Firstly, NR-modified ER was prepared by mixing NR with ER at mass proportions of 0, 10, 20, 30, 40, and 50% in a three-necked flask equipped with a stirrer, a condenser, and a thermometer. The flask was immersed in water bath at 80°C for 30 min to obtain liquefied NR-modified ER.

Secondly, 1 kg urea (2–4 mm, 46% N) was loaded into a rotary drum blender (WKY-400, China) and preheated at 80 ± 2°C for 10 min. Then, 10.0 g liquefied NR-modified ER were poured onto the rotating urea prills and cured for 8 min. This step was repeated three times so that a total of 40.0 g liquefied NR-modified ER were used. Six CUs, i.e., CU0, CU1, CU2, CU3, CU4, and CU5, were prepared with NR mass proportion in coating of 0, 10, 20, 30, 40, and 50%, respectively.

The surface and cross-sectional morphologies of the CUs were observed with a scanning electron microscope (SEM; JSM-6610LV, Japan) (Tian et al., 2019).

A nitrogen release experiment was conducted to learn the appropriate mass proportion of NR for ER modification. Nitrogen release characteristics of the CUs were evaluated using the national standard method GB/T 23348-2009 (GAQS, IQPRC, SA, 2009). Briefly, 10.0 g CU were placed into a 250-ml glass bottle containing 200 ml distilled water at 25°C. Nitrogen concentration in the solution was determined using the Kjeldahl (1883) method after 1, 3, 5, 7, 10, 14, 28, 42, 56, and 84 days until cumulative N release rate was ≥80%, which is a common benchmark for complete release (Yang et al., 2013). Three replicates were set up for each CU. The CU with the best N release performance would be identified according to the controlled release period and its NR proportion in coating would be adopted in the subsequent preparation of CSUs.

In order to test the effect of different formulations of inhibitor and urea on inhibitor activity, two types of uncoated, stabilized EENFs (referred to as SUs) were prepared: SU1 and SU2. The first type, SU1, was prepared by mixing 1 kg urea thoroughly with 1.15 g NBPT and 2.30 g DMPP. The preparation procedure of SU2 was similar to that of the CUs. Briefly, 1.15 g NBPT and 2.30 g DMPP were dissolved in 80 ml 75% ethanol solution. After 1 kg urea was loaded into a rotary drum blender (WKY-400, China) and preheated at 80 ± 2°C for 10 min, 20 ml of the NBPT and DMPP containing ethanol solution was poured onto the rotating urea prills and cured for 8 min. This step was repeated three times and SU2 was obtained.

Two CSUs (i.e., CSU1 and CSU2) were prepared in the same way as the CUs except that for CSU1, the first 10 g liquefied NR-modified ER contained 1.15 g NBPT and 2.30 g DMPP, and for CSU2, SU2 was used instead of urea. That is, NBPT and DMPP were homogeneously distributed in the innermost layer of the coating in CSU1, whereas they were sandwiched between the coating and the urea in CSU2. The surface and cross-sectional morphologies of the CSUs were observed with a SEM (JSM-6610LV, Japan) (Tian et al., 2019).

Two experiments were conducted to evaluate the N loss potential, including volatilization and leaching loss, of the SUs and CSUs.

The soil used in this study, including these two experiments and the pot experiment described later, was collected from the experimental station of the College of Resources and Environment, Shandong Agricultural University, China. The basic physical and chemical properties of the soil are shown in Table 1.

For NH₃ volatilization loss evaluation, urea, CU3, the SUs, and the CSUs were each mixed thoroughly with 250 g soil at 0.6 g N/kg dry soil in plastic boxes. In addition, boxes containing only 250 g soil without fertilizer (CK) were also prepared. After soil moisture was adjusted to 60% of field water holding capacity with distilled water, a Petri dish containing 10 ml 3% boric acid indicator solution was placed on the soil as a trap for volatilized NH₃. The boxes were sealed and incubated at 25°C in the dark. The boric acid traps were replaced with new ones at regular intervals during the 40 days of incubation and titrated with sulfuric acid standard solution (0.005 mol/l) for NH₃ quantification (Zhou, 2017).

A column leaching experiment was conducted for evaluation of ammonium and nitrate leaching loss potential of the CU3, SUs, and CSUs. PVC columns (6.0 cm in diameter and 15.0 cm in height) were first packed with a thin layer of quartz sand, then 2000 g soil, and finally a thin layer of quartz sand again. The SUs and CSUs were each mixed thoroughly with the top 5 cm of soil at 0.6 g N/kg dry soil. Columns without fertilizer added were also set up as control (CK). Distilled water was added to saturate the soil for 24 h. Then, 100 ml distilled water was added to the column every 4 days for a total of 1000 ml water in 40 days. Leachate was collected for NH₄⁺-N and NO₃^–-N determination using an AA3 continuous flow analyzer (BL-TECH, Germany). Soil inorganic N content was calculated as the sum of NH₄⁺-N and NO₃^–-N.

The pot experiment adopted a completely randomized block design. Seven treatments of different N fertilizations were set up: CK (no application of N fertilizer), urea, CU3, SU1, SU2, CSU1, and CSU2. Plastic pots (lower diameter 23.0 cm, upper diameter 35.0 cm, and height 43.5 cm) were used, and 15 kg of soil were put in each pot. The N fertilizers were applied at 0.15 g N/kg dry soil (equivalent to 337.5 kg N/ha). Calcium superphosphate was used as phosphorus fertilizer at 0.1 g P₂O₅/kg dry soil (equivalent to 225 kg P₂O₅/ha) and potassium chloride as potassium fertilizer at 0.1 g K₂O/kg dry soil (equivalent to 225 kg P₂O₅/ha). The phosphorus and potassium fertilizers, SUs, and CSUs were applied as basal fertilizers in the subsurface soil layer, while urea was split-applied as basal fertilizer (50%) and topdressing (50%, at the jointing stage). Five seeds of maize (Zea mays Ziyu 2) were sown in each pot on June 10, 2020 and thinned to one seedling at the 5-leaf stage.

At the seedling (June 20), jointing (July 10), tasseling (August 10), flowering (August 17), and mature (September 22) stages of maize, three pots were randomly taken from each treatment. Plant heights were measured. After maize roots were removed, fresh soil samples were taken and extracted with 1 mol/l KCl for determination of NH₄⁺-N and NO₃^–-N with an AA3 continuous flow analyzer (BL-TECH, Germany). Urease activity was measured by the sodium phenate-sodium hypochlorite colorimetric method (Santos et al., 2020) using air-dried soil (<1 mm). Maize was harvested on October 22. Ears per plant, grains per ear, and 100-grain weight were recorded. Plant and grain samples were first oven-dried at 105 °C for 30 min and then at 75°C to constant weight. The samples were digested with H₂SO₄/H₂O₂ and N content was determined by the Kjeldahl (1883) method. Maize grain yield, N utilization efficiency (NUTE), partial factor productivity of applied N (NPFP), N fertilizer apparent utilization efficiency (NFUE), and N use efficiency (NUE) were calculated as follows:

Soil apparent nitrification rate was calculated as:

Microsoft Excel 2016 was used for data processing and Origin 2021 software was used for figure drawing. Data were subjected to analysis of variance (ANOVA) using IBM SPSS Statistics 21.0 and means were separated by Duncan’s multiple range test (P < 0.05).

The surface and cross-sectional morphologies of the CUs are shown in Figure 1. The surface of the CUs became rougher with more NR in the coating. Pin holes of various sizes were observed on the surface of CU0 (Figure 1A). There were tiny bumps on the surface of CU3 (Figure 1D), whereas the surfaces of CRU4 and CRU5, the latter in particular, showed sheet structure (Figures 1E,F). The coating of CU0 displayed a compact structure, whereas that of CU1 was porous (Figures 1G,H). Loose sheet structure of the coating became more prominent with more NR (Figures 1I–L).

Nitrogen was released rapidly from CU4 and CU5, with first-day release of 44 and 42%, respectively (Figure 2). After only 7 days of incubation, over 70% of total N had been released from the two CUs, and after 28 and 42 days, 80% of total N had been released from CU4 and CU5, respectively. In contrast, N release from CU2 and CU3 was much slower in the early stage of incubation, with merely approximately 15% being released in the first week. The N release period of CU2 and CU3 was 75 and 73 days, respectively, comparable to that of CU0 (84 days). For CU1, over 30% of total N was released in the first day of incubation, nearly 50% was released in the first 7 days, and 80% was released in 61 days. In a word, of the five NR-modified ER-coated EENFs, CU2 and CU3 presented the most comparable N release performance to that of CU0. Their first-day N release rates were far lower than 15% and their N release periods were approximately two and a half months, well meeting the requirement for a controlled release N fertilizer stated in the national standard method GB/T 23348-2009. The N release characteristics of the CUs were closely related to the structure of their coatings. The rapid N release from CU4 and CU5 is attributed to the loose sheet structure of their coatings, whereas the rapid N release from CU1 is due to the porous structure of its coating (Figure 1). Both loose sheet and porous structures are favorable for the diffusion of water through the coating and subsequent diffusion of N solution out to the soil (i.e., N release). Considering that the N release pattern of CU3 was more similar to that of CU0 and that a smaller proportion of ER would be better for the environment, the NR mass proportion of 30% in coating was adopted for the subsequent preparation of CSUs.

A large amount of NH₃ volatilized from urea was measured in the first week of incubation, with peak volatilization of 1.13 mg NH₃-N at day 3 (Figure 3A), indicating rapid hydrolysis of conventional urea after applied to the soil. Ammonia volatilization in the conventional urea treatment decreased continuously from day 4 to 9, fluctuated in day 10–19, and then decreased continuously again, with daily volatilization lower than those of the CU3, SU, and CSU treatments from day 16 on. In contrast, NH₃ volatilization from the CU3, SU, and CSU treatments, which mainly occurred during days 7–17, was small during the entire incubation period, with even the peak daily volatilization smaller than 0.4 mg NH₃-N. The NH₃ volatilization in the CU3 reached the peak on the 14th day, which was 4 days later than the peak in the SU2. The NH₃ volatilization rate of EENFs was higher than that of conventional urea during days 23–40. On the 40th day, the NH₃ volatilization rate of CU3 was significantly higher than that of SUs and CSUs. During the cultivation stage, the NH₃ volatilization accumulation of conventional urea was higher than that of EENFs (Figure 3B). Compared with the conventional urea treatment, the NH₃ volatilization was significantly decreased by 10, 12, 13, 14, and 20% in CU3, SU1, SU2 CSU1, and CSU2, respectively. In addition, NH₃ volatilization from the CSU2 treatment was significantly lower than those from the CU3, CSU1, and SUs treatments, whereas there were no significant differences between CSU1 and SUs.

Leaching loss of NH₄⁺-N, NO₃^–-N, and inorganic N from the CU3, SUs, CSUs, and conventional urea are shown in Figure 4. The peak NH₄⁺-N leaching rate occurred at day 8, 16, 12, 16, 12, and 24 in the urea, CU3, SU1, SU2, CSU1, and CSU2 treatments, respectively (Figure 4A). Compared with the urea treatment, the peak NH₄⁺-N leaching rate decreased by 52, 20, 39, 57, and 66% in SU1, SU2, CSU1, and CSU2, respectively. Throughout the leaching experiment, NH₄⁺-N was the major inorganic N in the leachates from the conventional urea, SU1, and SU2 treatments. In addition, the leached amount of NH₄⁺-N from the CSUs treatments was significantly lower than those from the SU1 and SU2 treatments, whereas there were no significant differences between CSU1 and CSU2 (Figure 4B). Furthermore, the leached amount of NH₄⁺-N from the SU treatments was significantly less than that from the conventional urea treatment, and the leached amount of NH₄⁺-N from the CSU treatments was significantly less than that from the SU treatments. In the 40 days of incubation, the cumulative leached amount of NH₄⁺-N was 46, 13, 22, 50, and 54% lower from CU3, SU1, SU2, CSU1, and CSU2, respectively, than from conventional urea.

The leaching dynamics of NO₃^–-N was different from that of NH₄⁺-N. It can be seen from Figure 4C that the NO₃^–-N leaching rate of conventional urea was higher than those of the CU3, SUs, and CSUs in the first 12 days. This is due to the rapid hydrolysis of urea, which generated a large amount of NH₄⁺-N, the substrate of nitrification, and accelerated the production of NO₃^–-N. The NO₃^–-N leaching rate in the conventional urea treatment decreased from day 20 to 28 and was lower than those in the SU and CSU treatments during day 24–28. In the 40 days of incubation, the cumulative leached amount of NO₃^–-N was 38, 47, 42, 43, 53, and 42% from U, CU3, SU1, SU2, CSU1, and CSU2, respectively (Figure 4D). The decrease was attributed to the rapid hydrolysis of urea and leaching of NH₄⁺-N out from the column, which led to weak nitrification and low NO₃^–-N in the later stage. In contrast, in the SU and CSU treatments, NO₃^–-N leaching rate was low in the early stage and increased in the later stage. This is attributed to the presence of DMPP, which inhibits nitrification. With the gradual dissolution of DMPP, its inhibition on nitrification became weaker and more NH₄⁺-N was converted to NO₃^–-N.

The leaching loss of soil inorganic N was calculated (Figure 4F). After the incubation, the leaching loss of soil inorganic N was significantly reduced in CU3, SU2, CSU1 and CSU2 by 17, 6, 13, and 26%, respectively, compared with the conventional urea treatment. There was no significant difference in the leached amount of inorganic N between the SU1 and conventional urea treatments. In terms of leached amount of applied N, the fertilizers were in the order of: conventional urea and SU1 > SU2 > CSU2 and CU3 > CSU2.

Nitrogen requirement for maize during the growth stage followed an “S-shaped” curve (Zheng et al., 2017). Therefore, N supplement according to the different nutrient demands in each growth stage is particularly important for a high and stable yield of maize. Conventional quick-acting N fertilizer needs to be topdressing during the crop growth period. Advantageously, the basic application of controlled release EENFs can meet N demand during the crop growth stage and significantly improve the matching degree between soil N supply and crop nutrient demand (Geng et al., 2016; Xiao et al., 2019), promote crop yield and NUE. It was found that the N release performance of controlled release EENFs is related to the characteristics of coating material and thickness, as well as environmental factors such as soil temperature and moisture (Zhang et al., 2018). The difference in coating material characteristics is the direct factor affecting the N release of controlled release EENFs. An et al. (2021) reported that the nutrient release of coated fertilizers mainly has three stages, including the transport of water into the capsule, dissolution of fertilizers and release of the nutrients through coating materials. This experiment used natural rubber (NR)-modified ER as coating material to prepare coated EENFs. The results of the SEM (Figure 1D) showed that the surface of CU3 was smooth, without apparent protrusions, and the surface film holes were moderate. There were a few nutrient channels in the cross section of CU3. It is speculated that the controlled release mechanism of CU3 was consistent with the nutrient diffusion mechanism of conventional resin-coated urea. After coated controlled release EENFs were applied to the soil, as soil moisture penetrates into the membrane shell and dissolves the urea core, the osmotic pressure difference between the inside and outside of the membrane gradually increases. The pressure on the membrane shell changes the microstructure (the density of nutrient channels on the membrane shell increases), and the N was slowly released through the nutrient channel under the action of osmotic pressure. The hydrostatic release curve showed that the nutrient release process of CU3 could be roughly divided into two stages. (1) The first stage (1–7 days) is the slow-release period, in which the cumulative release of N was 20%; (2) The second stage (8–56 days) is the rapid release period, and the cumulative amount of N was 71%. The “N backward shift” phenomenon in hydrostatic release was consistent with high-yield maize still needed to absorb more N to meet crop material synthesis demand in the mid-late growth stage (Wang et al., 2010). The results demonstrated that the content of NR would reduce the controlled release performance as compared with CU0. However, controlling the amount of NR could reduce the amount of ER and reduce the pressure of resin on the soil environment without significantly changing the controlled-release performance of CU0.

To strengthen the matching degree between crop nutrient requirements and soil nitrogen supply, NR mass proportion of 30% in coating was adopted in the subsequent preparation of two EENFs: CSU1 (inhibitors homogeneously distributed in the innermost layer of the coating) and CSU2 (inhibitors sandwiched between urea and coating). Theoretically, the slow dissolution of N and NBPT + DMPP could be controlled simultaneously to realize the dual regulation of conventional urea dissolution and transformation. Li et al. (2020a) reported that the slow-release periods of coated controlled-release urea for N, hydroquinone (HQ), and dicyandiamide (DCD) were 56, 42, and 14 days, respectively. However, because determination of NBPT and DMPP content requires high-performance liquid chromatography (HPLC) (Santos et al., 2020). Therefore, it is difficult to directly determine the activity of NBPT and DMPP in EENFs and the controlled release performance of the organic membrane shell on the inhibitor. Furthermore, there is uncertainty about the true DMPP and NBPT application rate with the coated fertilizers due to eventual loss of inhibitor while treating the fertilizer with the coating. For this reason, we used other methods to prove the sustained release effect of the physical film on N, NBPT, and DMPP. Firstly, we compared the SEM of CU3 and CSUs to clarify the effect of adding inhibitors on the envelope. Secondly, we compared the effects of CU3, SUs, and CSUs on soil N regulation ability, corn growth and yield to clarify the combined effect of coating, inhibitor, and coating + inhibitor. The SEM images of the surface and cross-section of CSU1 and CSU2 showed that spraying the inhibitor on the urea surface could improve the smoothness of the urea surface and make the film closely combined with urea (Figure 8D). Compared with CU3 (Figure 1D), the surface of the membrane shell of CSU2 was smooth, and the adhesion between the membrane shell and the urea surface was closer (Figures 8C,D). Combined with the CU3 nutrient release characteristic curve (Figure 2) and the scanning electron microscope test of CSUs (Figure 8), CSU1 and CSU2 could simultaneously realize the slow dissolution of N and inhibitors, and the controlled release performance of CSU2 was better than that of CSU1.

Both NH₄⁺-N and NO₃^--N are the primary N forms absorbed by the plants (Zheng et al., 2017) and also the primary source of N loss (Xiao et al., 2019). After urea is applied to soil, it is rapidly hydrolyzed to (NH₄)₂CO₃ under the soil urease, decomposition of the (NH₄)₂CO₃ produces NH₄⁺-N. The rapid hydrolysis of urea will cause a large amount of surplus NH₄⁺-N, which is converted to NH₃ (Adotey et al., 2017). However, this phenomenon provides sufficient substrate for nitrification so that soil nitrification is active (Xiao et al., 2019), resulting in the loss of NO₃^--N, N₂O, and N₂. Urease and ammonia-oxidizing bacteria are the main factors of the soil N cycle. Inhibiting urease and ammonia-oxidizing bacteria can delay urea hydrolysis and transformation. Many studies have improved the duration of inhibitors by controlling the slow release of inhibitors. Li et al. (2020a) reported that the combined coating of HQ and urea could significantly reduce soil urease activity and prolong the action time of urease inhibitors during the growth period of wheat. Santos et al. (2020) found that the combined NBPT and biodegradable biofilm materials prepare controlled release urea, which could prolong the action time of NBPT and reduce the loss of N and NH₃ volatilization. Zhang (2004) reported that the controlled dissolution of urea and DCD could significantly improve the nitrification inhibition effect. In this study, the results of the EENFs on NH₃ volatilization loss and leaching loss were consistent with previous research conclusions. In addition, the leaching and NH₃ volatilization tests were carried out without crop interference. The soil NH₃ volatilization curves of SU1, SU2, CSU1, and CSU2 changed steadily throughout the culture stage (Figure 3). In the later stage of the leaching test, the leaching rate of NO₃^--N of CSU1 and CSU2 was higher than that of SU1, SU2, and U (Figure 4). The changing trend of Figures 4E,F showed that the NBPT + DMPP could reduce N loss at the initial stage of fertilization, ensure sufficient N supply in the later stage of soil, which was also confirmed in the maize pot experiment (Figure 5). The results of the N loss potential clearly demonstrated that the urease and nitrification inhibitors effectively reduced N loss from urea, but their effectiveness weakened and N loss accelerated over time. Furthermore, compared with SU, CU3 was better in reducing N leaching loss, because the existence of the coating layer hinders the direct contact of nitrogen with the soil. When inhibitors treated urea were coated, both the fertilizer and the inhibitors were protected and the effective period of the inhibitors was prolonged. In addition, the inhibitors were more effective when sandwiched between the coating and the urea (CSU2) than when treated in the inner layer of the coating (CSU1).

Consistently, the higher availability of nutrients in the soil, the greater nutrients utilization by the crop plant (Alhaj Hamoud et al., 2019a). The change trend of inorganic N in maize pot experiment was found that different additional methods of NBPT + DMPP would affect soil N loss and soil N supply capacity. SU1 and SU2 reduced N leaching loss, while the accumulation of leaching loss increased rapidly in the later stage of culture as compared with the conventional urea. The leaching rate curves of CSU1 and CSU2 were stable during the culture period, these may be due to the physical film on the outer layer of the inhibitor, CSU1 and CSU2 could maintain the appropriate content of soil nutrients during the growth stage of maize (Figure 5), which might be due to CSU2 having two main regulatory effects on soil N transformation. Firstly, the physical film avoided the direct contact between urea and inhibitor and soil and controlled the slow release of urea and inhibitor. Secondly, the combination of NBPT and DMPP can, respectively, inhibit urease activity (Figure 6) and reduce the apparent nitrification rate of soil NH₄⁺-N (Table 2), and regulate the transformation process of dissolving N. In addition, the results of the maize pot experiment showed that the NH₄⁺-N in CSU2 soil reached the maximum at the heading stage of maize (Figure 5) to ensure sufficient N supply in the middle and late stage of maize growth.

Crop yield and NUE are essential indicators for rational fertilization (Feng et al., 2020). The researchers (Yang et al., 2011; Geng et al., 2016; Zheng et al., 2016) reported that controlled-release urea met the long-term N demand of crops and improved crop yield and NUE. Li et al. (2020b) discussed that the combination of urea and HQ coated controlled-release urea could increase wheat yield by 56% compared with U. This experiment found that EENFs increased production significantly compared with conventional urea, consistent with the above report. Compared with U, CU3, SUs, and CSUs could promote the growth of maize and significantly increase maize yield and NUE 12–79% (Table 3) and 10–59% (Table 4), respectively. The growth stage of summer maize is from June to October each year, and the soil temperature and moisture are relatively high, which can accelerate the process of soil N conversion. Uncoated, stabilized EENFs (SUs) could regulate N transformation in the early stage of maize growth, however, the regulation ability of N delayed in the late stage of maize growth. In addition, CU3 could prevent the dissolution of N to a certain extent, but it was also greatly affected by the soil environment. The inhibitor and urea were co-coated (CSUs), and the hindering effect of the physical film layer could prolong the dissolution time of the inhibitor and N, and reduce the degradation and fixation of the inhibitor in the soil. The combined effect of coating and inhibitor strengthens the N supply capacity of the soil, guarantees effective nutrient supply in the middle and late stages of maize growth. Furthermore, the application of coated, stabilized EENFs in maize pot experiment was conducive to improve the matching degree between soil nutrient supply and maize nutrient demand, so that CSUs was better than SUs and CU3 in improving maize yield and NUE.