We have developed nano/microscale thermal impulse (TI) (i.e., temperature and duration) sensors for use in extreme environments such as explosions, furnaces, and combustion experiments. These sensors consist of chemically bonded precursor and precursor , which are prepared via mixed coprecipitation. In this study, we demonstrate their functionality for subsecond pulsed laser heating (isothermal duration of ) up to 1773 K, with evidence suggesting functionality to even higher temperatures. This large range is due to the formation of a new mixed-material phase (similar to yttria-stabilized zirconia) that results in spectral features that are not seen in the component materials on their own. To the best of our knowledge, this material has the largest functional temperature range of any ex situ TI sensor reported in the literature.
I. INTRODUCTION
Accurate measurements of the thermal impulse experienced by nano/microscale particles in extreme environments (e.g., explosions, furnaces, combustion experiments) are difficult to obtain with conventional means, such as thermocouples and pyrometers. The accuracy of these techniques is limited due to these extreme environments having high heating rates, high peak temperatures, opaque gases, and fluid flow of the nano/microparticles.1,2
To address these challenges, we have developed an ex situ technique, which can either determine the temperature3–6 or thermal impulse (i.e., temperature and duration)7–10 of a heating event. Both sensor types are based on lanthanide-doped metal oxide precursor nanoparticles, which undergo irreversible phase transitions when heated.10 These transitions modify the crystal field experienced by the lanthanides, which results in spectroscopic properties unique to the phase of the material. Using lab-based calibration to correlate a known temperature or TI to a specific spectroscopic signature, we are then able to determine the temperature or TI of an unknown heating event.
For our temperature-only sensors, we use a single lanthanide-doped precursor material. Calibration of a single material gives us one kinetic equation that we can solve for the isothermal equivalent temperature. However, in the case of determining the TI—which consists of two variables—we need to use two materials with unique kinetic equations. These kinetic equations can be solved simultaneously to give both the isothermal equivalent temperature and duration. For our TI sensor cocktail, we use a mixture of precursor () and precursor (). To date, we have demonstrated the functionality of the TI sensors when they have been heated using a laser,7–9 blow torch,9 structural fire,11 and an explosion.10,12
While these studies demonstrated the TI sensors' functionality, our explosion tests identified a fundamental limitation of our TI sensor mixture in explosive environments. Namely, the two sensor components are found to be separated during the explosion such that sensor materials collected in the same area may have experienced drastically different thermal histories.10,12 This separation is produced by the explosive shock front spreading and mixing nanoparticles seeded into the explosion.10,12,13 With the particles separated and mixed, we are then unable to produce a precise TI determination as our spectral measurements will see sensors with a wide range of thermal histories. This leads to us only being able to obtain approximate TI determinations.
To overcome these difficulties, we have developed a chemical blending technique (which we call mixed coprecipitation) in which the two TI sensor components are chemically prepared together such that the resulting microparticles are a chemically bonded mixture of the two materials. This technique was settled upon after parallel testing of three other blending techniques (core–shell morphology, microsheet attachment, and cross-linking) produced particles having poor photoluminescence.14
In this study, we report optical characterization of our chemically blended TI sensors prepared by mixed coprecipitation and perform a temperature calibration (up to 1773 K) of the material using rapid pulsed laser heating (isothermal durations of ). Note that the long term plan for our mixed coprecipitation sensors is to develop a calibration map covering a wide range of durations and temperatures. However, for this study, we focus on the temperature response of our sensor material as temperature is found to have a significantly larger impact on the sensors optical properties than the duration.7,9 Based on these measurements, we discover that the mixed coprecipitation sensors have a significantly larger operating temperature range than the mechanically mixed sensors, with the mechanically mixed sensors having an approximate operational temperature range of 673 K–1273 K,7,9 while the chemically blended sensors have an operational temperature range from 673 K to over 1773 K. To the best of our knowledge, this is the largest functional temperature range of any ex situ temperature sensing material reported in the literature.15–44
II. METHOD
Mixed coprecipitation of our TI sensors is achieved by coprecipitation of in the presence of . This method proceeds as follows. First, is synthesized according to the previously described method.6 Next, we disperse 0.9586 g of the pEuZ powder in an aqueous solution of (0.0594M) and (0.6 mM) and stir the solution. During stirring, we add an aqueous solution of (1/1, 1.25M), which causes precipitation of the product.45 This suspension is then allowed to age for 2 h and the resulting solid product is washed three times under water followed by acetone. The washed product is dried in vacuum oven for 12 h at room temperature followed by 12 h at .
After drying, the material is found to form a dried cake, which is then crushed with a mortar and pestle and run through a 100 mesh sieve. This sieving process sets a maximum particle size of . Figure 1 shows an SEM image of the as-prepared sensors and energy dispersive spectroscopy (EDS) images, demonstrating successful blending of the two materials.
(a) SEM, (b) EDS of Y, and (c) EDS of Z of precursor coprecipitated samples. Reproduced with permission from Anderson et al. AIP Conf. Proc. 1979, 130001 (2018). Copyright 2018 ■.
(a) SEM, (b) EDS of Y, and (c) EDS of Z of precursor coprecipitated samples. Reproduced with permission from Anderson et al. AIP Conf. Proc. 1979, 130001 (2018). Copyright 2018 ■.
After the precursor is prepared, we perform subsecond laser heating of small amounts of sample in order to perform temperature calibration. We first take of the TI sensor blend and press it into a 3.175 mm pellet using 281 MPa of pressure for 1 min in a dry pellet pressing die (MTI Corp., EQ-Die-03D). The pellet is then fragmented and one of the resulting flakes is placed onto the surface of a Resbond 920 adhesive bead (diameter of ) that is on the tip of a thermocouple (Type B, OMEGA Engineering, Inc. P30R-008). Laser heating of the bead is achieved by gently focusing the beam of a laser (Synrad Firestar Ti100, 150 W, ) onto the opposite (from the sensor flake) side of the bead. This configuration is used to avoid shattering the sensor flake due to direct laser heating.46
During heating, the bead’s temperature is controlled using a custom on/off algorithm, which takes feedback from a B-type thermocouple (recorded using a LabJack T7 DAQ), compares it to the set point, and then outputs a time varying control signal to the laser. Once heated, we perform optical characterization of the sensors using a custom PL spectroscopy system. This system uses a frequency-tripled Nd:YAG laser (Continuum Powerlite, 10 Hz, 8 ns, 355 nm) for optical excitation, various focusing/collection optics to collect the PL signal, and a fiber-coupled spectrometer (Acton Spectrapro 2500 monochromator and PI-Max 3 ICCD) to record the spectra.
III. RESULTS AND DISCUSSION
A. Spectral results
Using short-duration laser heating, we heated our TI sensor particles to temperatures ranging from 773 K to 1773 K for “isothermal” durations of . After heating, we measured the heated sensor's PL spectra using 355 nm excitation, with Fig. 2(a) showing the normalized PL spectra over the whole visible range and Fig. 2(b) showing a zoomed in view of the 540–640 nm range, which is the primary region of interest for our TI calibrations. Additionally—to more clearly demonstrate the emission from the ions—we plot the PL spectra normalized to the peak transition (either 606 nm or 613 nm) in Fig. 3.
(a) Full wavelength range and (b) narrow wavelength range PL spectra of pulse heated coprecipitated samples.
(a) Full wavelength range and (b) narrow wavelength range PL spectra of pulse heated coprecipitated samples.
From Figs. 2(a) and 3, we find that the PL spectra remain broad for temperatures up to 873 K with sharp PL peaks first observed at 1073 K. The broad emission peaks up to 873 K are due to inhomogeneous broadening,47,48 as both the Dy and Eu ions reside in a wide distribution of sites for the noncrystalline material. As the temperature is further increased, the precursor material undergoes decomposition, nucleation, and grain growth, resulting in the Dy and Eu ions residing in crystalline sites with a narrow distribution of crystal field strengths. This narrowing distribution results in the appearance of sharp peaks at 573 nm (corresponding to ) and also at 592 nm and 606 nm (corresponding to ).
The formation of sharp peaks at 573 nm, 592 nm, and 606 nm has been previously observed in our work with our mechanically mixed TI sensors,7,9 where heating results in the precursor forming the cubic phase and the precursor forming the metastable tetragonal phase.6 While these sharp peaks (observed at 1073 K and 1273 K) are consistent with our work using mechanically mixed TI sensors, we find that as the temperature is further increased (1473 K and above), new spectral features are formed in the PL spectra that have not been previously observed for our mechanically mixed sensors.
Upon the observation of new emission peaks from our TI sensors, we first hypothesized their source to be due to contamination from the Resbond adhesive, which was used during pulsed laser heating. To test this hypothesis, we heated mechanically mixed TI sensors, coprecipitated TI sensors, , and in a furnace at 1773 K for 30 min. After heating, we measured the emission spectra of all four samples with the resulting spectra shown in Fig. 4. From Fig. 4, we find that the heated mechanically mixed TI sensors emit a PL spectrum which consists of a sum of the component spectra, while the chemically blended sensors produce a unique spectral signature, which is significantly different. Additionally, we tested pulsed laser heating of the mechanically mixed sensors using the resbond adhesive and found that the heated mechanically mixes sensors did not display different spectral features than expected. These tests demonstrate that the new spectral features are due to the chemical blending of the sensor components.
PL spectra from the individual TI sensor components, the mechanically mixed TI sensors, and the coprecipitated TI sensors, heated to 1773 K for 30 min. The mechanically mixed TI sensor spectrum is found to be a sum of the component spectra, while the coprecipitated sensors’ spectra are drastically different.
PL spectra from the individual TI sensor components, the mechanically mixed TI sensors, and the coprecipitated TI sensors, heated to 1773 K for 30 min. The mechanically mixed TI sensor spectrum is found to be a sum of the component spectra, while the coprecipitated sensors’ spectra are drastically different.
Having ruled out experimental abnormalities as the source of the new peaks, we next considered the possibility of the peaks being a combination of emission from , , and , with the hypothesis being that some of the Dy ions may have migrated into the host during synthesis. Note that this discussion primarily focuses on the ions’ PL, as its intensity is significantly higher than the ions’ PL when using 355 nm excitation. However, this lanthanide migration can also occur with the Eu ions moving into the host. Migration of the ions into the host would lead to the heated sensors showing emission from Dy-doped phases as well as the phases. To consider this possibility, we compare the new PL emission spectra to the known spectra of Dy in each host phase, which is shown in Fig. 5. From Fig. 5, we find that the new emission peaks do not line up with Dy emission from the known phases. This inconsistency between the emission peaks’ locations and relative intensities suggests that the new emission spectra are not primarily due to any of the well known phases of or and are instead due to the formation of a new mixed phase (or phases). Note that while the majority of the emission appears to come from a new mixed phase (or phases), the usual phases of and may be present, but at a lower fraction resulting in their emission being “hidden.”
Comparison of Dy PL emission from different host phases under 355 nm illumination. The newly observed phase is found to have emission peaks distinct from the four previously observed phases.
Comparison of Dy PL emission from different host phases under 355 nm illumination. The newly observed phase is found to have emission peaks distinct from the four previously observed phases.
While a full microstructural characterization (e.g., TEM and XRD) of these new phase(s) is beyond the scope of this paper, we can speculate as to the nature of these phases based on the PL emission spectra obtained from the sensor material heated to 1773 K for 30 min (shown in Fig. 4). The mixed phase spectra from Fig. 4 contain spectral features identical to those seen from Dy- and Eu-doped yttria-stabilized zirconia (YSZ).49–51 This strongly suggests that for high temperatures and long durations, our coprecipitated samples form a phase of YSZ doped with both Dy and Eu ions. However, while the spectra for long duration heating are consistent with Dy:YSZ and Eu:YSZ, the spectra for short duration heating are significantly different. This suggests that for short durations, there is an intermediate phase formed.
B. Temperature calibration
Having measured the calcination temperature-dependent spectra of the mixed coprecipitation TI sensors, we next turn to computing calcination temperature calibration curves. To do so, we compute two different intensity ratios: 573/574.5 nm for the and 606/613 nm for the , which are shown in Fig. 6. From Fig. 6, we find that the ratio curves are nonmonotonic, with each curve displaying three separate transitions. For fit functions, we use the sum of three modified Arrhenius functions,
where is an offset, are amplitude parameters, are the characteristic temperatures of each transition, and are stretch exponents. Note that this fit function is not intended to represent underlying mechanisms and is instead phenomenological.
573/574 nm and 606/613 nm intensity ratios as a function of temperature for pulse heating. In contrast to the mechanically mixed TI sensors, the calibration curves are found to be nonmonotonic.
573/574 nm and 606/613 nm intensity ratios as a function of temperature for pulse heating. In contrast to the mechanically mixed TI sensors, the calibration curves are found to be nonmonotonic.
From Fig. 6, we find that ratio’s (573/574.4 nm) curve has its first transition occurring near 1000 K, with the ratio peaking at 1273 K, after which the second transition occurs at and the ratio quickly drops. The final transition occurs near 1700 K as the ratio begins to increase again. The first transition corresponds to the previously observed amorphous-to-cubic crystalline phase transition seen for , while the second transition marks the formation of the new mixed material phase not seen previously. Interestingly, this second transition is not the only novel transition as the material undergoes a second new transition above approximately 1700 K.
In the case of (606/613 nm ratio curve), the first transition occurs near 850 K and corresponds with the amorphous-to-metastable tetragonal phase transition.6 As the temperature is further increased, the Eu emission displays two more transitions which line up with the transitions seen for the Dy ratio. This suggests that the new mixed-material phase not only incorporates Dy ions but also the Eu ions as well. These new Eu phases are also evidenced by the spectra in Fig. 3, where we see new emission peaks at 611.5 nm, 616.6 nm, and 619.6 nm.
Having determined the two ion’s ratio curves as a function of temperature, we next determine their temperature sensitivity , which is defined as52
The sensitivity curves are shown in Fig. 7, with the TI sensors found to remain sensitive to at least 1773 K (albeit with low sensitivities occurring at the two inflection points corresponding to the two new phase transitions). This increased temperature range is a significant improvement over the mechanically mixed sensors, which have sensitivities dropping off at 1000 K for and 1100 K for .10 Additionally, this temperature range is also larger than any other ex situ TI sensor reported in the literature.15,16,16–44
Sensor sensitivity as a function of calcination temperature for an isothermal duration of 10 ms calculated using Eq. (2) and fits to the data in Fig. 6. Both ratios are found to display sensitivity in the 800 K–1773 K temperature range.
C. Sensor integrity
Thus far, we have characterized our chemically blended TI sensors’ optical properties using in-lab laser heating. While these results present promising improvements over our mechanically mixed TI sensors, they are only practically useful if the two materials remain together during an actual explosion. To test this adhesion, we placed some of our sensor materials in a closed-chamber explosion test using 4 lbs of C-4.
After the blast, we used a vacuum cleaner with forensic filters to collect debris from the blast chamber. As our sensor particles are smaller than , we sieved the debris to separate out larger particles, followed by an additional magnetic separation to remove any magnetic particles from the debris. The resulting separated debris was found to consist of sand, our sensor particles, and other small refuse.
Once isolated, we then placed our sensor particles in a SEM for characterization. Figure 8 shows example SEM images and EDS spectra of the as-prepared sensors (a) and (b) and the sensors recovered from the explosions. From the recovered sensor EDS spectrum, we find that each analyzed sensor particle displays EDS peaks at 2.042 keV and 14.931 keV, which corresponds to Zr and Y, respectively. The observation of both elemental signatures in each analyzed particle confirms that the two components remained together during the explosion.
SEM images [(a) and (c)] and EDS spectra [(b) and (d)] demonstrating adhesion of the sensor particles collected after an explosive event.
SEM images [(a) and (c)] and EDS spectra [(b) and (d)] demonstrating adhesion of the sensor particles collected after an explosive event.
IV. CONCLUSION
Previously, we developed an ex situ thermal impulse sensing technique relying on irreversible phase changes in two lanthanide-doped oxide hosts. Our original formulation consisted of a mechanically mixed blend of sensor particles, which were found to separate during an explosive event, thus limiting their functionality. To overcome this difficulty, we have developed a new technique to chemically bind the two components, which we call coprecipitation.
By chemically binding the two components, we find that their behavior due to heating is modified from our mechanically mixed sensors, with the chemically bound components forming new mixed phases, with unique PL emission spectra. These new phases are found to form at high temperatures () and provide new functionality to our technique as they extend the sensitivity range of our sensor particles.
In addition to performing preliminary optical characterization of the sensor particles, we also test their adhesion in a closed-chamber explosion. Performing SEM and EDS on the sensor particles recovered from the explosion, we find that the two components remain together during the explosion and are recovered intact.
Our results of higher temperature range and binding of components suggest that the new design using mixed coprecipitation is a significant improvement over the original mechanically mixed sensors. These improvements make our sensors more capable for use as thermal impulse sensors in extreme environments. Currently, we are in the process of performing further field tests of these sensors to demonstrate their performance in real world applications. Based on their performance in real world tests, we will make modifications as needed to further improve their performance, with our long term goal to commercialize our approach for use in a wide range of extreme environments.
ACKNOWLEDGMENTS
This work was supported by the Defense Threat Reduction Agency (Award No. HDTRA1-15-1-0044) at Washington State University.