1. Introduction

Ethylene (C₂H₄) is a hormone naturally produced by plants that regulates the metabolism, growth, development, and senescence of plants. However, fruits, vegetables, and flowers also contain receptors that serve as bonding sites to absorb free ethylene molecules present in the atmosphere. They respond to atmospheric ethylene the same way they respond to internally produced ethylene. Depending on the context, ethylene can therefore either be a useful tool or a contaminant.

Ethylene can cause undesired ripening and softening of fruits in storage, accelerate senescence and loss of green color in leafy fruits and vegetables [1], and cause the abscission of leaves of foliage plants [2]. It can also stimulate or retard sprouting, introduce phenolic synthesis, and cause russet spotting or lignification [3]. Ethylene can also damage ornamental crops [4, 5].

Ethylene pollution is a common source of crop loss in greenhouses, typically caused by the malfunction of heating units. Propane, diesel, and gasoline powered engines all produce enough ethylene to damage ethylene-sensitive fruits and vegetables. Fumes from welding, auto exhaust, and trash burning can also produce ethylene and thus lead to damage. The agricultural industry tries to protect ethylene-sensitive plants from ethylene exposure [6]. For example, they prevent ethylene production by using electric forklifts or actively remove it by installing ethylene absorbers on forklift exhaust systems.

The post-harvest industry avoids transporting and storing ethylene-sensitive fruits and vegetables with other fruits and vegetables that produce significant amounts of ethylene. Most fruits and vegetables have unique storage requirements for temperature and relative humidity as well as ethylene exposure [7]. To satisfy all these requirements, most compatibility charts for mixing post-harvest products divide fruits and vegetables into eight compatible groups. However, few wholesale or retail handling facilities have eight different temperature-controlled rooms. Scientists at UC Davis have devised an innovative set of conditions that make it possible to segregate fruits and vegetables into only three compatible groups for storage [8]. Each group requires ethylene concentrations below 1 ppm. While even lower levels, down to 100 ppb, would result in reduced product losses, the lack of a high sensitivity, rugged ethylene sensor currently prevents UC Davis from lowering the recommended storage ethylene concentration level [9]. Large wholesalers and retailers of fruits and vegetables typically remove ethylene from storage rooms by making sure they have adequate ventilation. They also rely on ethylene absorbers such as potassium permanganate (www.ethylenecontrol.com).

A range of methods for measuring ethylene levels have been implemented in both agricultural research and commercial storage facilities [10]. Gas Chromatography (GC) is presently the gold standard in post-harvest research GC systems with specialized detectors can measure ethylene concentrations of a few parts per billion (ppb) in the atmosphere [11]. However, highly trained personnel are needed to operate these systems in laboratories, making GC systems unsuitable for operation in storage or distribution facilities. Additionally, they require consumables such as chromatography columns or carrier gases. Some research groups have measured ethylene at sub-ppb levels by photo acoustic detection [12]. This technique is also impractical for in situ real-time measurements and is too costly for practical post-harvest applications. In practice, simple and inexpensive colorimetric tubes are used in the post-harvest industry, even though they lack basic sensitivity. Indicator tubes from companies such as QA Supplies (http://www.QAsupplies.com/2002040.html) or Draeger (www.draeger.com) can detect ethylene, but are unable to detect levels less than 1 ppm.

We describe in this paper an alternative measurement technology for the detection of trace levels of ethylene in air based on cavity ring-down spectroscopy (CRDS). CRDS is a laser-based optical absorption spectroscopy technique capable of achieving high sensitivities in a transportable package with high reliability and no consumables. CRDS has been automated so that it can operate without highly trained personnel or regular maintenance. Moreover, CRDS can perform ethylene detection in situ, and in real-time, thereby allowing a user to directly control the process, such as ventilation in a cold storage room, or removal of ethylene generating fruits that are inadvertently stored in the same area as ethylene sensitive produce.

2. Cavity ring-down spectroscopy combined with fluorescence

Cavity ring-down spectroscopy (CRDS) is a technique for detecting trace quantities of a wide variety of gases. Cavity ring-down spectrometers are the best instruments for making atmospheric measurements because they are highly sensitive, easy to operate, and provide an absolute measurement [13]. Figure 1(a) is a schematic showing the essential components of a CRDS system. In the basic CRDS technique, a laser beam shines into an optical cavity formed by at least two high-reflectivity mirrors, filling it with photons [14]. The photons bounce between the mirrors up to 10,000 times. Most of the photons stay within the cavity for about 30 microseconds, traveling several kilometers. On each round trip, about 0.01% of the photons are transmitted through the right-hand mirror, and strike a photo detector.

At some time t ₀ the laser beam is shut off, and all the photons eventually escape from the cavity (“ring-down”). Figure 1(b) shows a ring-down curve, which plots the intensity of light inside the cavity as a function of time (in microseconds) after the laser is shut off. The intensity of light inside the cavity decreases exponentially with a time constant τ. The equation in Fig. 1(b) indicates that the intensity of light (I _circ) circulating inside the cavity at any given moment equals the intensity of light which was circulating in the cavity when the laser beam is shut off at t ₀, multiplied by e ^-t/τ. That mathematical relationship also holds for the light escaping from the cavity, since it is related to the circulating intensity by the transmissivity of the mirror, which is a constant. All the light escaping from the cavity through the right-hand mirror strikes the photo detector, which responds with a voltage. The voltage across the detector in Fig. 1(a) therefore also decays exponentially starting at t ₀, when the laser beam shuts off, and at time t equals it value at t ₀ multiplied by e ^-t/τ.

Trace quantities of gases affect the cavity’s decay rate (1/τ). The decay rate is greater if molecules of the targeted substance are inside the cavity, because they absorb some photons that would otherwise hit the detector. By measuring the decay rate of the cavity, a CRDS instrument can determine the concentration of the molecule(s) of interest with extreme sensitivity. Equation (1) defines the decay rate as the losses per unit time, or the losses in the cavity divided by the time over which those losses occur.

The three terms in the numerator represent three types of losses. The first term, ε/C, refers to sample absorption losses: ε (the extinction coefficient of the absorbing molecule) times l (the cavity round-trip length) times C (the concentration of the absorbing gas). The second term, n(l - R), describes the losses due to transmission through a mirror or absorption within a mirror. The n refers to the number of mirrors in the cavity, and R is the mirror reflectivity. The third term, L stands for any scattering or randomized losses in the cavity. The denominator, T _rt, represents the time required for photons to make a round trip within the cavity. We define the decay rate of an empty cavity (or a cavity containing zero molecules of the target gas) as 1/τ₀. Since the cavity is empty, the concentration is zero, and the first term (ε l C) in the numerator above is also zero.

 

Fig. 1. (a) Schematic of cavity-ring down optical train, (b) exponential cavity ring-down decay curve and associated equation, where t ₀ is the laser shut off time and τ is the ring-down decay constant, (c) spectral acquisition of ring-down waveforms as a function of wavelength, where the rate is proportional to absorption, and, (d) optical loss (1/cτ) spectrum as a function of wavelength covering an absorption feature.

Download Full Size | PDF

If we solve Eq. (1) for the concentration C by substituting this value of τ₀, we arrive at Eq. (2).

Equation (2) indicates that CRDS can determine the concentration of target molecules in the cavity by simply subtracting the decay rate of an empty cavity from that of a filled cavity, then dividing by the extinction coefficient of the target molecule and the speed of light. To measure the concentration of a gas in the cavity, the system sets the laser to the starting wavelength λ _n=1, normally below the absorption feature of interest. The system dithers a piezo-electric transducer (PZT) to move one of the cavity’s end mirrors, thereby adjusting the cavity length, until the laser and the cavity come into resonance. When sufficient light intensity builds up in the cavity, the system turns the laser off. It monitors the detector signal and records the ring-down curve. It converts the ring-down time to an absorbance. It then moves the laser from wavelength λ_n to wavelength λ_n+1 and repeats the process. By repeating the process several times at wavelengths covering the absorption feature, the instrument collects a series of ring-down curves. Figure 1(c) shows the succession of ring-down curves collected as the laser is scanned across the absorption feature. The ring-down curves collected can be converted into an absorption spectrum, by computing the optical loss, as shown in Fig. 1(d). CRDS measures the absolute optical losses (1/cτ) directly, so that the concentration of a target sample is simply the difference between optical losses at the sample absorption peak and at a reference baseline wavelength, divided by the extinction coefficient.

Therefore, CRDS provides a very accurate measurement of the concentration of the target molecule. It relates the concentration to the decay rate through two fundamental constants. The measurement of the concentration of the target molecule does not depend on cavity length, mirror reflectivity, or any other physical parameter of the instrument. A CRDS system calibrates itself by measuring the ring-down decay constant for an empty cavity. In addition, it is insensitive to random intensity fluctuations in the laser light source because it detects the rate of the decay of the transmitted intensity rather than its actual value.

In comparison with other detection techniques, CRDS offers the following additional advantages for the detection of trace ethylene in post-harvest applications:

3. Experimental section

3.1 CRDS sensor System

We designed and built a CRDS system to target the ethylene absorption peak at 1618 nm. We chose 1618-nm because the HITRAN database and other sources indicated that other gases (such as carbon dioxide) present in greenhouse and storage environments have no nearby absorption peaks and thus would not interfere with our measurements. Figure 2 shows the functional design of the CRDS prototype system.

 

Fig. 2. Schematic diagram of a cavity ring-down optical engine using a three-mirror optical cavity and direct laser modulation.

Download Full Size | PDF

We selected a commercially available 20 mW distributed feedback laser (DFB) which emits at 1618 nm. We built a compact cavity ring-down cell using our patented, substantially self-aligning design [16]. The cavity was 21 cm long, had a 25 mL sample volume, and an empty ring-down time of 32.4 μs. A pressure transducer was placed near the center of the cavity, and a temperature sensor was mounted to the cavity sidewall. A wavelength monitor was optimized for use near 1618 nm. The noise level of the electrical signals enabled a resolution of 2 MHz over the 30 GHz measurement range. We also built a customized photo detector for the CRDS instrument. The detector had a responsivity of 0.23 V/μW, a bandwidth of 2.3 MHz, and a root mean square (RMS) noise of 0.53 mV. The high gain and low RMS noise increased the usable electrical signal level available at the digitization electronics, resulting in significantly lower digitization noise.

We integrated all of the opto-mechanical components into a compact 10” × 5” platform. We mounted the opto-mechanical platform with vibration isolators within a thermally controlled enclosure. We then integrated this enclosure into a 3U (5.25” high) rack. The electronics boards were attached to the enclosure, and the vacuum and sample lines with fittings were mounted on the front panel. Power enters through the back panel. We integrated the prototype subassemblies into rack-mount enclosures. The first enclosure contained the cavity ring-down optomechanical platform, sample handling valves, and electronics boards. A second 19” rack-mount enclosure contained the pump and power supply assembly. Figure 3 shows a photograph of a fully assembled and integrated CRDS instrument.

The system used in these experiments was based on our patented three-mirror ring cavity design, which eliminates optical feedback into the laser [16]. The basic sequence to obtain a single ring-down measurement is as follows:

 

Fig. 3. Photograph of assembled and integrated cavity ring-down trace ethylene gas sensor.

Download Full Size | PDF

The ring-down detector measured cavity power versus time. A 14-bit A/D card digitized the output of the detector. A DSP card analyzed the digital output to determine the ring-down time. In order to minimize instrument memory to ethylene, the CRDS instrument was designed to operate under continuous flow. A sample handling system with appropriate safety features (to protect the ring-down cavity both from sudden changes in pressure and backflow from the pump) was implemented.

3.2 Ethylene Spectroscopy

The CRDS engine was used to measure the spectra of ethylene and common atmospheric background gases (carbon dioxide, water, and methane) in the 1615.5 to 1618.0 nm region. The ethylene peaks between 1617.6 and 1617.8 nm appeared to be a good choice since they were relatively strong lines and, in addition, were well separated from the adjacent, large carbon dioxide spectral features. Therefore, the 1617.6 nm to 1617.8 nm ethylene features shown in Fig. 4(a) were selected.

Additional work was done to select the final instrument operating pressure. Figure 4(b) shows the dependence of the largest ethylene peak (1617.67 nm) on pressure. The peak width continues to increase linearly with pressure, while the peak height saturates around 150 Torr. Note that when one compares operation at 90 Torr and at 150 Torr, the peak height is only 20% weaker at 90 Torr, but the peak width is reduced by almost 50%. We therefore chose an operating pressure of 120 Torr as a good compromise.

At a fixed temperature and pressure, the ethylene concentration was proportional to the height of the absorption peak above the baseline level. The absorption level at the top of the peak was determined by fitting spectral data points in the vicinity of the peak to a low order polynomial. The baseline loss represented all the other losses in the system, including mirror loss, scattering, interfering spectral lines, and continuum absorption due to the matrix gas.

 

Fig. 4. (a) Target spectral region showing isolated lines of ethylene in the presence of carbon dioxide and, (b) pressure dependence of ethylene peak absorption and ethylene peak width.

Download Full Size | PDF

Figure 5 shows spectral measurements of ethylene at concentrations of 310, 160, 86, and 50 ppb taken without averaging. At an ethylene concentration of 50 ppb, the signal-to-noise ratio was 5:1 without averaging. We determined the measurement precision of the prototype instrument by repeated measurement of gas flowing from a 82 ppb ethylene in air cylinder.

 

Fig. 5. Target spectrum of four different concentrations of ethylene.

Download Full Size | PDF

The measured concentration over 90 hours of continuous operation with a continuous sample flow through the cavity appears in Fig. 6. Each spectral scan took 7 minutes. The data shown is a running average of 5 scans, for a total measurement time of 35 minutes. The concentration of ethylene was measured to be 82.8 ppb with a standard deviation of 1.1 ppb. The detection limit is typically defined as twice the instrument precision. Therefore, the CRDS instrument achieved a detection limit of 2 ppb of ethylene.

The CRDS instrument was transported to UC Davis for benchmarking against a gas chromatography system. The instrument was then moved to a post-harvest packing and storage facility in Selma, CA, and installed to monitor the ambient air in a cold storage room over a period of a month.

 

Fig. 6. Measurement of ethylene concentration (82.8 ppb) over 92 hours of sampling showing repeatability (precision) of 1.1 ppb.

Download Full Size | PDF

4. Results and discussion

4.1 Comparison with Gas Chromatography

The CRDS instrument was installed in the laboratory of Prof. Adel Kader at UC Davis and tested using four different ethylene gas standards: 7.4 ppm, 1.77 ppm, 1.00 ppm, and 75 ppb. The standards were kept in large gas cylinders. Each concentration sample was run through the CRDS system in order to obtain at least five (5) consecutive concentration measurements. The gas bottle ethylene concentration was then measured on the UC Davis reference GC system. Note that the CRDS system was flushed before and after the 7.4 ppm (highest ethylene concentration) sample to prevent excessive sample memory.

A comparison between the UC Davis GC system and the Picarro CRDS instrument measured values is plotted in Fig. 7. Note the linear relationship between the measurements. When a linear fit was made to the data, an offset of 8 ppb and a slope of 0.94 (ppb CRDS)/(ppb GC) were obtained. The R² of the linear fit was 0.99, which indicates excellent linearity. Given that the reference ethylene bottles used for CRDS system calibration had a 5% concentration certification from the supplier, and that dilution introduced another 2% to 5% error (depending on the degree of dilution) in “true” concentration, the expected differences in reading between the two systems was 5% to 10%. In fact, the difference in readings between the CRDS instrument and the GC ranged from 4% at the highest ethylene concentration to 12% for the lowest ethylene concentration. The results obtained were within the expected range of error between the two systems. Moreover, the UC Davis calibration results validated our internal CRDS calibration scheme.

One benefit of CRDS is that it measures absolute optical loss. Therefore, once a calibration is made relating optical loss to concentration, all CRDS systems using that calibration will produce the same concentration readings. The results at UC Davis illustrated the difficulties inherent in using relative measurement methods that require calibration, such as GC.

 

Fig. 7. Comparison of cavity ring-down and gas chromatography measurements of four standard concentrations of ethylene at UC Davis.

Download Full Size | PDF

A calibration between the CRDS and the packing house third party GC system was also made. Fig. 8 illustrates that although the calibration was linear and had a comparable offset to the UC Davis calibration (-2 ppb at Enns versus 8 ppb at UC Davis), the slope was only 0.68. Thus, the third party GC used for the packing house samples also had poor accuracy relative to the UC Davis GC. The third party GC would typically read higher values. In fact, the calibration between the CRDS and UC Davis instrument was distinctly superior to the calibration between the UC Davis GC and the third party GC used by the packing facility.

 

Fig. 8. Comparison of cavity ring-down and gas samples analyzed by gas chromatography of ambient ethylene concentrations at the Selma packing facility.

Download Full Size | PDF

4.2 Application to a Cold Storage Facility

Both prior and subsequent to the measurements at UC Davis, the CRDS instrument was installed in one of the cold storage rooms for kiwi fruit at a packing facility in Selma, CA. We took continuous CRDS measurements of the ambient air and periodically obtained an air sample from the storage room and sent it for analysis to a third party GC laboratory. Figure 9(a) shows the ethylene concentration in the room over a period of 13 days (prior to the UC Davis calibration) while Fig. 9(b) shows the ethylene concentration in the same room for 26 days (after the UC Davis calibration).

The targeted ethylene concentration was expected to be 10 ppb or lower. When the initial ethylene measurements were made [Fig. 8(a)], the packing facility noted the abnormally high ethylene levels, and the CRDS measurements were verified using independent GC measurements. The room was initially vented with outside air for several hours (dips in ethylene concentration) and then closed again. The CRDS measurements showed in real time (unlike the GC) that the ethylene concentration recovered to its high value by the next day. The active component (potassium permanganate) in the scrubber was replaced, in case it was no longer removing excess ethylene. However, the ethylene levels still kept returning to high levels even after additional venting.

Finally, an inventory was made of the room contents. Originally, only kiwi fruit was intended to be stored in the cold room. However, prior to the start of the experiment, Asian pears had been added to the cold room, because of limited space availability in the packing facility. Asian pears are a known source of ethylene production, and even a few small crates in a large cold room were sufficient to increase the ethylene levels beyond the 10 ppb deemed appropriate for long-term kiwi fruit storage.

The Asian pears were removed from the room, and, after two days of venting, the ethylene concentration was again measured, and it was found that the ethylene level had dropped to (and remained) below 10 ppb. The ethylene level was again validated by GC measurements. The instrument was then removed for the UC Davis calibration experiments. For the remaining period of time after the instrument was re-installed, the ethylene concentration stayed below the desired target, with the exception of a small excursion to 13 ppb on days 16 and 17 in Fig. 9(b).

 

Fig. 9. Measured ethylene concentration in a kiwi fruit cold storage room at a packing facility in Selma, CA, over a period of 5 weeks. (a) Note the difference in ethylene levels prior to and post removal of Asian pears, on day 10, which were stored in the same room. The sharp changes in ethylene concentration are due to venting the room. (b) Shows 25 days of monitoring of the storage room containing only kiwi fruit.

Download Full Size | PDF

The packing house observed the utility of CRDS due to its ability to provide real-time feedback. Not only could diurnal changes in ethylene concentration be monitored to see the longer term effects of storing different kinds of fruit together, but the CRDS measurement were so fast, that the immediate benefits of room venting were clearly apparent. Furthermore the real-time data allowed us to observe the effect of replacing the potassium permanganate in the ethylene scrubber (which “cleaned out the room” after the Asian pears were removed), and understanding the capacity for ethylene removal of the scrubber system (which could not sustain 10 ppb of ethylene in the room in the presence of Asian pears).

5. Concluding remarks

We have demonstrated that CRDS has the capability of achieving a detection limit of 2 ppb for ethylene in air. The sensitivity of the instrument was shown to be independent of interferences from ambient water, carbon dioxide, or methane. The instrument measurement precision (or repeatability) was shown to be better than 1 ppb of ethylene. When CRDS was compared to a GC and calibration standards, a linear calibration was obtained, indicating that these two approaches are equivalent in linearity.

The CRDS ethylene monitor performed satisfactorily in a fruit packing facility, and produced comparable measurement values to a manually sampled GC approach. Furthermore, CRDS provides real-time measurements, so that problems arising from incompatible fruits being stored together can be detected early, before significant damage caused by ethylene exposure can occur.

The successful development of a robust ethylene monitoring instrument can provide significant benefits to post-harvest research and to the agricultural industry. By enabling continuous monitoring of extremely low ethylene levels, new research could be performed to assess the impact of ethylene on horticultural products. This information could be used to develop improved growing and storage guidelines and/or improved packaging materials. The deployment of a CRDS ethylene monitor into greenhouses would improve yields and reduces product losses to the agricultural production industry.