Practical Strategies for Reducing Ammonia Volatilization from Urine-Derived Fertilizers

Ammonia Volatilization Trials 

The purpose of this experiment was to measure how much of the nitrogen in urine was lost into the air through the process of ammonia volatilization, following the application of urine to perennial grass under different scenarios (including different field conditions, application methods, and urine pre-treatment methods). We conducted four trials testing four variables: 

1. 1. Urine application method
   2. Ground wind speed
   3. Irrigation/rainfall
   4. Urine pre-treatment methods 

We measured ammonia loss by first applying urine to small subplots in a hayfield, and then immediately placing a plastic container over each subplot so that any ammonia that volatilized would be held in the air within the container. Inside each container was an ammonia trap, which absorbed the ammonia from the air and saved it for later analysis.

There are many ways of measuring ammonia volatilization in the field, each with advantages and disadvantages. We chose a closed dynamic chamber method, which uses an acid trap placed inside a transparent chamber covering the test area. The acid trap contains dilute acid, which catches any ammonia gas that comes into contact with it. Air circulation is facilitated by a small fan suspended over the trap, blowing air down onto the trap. The acid trap can be replaced and analyzed periodically to determine the amount of ammonia captured over time. The method we used was based on the method described by Campbell-Nelson (2009). 

Pre-trials in 2018 and 2019

Before starting the field trials, we tested the method in a series of pretrials to validate and optimize the design. An acid trap (containing H2SO4) and an ammonia emitter (containing a known quantity of ammonia solution) were placed on a table, covered with an enclosure, and left for 24 hours. The acid trap and ammonia emitter were both then removed and analyzed. The emitter contents were titrated with 0.1N H2SO4 to confirm that the ammonia had fully volatilized. The acid trap contents were analyzed to determine the amount of ammonia captured. This tabletop testing was repeated in a variety of different configurations, including with and without fans blowing air onto the acid traps, and with and without air exchange holes in the sides of the chambers.

Analysis of the acid traps was initially attempted colorimetrically in 2018 using the Hach High Range Ammonia Nitrogen Test ‘N Tube method. Unfortunately, the colorimetric quantification produced inconsistent results that we were not able to resolve, which delayed the rest of the experiment until the field season of 2019. In 2019 we developed a new quantification method using titration with 0.1N NaOH to measure the amount of acid that had been neutralized as the trap absorbed ammonia. This worked well and is the method used and described in the rest of this report. 

In preliminary indoor table-top testing of the ammonia capture system, using a chamber without holes and a fan circulating air over the ammonia trap, we measured a 93% ammonia capture rate. However, when we tested the ammonia capture chamber outdoors on turfgrass, the capture rates declined dramatically. We again tested chambers with and without ventilation holes, with different size fans, with weights or stakes securing the chamber (to minimize gaps between the chamber and the ground), and with a hard surface placed over the grass inside the chamber to eliminate absorption into the grass our soil. Chambers without holes had slightly higher capture rates than chambers with holes, but still only captured 30-40% of volatilized ammonia regardless of any modifications made. 

To account for this low capture rate, we added a set of positive controls to each field trial, in order to calculate what fraction of the volatilized ammonia was actually captured in the trap. This positive control treatment involved placing urine in a dish, rather than applying it to the ground, thus ensuring that all the ammonia would volatilize. By dividing the amount of ammonia captured by the amount of ammonia known to have been volatilized from the urine in the dish, we calculated the ammonia capture rate. We then used the ammonia capture rate from the positive control as a reference for calculating the percentage volatilized from the other treatments. For example, if the trap in the positive control captures 40% of ammonia in the urine, and a soil application captures 10%, the adjusted capture rate for the soil application is 10/40 = 25% volatilization. A weakness of this method is that it assumes that the unexplained loss is the same in all chambers. However, the values that we obtained in the field trials are within the ranges that we expected, and we feel it is reasonably accurate. 

Field Trials in 2019

Experiment Location and Soil Type

All field trials were conducted on the back hay field at Fair Winds Farm in Brattleboro, VT. Different parts of the field were cut at different points during the summer, allowing us to place each trial on a different section of newly-cut hay. 

The soil in this field is a sandy loam, with 72% sand, 21% silt, and 7% clay, and a pH of 6.6. The top two inches of soil contain 7% organic matter and have a CEC of 11.1me/100g. 

We would have liked to repeat the irrigation trial on a second farm with a less sandy soil, but because of the extensive method development phase, there was insufficient time to do this.

Equipment and experiment setup 

Chambers

The volatilization chambers were Sterilite brand 58-quart / 55-liter plastic storage tubs turned upside down. The dimensions of each chamber were:

23 ½” L x 16 ⅞” W x 12 ¼” H

59.7 cm x 42.9 cm 31.1 cm 

Chambers were placed in 3 rows of 6 each. Rows were spaced approximately 1 meter apart, and chambers within each row were spaced approximately 50cm apart. Each chamber was held in place by two 5.7” plastic tent stakes, one on each end. 

Experimental setup with chambers and weather station 

Acid traps

Acid traps contained 250 mL of 0.1N H2SO4. The traps were clear, round, 500-mL polypropylene food containers with snap-on lids. All acid traps were filled and weighed in the lab and capped for transport to the field. These same containers were also used to transport urine from the lab to the field.

Each trap was held approximately 1.5” off the ground with a stand made of rigid plastic mesh. A 6x6cm Sunon brand fan was placed on top of each trap to circulate air to the surface. At 12V, each fan drew 0.18 amps and produced a wind speed of 9.7mph. Fans were attached to small, stiff wire fixtures that rested over the acid traps.

The traps were placed in the application area, avoiding placing them directly on top of the applied urine (except when broadcast), and the chamber placed over the entire area as quickly as possible after application. 

The first two trials used a 12 volt, 75 amp hour lead-acid battery to power the small fans and weather station. The battery was placed in the center of chamber arrangement underneath a smaller plastic chamber and swapped with a fully charged battery at each sampling time. Therefore two of the experiment chambers were right next to this smaller tub, which may have slightly changed the microclimate in those two chambers. 

Urine application

The urine used in trials 1-3 was collected from our portable urine collection toilets and urinals at a large public festival, and pasteurized at 80°C for 1.5 minutes using custom pasteurization equipment. 

108 g of urine was applied to a 12” x 15” area within the footprint in each chamber. The urine had an N content of 6.13 g/L, resulting in an application rate of 50.8 lbs N/acre. The 12” dimension was chosen to match the spacing of the custom applicator that we currently use for farm field application of urine. The 15” length was chosen arbitrarily to fit within the chamber.

All trials included two sets of controls: 

1. 1. Blank control: acid trap with no urine applied. In all trials, all controls yielded an ammonia capture rate of 0 or slightly below 0. 
   2. Positive control: Urine was poured into the cut-off bottom of a gallon plastic water jug - the “emitter.” The emitter was placed on a support and adjusted until it was roughly level, with the liquid covering the bottom of the emitter. The positive control was included as a reference in order to calculate the capture rate, as described in the pretrial section.

Titrations

The amount of ammonia caught in each trap was calculated by measuring the quantity of acid neutralized in the trap. To calculate this, we recorded the volume and concentration of acid that was added to the trap before it was placed in the enclosure, to quantify the total moles of acid in the trap. After the trap was exposed to ammonia, we determined the volume of the trap contents again and titrated a subsample with 0.01 M NaOH, to determine the total moles of acid remaining. This allowed us to calculate the moles of acid (H+) consumed during the experiment, which equals the moles of ammonia captured, assuming no other sources of alkalinity entered the trap. 

Sampling times

Due to the very rapid ammonia volatilization rate from liquid manure (and presumably urine), with the great majority of ammonia loss occurring within the first two days (Meisinger and Jokela, 2000), trials 1-3 were sampled after 24 hours and 48 hours. These samples accounted for most of the ammonia captured. A final sample was taken on day 3 or 5, depending on scheduling availability, but the amount collected in the final sample was always very small (<10% of total).

We used a longer sampling window in the fourth trial testing urea and urine treatments, because urea is known to volatilize ammonia over a longer time period. This trial was sampled on days 4, 8, and 12. 

Weather data collection

We set up a weather station next to the chambers to record temperature and rainfall. There were two thermometers, one measuring the ambient air temperature, and one inside a control chamber, measuring the interior chamber temperature. This data was logged on an SD card and downloaded at each sampling time for the first two trials. The Arduino was damaged during setup of the third trial, so there was no weather data for that trial. For the final trial, the data was recorded directly from the Arduino. 

Soil moisture sampling method

After setting up each experiment, we used a homemade drill-soil corer with a 2 centimeter diameter to take soil samples from in between the chambers and the area immediately around them (no further than 1m away from any chamber). 

For each trial we collected three replicates, each consisting of six soil cores of the top 2” of soil collected from random locations in the setup area. (18 total 2” soil cores taken). Each replicate was oven-dried at 140°F until a stable weight was achieved, which was used to calculate the starting soil moisture level. 

Data analysis

The charts display the average amount of ammonia captured for each treatment adjusted by the average amount captured in the positive control, with error bars indicating compound error from this calculation. To calculate the compound error when dividing the average value for each treatment (a) by the average value for the positive control (b)

Compound error formula:
Where a is the standard deviation from average a, b is the standard deviation of average b, and c is the resulting number from the calculation involving a and b.

Significant difference between treatments was analyzed using the actual total amount of ammonia captured, not the amount adjusted by the positive control. These data were tested for homogeneity of variance using the Levene test and for normality using the Shapiro-Wilk test (both tests accepted at p>0.05). Non-normal data were transformed and reanalyzed for normality. Data were then analyzed for significant differences using a one-way analysis of variance (ANOVA) and Tukey post hoc test, with significance determined at p<0.05. Significant difference is indicated with letters in each chart. Statistical analysis was carried out with R version 3.6.2 using RStudio version 1.2.5033, and plots were created using Microsoft Excel 2016.

Details on the setup of individual trials

1. Application method trial

Description of each treatment: 

• • Broadcast: Urine was poured into a 120 mL plastic bottle with a lid perforated with holes. Bottle was squeezed evenly over 12”x15” application area. 

  • Band: Urine was poured into a 120 mL plastic bottle and poured onto the ground in two 15” rows 12” apart. 

  • Spot: Urine was poured directly onto the ground in one place within the application area.

  • Injection: Two 5” holes were made in the ground using a plastic rod. Approx 54 mL of urine was injected into each hole using a syringe pressed firmly into the ground. A small fraction of the urine rose and puddled around the surface, but quickly sank back into the soil. 

Sampling times: 

Traps were changed and titrated after 24, 49, and 117 hours, except the control, which was only titrated after 117 hours, due to a miscommunication. 

Setting up Application Method trial 

2. Irrigation trial

Setup: 

This trial was originally conceived to test different starting soil moisture levels, but due to the challenge of creating different starting soil moisture levels within in each chamber area, we modified this trial to test simulated rainfall/irrigation before and after urine application. 

We started this experiment after a period of warm, dry weather, with soil that had a much lower moisture level than in the previous trial. The urine was applied in bands by pouring from a 120 mL bottle in two 15” lines within a 12”-wide irrigation area.

Description of each treatment:

Irrigation water was applied 1 hour before urine application, immediately after urine application (within less than a minute), 1 hour after urine application, or not at all. 

To apply the irrigation water, we measured approximately 500mL tap water into a 1000mL bottle with holes drilled into the lid. We squeezed the water out over the application area, marked by flags, attempting to cover it evenly. 

Irrigation calculation:

We had planned to simulate typical July rainfall events by irrigating with 0.5 inches of water, which was determined to be a typical rainfall event by referencing data for July from the Advanced Hydrologic Precipitation Service for Vermont on NOAA’s webpage. This would have required 1.5L of water per application area (12 x 15 x 0.5 inches = 90 cubic inches, divided by 60 cubic inches per liter). However, we made a mistake while applying, and instead applied the equivalent of a light rain of 0.17 inches (⅓ the intended rate). 

Sampling times:

Traps were changed and titrated after 24, 47, and 71 hours. 

3. Wind speed trial

Setup:

We used 120mm tube-axial fans (called “large fans” in this description to differentiate from the small fans on the acid traps) mounted on dowels driven into the soil to create high, medium, and low wind speeds at ground level inside each chamber. Fan orientation was checked using a square to ensure it was pointed parallel to the ground. We used a handheld Kestrel 2000 wind speed meter to measure the wind speeds. These measurements were taken at both ends of the application strip (where the near end is defined as the point on the application strip closest to the large fan). The wind speeds we observed are in the table below.

NOTE: The wind speeds in the table below were measured within about 2” of the ground. In field measurements taken during experimental design to assess typical field conditions, wind speed at chest height was several times higher than wind speed at 2”.

The urine was poured in a single 12” strip in front of the fan (starting ~2 inches from the fan) and marked by flags. The acid trap was set up in the same corner of every chamber. Each chamber was placed so that there was approximately 2” between the fan and the rear wall of the chamber. Each chamber had this layout: 

The large fans only operated for the first 48 hours, after which they were removed from all chambers. 

The previous two experiments had been logging very high temperatures (up to 54°C) inside the chambers. To help keep the internal temperatures more realistic, we draped shade cloth over the bins after they were set up and secured it in place with plastic tent stakes. 

Sampling times:

Traps were changed after 24, 48, and 117.5 hours.

Setup of trials 3 and 4 with solar panels and shade cloth 

4. Urine pre-treatment trial

Background: 

When urine leaves the body, nearly all of the nitrogen is in the form of urea. Once it enters the environment, the urea is rapidly hydrolyzed into ammonia (NH4) and CO2, raising the pH of the urine to around 9. Some urine collection methods seek to prevent this hydrolysis, because preventing hydrolysis helps prevent precipitation of minerals in plumbing and also minimizes odor and NH4 emissions in handling, storage, and transportation of urine. Because the urease enzyme only functions in a narrow pH range, modifying the pH either above or below this range is one way to prevent hydrolysis. Adding a small amount of vinegar is sufficient to lower the pH and is currently a common practice in our urine collection program. We were interested in testing whether this could also help reduce ammonia emissions in the field. 

Setup: 

This trial compared ammonia losses from untreated urine, in which urea had been fully hydrolyzed to ammonia, with losses from urine that had been treated with acid or base to prevent urea hydrolysis.

The urine used in this trial was different than that used in the other trials. It was collected in the bathroom at our research center, using a process that split each urination into equal fractions, and subjected each fraction to a different treatment. The low pH fraction was treated with 5% acetic acid (vinegar) at a rate of 50ml/L, the high pH fraction with slaked lime at a rate of 10 g/L, and the hydrolyzed fraction with no additions (to encourage urea hydrolysis).

Once this urine was collected, we took initial pH and conductivity measurements, and then stored the high and low pH samples at 4°C, leaving the untreated urine at room temperature to hydrolyze. We monitored this fraction periodically and decided it was fully hydrolyzed when the conductivity stabilized around 43 mS. Approx 1 L of each type of urine was then pasteurized in a volumetric flask on a hot plate in the lab, reaching maximum temperatures of 83-88 °C. 

Because these urines had a slightly higher N content than the urine used in previous trials, (determined by measuring the ammonia content of the hydrolyzed fraction in our lab,) we used a slightly smaller amount, 103 mL. We used the same band application method as in trials 1 and 2, and the shade cloth and solar panels used in trial 3. 

Sampling times: 

Traps were changed after 4, 8, and 12 days. 

Site Visits 

The purpose of this portion of the study was to engage in dialogue with farmers from a diverse array of livestock farms, in order to explore the best strategies and opportunities for using current or slightly modified equipment to use UDF effectively on their farms for forage crops, while minimizing ammonia losses. We conducted site visits and interviews on ten farms in southern and central Vermont in the late winter of 2019. Some of the interviews included more than one farmer, for a total of 14 farmer participants. Our focus was farms raising livestock and growing forage crops, because our earlier research has suggested that using urine as a fertilizer on animal feed (as opposed to vegetable crops) is currently preferable in terms of public acceptance. We also sought a diversity of farm size, livestock type and management style. Participating farms ranged in size from ten to 1700 acres, with several between 300 and 600. Crops included vegetable gardens, hay and pasture, corn silage, corn grain, as well as other enterprises such as maple sugaring, hemp and cut flowers. Livestock included cows, goats, sheep, horses, alpaca, chickens and pigs. Numbers of livestock ranged from a few to 400 goats, 500 sheep and 1600 cows on the larger farms. Of the ten, two are certified organic, four are “conventional”, and four are “something else” such as “animal-based.” 

Site visits lasted approximately 90 minutes and included dialogue based on an interview guide. An extensive list of both quantitative and open-ended questions had been developed from topics and ideas raised in previous farmer surveys (SARE ONE14-218). These were used to create a guideline for dialogue at each site visit, with a focus on allowing each site’s specific characteristics and each farmer’s ideas to be documented, while creating a consistent format for each visit following standard qualitative research procedures. The guide was modified for this study in consultation with agricultural educators Drew Adams and Heather Darby. The interview guide included basic information about the nutrient value of urine, and research results with regard to hay yield, dilution levels, and levels of microconstituents (i.e. pharmaceuticals) based on Rich Earth data.

The documentation and dialogue included observation and discussion of current soil fertility management practices; equipment used; and application methods for synthetic fertilizers, organic amendments, manures, liquid manures and digestates. In addition, the visits included conversation with farmers concerning the potential for incorporating urine-based fertilizers in their management plans using existing or modified equipment, as well as farmer observations and beliefs about the best timing and weather conditions for application of nitrogen-based fertilizers; the potential obstacles farmers foresee in using UDF (including social issues such as customer perception, and best methods for addressing these); recommendations and ideas for equipment modification; and suggestions for additional research, such as recommendations for soil health parameters that could be evaluated in future trials.

At the start of each visit, farmers and interviewer orally went over an “informed consent” document, which farmers signed. The consent form included consent for some portions of the interviews to be published in various media. They received a copy of the form. Farmers received a stipend of $100 for their participation, with the exception of one farmer who refused the stipend.

All site visits were recorded and later transcribed. Given the relatively small number of interviews, this data was not subject to formal content analysis. Instead, the interviews were reviewed for key ideas relevant to the study objectives. We are cognizant of the fact that the results represent the ideas of each farmer and apply only to his or her specific situation. However, we chose a wide range of farm types and sizes to gain as much diversity in perspective as possible within the geographic region of the study.

We are providing farmer participants with copies of the current report. We will solicit comments and feedback which will then be incorporated into articles produced for wider distribution to other farmers and the general public.