Reaction Mechanism of Nitrogen Compounds in Gaseous Ethylene Followed by Catalytic Combustion‐Chemiluminescence

ABSTRACT

Herein, a practical method for the determination of various organic and inorganic nitrogen (N) compounds in gaseous ethylene was established via the catalytic combustion–chemiluminescence technique. First, the regularity of equal N mass/molar response for N compounds was summarized by evaluating the conversion behavior of different speciation (bipyridine, acetonitrile, NO, NO₂, NH₃, N₂O). Then, based on the assumption that N response is unaffected by speciation, the actual N mass in gaseous samples was accurately calculated using the Van der Waals equation (to correct gas mass), and the theoretical N mass was derived from standard curves. As a result, the recoveries of N compounds at different levels were in the range of 106.0%–108.8% (15.0 μmol/mol acetonitrile), 103.2%–106.0% (24.1 μmol/mol NO), 100.1%–102.5% (10.1 μmol/mol N₂O + 10.1 μmol/mol NO₂), and 90.1%–95.5% (51.0 μmol/mol NH₃), successfully verifying the above assumption without the need to rebuild quantitative relationships for different N systems. In addition, the N detection limit (S/N = 3) in ethylene was 0.03 mg/kg, with intraday and interday relative standard deviations (RSDs) of 1.0%–2.3%. Finally, the reaction mechanism based on system passivation and catalytic combustion was successfully applied to unify various nitrogen compounds analysis, achieving gas–liquid equal mass/molar response.

A catalytic combustion–chemiluminescence approach is constructed for nitrogen measurement in gaseous ethylene, with Van der Waals‐based precise correction adopted to elucidate the reaction mechanism of nitrogen compounds.

1. Introduction

Ethylene is the basic raw material of the petrochemical industry. It is mainly used to produce polyethylene [1, 2] and downstream products including ethylene glycol, polyvinyl chloride, styrene, and acetic acid [3, 4]. Moreover, it is widely used in plastics, fibers, synthetic rubber, and other materials, as well as surfactants, adhesives, coatings, and other fields [5, 6]. However, the presence of organic and inorganic nitrogen (N) in ethylene products can be a concern in various cracking processes. N compounds in ethylene include speciation of NO, NO₂, N₂O, NH₃, or nitrile compounds. Among them, NO, NO₂, and N₂O are mainly derived from the reaction of N₂ and O₂ at high temperature. NH₃ is mainly produced by high‐temperature decomposition of N compounds. Nitrile compounds are mainly produced by the reaction of ethylene and cyanide. These N compounds will have a certain impact on ethylene quality, catalyst activity, and industrial device safety. The control and analysis of N compounds have received more and more attention. For example, NH₃ can form flammable or explosive mixtures. If leakage or pressure loss occurs, it may lead to a container explosion or equipment failure. The presence of NO or NO₂ may not only cause some N oxides to be adsorbed in ethylene, affecting the purity and quality, but also adversely affect downstream chemical processes. Under certain conditions, NO or NO₂ may form combustible mixtures, increasing the risk of fire and explosion. Therefore, it is of great social significance to strictly monitor the N content in the product gas.

The Kjeldahl [7, 8] and colorimetric [9] methods are well recognized as traditional approaches for N determination. However, as chemical analysis methods, they require complex pretreatment processes prior to titration or colorimetry, exhibit low detection sensitivity, and are thus restricted to a narrow application scope. Since the 1990s, the development of chemiluminescence has achieved a qualitative leap in N analysis. Several methods have been published successively with greatly improved sensitivity and greatly reduced detection limit in the petrochemical industry [10, 11, 12, 13, 14, 15, 16, 17, 18]. This technology was used for the rapid analysis of total N content in samples and was further developed to be coupled with GC for the analysis of N speciation in samples [15, 16, 17]. However, the application of the combustion–chemiluminescence method for the determination of total N was largely limited to petroleum matrices [10, 11]. Existing studies mostly recommend the application of such methods to liquid petroleum products with different compositions or boiling points [13, 14]. These methods are not suitable for light olefins, and systematic research on various N compounds is still insufficient, especially in terms of inorganic gases that are prone to forming explosive mixtures, such as NO₂, N₂O, and NH₃. Light olefin samples differ significantly from liquid petroleum hydrocarbon and aromatic hydrocarbon samples in terms of speciation, sample injection methods, and quantitative methods.

The sample enters the combustion tube through the carrier gas and mixes with oxygen. Only when the nitrogen compounds are converted into NO can they further react with O₃ to generate excited N dioxide (NO₂*). When NO₂* returns to the ground state, a light signal will be detected [10, 18], which is the basic principle of the chemiluminescence method (Figure 1). However, there are few systematic studies on the mechanism of the conversion pathways or the conversion rates of various N compounds. It is worth noting that the chemiluminescence method has certain limitations in the determination of azides [19]. This is because the decomposition of azides usually produces N₂ [20], which has extremely stable properties and is difficult to determine by the chemiluminescence method. This paper focuses on the research of common organic or inorganic N impurities in ethylene but does not involve azides.

FIGURE 1.

Principle of N determination by chemiluminescence.

Ethylene is normally gaseous, and gas standard samples can be used to achieve accurate quantification through experiments. However, a single commercial gas standard sample bottle is bulky and costly and has a short validity period. By contrast, liquid standard samples have the advantages of small volume, low cost, and long validity period and are more economical and simpler in practical use. Therefore, the use of liquid standards for method development has significantly reduced the cost of quality control, and what follows is the difficulty in gas–liquid conversion computation. This study developed an analytical method for quantifying gaseous ethylene using economically feasible liquid standards. The molecular behaviors of various N compounds on the detector were also explored to determine whether a unified calculation method can be established, providing a detailed basis for the high‐sensitivity determination and the clarification of its reaction characteristics following chemiluminescence.

2. Materials and Methods

2.1. Reagents and Materials

N concentrations of 0.2, 0.5, 1.0, 5.0, 10.0, 20.0, 40.0, 50.0, and 100.0 mg/L (the primary N compound is bipyridine; the matrix is isooctane; qualitative analysis is referred to in Section 3.1) were purchased from Sinopec Research Institute of Petrochemical Technology Co. Ltd. for quantification. The commercially available standard gases, 15.0 μmol/mol acetonitrile (matrix of ethylene), 24.1 μmol/mol NO (matrix of ethylene), 9.9 μmol/mol N₂O (matrix of ethylene), a mixture of 10.1 μmol/mol N₂O and 10.1 μmol/mol NO₂ (matrix of N₂), and 51.0 μmol/mol NH₃ (matrix of ethylene), used during the experiments were all obtained from Dalian Special Gases Co. Ltd. for verification. The actual samples of ethylene1# and ethylene2# were taken from the ethylene plants of Sinopec Shanghai Research Institute of Petrochemical Technology Co. Ltd. Sample bottles (4 mL) used for sample injection were obtained from Dikma. Air bags (2 L) for sample collection were purchased from DuPont, Tedlar PVF.

2.2. Instruments and Conditions

The response to N was characterized using a sulfur‐nitrogen analyzer model NSX‐2100V from Mitsubishi Chemical Group. The liquid samples were injected using the automatic injection device, and the gaseous samples were injected using the gas sampling equipment and brought into the sulfur‐nitrogen analyzer combustion tube with a reaction temperature of 900°C, Ar flow rate of 100–130 mL/min, O₂ flow rate of 500 mL/min, O₃ flow rate of 300 mL/min, injection temperature of 25°C, and a volume of 20 μL or 10 mL for liquid or gas injection. After full combustion, the products were transferred to the detector. Qualitative analysis of the raw material (liquid standard samples) was performed using a gas chromatography–mass spectrometer (GC–MS, model 7890A‐5977C) from Agilent Technologies Inc., with the following operating conditions: inlet temperature of 300°C, split ratio of 5:1, injection volume of 1 μL, column of HP‐1 (30 m × 0.25 mm × 0.25 μm, Agilent Technologies Inc.), flow rate of 1 mL/min, and a temperature gradient from 50°C to 300°C.

2.3. Quantification

The quantitative method examines the accuracy of quantifying gaseous samples using economically viable liquid standard samples. Investigating whether N compounds in gaseous or liquid standard samples are converted to the same speciation (NO) with equimolar amounts of N during combustion is a prerequisite for establishing gas–liquid conversion calculation methods and a unified quantification equation. Additionally, regardless of whether the matrix is gaseous or liquid, the conversion rates of different N compounds and solutions for improving low conversion rates constitute another focus of this study. It is assumed that under gas or liquid injection conditions, the aforementioned N compounds can be quantitatively or completely converted to NO, and the relationship between N mass and response value is linear within a certain range. If this assumption holds, it implies that for different N compounds, matrices, and gas or liquid states, the response value is only related to the N mass or N molar—as exemplified by the two injection modes shown in Figure 2—which would enable the establishment of a unified quantification formula. If the assumption is invalid, it will be necessary to separately explore the response laws of different N compounds under various matrix conditions.

FIGURE 2.

Combustion process of liquid or gas sample.

3. Results and Discussion

3.1. Liquid Standard Samples

GC–MS analysis was performed on externally purchased liquid standard samples mentioned in Section 2.1. The results showed that the matrix was isooctane with trace amounts of 2,5‐dimethyl‐hexane, 2,4‐dimethyl‐hexane, and 2,3‐dimethyl‐pentane. The primary speciation identified was bipyridine, accompanied by small quantities of 2,4‐diisocyanato‐1‐methyl‐benzene (Figure 3).

FIGURE 3.

GC–MS analysis of purchased standard sample.

Replicate measurements were conducted twice on commercially purchased series standard solutions at N concentrations of 0.2, 0.5, 1.0, 5.0, 10.0, 20.0, 40.0, 50.0, and 100.0 mg/L. N mass and corresponding mean net peak area standard curve were established. Within the range of N mass from 4.0 to 2000.0 ng (0.2–100.0 mg/L with an injection volume of 20 μL), a standard curve was constructed, showing excellent linearity with a linear equation of y = 9.3774x + 75.0069 and R ² = 0.9996 (Figure 4a), which indicated that N in these organic liquid standards themselves followed an equal mass (or equal molar) response by chemiluminescence. However, when the net peak area of each standard concentration point was used to back‐calculate the N mass via the aforementioned equation and compared with the theoretical N mass, it was found that the recoveries in the low concentration range were below 80% (Figure 4d). The low‐concentration points were forced into linearity, making it impossible to accurately quantify the N content at these points. Due to the forced linearity, a single standard curve cannot accurately quantify the N content at low concentrations. Thus, segmented standard curves were established for the ranges of 4.0–200.0 ng and 200.0–2000.0 ng. The linear equations were y = 10.6541x + 28.1267 with R ² = 0.9998 (Figure 4b) and y = 9.3634x + 92.3772 with R ² = 0.9993 (Figure 4c), respectively. When the net peak area values for each point were used to calculate the recoveries for the standard samples, the results were controlled at 96.9%–119.6% (Figure 4e) and 97.8%–110.8% (Figure 4f), respectively. Therefore, using segmented standard curves allows for accurate quantification of a wider range.

FIGURE 4.

One curve (a) with standard sample recoveries (d), and segmented standard curves (b and c) with standard sample recoveries (e and f).

3.2. Quantification of Gaseous Samples

3.2.1. Acetonitrile

Neglecting the volume of gas molecules and mutual attractive forces, the theoretical N mass was calculated using Equation (1) based on the ideal gas equation PV = nRT {101,325 (Pa) × 22.414 (L) = 1 (mol) × R × 273.15 (K), where R = 8314.47 [Pa•L/(mol•K)]}. The calculated theoretical N mass was 85.8 × 10⁻⁹ g, equivalent to 85.8 ng. Herein, m ₁, P, V ₁, M ₁, R, and T represent the theoretical N mass (g), pressure of gas (101,325 Pa), injection volume of the gas sample (0.01 L), molar mass of N (14.0 g/mol), conversion coefficient {R, 8314.47 Pa•L/(mol•K)}, and temperature of the sample (298.15 K), respectively.

$$m_1={\mathit{PV}}_1{M}_1/\left(\mathit{RT}\right)\times 15.0\times {10}^{-6}$$ (1)

For the nonideal behavior of gases, corrections were applied to the pressure term (n ² a/V ²) and the volume term (nb), resulting in the Van der Waals Equation (2). The Van der Waals equation takes into account the attractive forces between molecules (through the parameter a) and the volume of gas molecules (through the parameter b). Substituting n = m/M into Equation (2) yields Equation (3), where m ₂, R, T, M ₂, P, a, V ₂, and b represent the theoretical N mass (precise mass, g), conversion coefficient [R, 8314.47 Pa•L/(mol•K)], temperature of the sample (298.15 K), molar mass of N (equal to M ₁, 14.0 g/mol), pressure of gas (101,325 Pa), Van der Waals constant related to intermolecular attraction (a _ethylene = 4.612 × 10⁵ Pa•L²/mol²) [16], volume of N‐containing compound counted by 1 N (L), and Van der Waals constant related to the volume of gas molecules (b _ethylene = 0.0582 L/mol) [21], respectively. After calculations, a cubic equation for m ₂ was obtained. The program code used was “import numpy as np ↵coefficients = [4347560889.4e5, −156870748.4e2, 17748973.9e‐2, −151987.5e‐7] ↵ roots = np.roots (coefficients) ↵print(“The approximate roots are:”, roots),” and the theoretical precise N mass was 86.3 × 10⁻⁹ g. The derivations of the mathematical models and equations presented have been relocated to Appendix A.

$$\left(P+n^{2}a/V^2\right)\times \left(V-\mathit{nb}\right)=\mathit{nRT}$$ (2)

$$\left(P+{m_2}^{2}a/{V_2}^2/{M_2}^2\right)\times \left(V_2-m_{2}b/M_2\right)=m_2\mathit{RT}/M_2$$ (3)

Nevertheless, during gas sample testing, the total flow rate was the sum of the combustion tube branch and the sample introduction branch (combustion tube branch flow rate of 40 mL/min and sample introduction branch flow rate of 60 mL/min), and the net peak area results were input into the segmented liquid standard curve equation to calculate the actual N mass of 93.0 ng. Therefore, the recovery N mass was 107.8% (Table 1). This preliminary finding indicated that whether in liquid or gas samples, regardless of the organic speciation including pyridine and acetonitrile, in different matrices including organic isooctane or ethylene, the combustion mechanism remained consistent. However, further investigation revealed that with an increase in carrier gas flow rate, the response of N gradually increased. Using the reference flow rate of the combustion tube branch as 40 mL/min and the sample introduction branch flow rate as 60 mL/min, various scenarios were explored (Table 1). It was found that when the total flow rate exceeded 100 mL/min, the recoveries increased with the increase in total flow rate (123.6%–155.2%, Table 1). This emphasizes that the six‐port valve gas sampling device needs to strictly control the total carrier gas flow rate.

TABLE 1.

Recoveries of acetonitrile by six‐port valve gas sampling device.

Standard sample	Ar flow rate (mL/min)	Actual N mass (ng)	Theoretical N mass (ng)	Recoveries (%)
Acetonitrile/matrix of ethylene	100.0 (40.0 + 60.0)	93.0	86.3	107.8
	100.0 (50.0 + 50.0)	93.0	86.3	107.8
	100.0 (40.0 + 60.0)	95.8	86.3	111.0
	110.0 (50.0 + 60.0)	106.7	86.3	123.6
	120.0 (60.0 + 60.0)	113.2	86.3	131.2
	130.0 (70.0 + 60.0)	114.3	86.3	132.4
	130.0 (40.0 + 90.0)	112.4	86.3	130.2
	160.0 (100.0 + 60.0)	135.0	86.3	156.4
	160.0 (40.0 + 120.0)	133.9	86.3	155.2

Further, the above‐mentioned six‐port valve gas sampling device was replaced with an injection‐type gas sampling device. The results indicated that with the injection‐type gas sampling device, the chemiluminescence reaction exhibited an equal molar (or equal mass) response over a wide range of Ar flow rate variations. As the total carrier flow increased to a certain extent, such as 160.0 mL/min, there was a slight decrease in the recovery (95.8%–97.6%, Table 2), which was consistent with the earlier experimental results that showed a slight decrease in the signal value of N when the total flow rate of liquid injection increased to 160 mL/min. Excessive carrier gas flow rate could lead to incomplete combustion and a lower recovery. Therefore, the optimal flow rate was 100–130 mL/min.

TABLE 2.

Recoveries of acetonitrile by injection‐type gas sampling device.

Standard sample	Ar flow rate (mL/min)	Actual N mass (ng)	Theoretical N mass (ng)	Recoveries (%)
Acetonitrile/matrix of ethylene	100.0 (40.0 + 60.0)	88.7	86.3	102.8
	130.0 (70.0 + 60.0)	87.1	86.3	100.9
	130.0 (40.0 + 90.0)	87.9	86.3	101.9
	160.0 (100.0 + 60.0)	84.2	86.3	97.6
	160.0 (40.0 + 120.0)	82.7	86.3	95.8

From the perspective of device structure, the mass of N injection is closely related to the Ar flow in a six‐port valve gas sampling device. During the load phase, the sample filled the quantification loop. Next, during the inject phase, valve switching flushed the sample from the quantification loop into the combustion tube (Figure 5a). When the flow rate exceeded a specific threshold (e.g., 120 mL/min), excessively rapid and large‐volume sample injection still led to carbon deposition, which in turn resulted in a high recovery rate. On the other hand, for the injection‐type gas sampling device (Figure 5b), during the load phase, a certain volume of the sample was extracted by the syringe; next, during the inject phase, the syringe injected the sample into the combustion tube (Figure 5b). This process has a simple flow path, and the injection pump itself operates at a relatively low flow rate (20 mL/min), which reduces the occurrence of carbon deposition. This also reasonably explains the data differences between the two sampling systems, as well as the need to control the flow rate range during their application.

FIGURE 5.

Schematic diagram of six‐port valve gas (a) and injection‐type gas sampling device (b).

3.2.2. N₂O + NO₂

Repeated measurements of 9.9 μmol/mol N₂O (matrix of ethylene) did not detect any response signal. N₂O with stable properties was more difficult to burn with O₂ or be oxidized to NO to participate in the chemiluminescence reaction. As the present system did not show a response to N₂O, the recovery calculation for this mixed gas (10.1 μmol/mol N₂O + 10.1 μmol/mol NO₂) considered only the contribution of NO₂ to the total N response. Referring to Equations (1)–(3), the program code used was “import numpy as np ↵coefficients= [2104564367.3e4, −2306863339.4, 177348607.8e‐3, −3070147.5e‐8]↵ roots=np.roots (coefficients) ↵print(“The approximate roots are:”, roots),” and the theoretical N mass was 173.5 ng (a _N2 = 1.370 × 10⁵ Pa•L²/mol², b _N2 = 0.0387 L/mol) [21]. The N recoveries were in the range of 92.6%–95.3%. These results indirectly confirmed that NO₂ and organic acetonitrile were almost similarly converted to NO and participated in the chemiluminescence reaction.

However, in further study, a platinum catalyst was added to the combustion tube, which could oxidize N₂O to NO and make the recovery of N₂O reach more than 95%. This is because the application of the catalyst facilitates the conversion of relatively stable N₂O to NO [22]. Figure 6 shows the spectrum change of this mixed gas before and after the addition of the catalyst, demonstrating the above statement. It is worth noting that the catalyst needs to be activated in the range of 800°C–1000°C for several hours to achieve good results. Otherwise, the reducing H₂S released by the partially activated catalyst will interfere with the oxidation and determination of N compounds. Besides, for trace N compounds, the suitable amount of catalyst used is 0.5–1.0 g. The application of the platinum catalyst provides strong support for the previously proposed hypothesis of “gas‐liquid equal mass (or equal molar) response.” We can continue to test more speciation along this line of thinking to verify whether the hypothesis holds or adopt auxiliary methods to make the hypothesis valid. Furthermore, the experimental results indicated that this method is also applicable to the analysis of N compound standards with N₂ as the matrix. This implies that the high‐sensitivity analysis of total N is not affected by matrix variations, even when the ethylene matrix is impure or mixed with partial inorganic gases.

FIGURE 6.

Spectrum of N₂O + NO₂ before (a) or after (b) the addition of the catalyst.

3.2.3. NO

Injecting 10 mL of 24.1 μmol/mol NO (matrix of ethylene), referring to Equations (1)–(3), the program code used was “import numpy as np ↵coefficients = [1684201717.7e5, −97637395.2e2, 17748973.9e‐2, −2441932.5e‐8]↵ roots = np.roots (coefficients) ↵print(“The approximate roots are:”, roots),” and the theoretical N mass obtained was 138.6 ng (a _ethylene = 4.612 × 10⁵ Pa•L²/mol², b _ethylene = 0.0582 L/mol) [21]. The integral net peak area was input into the liquid standard curve, and the actual mass was calculated based on the curve equation. The N recoveries were in the range of 103.5%–106.0%. This indicated that NO came into direct contact with O₃ and converted to the excited state NO₂*. This proved that the aforementioned N compounds were converted to NO with a conversion rate of nearly 100%, laying a foundation for the unified determination.

3.2.4. NH₃

Injecting 10 mL of 51.0 μmol/mol NH₃ (matrix of ethylene), similarly, the program code used was “import numpy as np ↵coefficients = [376086582.0e5, −4613845538.0, 17748973.9e‐2, −5167575.0e‐8]↵ roots = np.roots (coefficients) ↵print(“The approximate roots are:”, roots),” and the theoretical N mass obtained was 293.4 ng (a _ethylene = 4.612 × 10⁵ Pa•L²/mol², b _ethylene = 0.0582 L/mol) [21]. The integral net peak area was input to the liquid standard curve, and the actual mass was calculated based on the curve equation. By examining the recovery, it was found that the strong adsorption of NH₃ has become a key challenge in N determination. The significant difference with other speciation determination lay in the gradual increase of recovery with the continuous measurements until stabilization, after which the sample was removed, and the signal in the successive determination of air gradually recovered. This demonstrated the adsorption characteristics of NH₃. Under conditions where accurate determination of other speciation was feasible, the maximum recovery stabilized at 73.9% (Figure 7). Upon inspecting the adjacent pipeline at the outlet of the combustion tube, carbon deposit contamination was found. After replacing the pipeline, the maximum recovery stabilized at 95.8% (Figure 7), indicating that NH₃ follows a unified behavior principle during the combustion process and that the cleaning of pipeline carbon deposits contributes to the accurate determination of NH₃. During the operation of the sulfur‐nitrogen analyzer, it is crucial to regularly inspect pipelines for carbon deposit contamination, which is of great importance for the accurate determination of highly adsorptive N compounds such as NH₃. In this study, by comparing a clean pipeline with a carbon‐deposited one, the recovery of the NH₃ standard sample increased from 73.9% to 95.8%.

FIGURE 7.

Optimization of NH₃ analysis technique.

3.2.5. Repeatability and Recoveries

The above results verified that the various N compounds in gaseous samples including acetonitrile, NO₂, NO, N₂O, and NH₃ can be accurately determined by the segmented liquid standard curves. Recoveries were 106.0%–108.8% (15.0 μmol/mol acetonitrile), 103.2%–106.0% (24.1 μmol/mol NO), 100.1%–102.5% (10.1 μmol/mol N₂O + 10.1 μmol/mol NO₂), and 90.1%–95.5% (51.0 μmol/mol NH₃), respectively. Table 3 shows the experimental data of five repeated measurements, and the results show that the repeatability was good. Intraday and interday relative standard deviations (RSDs) were in the range of 1.0%–2.3%. The implications of these data lie in significantly improving the sensitivity of total N content control in industrial applications, and it particularly contributes to overcoming the potential limitations of highly hazardous N compounds like N₂O and NH₃.

TABLE 3.

Recoveries and repeatability.

Standard samples	Calculation	Repeats					RSD (%)
		1	2	3	4	5
Acetonitrile 15.0 μmol/mol	Actual N mass (ng)	91.5	92.4	93.8	93.9	92.5	1.1
	Theoretical N mass (ng)	86.3	86.3	86.3	86.3	86.3
	Recoveries (%)	106.0	107.1	108.7	108.8	107.2
NO2 + N2O 10.1 + 10.1 μmol/mol	Actual N mass (ng)	177.9	174.7	175.2	177.4	173.7	1.0
	Theoretical N mass (ng)	173.5	173.5	173.5	173.5	173.5
	Recoveries (%)	102.5	100.7	101.0	102.2	100.1
NO 24.1 μmol/mol	Actual N mass (ng)	146.8	146.9	146.9	143.0	143.5	1.4
	Theoretical N mass (ng)	138.6	138.6	138.6	138.6	138.6
	Recoveries (%)	105.9	106.0	106.0	103.2	103.5
NH3 51.0 μmol/mol	Actual N mass (ng)	274.2	277.6	280.2	277.2	264.3	2.3
	Theoretical N mass (ng)	293.4	293.4	293.4	293.4	293.4
	Recoveries (%)	93.5	94.6	95.5	94.5	90.1

3.3. Reaction Mechanism

Above, we have investigated the molecular behaviors of organic speciation presented as pyridine and acetonitrile, as well as inorganic speciation presented as NO, NO₂, N₂O, and NH₃ in the chemiluminescent reaction. The behavior mechanism indicated that NO directly underwent a chemiluminescent reaction with O₃ (Figure 8). Pyridine, acetonitrile, and NH₃, NO₂ first reacted with O₂ in the combustion tube, almost completely converting into NO, which then reacted with O₃ by chemiluminescence. (Figure 8). The stable N₂O was less likely to undergo a combustion reaction with O₂ or be oxidized to NO and thus participated less in the chemiluminescent reaction. By optimizing the reaction system, it was first oxidized to NO in the presence of a platinum catalyst and then underwent a chemiluminescent reaction (Figure 8), which almost unified all the N in the aforementioned common N compounds into one method.

FIGURE 8.

Molecular behaviors of NO, bipyridine, acetonitrile, NH₃, NO₂, and N₂O followed by oxidation combustion–chemiluminescence method.

3.4. Quantification of Actual Samples and Detection Limit

Similarly, the total mass of ethylene (10 mL) calculated by Equation (4) or Equation (5) was 11.5 × 10⁻³ g. Here, m ₃, P, V ₃, M ₃, R, T, a, and b represent the mass of ethylene (g), pressure of gas (101,325 Pa), injection volume of gas sample (0.01 L), molar mass of ethylene (28.1 g/mol), conversion coefficient [8314.47 Pa•L/(mol•K)], temperature of the sample (298.15 K), Van der Waals constant related to intermolecular attraction (a _ethylene = 4.612 × 10⁵ Pa•L²/mol²) [21], and Van der Waals constant related to gas molecule volume (b _ethylene = 0.0582 L/mol) [21]. The program code used was “import numpy as np ↵coefficients = [120974335.3×10^‐4, −584085814.5×10^‐4, 884290514.1×10^‐4, −10132.5×10^‐1] ↵ roots = np.roots (coefficients) ↵print(“The approximate roots are:”, roots).” N content in the actual samples was calculated according to Equation (6), where ω ₁, m ₄, and m ₅ represent the actual N content of actual samples (mg/kg), N mass calculated by standard curve (ng), and mass of ethylene (equal to m ₃, g), respectively.

$$m_3={\mathit{PV}}_3{M}_3/\left(\mathit{RT}\right)$$ (4)

$$\left(P+{m_3}^{2}a/{V_3}^2/{M_3}^2\right)\times \left(V_3-m_{3}b/M_3\right)=m_3\mathit{RT}/M_3$$ (5)

$${\omega }_1=m_4/m_5\times {10}^{-3}$$ (6)

The results showed that the average N content in ethylene1# (Figure 9a, Table 4) was 4.4 mg/kg, and the average N content in ethylene2# (Figure 9b, Table 4) was 3.6 mg/kg. The average peak height of the five repeated measurements for the signal noise and the 0.2 mg/L N liquid standard sample was 300 (Figure 9c) and 7220 (Figure 9d), respectively. The calculated detection limit (LOD, S/N = 3) was 0.02 mg/L. Combining the liquid injection of 20 μL and Equations (4)–(6), the detection limit for N in 10 mL of gaseous ethylene was 0.03 mg/kg (based on LOD), which is significantly lower than the minimum detection limit of 0.1 mg/kg for N content in liquid hydrocarbons [12, 13, 14].

FIGURE 9.

Actual samples (a and b), signal noise (c), and 0.2 mg/L N liquid standard sample (d).

TABLE 4.

Quantification.

Actual samples	Analyte	Repeats					RSD (%)
		1	2	3	4	5
Ethylene1#	Injection volume (mL)	10					1.0
	Net peak area	574.9	573.5	565.0	563.9	565.5
	Standard curve equation	y = 10.6541x + 28.1267, R 2 = 0.9998
	N mass calculated by standard curve (ng)	51.3	51.2	50.4	50.3	50.4
	Mass of ethylene (mg)	11.5 by Equation (4) or Equation (5)
	N content of actual samples (mg/kg)	4.5	4.5	4.4	4.4	4.4
	Average (mg/kg)	4.4
Ethylene2#	Injection volume (mL)	10					1.4
	Net peak area	463.6	477.4	467.5	468.5	461.1
	Standard curve equation	y = 10.6541x + 28.1267, R 2 = 0.9998
	N mass calculated by standard curve (ng)	40.9	42.2	41.2	41.3	40.6
	Mass of propylene (mg)	11.5 by Equation (4) or Equation (5)
	N content of actual samples (mg/kg)	3.6	3.7	3.6	3.6	3.5
	Average (mg/kg)	3.6

Note: When the N content is less than 0.1 mg/kg, the calculation results are accurate to 0.01 mg/kg; when the N content is higher than 0.1 mg/kg, the results are accurate to 0.1 mg/kg.

4. Conclusion

A sensitive and accurate method for the determination of total N in gaseous ethylene via catalytic combustion–chemiluminescence was established and applied to industrial ethylene quality control. The method was evaluated by systematically studying the effects of N speciation (bipyridine, acetonitrile, NO, NO₂, NH₃, N₂O) and matrices (N₂, ethylene) on the detection results and validated to show good linearity, high accuracy, excellent precision, and a low detection limit. The study highlights the challenges in achieving complete conversion for certain compounds, providing an opportunity to challenge lower detection limits. Key conversion conditions for different N compounds were clarified: Most ON and inorganic NO₂ were directly oxidized to NO by high‐temperature O₂, whereas NH₃ required system cleaning treatment to avoid adsorption loss, and N₂O needed a platinum catalyst for oxidation to NO—this solved the problem of incomplete conversion of recalcitrant N speciation (N₂O) and strongly adsorptive N (NH₃). A unified quantitative approach was established by combining the Van der Waals equation (for precise gas mass calculation) with equal N mass/molar response, enabling simultaneous determination of organic and inorganic N without developing separate calibration systems for different species. The N detection limit in ethylene was 0.03 mg/kg, reaching advanced levels and improving the analysis capability for trace N in light olefins. By comparing inorganic (N₂) and organic (ethylene) matrices, it was concluded that N response was almost undisturbed by matrix type, making the method widely applicable to actual industrial samples with complex matrices. This method provides reliable technical support for petrochemical processes, particularly for the detection of hazardous and recalcitrant N compounds (N₂O and NH₃) in gaseous light olefins, and contributes to ensuring catalyst activity and industrial device safety.

Author Contributions

Xiaoyu Wang: conceptualization, investigation, methodology, visualization, writing – original draft, writing – review and editing. Jiwen Li: conceptualization, writing – review and editing, investigation. Liyan Jiang: resources, supervision. Chuan Wang: resources, funding acquisition.

Funding

This work was supported by the Sinopec.

Conflicts of Interest

The authors declare no conflicts of interest.

Appendix A.

The derivations of the mathematical models and equations presented in Equations (1) and (3)–(5) are as follows:

$$m_1={\mathit{PV}}_1{M}_1/\left(\mathit{RT}\right)\times 15.0\times {10}^{-6}=\mathrm{101,325}\times 0.01\times 14.0/\left(8314.47\times 298.15\right)\times 15.0\times {10}^{-6}=85.8\times {10}^{-9}$$ (A1)

$$\begin{array}{c}0=m_2\mathit{RT}/M_2-\left(P+{m_2}^{2}a/{V_2}^2/{M_2}^2\right)\times \left(V_2-m_{2}b/M_2\right)=m_{2}8314.47\times 298.15/14.0-\left[\mathrm{101,325}+{m_2}^{2}4.612\times {10}^5/{\left(0.01\times 15.0\times {10}^{-6}\right)}^2/{14.0}^2\right]\hfill \\ \times \left(0.01\times 15.0\times {10}^{-6}-m_{2}0.0582/14.0\right)=m_2\mathrm{177,068,516.5}\times {10}^{-3}-\left(\mathrm{101,325}+{m_2}^2\mathrm{104,580,498.9}\times {10}^9\right)\times \left(1.5\times {10}^{-7}-m_2\mathrm{415,714.3}\times {10}^{-8}\right)\hfill \\ =m_2\mathrm{177,068,516.5}\times {10}^{-3}-\left(\mathrm{151,987.5}\times {10}^{-7}-m_2\mathrm{4,212,225,144.8}\times {10}^{-7}+{m_2}^2\mathrm{156,870,748.4}\times {10}^2-{m_2}^3\mathrm{4,347,560,889.4}\times {10}^5\right)\hfill \\ =m_2\mathrm{17,748,973.9}\times {10}^{-2}-\mathrm{151,987.5}\times {10}^{-7}-{m_2}^2\mathrm{156,870,748.4}\times {10}^2+{m_2}^3\mathrm{4,347,560,889.4}\times {10}^5\hfill \end{array}$$ (A2)

$$m_3={\mathit{PV}}_3{M}_3/\left(\mathit{RT}\right)=\mathrm{101,325}\times 0.01\times 28.1/\left(8314.47\times 298.15\right)=11.5\times {10}^{-3}$$ (A3)

$$0=m_3\mathit{RT}/M_3-\left(P+{m_3}^{2}a/{V_3}^2/{M_3}^2\right)\times \left(V_3-m_{3}b/M_3\right)$$

$$\begin{array}{c}0=m_{3}8314.47\times 298.15/28.1-\left(\mathrm{101,325}+{m_3}^{2}4.612\times {10}^5/{0.01}^2/{28.1}^2\right)\times \left(0.01-m_{3}0.0582/28.1\right)\hfill \\ =m_3\mathrm{88,219,189.7}\times {10}^{-3}-\left(\mathrm{101,325}+{m_3}^2\mathrm{584,085,814.5}\times {10}^{-2}\right)\times \left(0.01-m_3\mathrm{207,117.4}\times {10}^{-8}\right)\hfill \\ =m_3\mathrm{88,219,189.7}\times {10}^{-3}-\left(\mathrm{10,132.5}\times {10}^{-1}-m_3\mathrm{2,098,617,055.5}\times {10}^{-7}+{m_3}^2\mathrm{584,085,814.5}\times {10}^{-4}-{m_3}^3\mathrm{120,974,335.3}\times {10}^{-4}\right)\hfill \\ =m_3\mathrm{884,290,514.1}\times {10}^{-4}-\mathrm{10,132.5}\times {10}^{-1}-{m_3}^2\mathrm{584,085,814.5}\times {10}^{-4}+{m_3}^3\mathrm{120,974,335.3}\times {10}^{-4}\hfill \end{array}$$ (A4)

Data Availability Statement

Data will be made available on request.