH rangepH of range ofoperation. coolant operation. coolant

3.3. Sensor Integration in Wireless Node Modules The objective of this study was to identify viable low cost and low complexity sensing platforms that could be integrated efficiently in autonomous sensing units capable of forming a distributed monitoring network. The development of such sensors for a challenging monitoring application like coolant aging demonstrates the expected applicability of the approach to other also sensing problems, e.g., for gas sensing in large scale environmental applications. The selection of large core optical fibers allows the use of low power light sources and photodetectors that could be integrated in the sensing unit with low power requirements. Initially a lab prototype of a Fiber Optic Driver Circuit Board (FODCB) [7] supplied by 2 AA batteries (3 V in total), for the integration of the LED source (1000 microwatts output optical power at 650 nm) and a phototransistor was developed, as shown in Figure 12. The driving circuit board and the interface of the optical sensor were further improved and optimized in terms of footprint dimensions in order to be accommodated in an environmentally robust wireless node unit. A new low cost PCB was designed, as can be seen in the inset photograph of

efficient and reversible operation of the proposed inexpensive sensors under harsh conditions, making them viable for industrial applications. Of course for real applications certain improvements of the sensors are needed for efficient standardization that would enable the easy replacement of the sensing head which is in any case of very low cost. We also have to note that due to the nature of Sensors 2017, 17, 568 15 of 20 aged coolants, the developed biological load increases considerably, even under normal laboratory storage conditions even without any presence of metal processing. Therefore it was not possible to directly13, measure and compare with inside the same heads of thethe previously examined as Figure in order to accommodate the sensing node module LED source (IF-E99,coolants Industrial after a few days/weeks period their composition had changed. However we have examined the Fiber Optics 650 nm with maximum output power of 1 mW) and the photodetector (Photodarlington response of the sensing heads Fiber in standard after 5 and 10 days for configuration IF-D93, Industrial Optics). solutions The chosen photodetector has anfollowing integratedtheir highuse optical coolant characterization and confirmed a very good and acceptable repeatability. gain and performs linearly over a broad range of temperature. The PCB hosting the LED source and the photodetector utilizes two separate voltage rails (Figure 13). 5 V is used to power the LED and 3.3. Sensor Integration in Wireless Node Modules 3.3 V is used to bias the output of the photodetector. The LED is fed through a 10 KOhm resistor resulting in 2.5 mWofofthis consumption, while the photodetector is and biased through a 120sensing Ohm (R2) resistor The objective study was to identify viable low cost low complexity platforms adding another 90.75 mW efficiently power. Theintotal power drain on theunits optical board of is 93.25 mW.aThe amount that could be integrated autonomous sensing capable forming distributed of dissipatednetwork. power isThe fairly small and would impact autonomy of a battery powered device. monitoring development of suchnot sensors forthe a challenging monitoring application like In order to avoid a noisy power supply, which could degrade the accuracy of the optical transceiver, coolant aging demonstrates the expected applicability of approach to other also sensing a low noisee.g., Low (LDO) DC-DC regulator from Texas Instruments TX, was problems, forDropout gas sensing in large scale environmental applications. The(Dallas, selection of USA) large core chosen. The LM3480IM3 canofexhibit almost 70 sources dB of Power Supply Rejection over a optical fibers allows the use low power light and photodetectors thatRatio could(PSRR) be integrated broad of unit frequencies, in a stable output withaminimal voltageofripple. This IC is used in the range sensing with lowresulting power requirements. Initially lab prototype a Fiber Optic Driver only forBoard the 5 V rail, while rest of by the2system is operating attotal), the 3.3for V level. This way, of thethe analog Circuit (FODCB) [7]the supplied AA batteries (3 V in the integration LED output the photodetector can be directly digitized and processed in the microcontroller unit (MCU) source of (1000 microwatts output optical power at 650 nm) and a phototransistor was developed, as of the wireless shown in Figurenode, 12. without the need of additional interface circuitry.

Figure 12. 12. Prototype Prototype development development of the Fiber Fiber Optic Optic Driving Driving Circuit Circuit Board. Board. Integration Integration of the LED LED Figure of the of the source and and the the photodetector photodetector in in the the same same battery battery operated operated PCB-board. PCB-board. source

The wireless node is hosted in an IP54 environmentally protected enclosure (Figure 14a) and transmits the sensors data through a ZigBee network in the 2.4 GHz by using a ZigBee to WiFi Gateway wireless communication unit [43]. All the components, including the power supply circuits, reside inside the device allowing this way an easy and reliable installation. The sensing head could be immersed directly in the coolant, however in order to avoid any potential damage by increased laminar or turbulent flow inside the coolant tank that could potentially degrade the mechanical robustness of the sensor, another scheme was developed and employed. A custom made solid glass enclosure with a dual communicating tube system was used. The sensing head was inserted and centered in the first inner tube where at the two ends of the tube the two associated fiber connectors were safely and hermetically attached. The sensing head was connected to the wireless unit with two POF extensions, as seen in the Figure 14a. The second tube, as can be seen more clearly in Figure 14b,c, was used to control the flow of the coolant fluid in the sensing head’s compartment. The coolant fluid was directed to the glass enclosure by a pump in a controllable manner, where after the successful assessment of the coolant and the corresponding reading of the sensor, the pump emptied the entire coolant content from the glass unit. This way any sudden and shocking contact of the fiber head with the coolant was

(R2) resistor adding another 90.75 mW power. The total power drain on the optical board is 93.25 mW. The amount of dissipated power is fairly small and would not impact the autonomy of a battery powered device. In order to avoid a noisy power supply, which could degrade the accuracy of the optical transceiver, a low noise Low Dropout (LDO) DC-DC regulator from Texas Instruments (Dallas, TX, Sensors 2017, 17,USA) 568 was chosen. The LM3480IM3 can exhibit almost 70 dB of Power Supply Rejection 16 of 20 Ratio (PSRR) over a broad range of frequencies, resulting in a stable output with minimal voltage ripple. This IC is used only for the 5 V rail, while the rest of the system is operating at the 3.3 V level. avoided and it was minimized total immersion time of the digitized head in the during This way, thealso analog output of the the photodetector can be directly andcoolant processed in the the sample monitoring procedure. microcontroller unit (MCU) of the wireless node, without the need of additional interface circuitry.

Figure 13. Schematic of the compact optical interface for the integration of the optical transceiver module in the wireless ZigBee unit. Inset: Photograph of the developed PCB with the attached transceivers.

A number of issues could be addressed for the further enhancement of sensors’ response in this demanding application. Optimizing the deposition conditions of the sensitive overlayer and also the quantity of Triton X-100 added, could further improve the overlayers’ uniformity and reduce the micro-crack formation in the deposited sol-gel derived film, especially in the case of polymer optical fiber sensors, thus improving their robustness and reusability. Furthermore, the sensors’ response can be improved by engineering the fiber optical sensitivity by the well-known diameter tapering technique [44] that enhances the evanescent field at the fibers’ surface and consequently the interaction strength with the overlayer and the measurands. An alternative approach for coolant monitoring could be also possibly considered based on their intrinsically different absorbance for different degrees of ageing, as indicated in Figures 4a and 5a. The approach of directly sensing their absorbance properties by an evanescent wave mechanism [45]

The wireless node is hosted in an IP54 environmentally protected enclosure (Figure 14a) and transmits the sensors data through a ZigBee network in the 2.4 GHz by using a ZigBee to WiFi Gateway wireless communication unit [43]. All the components, including the power supply circuits, reside inside the device allowing this way an easy and reliable installation. The sensing head could be immersed directly in the coolant, however in order to avoid any potential damage by Sensors 2017, 17, 568 17 of 20 increased laminar or turbulent flow inside the coolant tank that could potentially degrade the mechanical robustness of the sensor, another scheme was developed and employed. A custom made using modifiedwith fibersa with removed cladding and not immersing the pH head sensitive solid suitably glass enclosure dualsimply communicating tube system wasbyused. The sensing was overlayers is probably a different the one employed here. By tube selecting the associated optimum inserted and centered in the first approach inner tubetowhere at the two ends of the the two interrogation wavelength that and corresponds to theattached. highest absorbance and the was betterconnected discrimination fiber connectors were safely hermetically The sensing head to the wireless unitaged withcoolants two POFa extensions, as seen incould the Figure 14a. The second tube, as canlinearity be seen of differently simplified approach be possible, although the expected more clearly in Figure waslower usedand to control of the dependent coolant fluid sensing head’s and responsivity would 14b,c, be much would the alsoflow be greatly on in thethe specific coolants’ compartment. coolant fluid was directed biological to the glass enclosure by a pump in a controllable composition (orThe other residues and absorbing load), thus limiting the applicability and manner, where after the assessment of the coolant corresponding reading of the flexibility of the sensor in successful real applications. In a further analysisand thethe different absorbance of coolants sensor, the pump emptied the entire coolant content from overlayer the glass unit. This way any sudden and at the operating light employed here in the pH responsive approach, is expected to have shockingorcontact of the head with tail the of coolant wasfield avoided and also itdirectly was minimized total minimal zero effect asfiber the evanescent a fiber’s cannot overlap with the the coolant immersion of the head infurther the coolant during the sample monitoring procedure. given that it time cannot penetrate than the entire thickness of the deposited overlayer.

Figure14. 14.(a) (a)Measuring Measuring apparatus with sensing head connected towireless the wireless sensing node Figure apparatus with thethe sensing head connected to the sensing node unit; unit; (b) Photograph of the glass measuring cell with the two interconnected cylindrical inner (b) Photograph of the glass measuring cell with the two interconnected cylindrical inner compartments compartments for the accommodation of sensing the fiberhead opticand sensing head and thesystem; coolant (c) flow system; (c) for the accommodation of the fiber optic the coolant flow Photograph Photograph showing in detail dual showing in detail the dual tube the glass cell.tube glass cell.

A number of issues could be addressed fοr the further enhancement of sensors’ response in this The optimization of the wireless sensing unit and its detailed systemic operational characterization demanding application. Optimizing the deposition conditions of the sensitive overlayer and also the in a wide wireless sensor network that monitors multiple parameters in the operation of the quantity of Triton X-100 added, could further improve the overlayers’ uniformity and reduce the entire coolant fluid management and machinery maintenance in a large scale infrastructure system, micro-crack formation in the deposited sol-gel derived film, especially in the case of polymer optical is currently in progress and will be reported elsewhere. fiber sensors, thus improving their robustness and reusability. Furthermore, the sensors’ response be improved by engineering the fiber optical sensitivity by the well-known diameter tapering 4.can Conclusions technique [44] that enhances the evanescent field at the fibers’ surface and consequently the For the successful penetration of predictive maintenance techniques in the management of interaction strength with the overlayer and the measurands. infrastructures and industrial assets, the provision of reliable, environmentally robust and low An alternative approach for coolant monitoring could be also possibly considered based on cost customizable sensing devices is absolutely critical. Photonics technology is well suited for their intrinsically different absorbance for different degrees of ageing, as indicated in Figures 4a and operation in industrial environments with increased electromagnetic noise and interference and a 5a. The approach of directly sensing their absorbance properties by an evanescent wave mechanism special class of low cost fiber-based sensors could provide a platform for such customizable solutions. The demanding problem of coolant fluid ageing has been selected and studied in order to design and implement efficient sensors for monitoring the ageing process as this could have a strong impact on manufacturing, environmental and occupational health issues related to coolant use. Furthermore the need to combinedly monitor coolants’ condition in remote places in an industrial infrastrucure establishes the requirements for wireless-enabled sensors able to operate in a wireless sensors network. Large core silica and polymer optical fibers have been identified as good candidate platforms for the development of sensors for pH monitoring as this was proved to be a parameter closely related to the rather complicated and multi-parametric ageing processes of such chemically complex fluids.

Sensors 2017, 17, 568

18 of 20

The developed sensors demonstrated very good and reliable behavior for pH monitoring in a wide range from pH 3 to 11. However, their response in real coolants which have a quite complicated chemical composition and content nature further complicated by the developed biological load and a number of inorganic impurities, was quite different. The silica-based sensor, due probably to the better adhesion of the sol-gel derived sensitive overlayer, withstood the attack of the harsh coolant and operated quite reliably, being able to monitor the critical area of coolant degradation with a response of a few seconds. The simplicity of the proposed scheme together with the low power consumption characteristics allowed the integration of the sensing head in a ZigBee- based wireless node demonstrating the capability of this sensing platforms in a wider range of monitoring applications by employing these hybrid photonic-based wireless sensing nodes. Acknowledgments: This work was funded by the R&D Project WelCOM 09SYN-71-856 (Wireless Sensors for Engineering Asset Life Cycle Optimal Management) funded by National Strategic Reference Framework NSRF 2007–2013/Hellenic General Secretariat of Research and Technology GSRT, and also partially funded by the project Polynano-Kripis 447963 by the Hellenic General Secretariat for Research and Technology. Authors would like to acknowledge also P. Velanas for technical help at the preliminary stage of the work, Stergios Pispas for valuable discussions on material aspects, and also Christos Emmanouilidis for discussions and suggestions regarding emerging predictive maintenance and asset management strategies. COST Action TD1001 “Novel and Reliable Optical Fibre Sensor Systems for Future Security and Safety Applications” is also acknowledged. Author Contributions: C.R., A.E.S., C.M. conceived and designed the experiments; A.E.S., C.M. and A.M. performed the experiments; A.E.S. and A.M. analyzed the data; A.P., A.S., S.K. and A.M. contributed to reagents/materials/analysis tools and prototype integration. A.E.S., A.M. and C.R. wrote the paper. Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2.

3. 4. 5. 6. 7.

8. 9.

10. 11. 12.

Emmanouilidis, C.; Liyanage, J.P.; Jantunen, E. Mobile solutions for engineering asset and maintenance management. J. Qual. Maint. Eng. 2009, 15, 92–105. [CrossRef] Emmanouilidis, C.; Riziotis, C. Wireless condition monitoring integrating smart computing and optical sensor technologies Engineering Asset Management. In Engineering Asset Management—Systems, Professional Practices and Certification, Lecture Notes in Mechanical Engineering; Tse, P.W., Mathew, J., Wong, K., Lam, R., Ko, C.N., Eds.; Springer International Publishing: New York, NY, USA, 2015; pp. 1389–1400. Kalligeros, S.S. Predictive maintenance of hydraulic lifts through lubricating oil analysis. Machines 2014, 2, 1–12. [CrossRef] Riziotis, C.; Eineder, L.; Bancallari, L.; Tussiwand, G. Fiber optic architectures for strain monitoring of solid rocket motors’ propellant. Sens. Lett. 2013, 11, 1403–1407. [CrossRef] Pruneri, V.; Riziotis, C.; Smith, P.G.R.; Vasilakos, A. Fiber and integrated waveguide-based optical sensors. J. Sens. 2009, 2009, 171748. [CrossRef] Athanasekos, L.; Vasileiadis, M.; El Sachat, A.; Vainos, N.A.; Riziotis, C. ArF excimer laser microprocessing of polymer optical fibers for photonic sensor applications. J. Opt. 2015, 17, 015402. [CrossRef] Athanasekos, L.; Dimas, D.; Katsikas, S.; Pispas, S.; Vainos, N.A.; Boucouvalas, A.C.; Riziotis, C. Laser microstructuring of polymer optical fibres for enhanced and autonomous sensor architectures. Procedia Eng. 2011, 25, 1593–1596. [CrossRef] Athanasekos, L.; El Sachat, A.; Pispas, S.; Riziotis, C. Amphiphilic diblock copolymer based multi-agent photonic sensing scheme. J. Polym. Sci. Part B Polym. Phys. 2014, 52, 46–54. [CrossRef] El Sachat, A.; Meristoudi, A.; Pispas, S.; Riziotis, C. Assessment of block and random copolymer overlayers on polymer optical fibers towards protein detection through electrostatic interaction. J. Polym. Sci. Part B Polym. Phys. 2015, 53, 327–334. [CrossRef] Sparrow, I.J.G.; Smith, P.G.R.; Emmerson, G.D.; Watts, S.P.; Riziotis, C. Planar Bragg grating sensors—Fabrication and applications: A review. J. Sens. 2009, 2009, 607647. [CrossRef] Riziotis, C.; El Sachat, A.; Markos, C.; Velanas, P.; Meristoudi, A.; Papadopoulos, A. Assessment of fiber optic sensors for aging monitoring of industrial liquid coolants. Proc. SPIE 2015, 9359. [CrossRef] Zhou, B.; Yang, S.; Nguyen, T.H.; Sun, T.; Grattan, K.T.V. Wireless sensor network platform for intrinsic optical fiber pH sensors. IEEE Sens. J. 2014, 14, 1313–1320. [CrossRef]

Sensors 2017, 17, 568

13.

14. 15. 16. 17. 18.

19. 20. 21. 22. 23. 24. 25. 26. 27.

28. 29.

30. 31.

32. 33.

34.

35.

19 of 20

Oikonomou, P.; Botsialas, A.; Olziersky, A.; Kazas, I.; Stratakos, I.; Katsikas, S.; Dimas, D.; Mermikli, K.; Sotiropoulos, G.; Goustouridis, D.; et al. A wireless sensing system for monitoring the workplace environment of an industrial installation. Sens. Actuators B Chem. 2016, 224, 266–274. [CrossRef] Lazarescu, M.T. Design and field test of a WSN platform prototype for long-term environmental monitoring. Sensors 2015, 15, 9481–9518. [CrossRef] [PubMed] Piyare, R.; Lee, S.R. Towards internet of things (IoTs): Integration of wireless sensor network to cloud services for data collection and sharing. Int. J. Comput. Netw. Commun. 2013, 5, 59–72. [CrossRef] Capella, J.V.; Campelo, J.C.; Bonastre, A.; Ors, R. A reference model for monitoring IoT WSN-based applications. Sensors 2016, 16, 1816. [CrossRef] [PubMed] Byers, J.P. (Ed.) Metalworking Fluids, 2nd ed.; CRC Taylor & Francis: Boca Raton, FL, USA, 2006. Anderson, J.E.; Kim, B.R.; Mueller, S.A.; Lofton, T.V. Composition and analysis of mineral oils and other organic compounds in metalworking and hydraulic fluids. Crit. Rev. Environ. Sci. Technol. 2010, 33, 73–109. [CrossRef] Irani, R.A.; Bauer, R.J.; Warkentin, A. A review of cutting fluid application in the grinding process. Int. J. Mach. Tools Manuf. 2005, 45, 1696–1705. [CrossRef] Gordon, T. Metalworking fluid—The toxicity of a complex mixture. J. Toxicol. Environ. Health 2004, 67, 209–219. [CrossRef] [PubMed] Li, K.; Aghazadeh, F.; Hatipkarasulu, S.; Ray, T.G. Health risks from exposure to metal-working fluids in machining and grinding operations. Int. J. Occup. Saf. Ergon. 2003, 9, 75–95. [CrossRef] [PubMed] Dilger, S.; Fluri, A.; Sonntag, H.G. Bacterial contamination of preserved and non-preserved metalworking fluids. Int. J. Hyg. Environ. Health 2005, 208, 467–476. [CrossRef] [PubMed] Howes, T.D.; Tönshoff, H.K.; Heuer, W.; Howes, T. Environmental aspects of grinding fluids. CIRP Ann. Manuf. Technol. 1991, 40, 623–630. [CrossRef] Castrol Syntilo 81BF Data Sheet. Available online: https://thelubricantoracle.castrol.com/product/syntilo81-bf (accessed on 3 June 2015). Multan-Henkel Data Sheet. Available online: http://www.henkel-adhesives.com/industrial/brands-5497. htm?nodeid=8797570924686 (accessed on 3 June 2015). Lemos, S.G.; Nogueira, A.R.A.; Torre-Neto, A.; Parra, A.; Alonso, J. Soil calcium and pH monitoring sensor system. J. Agric. Food Chem. 2007, 55, 4658–4663. [CrossRef] [PubMed] Yin, M.; Gu, B.; Zhao, Q.; Qian, J.; Zhang, A.; An, Q.; He, S. Highly sensitive and fast responsive fiber-optic modal interferometric pH sensor based on polyelectrolyte complex and polyelectrolyte self-assembled nanocoating. Anal. Bioanal. Chem. 2011, 399, 3623–3631. [CrossRef] [PubMed] Gu, B.; Yin, M.-J.; Zhang, A.P.; Qian, J.-W.; He, S. Low-cost high-performance fiber-optic pH sensor based on thin-core fiber modal interferometer. Opt. Express 2009, 17, 22296–22302. [CrossRef] [PubMed] Michie, W.C.; Culshaw, B.; McKenzie, I.; Konstantakis, M.; Graham, N.B.; Moran, C.; Santos, F.; Bergqvist, E.; Carlstrom, B. Distributed sensor for water and pH measurements using fiber optics and swellable polymeric systems. Opt. Lett. 1995, 20, 103–105. [CrossRef] [PubMed] Jones, T.P.; Porter, M.D. Optical pH sensor based on the chemical modification of a porous polymer film. Anal. Chem. 1998, 60, 404–406. [CrossRef] Alvarado-Méndez, E.; Rojas-Laguna, R.; Andrade-Lucio, J.A.; Hernández-Cruz, D.; Lessard, R.A.; Aviña-Cervantes, J.G. Design and characterization of pH sensor based on sol–gel silica layer on plastic optical fiber. Sens. Actuators B Chem. 2005, 106, 518–522. Lee, S.T.; Gin, J.; Nampoori, V.P.N.; Vallabhan, C.P.G.; Unnikrishnan, N.V.; Radhakrishnan, P. Sensitive Fiber Optic pH Sensors Using Multiple Sol-Gel Coatings. J. Opt. A Pure Appl. Opt. 2001, 3, 355–359. [CrossRef] Yin, M.; Qian, J.; An, Q.; Zhao, Q.; Gui, Z.; Li, J. Polyelectrolyte layer-by-layer self-assembly at vibration condition and the pervaporation performance of assembly multilayer films in dehydration of isopropanol. J. Membr. Sci. 2010, 358, 43–50. [CrossRef] Itano, K.; Choi, J.Y.; Rubner, M.F. Mechanism of the pH-induced discontinuous swelling/deswelling transitions of poly(allylamine hydrochloride)-containing polyelectrolyte multilayer films. Macromolecules 2005, 38, 3450–3460. [CrossRef] Goicoechea, J.; Zamarreño, C.R.; Matias, I.R.; Arregui, F.J. Utilization of white light interferometry in pH sensing applications by mean of the fabrication of nanostructured cavities. Sens. Actuators B Chem. 2009, 138, 613–618. [CrossRef]

Sensors 2017, 17, 568

36. 37. 38. 39. 40. 41. 42. 43. 44.

45.

20 of 20

Corres, J.M.; Villar, D.I.; Matias, I.R.; Arregui, F.J. Fiber-optic pH-sensors in long-period fiber gratings using electrostatic self-assembly. Opt. Lett. 2007, 32, 29–31. [CrossRef] [PubMed] Shao, L.-Y.; Yin, M.-J.; Tam, H.-Y.; Albert, J. Fiber optic pH sensor with self-assembled polymer multilayer nanocoatings. Sensors 2013, 13, 1425–1434. Aspiotis, N.; El Sachat, A.; Athanasekos, L.; Vasileiadis, M.; Mousdis, G.; Vainos, N.A.; Riziotis, C. Diffractive ammonia sensors based on sol-gel nanocomposites materials. Sens. Lett. 2013, 11, 1415–1419. [CrossRef] Gupta, B.D.; Singh, C.D. Evanescent-absorption coefficient for diffuse source illumination: Uniform- and tapered-fiber sensors. Appl. Opt. 1994, 33, 2737–2742. [CrossRef] [PubMed] Gupta, B.D.; Singh, C.D.; Sharma, A. Fiber-optic evanescent field absorption sensor: Effect of launching condition and the geometry of the sensing region. Opt. Eng. 1994, 33, 1864–1868. [CrossRef] Gupta, B.D.; Sharma, S. A long-range fiber optic pH sensor prepared by dye doped sol-gel immobilization technique. Opt. Commun. 1998, 154, 282–284. [CrossRef] Dong, S.; Luo, M.; Peng, G.; Cheng, W. Broad range pH sensor based on sol-gel entrapped indicators on fibre optic. Sens. Actuators B Chem. 2008, 129, 94–98. [CrossRef] Prisma Electronics S.A. Available online: http://www.prismaelectronics.eu (accessed on 8 March 2016). Arrue, J.; Jiménez, F.; Aldabaldetreku, G.; Durana, G.; Zubia, J.; Lomer, M.; Mateo, J. Analysis of the use of tapered graded-index polymer optical fibers for refractive-index sensors. Opt. Express 2008, 16, 16616–16631. [PubMed] Kumar, P.S.; Vallabhan, C.P.G.; Nampoori, V.P.N.; Pillai, V.S.; Radhakrishnan, P. A fibre optic evanescent wave sensor used for the detection of trace nitrites in water. J. Opt. A Pure Appl. Opt. 2002, 4, 247–250. [CrossRef] © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).