Evaporation of anhydrous ammonia from small concrete coupons and implications regarding evaporation from a large accidental spill on concrete

Anhydrous ammonia is transported via ship, rail, and road everyday in the United States as a refrigerated liquid at ambient pressure. As a result, unintential release of a large amount of anhydrous ammonia could result from an accident during transportation. The aim of this paper is to provide a better understanding of the evaporation of anhydrous ammonia from porous media. In our investigation, laboratory‐scale concrete coupons were used as a surrogate for a larger concrete pad that could be present in an event involving an unintentional release of liquid anhydrous ammonia. Concrete coupons sized 5 cm × 5 cm × 1.9 cm were saturated in liquid anhydrous ammonia, and measurements of the subsequent evaporation from the coupons were made in an environmental chamber. The ambient air temperature within the chamber varied from 5 to 45°C, and the relative humidity varied from 5% to 75%. Mass‐difference calculation and Berthelot's reaction were used to determine the average evaporation rate from a concrete coupon across all trials, which was found to be 6.5 ± 1.9 mg/s. To validate these evaporation rates, the remaining ammonia in the concrete coupon was measured for each trial. We found that the time‐integrated calculated evaporation rates correlated well with the total mass of ammonia that was lost from the coupons. In addition, it was found that ambient air temperature and relative humidity had little influence on anhydrous ammonia evaporation from the concrete coupons.

containment can react with organic materials and negatively impact the environment. 5Exposure to ammonia has resulted in corrosive damage of the respiratory system and mucosal membranes in humans, as well as lethal injuries. 6,7Therefore, there is a need to understand not only the immediate primary impact of an ammonia release but also longer term secondary effects such as evaporation of ammonia out of porous media back into the atmosphere after an initial plume has moved away.
Numerous studies have been carried out regarding the modeling of atmospheric transport, dispersion, and deposition of anhydrous ammonia releases.10][11][12][13][14][15] There is also an abundance of research regarding the environmental impact of routine long-term ammonia releases, mostly in the form of fertilizing (agriculture) or organic matter breakdown (natural processes), showing that there are many different pathways that ammonia can take through the environment. 16The Committee on the Environment and Natural Resources Air Quality Research Subcommittee has suggested that there is a research need for progress in atmospheric chemistry and transport, deposition, and models, among other areas, to address key gaps in understanding how ammonia interacts with the environment. 1However, these research needs are associated with ammonia as an air pollutant, which typically has concentrations less than 1 ppb and seldom larger than 100 ppb.In the case of accidental releases of anhydrous ammonia, concentrations in the air during the time period a few minutes after the release are larger by several orders of magnitude (as high as 1000-10,000 ppm).Although there have been some investigations on these areas of interest over the past several years, there is still more to be known regarding an accidental release of ammonia and the resulting impact on the environment and surrounding population.
8][19][20][21] Our work here seeks to evaluate the evaporation of anhydrous ammonia from a saturated porous substrate, which could result in a secondary threat to first responders, the population at large, and the environment.[25] A significant amount of research has been conducted related to water evaporation from porous media, ranging from large amounts of water evaporation from soil 26 to evaporation from wetted pharmaceutical slurries. 27Additionally, mathematical models have been developed to demonstrate the evaporation of liquids from porous media. 27,28There have been a limited number of studies that investigated the evaporation of ammonia from porous media, although these studies typically involve the release of ammonia from aggregate within the concrete as it cures and not concerning anhydrous ammonia that has been introduced after curing the concrete. 29,30frigerated liquid spills, such as occurring after the rupture of an anhydrous ammonia tank or pipeline, can occur in two phases-vapor or liquid.Because the boiling point of ammonia is À33 C, liquefied ammonia requires refrigeration, pressurization, or a combination of both (e.g., terminal storage tanks are refrigerated to a point of 1 psi overpressure).Additionally, the storage pressure of the ammonia will influence the ratio of the liquid to vapor phase.Refrigerated ammonia at ambient pressure will be almost entirely in the liquid phase and form a liquid pool, which subsequently evaporates.At the other extreme, a spill of highly pressurized liquefied ammonia at ambient temperature will result in two phases.One phase presents as the liquid anhydrous ammonia flashes and occurs at the hole opening, with the initial jet consisting of about 20% gas and 80% liquid by mass.The liquid portion presents as tiny drops with a mass-median diameter of about 50 μm. 31pending on the initial direction of the jet, some of the liquid drops will "rain-out" on the surface near the release point and may form a liquid pool.This fraction was found to be about 30% in the Jack Rabbit II (JR II) chlorine release trials 32 and is assumed to be similar for ammonia based on their similar boiling points and vapor pressures.For this scenario, at any distance downwind, there are therefore two ammonia "plumes" that go by: (i) the initial vapor plume, with duration equal to the duration of the pressurized jet, and (ii) the plume made up of vapor evaporated from the rained-out pool.The releases conducted in the JR II trials had an initial jet release duration of about 1 min, with a similar plume duration at downwind distances of a few hundred meters.
The secondary evaporation plume was observed to have a duration of about 5 or 10 min, during/after which the rained-out pool was seen to have mostly evaporated.The observed maximum concentrations in the air for the initial plume were about 5-10 times greater than the maximum concentrations seen from the evaporation stemming from the rained-out pool.However, the concentrations from evaporation were still high enough to potentially cause adverse health effects.
4][35][36] However, the parameters in the models are often based on limited observations.Therefore, a purpose of the current study was to improve parameterizations for the evaporation of ammonia from a specific substrate (concrete).
The spectrophotometric technique utilized in this study to investigate the amount of ammonia present at specified timepoints involves the Berthelot reaction, which is a widely employed reaction for the quantification of ammonia in water.In particular, the indophenol blue (IPB) method is used in this study.Through this method, it is possible to force the equilibrium within an aqueous system to favor the ammonia species (NH 3 ) by increasing the pH of the solution.By doing so, over 99% of available ammonia is able to be quantified. 24,25,37,38A reaction carried out with a sample containing any amount of ammonia above a certain threshold will produce a blue color in varying degrees of saturation.The degree of saturation within the sample is linearly dependent on the amount of ammonia that is present.Therefore, the IPB method provides us with a colorimetric (and orthogonal) technique that allows for accurate determination of ammonia within a sample. 24,39o determine the amount of anhydrous ammonia that would be absorbed in a concrete substrate, an experiment was performed involving the saturation of 5 cm Â 5 cm Â 1.9 cm concrete coupons in a beaker of liquid anhydrous ammonia.The coupon was then removed and immediately placed in an environmental chamber where the ammonia was allowed to evaporate for a specified period of time (10 min).Afterward, the coupon was removed from the chamber and placed in a jar of water in order to quench the evaporation process.This allowed for any remaining ammonia within the concrete to diffuse out into the water.Samples of this water were then taken at predetermined timepoints.Since the IPB reaction is well documented in the literature to quantify the amounts of ammonia in water, it was employed here as a method to accurately determine the amount of ammonia remaining in the concrete coupon.
The masses and dimensions of each coupon in the study were measured and recorded prior to starting each trial.When transferring the ammonia-saturated coupon from the beaker into the chamber, the coupon was placed with the non-brushed surface facing down and brushed surface facing up.This was done to mimic the intended/ typical placement of concrete used in sidewalks or roads.The size of the concrete coupons used in this study was smaller than one would expect to see employed in civil engineering applications.However, in some homogenous and porous materials, the evaporation rate of water remains high and constant, starting from the surface of the material through 15 cm (Stage I, diffusion-limited evaporation). 26age I is largely dependent on the viscous capillary flow through the pores.Here, anhydrous ammonia exists in a less viscous state (almost an order of magnitude less) than water.The change in viscosity allows the ammonia to flow more freely through the pore system.The standard concrete in sidewalks (from which the concrete coupons were originally modeled) is typically 10-15 cm thick, making the findings from this study directly comparable to and therefore representative of common concrete sidewalks. 40,41The pavements that comprise most roads and military airport runways are ideally slightly thicker, approximately 25-30 cm, but this is still near the Stage 1 limit for water in homogenous porous media. 42,43Despite the difference in thickness between those materials and the coupons used in this study, the evaporation rate is still expected to be at least representative of not only that found in common concrete sidewalks but likely other civil engineering applications as well, given that the dynamics of water is very different from that of anhydrous ammonia.concrete coupons conform to ASTM C928 R2 and ASTM C387 and were cured to 6000 psi.All concrete coupons tested were chosen at random from the delivered supply, and multiple replicate measurements were made for each scenario studied.To remove any water absorbed by the concrete that could affect the mass readings, the coupons were kept in an oven at 80 C for at least 2 h before the start of the experiment.
Liquid anhydrous ammonia (Prax Air) was purchased and used as received.The gas cylinder, with an internal educator tube, was connected directly to a jacketed addition funnel (25 mL), which was cooled with a bath of dry ice and isopropyl alcohol (IPA).A 16-oz.glass jar holding the concrete coupon was kept partially submerged in the dry ice-IPA bath and then filled with liquid anhydrous ammonia to completely submerge the coupon.The dry ice-IPA bath prevented significant ammonia vaporization while the coupon was getting saturated with liquid anhydrous ammonia.
Environmental conditions at a site of anhydrous ammonia release or spill may be important to understand and must be taken into account.Specifically, liquid anhydrous ammonia mixes rapidly with water, and the mass content of water in the extremes of the conditions that were tested is different by 100 times and can be calculated by water vapor mixture charts (or psychrometric charts).Within this region, condensation from humid air can occur, and frosting within this temperature range has been evaluated. 44,45We wanted to address any effects of refrigerated liquid condensation on the surface of the concrete thereby reducing the evaporation rate.Therefore, an environmental chamber was employed.
An environmental chamber (Electro-Tech Systems, Inc., Perkasie, PA) (see Figure 2, which is a 3D rendering of the chamber) was used in all conditioned experiments.The internal dimensions of the chamber were 0.99 m Â 0.51 m Â 0.54 m (length Â height Â depth), resulting in an internal volume of approximately 0.32 m 3 .The chamber allowed the desired temperature and relative humidity (RH) to be set between 0 and 50 C and 0% and 100%, respectively, which were maintained throughout the experiments.These environmental conditions are listed in Table 1.The temperature within the environmental chamber was regulated by heating or cooling the air via coils.The humidity was managed by misting water into the air via ultrasonic action.In all the trials, both the temperature and relative humidity were actively monitored and maintained, with automatic adjustments being made as needed.The values in Table 1 were chosen as the bounds of conditions one would expect in a typical temperate climate, as well as being achievable and allowing for a consistent level of humidity to be maintained in the atmosphere within the chamber in a noncondensing manner.
The air within the chamber was actively circulated via fans.During each run, the air was closely monitored, with the temperature and relative humidity maintaining values within 5 C and 7%, respectively, of the set points.The flow speeds were measured at several locations at the evaporation site and were found to be small (<0.1 m/s maximum; Figure 3).Within the shielded area of the balance, flow speeds were found to be 0.02 m/s at their maximum (Figure 3B) and would have very little effect on the evaporation.Because the vapor pressure of ammonia is very large, the air flow over the concrete coupon was expected to have little effect on the evaporation rate.Sec.4.25 in Hanna et al. 12 discusses pool evaporation for chemical spills and states "The evaporation rate from cryogenic pools depends almost entirely on the heat balance affecting the boiling liquid pool," and "conduction heat transfer from the ground or soil is the primary source of heat for cryogenic spills."Thus, for the concrete coupons in our experiments, the heat for evaporation would come primarily from the coupon itself and possibly from whatever the coupon is sitting on.In other words, as the coupon is sitting in an ambient environment, the additional heat will come from the surroundings.
Before each replicate trial (three total), the chamber was set at the specified, predetermined temperature and relative humidity for at least 30 min prior to the introduction of ammonia.To start the experiment, a concrete coupon was submerged in a beaker of liquid F I G U R E 2 A 3D rendering of the experimental set up.A balance was enclosed in an environmental chamber with controllable temperature and relative humidity, and a digital camera recorded the balance reading to provide the mass of the remaining anhydrous ammonia.The white block within the aluminum tray is a representation of the 5 cm Â 5 cm concrete block (coupon) that was used in these experiments.
T A B L E 1 Environmental conditions that were tested.

Condition number
Temperature ( C) Relative humidity (%) anhydrous ammonia until fully saturated, which took approximately 10 min and was confirmed visually.Once the coupon was submerged in liquid anhydrous ammonia, small bubbles were observed streaming away and collecting on the surface of the coupon.The coupon and liquid anhydrous ammonia were manually agitated until the bubbling ceased.The bubbling is likely due to a combination of air trapped within the interstices of the coupon resulting in effervescence and the liquid anhydrous ammonia evaporating from the warmer interstices and bubbling out.After each day, the environmental chamber was passively cleared of ammonia overnight.
The masses of the coupons (with an accuracy of 0.4 g) were recorded prior to the start of the experiment.Next, each concrete coupon was submerged in liquid anhydrous ammonia for at least 10 min to saturate it.The coupon was then removed from the liquid anhydrous ammonia, and ammonia was allowed to flash off the surface (approximately 5 s), after which the coupon was transferred into a shallow tray on a balance (Sartorius, Göttingen, Germany) within the environmental chamber.The mass of the concrete coupon was digitally recorded at 0.25 Hz for at least 10 min within the environmental chamber prior to coupon removal and extraction of any remaining anhydrous ammonia from the coupon.In our experiments, we found that an interval of 10 min was required for evaporation from the concrete coupon to decrease by 30 ± 1%.A 3D rendering of the experimental apparatus is shown in Figure 2. The concrete coupon was placed in a small aluminum tray to avoid any potential damage to the balance.
To validate the mass of ammonia left in the coupon, the concrete coupon was removed from the environmental chamber after the 10-min interval and immediately submerged into a closed jar of 150 mL deionized (DI) water, which had been prepared before the start of the experiment.The DI water was brought to pH ≥11 by the addition of sodium hydroxide (Thermo Fisher Scientific, 99.99%) to ensure that at least 99% of the available ammonia was in the form of NH 3 (as discussed in the introduction). 37After 10 min, a 10-μL aliquot was obtained from the concrete-water solution, and Berthelot's reaction was promptly carried out on the sample.Before performing the reaction, the 10-μL aliquot of the concrete-water sample was diluted to be within the linear dynamic range of the spectrometer used via serial dilution of 10,000Â.
For the Berthelot reaction to form IPB, the sample was mixed (with manual agitation between each addition; see Table 2) with Berthelot's reaction to form IPB, it was used in this case to reduce any effects from other ions that may leach from the concrete coupons. 24ter completing the reactions, the samples were stored for at least overnight and up to 24 h at room temperature in a closed cabinet free from any light source.The storage time between mixing and analysis was variable between trials (due to convenience) but always within 24 h, as Ma et al. have shown that the reaction was stable after 20 min and until at least 24 h. 24A calibrated UV-visible spectrum was then obtained from the reacted samples to accurately measure the absorbance of the solution to calculate the amount of NH 3 (aq).Each sample ($1.5 mL) was analyzed on a Varian Cary 50 spectrometer (Agilent Technologies, CA, USA) from 500 to 800 nm.The peak absorbance was measured at 670 nm.

| RESULTS AND DISCUSSION
The concrete coupons had, on average, 4.3 g ± 1.1 g of liquid anhydrous ammonia soaked into the complex matrix of the concrete as determined by the mass difference.This implies that the concrete coupons had 6 ± 1% internal free space for the liquid ammonia to occupy.This was determined by finding the volume of space that the mass of anhydrous ammonia occupies.Since the dimensions of the concrete coupon stay constant, the change in volume is the estimated free space within the concrete coupon.In an actual event where there is a release of liquid anhydrous ammonia on a concrete or porous surface, if 6% of the volume of the surface material is free space, there may be significant secondary emissions.For example, one could consider an unsealed 1 m Â 1 m Â 0.2 m piece of brushed concrete sidewalk that becomes completely saturated as a result of a release.If within that concrete area there is 6% free space, that could potentially result in more than 1.5 kg liquid anhydrous ammonia present within that volume and constitute a significant threat as it evaporates.
The evaporation of liquid anhydrous ammonia from the coupons was measured in replicate (at least three per trial) for each condition.
The mass of ammonia in a coupon was found to decrease with time in an approximately linear manner over the course of the experiment (10 min).Figure 4 shows a representative evaporation plot at 40 C and 5% RH.For this condition, during the first approximately 10 min, over 1.5 g of ammonia was lost/evaporated from the concrete coupons (solid lines, left axis), which is equivalent to a 30% ammonia loss from the coupon (dashed lined, right axis).
It is expected that the fractional ammonia lost/evaporated from the 5 cm Â 5 cm Â 1.9 cm coupon would scale with the typical sizes of real concrete pads and accidental releases of liquid ammonia (1-20 tons in historic accidents).This scaling would be valid because imbibition is expected with refrigerated liquid materials in porous structures, 46,47 and liquids on surfaces will naturally spread (and therefore be likely to imbibe). 28,48At thicknesses of up to $15 cm, which is comparable to some civil engineering scenarios, the evaporation rate is diffusion-limited and is therefore expected to scale similarly.
With the validity of the scaling assumption, the 30% loss of ammonia found in the small concrete coupons would be also expected in a full-scale accident.However, for thicker slabs of concrete capable of retaining more ammonia and subject to molecular diffusivity, one could expect the evaporation process to take much longer-which is consistent with what is found in the literature. 49In any event, this type of potential threat has long-term implications for first responders and people who come upon the site after an initial event is over.
Figure 5 shows a time series of the average evaporation rate for the first 600 s of the experiment for each temperature and RH scenario.
The average evaporation rate of ammonia from the concrete coupons for the initial 600-s period for all conditions is 6.5 ± 1.9 mg/s (error is the standard deviation of the data).Considering that the area of the coupons is 65 cm 2 (the top surface and the four exposed sides), this is equivalent to an evaporation rate per unit area of 0.1 mg/cm 2 s or 1 g/m 2 s.As seen from the bar graph in Figure 5, the error in each of the bars overlaps with that of all the others, suggesting that the results for each scenario are not statistically different from the other scenarios.Furthermore, there are no clear trends seen from the results in Figure 5.
Additionally, given the short distances that anhydrous ammonia must travel through the coupon, the expected mechanism of evaporation is via diffusion through the capillary to the surface. 26If taking into account that the heat flux from the atmosphere or the surface on which the concrete coupon is positioned and Knudsen diffusion, the contribution due to the vapor pressure is much larger than the effect of temperature or relative humidity. 46,50,51Therefore, temperature and relative humidity at this scale (which is relevant for most scenarios; see introduction) will not dominate the evaporation process of liquid anhydrous ammonia from a porous network such as concrete.This work and the data shown in Figure 4 (diffusion-limited evaporation) and Figure 5 (limited or no effects due to temperature and relative humidity) support this claim.
Additionally, diffusion-limited evaporation from concrete or other porous media potentially prolongs a secondary threat for first responders and the general public.
As described in the introduction, the IPB method can be used as an alternate and orthogonal way of calculating the amount of ammonia remaining in the concrete coupon.Figure 6A shows the visible spectrum from the IPB as a function of an increase in available ammonia in solution.There is a peak at $670 nm-consistent with what is found in the literature-that linearly increases with the ammonia concentration. 24,39These spectra are also representative of samples extracted from the liquid anhydrous ammonia-soaked concrete coupons.Figure 6B shows the UV-visible calibration using ammonium chloride and provides a slope of 0.0152, intercept of 0.0892, and an R 2 of 0.954.The remaining amount of liquid anhydrous ammonia in each coupon was quantified, which gave an average recovery of 93 ± 33% for all trials.This therefore validates the observed evaporation rates and our methodology for determining these values.
Understanding the amount of ammonia that evaporates from a porous substrate is important from a modeling perspective as well.
While the initial plume may be most impactful to the surrounding population and environment, understanding the potential of secondary impacts and the rate at which ammonia evaporates from a material is also important.
It was stated in the introduction that current comprehensive operational hazard assessment models need improved parameterizations for evaporation of anhydrous ammonia from various substrates.
The current study provides observations that justify the assumption of an evaporation rate of anhydrous ammonia from a saturated concrete of approximately 0.1 g/m 2 s.Thus, for evaporation continuing for 600 s from a 100 m 2 concrete pad, it can be calculated that approximately 6 kg of ammonia would evaporate into the air passing over the pad at this rate.The transport and dispersion model can use this information to calculate downwind ammonia concentrations.

| CONCLUSIONS
As expected, because of the large vapor pressure of ammonia, air temperature and relative humidity are not the dominating factors influencing evaporation of anhydrous ammonia from concrete coupons.The dominating factor is the rate at which heat can be drawn out of the concrete and the surface on which it is lying.This work demonstrates that the average evaporation rate is 6.5 ± 1.9 mg/s from a saturated 5 cm Â 5 cm Â 1.9 cm concrete coupon.Additionally, we observed a linear evaporation rate over 600 s after liquid anhydrous ammonia saturates the coupon.The rate of evaporation is one factor in calculating secondary effects of ammonia release over a larger area and for a longer period on the environment and the surrounding population.
F I G U R E 6 Visible spectra of the indophenol blue reaction with increasing amounts of available ammonia in solution (panel A) and a calibration curve of available ammonia in solution ranging from 0.2 to 20 mg/mL (panel B).The vertical lines in panel B are the standard deviations of the measurements at each concentration.
Figure 1 shows the brushed front surface (A), the non-brushed surface (B), and the cut edge surface (C) for a representative coupon.The

F I G U R E 3
Radial air speed above the glass enclosure surrounding the sampling area (A, yellow); the glass enclosure surrounding the sampling area (B, blue); and a digital image showing where the two measurements were taken (C).160 μL of a citrate buffer (500 g/L trisodium citrate dihydrate, Sigma-Aldrich), 40 μL of an alkaline o-phenylphenol solution (20 g/L, Sigma-Aldrich), 40 μL of alkaline sodium dichloroisocyanurate (10 g/L, Sigma-Aldrich), and 40 μL of alkaline sodium nitroprusside (7 g/L, Sigma-Aldrich) in that order.The pH of each reagent solution was increased (made alkaline) with optimal amounts of NaOH, with the exception of the citrate solution.NaOH concentrations were 20 g/L NaOH for the o-phenylphenol solution, 10 g/L NaOH for the dichloroisocyanurate solution, and 70 g/L NaOH for the alkaline sodium nitroprusside.Though the citrate buffer is not traditionally used in

F I G U R E 4
Time series, ending at 600 s, of the mass of ammonia evaporated from the concrete coupon (left axis, solid lines) and the percent of initial ammonia mass lost (right axis, dashed lines) for three different trials (i.e., coupons) for T = 40 C and RH = 5%.F I G U R E 5 The average rate of ammonia loss over the first 600 s from the concrete coupons (in mg/s) at each T and RH scenario.T is temperature ( C) and RH is relative humidity.Vertical bars indicate the standard deviation of the measurements.The methodologies used here can be used to develop evaporation parameterizations for other chemicals and other materials.However, as mentioned earlier, this methodology is best suited for refrigerated liquid chemicals with large vapor pressures and evaporation rates.If the chemical has a small evaporation rate, the evaporation formulation must include the influence of the atmospheric boundary layer (i.e., wind speed and turbulence profiles).
T A B L E 2 Berthelot's reaction addition order and reagent concentrations.