Journal of Oleo Science Copyright ©2018 by Japan Oil Chemists’ Society J-STAGE Advance Publication date : June 7, 2018 doi : 10.5650/jos.ess17212 J. Oleo Sci. Effect of Roasting Temperatures on the Properties of Bitter (Armeniaca sibirica L.) Kernel Oil Feng Jin1, Ji Wang1, Joe M. Regenstein2, and Fengjun Wang1＊ 1 Department of Food Science and Engineering, College of Biological Sciences and Biotechnology, Beijing Key Laboratory of Forest Food Processing and Safety, Beijing Forestry University, Beijing 100083, P. R. CHINA 2 Department of Food Science, Cornell University, Ithaca, NY 14853-7201, USA

Abstract: Volatile compounds and quality changes of bitter apricot (Armeniaca sibirica L.) kernel oil (AKO) with different roasting conditions were determined. Bitter apricot kernels were roasted at 120, 130, 140 and 150℃ for 15 min. Unroasted bitter apricot kernel oil was used as the control. Quality indicators included color, acid value and peroxide value, fatty acids, total phenols and oxidative stability. Peroxide values of the tested oils were 0.46-0.82 meq/kg, acid values were 0.60-1.40 mg KOH/g, and total phenol contents were 54.1-71.5 µg GAE/g. was the major fatty acid, followed by linoleic, palmitic, stearic and palmitoleic acids. Roasting increased the oxidative stability of bitter AKO. Volatile compounds were tentatively identified and semi-quantified. Among the 53 volatiles identified, benzaldehyde and benzyl alcohol were the major components. These two aroma compounds increased significantly during roasting and contributed sweet and flavors. Pyrazines were also prevalent and significantly increased with roasting. Sensory evaluation showed that roasted, nutty, sweet and oily aromas increased as roasting temperature increased. Practical applications: Bitter apricot kernels cannot be consumed directly, thus it is potentially beneficial to find uses for them, especially in China where bitter apricot processing is a significant industry. Roasted bitter AKO with a pleasant aroma could be prepared and might find use as an edible oil. The roasting process gave the bitter AKO a pleasant flavor. This study provided preliminary information on production parameters and potential quality control parameters.

Key words: bitter apricot kernel oil, Armeniaca sibirica, benzaldehyde, benzyl alcohol, pyrazines

1 INTRODUCTION has shown beneficial effects such as promoting respiration Bitter （Armeniaca sibirica L.）belong to the and improving digestion, although it will lead to respiratory Rosaceae family. The genus Armeniaca is widely cultivat- failure and even death when taken in excess5）. However, ed in the mountainous areas of northern and northeastern is removed during oil production and is used in China, eastern Siberian, and Mongolia. In China, the total China. Based on previous report, amygdalin ＜5 mg/kg in area for bitter apricots was about 1,700,000 ha, and the bitter apricot oil is considered safe and the amygdalin in amount of kernels available is estimated at about 192,000 refined bitter apricot oil ranged from 0.17 to 2.15 mg/kg tonnes in 20111）. The oil content of apricot kernel ranged oil50）. from 40 to 56％2）. The major fatty acids reported in apricot Bitter apricot kernel oil is not only used as an edible oil, kernels were oleic（58.3-73.4％）and linoleic（18.8-31.7％）3）. but also be used as a component of lubricants, cosmetics, Apricot kernels contain a variety of nutritional components and surfactants; and to prevent cardiovascular diseases such as high concentrations of flavonoids and phenolic acid antioxidants4）. Bitter apricot cultivars have a high concen- Abbreviations: AKO, apricot kernel oil; PV, peroxide value; AV, acid value; CDA, conjugated dienoic acid; RI, retention tration of amygdalin（～5.5％）, while no amygdalin was de- 2） index; FAME, fatty acid methyl ester; IS, internal standard; tected with sweet apricot cultivars . The β-glucosidase MS, mass spectrum; GAE, gallic acid equivalents; GC-MS, gas enzyme decomposes amygdalin to glucose, benzaldehyde chromatography-mass spectrometry; HS-SPME, headspace and hydrocyanic acid. Hydrogen cyanide in small quantities solid-phase micro-extraction.

＊Correspondence to: Fengjun Wang, Beijing Forestry University, Beijing 100083, P. R. CHINA. E-mail: [email protected] Accepted February 5, 2018 (received for review September 28, 2017) Journal of Oleo Science ISSN 1345-8957 print / ISSN 1347-3352 online http://www.jstage.jst.go.jp/browse/jos/ http://mc.manusriptcentral.com/jjocs

1 F. Jin, J. Wang, J. M. Regenstein et al.

and reduce plasma cholesterol levels5）. AKO is popular at 2504×g for 15 min, the supernatant was recovered and among consumers because of its high nutritional value4）. It sealed in 100 mL flasks and immediately refrigerated for a was found that AKO can positively affect rat growth and it maximum of 7 days before further analysis. No further oil inhibited cyclophosphamide-associated organ degenera- purification was used. tion6）. Although the volatile profiles during roasting were investigated for palm kernel, peanut, sesame and pumpkin 2.3 Volatile compounds seed oils7－10）, roasted bitter AKO has not been studied. 2.3.1 Extraction of volatile compounds During roasting, the amount of Maillard reaction occur- The volatile compounds were extracted using HS-SPME ring is affected by roasting temperatures and times11）. Also coupled with GC-MS. The conditions were adapted from a during heating oxidative reaction may occur that give oils method reported earlier with minor modifications8）. A 65 an undesirable flavor and produce free radicals12）. Thus, it μm diameter polydimethylsiloxane/divinylbenzene fiber is important to evaluate the oxidative stability of oil was used（Supelco, Bellafonte, PA, USA）for the absorption samples. Peroxide values（PV）, acid values（AV）, total of volatile compounds. The fiber was conditioned at 250℃ phenols, fatty acid composition and conjugated dienoic for 30 min prior to the first measurement. The bitter AKO acid（CDA）values were previously used as quality indica- sample（2.00 g）and 10 μL of 1,2,3-trichloropropane（1.000 tors with roasted seed oils13－15）. mg/mL in methanol）was used as an internal standard（IS）. The aim of the present work was to determine the The samples were weighed into a 15 mL glass vial and changes in volatile compounds of crude extracts of bitter sealed with a Teflon rubber septum and an aluminum cap AKO before removal of amygdalin using headspace solid- （Beijing Kebio Biotechnological Co. Ltd., Beijing, China）. phase micro-extraction（HS-SPME）and gas chromatogra- The sample vial was heated on a hot plate（laboratory phy-mass spectrometry（GC-MS）. Indicators like color, AV, stirrer/hot plate PC-620D, Corning Inc., Corning, NY, USA） PV, fatty acid composition, total phenol content and CDA at 60℃. After 20 min equilibration, the fiber was insert into values were used to evaluate oil quality. the vial and volatile compounds extracted at 60℃ for 30 min. When the headspace extraction was completed, the fiber was immediately transferred to the injection port of the GC（GCMS QP2010, Shimadzu, Kyoto, Japan）, and vola- 2 EXPERIMENTAL PROCEDURES tile compounds desorbed at 250℃ for 5 min. 2.1 Materials and reagents 2.3.2 GC-MS The bitter apricots used in this study were bought from Volatile compounds were separated and identified using Chengde Yaou Nuts and Seeds Co., Pingquan, Hebei Prov- a capillary column（Rtx-5MS, 30 m×0.25 mm i.d., film ince, China. The apricots were harvested in August, 2016. thickness 0.25 mm）coated with diphenyl dimethyl polysi- Bitter apricot kernels were vacuum-sealed in plastic bags loxane（Restek, Bellafonte, PA, USA）. The injection port and kept in a freezer at 4℃ for a maximum of 12 wk to temperature was 250℃ in a split mode of 1:10. The column minimize oxidation. 1,2,3-Trichloropropane（chromatogra- temperature was programed as: 45℃ for 10 min, increasing phy grade）was purchased from AccuStandard Inc（New 3℃/min up to 100℃ and held for 1 min, then increasing at Haven, CT, USA）. Other chemicals were analytical grade 5℃/min up to 150℃ and held for 5 min, 10℃/min up to a and purchased from Yixiubogu Biological Tech Co.（Beijing, final temperature of 250℃, and held for 10 min. Purified China）. Deionized water was obtained using a Milli-Q water helium gas（≥ 99.999％, Beijing AP BAIF Gases Industry purification system（Canrex Analytical Instrument Co., Co., Ltd., Beijing, China）was used as the carrier gas with a Ltd., Shanghai, China）. constant flow rate of 1 mL/min. The MS was operated in the election impact mode at an ionization energy of 70 eV 2.2 Roasting conditions and extraction of AKO with a scan range of 35-500 atomic mass units. The tem- Bitter apricot kernels（～500 g）were roasted for 15 min perature of the ion source and the transfer line were set at at different roasting temperatures（120, 130, 140 and 230 and 250℃, respectively. 150℃）using an oven（T3-L383b, Media Kitchen Appliance 2.3.3 Identification and quantification Manufacturing Co., Ltd., Foshan, Guangdong, China）. Volatile compounds in bitter AKO samples were tenta- Bitter apricot kernels were put on pans with tin foil and tively identified by matching their mass spectrum with spread evenly on the pan for even roasting. Roasted bitter those in the National Institute of Standards and Technology apricot kernels were pressed for oil production using a lab- （NIST）11 library and comparing their retention indices as oratory-scale oil press（YKY-6YL-550, Longyan Machinery found in the NIST Chemistry WebBook（Gaithersbug, MD, Manufacturing Co., Ltd., Longyan, Fujian, China）. Oil from USA）. The compounds having a ≥ 80％ similarity of their unroasted bitter apricot kernels was also obtained mass spectrum with the NIST library data were considered （control）. Crude bitter AKO was centrifuged（HC-2518R, present. The RI was calculated in relation to the retention

Zonkla Scientific Instrument Co., Ltd., Hefei, Anhui, China） times of C7 – C24 n-alkanes（Supelco）under the same condi-

2 J. Oleo Sci. Roasting of Bitter Apricot Kernel Oil

tions. 2.5.4 Fatty acids composition 1,2,3-Trichloropropane（IS）was not found in pistachio Fatty acid methyl esters（FAME）of oil samples were pre- seeds oil and peanuts8, 22）which made it appropriate as the pared based on ISO standard 550818）using the GC. The internal standard. Accurate quantification of volatiles Rtx-5MS column was used. The temperatures of the injec- cannot be done using a single internal standard10）. Thus, tor and detector were 200 and 230℃, respectively. Initial concentrations of volatiles are reported relative to the IS temperature of the column was 120℃ and was held for 1 rather than as absolute concentrations. min, then increased at a rate of 10℃/min from 120 to The relative concentration was: 200℃ and held for 16 min. Finally, the temperature gradi- Relative concentration（mg/kg）＝（ Concentration of the ent was 20℃/min to 220℃ and held for 25 min. Purified IS / Peak area of the IS）× Peak area of the particular com- nitrogen（≥ 99.999％, Beijing Ruyuanruquan Technology pound Co., Ltd., Beijing, China）was used as carrier gas with a flow Peak areas were calculated using the computer program rate of 1 mL/min. A split injector with a split ratio of 1:30 supplied by the manufacturer. and an injection volume of 1 μL was used. Peaks of FAME were identified by comparing their retention times with au- 2.4 Sensory thentic standards from Sigma-Aldrich Co.（St. Louis, MO, A 10-person trained sensory panel（between 20 and 30 USA）, and relative area percentages as determined by the years old, 5 males and 5 females）was used. All of the pan- computer program with the GC among the peaks obtained elists had sensory evaluation experience in descriptive was used for estimating the amount of each fatty acid. evaluation of roasted, grassy, sweet and other aromas. The 2.5.5 Oxidative stability single characteristic flavors including roasted, grassy, Oxidative stability was determined by measuring sweet, nutty, oily and caramel intensity were scored from 0 changes in PV and CDA on days 1, 3, 6, 9 and 12. A 50 g （no perception of aroma）to 9（extremely intensive aroma）. sample of bitter AKO was placed into a 100 mL flask and An oil sample（40 g）was weighed and put into a 60 mL vial, sealed with Parafilm（Bemis Company, Inc., Neenah, WI, then sealed and equilibrate in a water bath at 60℃ for 10 USA）that according to the manufacturer permitted gas min. Before sensory evaluation, each panelist needed to transmission. The flask was covered with foil and put into shake the vial before removing the cap. Sensory evaluation 60℃ oven for 12 days. CDA values were measured accord- was carried out at room temperature. To compare the data ing to AOCS official method Ti1a-6416）and calculated using for different roasting temperatures, spider plots were the following equation: drawn. CDA（％）＝（0.84×A）（/ bc-K）, where A was the absor- bance of bitter AKO samples at 233 nm, b was the length of 2.5 Quality indicators cell（1 cm）, c was the concentration of the sample（g/L）, 2.5.1 Color and K＝0.03, which is a correction factor for the back- The color was measured at room temperature（～20℃）, ground absorptivity of the fatty acid groups using the value using a colorimeter（DTQC-10A, Beijing Oriental Precision for dienoic acid14）. Technology Co., Beijing, China）. Lightness L*, and the chromatic coordinates a（* green-red）and b（* blue-yellow） 2.6 Statistical analysis were measured on the oil surface. The colorimeter was re- All the data are reported as mean±standard deviation of calibrated using a standard white plate after each measure- triplicate determinations. Results of the volatile com- ment. All the analyses were done in triplicate. pounds were statistically analyzed using analysis of vari- 2.5.2 Peroxide value and acid value ance and Duncan’s multiple range test using the statistical PV and AV of AKO were determined using AOCS Official software SPSS 17.0（SPSS Inc., Chicago, IL, USA）with a 5％ Methods Cd 8-53 and Cd 3d-63, respectively16）. significance level. 2.5.3 Total phenols The determination of total phenols was done using the methods of a previous study17）with some modifications. A 0.1 mL sample of oil was mixed with 0.5 mL of Folin-Cio- 3 RESULTS AND DISCUSSION calteu reagent, 1.5 mL of sodium carbonate solution（10％, 3.1 Volatile compounds w/v）and 1.5 mL of deionized water. The absorbance at 765 A total of 53 volatile compounds were identified using nm of the mixture was measured after storage at room the NIST library and RI（retention index）in the bitter AKO temperature for 2 hr（L6, Inesa Analytical Instrument Co., samples, including N-heterocyclic compounds, O-heterocy- Ltd., Shanghai, China）. Gallic acid was used to prepare a clic compounds and non-heterocyclic compounds. There standard curve and the results were expressed as μg of were 8 pyrazines, 3 pyridine derivatives, 4 pyrrole deriva- gallic acid equivalents（GAE）/g of sample. tives, 1 pyran derivative, 4 furan derivatives, 5 aldehydes, 4 ketones, 6 alcohols, 6 alkanes, 1 ester and a few other mis-

3 J. Oleo Sci. F. Jin, J. Wang, J. M. Regenstein et al.

cellaneous compounds. The identified volatiles are listed in apple and strawberries, and also has been detected in heat- Table 1 including their RI and the relative concentration processed foods like coffee and roasted plantains24）. 2,5-Di- changes with different roasting temperatures. methyl-4-hydroxy-3（2H）-furanone contributed to the Pyrazines were formed during the Maillard reaction sweet aroma of roasted almonds25）. Furfuryl alcohol was through Strecker degradation, and contributed nutty, detected at low concentrations, 0.37 and 1.01 mg/kg oil at roasty, potato-like, popcorn-like and earthy aromas8）. Free 140 and 150℃, respectively. Furfuryl alcohol has been re- amino acids and monosaccharides are considered to be im- ported in roasted perilla seed oil and roasted sesame seed portant flavor precursors for the formation of unique pastes, and contributed to the sweet, nutty and caramel- flavors like pyrazines during roasting19）. A previous study like aromas26）. Although most of the furan derivatives gen- reported that various pyrazine provided the typical aroma erated with roasting do not account for a large proportion and was a crucial factor for roasted pumpkin seed oils20）. of the total volatiles, they made a contribution to the Eight pyrazines were identified in this study, most of them sweet, nutty and roasted aromas of bitter AKO. were only detected when the roasting temperature reached Maltol is a pyran derivative and was detected when 130℃. Pyrazines which were not detected in unroasted roasting temperatures were 140 and 150℃ and is responsi- bitter AKO samples accounted for 53.7％ of the total vola- ble for a burnt flavor27）. tiles when the roasting temperature was 150℃. The higher Aldehydes included hexanal, benzaldehyde, nonanal, the roasting temperature, the higher the content of pyr- phenylacetaldehyde and 2-phenyl-2-butenal. In general, al- azines with ethylpyrazine, 2,5-diethylpyrazine and cyclo- dehydes are responsible for green, painty, metallic, beany hexapyrazine only being detected when the roasting tem- or rancid aromas, and are often related to the undesirable perature reached 140 and 150℃. The 2-methylpyrazine, flavors in and oils8）. Hexanal and nonanal are lipid oxi- 2,5-dimethylpyrazine and 2,6-diethylpyrazine showed a dation products9）. Hexanal increased significantly as tem- significantly increasing trend（p＜0.05）. 2,5-Dimethylpyr- perature increased（p＜0.05）and was responsible for the azine and 2,6-diethylpyrazine were the most abundant pyr- green and herbaceous aromas in cashew nuts28）. Phenylac- azines and accounted for a great proportion of the total etaldehyde was found in all roasted samples reaching its pyrazines. 2,5-Dimethylpyrazine was reported as the best highest concentration at 130℃. Phenylacetaldehyde con- overall pyrazine to predict roasted peanut flavor20）. Cyclo- tributes green, floral and honey flavors, and was reported hexapyrazine is a pyrazine derivative and is being reported as one of the important flavors in roasted virgin rapeseed in thermally-heated edible oil for the first time. oil23）. Benzaldehyde was identified in all samples. It first in- Several other nitrogen-heterocylic compounds were iso- creased and then decreased, which was opposite of previ- lated and identified in bitter AKO samples including pyri- ous studies4）. Benzaldehyde is an aromatic aldehyde re- dine derivatives and pyrrole derivatives. Most of these vola- sponsible for a pleasant almond-like aroma and is crucial to tiles were not detected until the roasting temperature was almond flavor. It is formed from the enzymatic breakdown 130℃. This may indicate that the formation of these com- of the diglucoside amygdalin as almond kernels are dis- pounds needs a higher roasting temperature. Pyridine and rupted29）. It has also been reported as one of the major vol- pyrrole derivatives have been reported as Maillard reaction atiles from the essential oils of the kernels of unripe Japanese products of amino acid-sugar model systems and contrib- apricots（Prunus mume Sieb. et Zucc）and Longwangmo uted to roasty/smoky aromas21）. In addition, 2-acetylpyrrole apricots（Prunus armeniaca L.）4）. Thus, the increase of was identified in this study. It is a typical Maillard reaction benzaldehyde during roasting could enhance the typical product and has been reported to enhance the roasty almond aroma of apricot kernel oil. 2-Phenyl-2-butenal was flavor of pumpkin seed oil10）. not detected until the roasting temperature reached 130℃ Furan-containing compounds were produced during the and increased significantly（p＜0.05）thereafter. It has also thermal degradation of fructose and glucose, and were been found in roasted peanut oil8）. generally described as having the caramel-like, sweet and On the other hand, three ketones were not detected nutty aromas of heated carbohydrates22）. In this study, four until the roasting temperature reached 140℃. 3-Methyl- furan derivatives were detected: 2,5-dimethyl-2,4-dihy- 2-hydroxy-2-cyclopenten-1-one was only detected at Hdroxy-3（2 ）-furan-3-one, furfural, 2,5-dimethyl-4-hy- 150℃. It was also found in roasted palm kernels oil with a droxy-3（2H）-furanone and furfuryl alcohol. Based on the relatively long roasting time of 20 min at 180℃7）. literature, furfural was one of the predominant compounds Six alcohols were found and reached a maximum at among the furan derivatives of a roasted rapeseed variety 120℃. Benzyl alcohol was a major component. Compared named“ Brandy” and was regarded as a typical product of to unroasted bitter AKO, the relative concentration of a thermal reaction, which is generated from 1-deoxyosone benzyl alcohol increased almost 4 – 5 fold. It was one of the in the Maillard reaction23）. 2,5-Dimethyl-4-hydroxy-3（2H） dominant volatiles in almond oil and is often found in the -furanone was identified at 140℃ and increased almost 10 essential oils extracted from plants30）. Phenethyl alcohol fold at 150℃. This compound was first identified in pine- was detected in all samples and showed a non-significant（p

4 J. Oleo Sci. Roasting of Bitter Apricot Kernel Oil

Table 1 Volatile compounds identified in bitter AKO using HS-SPME-GC-MS.

5 J. Oleo Sci. F. Jin, J. Wang, J. M. Regenstein et al.

＞0.05）upward trend as roasting temperature increased. Phenethyl alcohol was described as having honey, spice, rose and lilac aromas22）. Small amounts of alkanes and alkenes were detected in all samples. Decane and dodecane were not detected until 130℃. Decane was also detected in flax seed and sesame seed oils31）. The relative concentration of toluene decreased and dis- appeared at 130℃ and above. Toluene was found in roasted cashews and was responsible for the green, fatty, and lard aromas28）. It was also found in roasted peanuts and was considered a high impact aroma-active com- pound32）. α-Pinene was only found in unroasted bitter AKO. The ester benzyl acetate was found in all samples Fig. 1 Spider plot of sensory profiles of bitter AKO with with the exception of 140℃. It is one of the most important different roasting temperatures. esters in ripe berries33）. 1,2-Dimethoxy-3-methylbenzene increased then decreased. Creosol, a guaiacol derivative, and 150℃ for 15 min gave the best sensory results. increased with temperature, but was relatively low However, the obvious caramel aroma at 150℃ might be ob- throughout. It was one of the most abundant volatile com- jectionable to some, so roasting at 140℃ was considered to pounds in olives. The volatile phenols in olive oils lead to be the most suitable for the production of bitter AKO. musty and muddy flavors34）. Other ethers disappeared at 130℃. However, acids were not detected, possibly because 3.3 In uence of roasting temperature on quality of their high volatility and heat lability7）. In addition, benzyl 3.3.1 Color cyanide was detected in all samples. After 15 min of thermal heating, the lightness L* of bitter AKO decreased quickly from 140 to 150℃（ Table 2）. 3.2 Sensory evaluation The a* value increased significantly（p＜0.05）with roasting The average scores for bitter AKO at different tempera- temperature, while the b* value also increased significantly tures are shown in Fig. 1. The major descriptor for un- （p＜0.05）. These experimental results were in agreement roasted bitter AKO were grassy and oily. This could be with a study of where the L* and a* values de- related to the volatile aldehydes and alcohols like nonanal creased and increased, respectively27）. The composite color and 1-hexanol. At 120℃, the grassy descriptor decreased of bitter AKO changed gradually from light yellow to deep while sweet, nutty and roast aromas were noted. The sweet brown, which was consistent with that found with cashew aroma could be explained by the significantly（p＜0.05）in- oils37）. This browning has been attributed to the formation creased benzaldehyde, which was related to almond, sugar, of Maillard reaction products and other non-enzymatic re- burnt and cherry flavors22, 27）. Other aldehydes, like phenyl- actions38）. acetaldehyde, which provided floral and honey aromas 3.3.2 Acid value and peroxide value were also found at 120℃. The relative concentration of According to the“ Hygienic Standard for Edible Vegeta- benzyl alcohol also increased significantly（p＜0.05）at ble of the People’s Republic of China39）, edible vegetable 120℃ and this compound contributed to the cherry and oils with an AV of ＜3 mg/g are acceptable. The AV of bitter walnut flavors. Roasted and nutty aromas were noted at AKO showed an increasing trend as roasting temperature 130℃ and intensified as roasting temperature increased. increased（Table 2）（ but was well below the standard）. AV The perception of roasted aroma was probably associated increased during thermal processing. PV also increased with the generation of pyrazine compounds. At 130℃ they with temperature but was also well below the value of 2.5 accounted for a large proportion of total volatiles, and meq/kg for almond oil that was suggested as an acceptabili- these compounds contribute roast and nutty aromas. ty cut-off. At about 5 meq/kg and above a rancid flavor was Pyrrole derivatives like 2-acetylpyrrole were correlated to noted25）. The PV of hazelnut oil increased slightly as roast- nut, walnut and bread flavors. Also the furan derivatives ing temperature went from 100 to 150℃40）. like furfural, furfuryl alcohol and 2,5-dimethy-4-hydroxy-3 3.3.3 Total phenols （2H）-furanone contributed caramel flavors to bitter AKO. Total phenols fluctuated（Table 2）which was also found As the roasting temperature increased, oily flavor also in- with Pistacia terebinthus（a snack food from the same creased. This could be explained by the increased alde- family as pistachios）oil. The increase was mainly related to hydes. the accumulation of relatively polar compounds44）. The in- The sensory evaluations were consistent with the result crease of total phenols could be related to their high resis- of the volatiles analysis. The roasting temperature of 140 tance to oil oxidation. Many Maillard reaction products

6 J. Oleo Sci. Roasting of Bitter Apricot Kernel Oil

Table 2 General properties of bitter AKO with different roasting temperature. Roasting temperature (℃) Parameter Control 120 130 140 150 L* 38.8±0.1d 38.4±0.1c 37.5±0.1b 37.2±0.2b 31.8±0.2a a* 4.2±0.1a 5.2±0.1b 5.7±0.1c 6.6±0.1d 7.3±0.2e b* 8.8±0.2a 10.2±0.1b 11.3±0.1c 11.7±0.2d 14.3±0.1e Acid value, AV (mg KOH/g) 0.60±0.03a 0.83±0.02b 0.93±0.03c 1.1±0.1d 1.4±0.02e Peroxide value, PV (meq/kg) 0.46±0.01a 0.49±0.01b 0.57±0.02c 0.64±0.01d 0.81±0.01e Total phenols content (µg GAE/g) 71.5±0.6d 54.1±0.4a 68.2±0.3c 55.0±0.4a 60.2±0.5b Different letters in the same row represented significant differences (p < 0.05). Control represented unroasted AKO sample.

Table 3 Changes in major fatty acid content of bitter AKO with different roasting temperatures. Roasting temperature (℃) Fatty acids (%) Control 120 130 140 150 , C16:0 4.45 ±0.07a 4.44±0.06a 4.41±0.03a 4.45±0.08a 4.41±0.07a Palmitoleic acid, C16:1 0.69 ±0.04a 0.69±0.04a 0.7±0.02a 0.69±0.01a 0.7±0.04a , C18:0 0.8±0.02a 0.83±0.03a 0.81±0.03a 0.78±0.08a 0.81±0.02a Oleic acid, C18:1 72.4±0.1a 72.1±0.52a 72.2±0.12a 72.4±0.16a 72.2±0.27a , C18:2 21.8±0.29a 22.1±0.18a 21.9±0.1a 22±0.28a 21.8±0.4a Different letters represented significant differences (p < 0.05).

have shown strong antioxidant activities38）. The control had perature had the lowest CDA during storage. Similar the highest total phenols followed by the samples at 130℃ results were found with soybean oils46）and mustard seed which suggested the possible volatilizing of these com- oil15）. These results probably reflect the previously dis- pounds at higher temperature. cussed production of anti-oxidant compounds at higher 3.3.4 Fatty Acids temperatures. Nine fatty acids were identified in apricot kernel oil in- cluding small amounts of arachidic and eicosanoic acids and some odd numbered fatty acids. Table 3 shows the changes in the major fatty acids. Oils with high levels of 4 CONCLUSIONS oleic and linoleic acid are commercially desirable. The The pyrazines as the major aroma volatiles contributed results were similar to those for Longwangmo and Prunus pleasant aromas like roast and nutty to the crude extract armeniaca apricot4, 49）. AKO（prepared using either baked of bitter AKO. Benzaldehyde which was detected in all （80℃）or sun-dried kernels, cold pressed and then heat tested samples although higher in roasted bitter AKO con- pressed at 40 and 120℃, respectively）contained 70.3- tributed almond and sugar aromas. Other aroma volatiles 71.3％ oleic acid and 22.3-23.0％ linoleic acid4）. Roasting like pyran, furan derivatives, aldehydes and alcohols also had little or no effect on the fatty acid composition similar probably contributed to the typical aroma of bitter AKO. to cashew oils37）. On the other hand, minor changes oc- From the sensory results and the relative concentration curred in hazelnuts, rice germs and sesame oils during changes in volatiles, 140℃ was considered as the most roasting41－43）. suitable roasting temperature for the production of bitter 3.3.5 Oxidative stability AKO. AV and PV were consistent with the requirements for PV and CDA values were measured during storage of the an edible oil. During the accelerated oxidation experi- oils（Fig. 2）. PV of the unroasted oil increased at 60℃ sig- ments, PV and CDA, both increased with the higher roast- nificantly（p＜0.05）in the first 12 days. However, the PV of ing temperature showing the slowest changes, i.e., roasting the roasted oils after 12 days was less than the control and would increase the oxidative stability of bitter AKO sug- inversely related to the roasting temperature. Similar gesting that this might be an economical and efficient way results were found for perilla seed oil45）. to use apricot kernels. CDA measures the primary oxidation products of poly- unsaturated fatty acids14（） Fig. 2B）. The initial CDA values of all samples were similar. Again the higher roasting tem-

7 J. Oleo Sci. F. Jin, J. Wang, J. M. Regenstein et al.

2） Femenia, A.; Rossello, C.; Mulet, A.; Canellas, J. Chem- ical composition of bitter and sweet apricot kernels. J. Agric. Food Chem. 43, 356-361（1995）. 3） Alpaslan, M.; Hayta, M. Apricot kernel: Physical and chemical properties. J. Am. Oil Chem. Soc. 83, 469- 471（2006）. 4） Zhou, B.; Wang, Y.; Kang, J.L.; Zhong, H.Y.; Prenzler P.D. The quality and volatile-profile changes of Long- wangmo apricot（Prunus armeniaca L.）kernel oil prepared by different oil-producing processes. Eur. J. Lipid Sci. Technol. 118, 236-243（2016）. 5） Wang, L.B.; Yu, H.Y. Biodiesel from Siberian apricot （Prunus sibirica L.）seed kernel oil. Bioresour. Tech- nol. 112, 355-358（2012）. 6） Tian, H.L.; Yan, H.Y.; Tan, S.W.; Zhan, P.; Mao, X.Y.; Wang, P.; Wang, Z.P. Apricot kernel oil ameliorates cy- clophosphamide-associated immunosuppression in rats. Lipids 51, 1-9（2016）. 7） Zhang, W.C.; Wang, R.; Yuan, Y.H.; Yang, T.K.; Liu, S.Q. Changes in volatiles of palm kernel oil before and after kernel roasting. LWT-Food Sci. Technol. 73, 432-441 （2016）. 8） Liu, X.J.; Jin, Q.Z.; Liu, Y.F.; Huang, J.H.; Wang, X.G.; Mao, W.Y.; Wang, S.S. Changes in volatile compounds of peanut oil during the roasting process for produc- tion of aromatic roasted peanut oil. J. Food Sci. 76, C404-C412（2011）. 9） Park, M.H.; Jeong, M.K.; Yeo, J.D.; Son, H.J.; Lim, C.L.; Hong, E.J.; Noh, B.S.; Lee, J.H. Application of solid phase-microextraction（SPME）and electronic nose techniques to differentiate volatiles of sesame oils pre- pared with diverse roasting conditions. J. Food Sci. Fig. 2 Changes in peroxide values（A）and conjugated 76, C80-C88（2011）. dienoic acid values（B）of bitter AKO during 10） Siegmund, B.; Murkovic, M. Changes in chemical com- accelerated oxidation experiments at different position of pumpkin seeds during the roasting process roasting temperatures. for production of pumpkin seed oil（Part 2: Volatile compounds）. Food Chem. 84, 367-374（2004）. ACKNOWLEDGMENTS 11） Agila, A.; Barringer, S. Effect of roasting conditions on This paper was supported by the“ Fundamental Re- color and volatile profile including HMF level in sweet search Funds for the Central University（No. 2015ZCQ-SW- almonds（Prunus dulcis）. J. Food Sci. 77, 1-8（2012）. 04）”. 12） Durmaz, G.; Alpaslan, M. Antioxidant properties of roasted apricot（Prunus armeniaca L.）kernel. Food Chem. 100, 1177-1181（2007）. 13） Neđeral, S.; Škevin, D.; Kraljić, K.; Obranović, M.; Conflict of Interest Statement Papeša, S.; Bataljaku, A. Chemical composition and The authors have no conflicts of interest with respect to oxidative stability of roasted and cold pressed pump- this manuscript. kin seed oils. J. Am. Oil Chem. Soc. 89, 1763-1770 （2012）. 14） Lee, S.W.; Jeung, M.K.; Park, M.H.; Lee, S.Y.; Lee, J.H. Effects of roasting conditions of sesame seeds on the References oxidative stability of pressed oil during thermal oxida- 1） Wang, L.B. Evaluation of Siberian apricot（Prunus si- tion. Food Chem. 118, 681-685（2010）. birica L.）germplasm variability for biodiesel proper- 15） Vaidya, B.; Choe, E. Effects of seed roasting on to- ties. J. Am. Oil Chem. Soc. 89, 1743-1747（2012）. copherols, carotenoids, and oxidation in mustard seed

8 J. Oleo Sci. Roasting of Bitter Apricot Kernel Oil

oil during heating. J. Am. Oil Chem. Soc. 88, 83-90 （Anacardium occidentale L.）from different geo- （2011）. graphical origins during roasting. J. Food Sci. 76, 768- 16） AOCS. Official Methods and Recommended Practic- 774（2011）. es of the American Oil Chemists’ Society, 5th ed. 29） Xiao, L.; Lee, J.Y.; Zhang, G.; Ebeler, S.E.; Wickramas- AOCS Press, Champaign, IL, USA（1997）. inghe, N.; Seiber, J.; Mitchell, A.E. HS-SPME GC/MS 17） Gutfinger, T. Polyphenols in olive oils. J. Am. Oil characterization of volatiles in raw and dry-roasted al- Chem. Soc. 58, 966-968（1981）. monds（Prunus dulcis）. Food Chem. 151, 31-39 18） International Organization for Standardization（ISO）, （2014）. Animal and vegetable oils and fats̶analysis by gas 30） Pićurić-Jovanović, K.; Milovanović, M. Analysis of vola- chromatography of methyl esters of fatty acids. Stan- tile compounds in almond and plum kernel oils. J. Am. dard No. 5508. ISO, Geneva, Switzerland（1990）. Oil Chem. Soc. 70, 1101-1104（1993）. 19） Alasalvar, C.; Shahidi, F.; Cadwallader, K.R. Compari- 31） Ivanova-Petropulos, V.; Mitrev, S.; Stafilov, T.; Markova, son of natural and roasted Turkish Tombul hazelnut N.; Leitner, E.; Lankmayr, E.; Siegmund, B. Characteri- （Corylus avellana L.）volatiles and flavor by DHA/GC/ sation of traditional Macedonian edible oils by their MS and descriptive sensory analysis. J. Agric. Food fatty acid composition and their volatile compounds. Chem. 51, 5067-5072（2003）. Food Res. Int. 77, 506-514（2015）. 20） Kwon, T.Y.; Park, J.S.; Jung, M.Y. Headspace-solid 32） Smith, A.L.; Perry, J.J.; Marshall, J.A.; Yousef, A.E.; phase microextraction-gas chromatography-tandem Barriger, S.A. Oven, microwave, and combination mass spectrometry（HS-SPME-GC-MS2）method for the roasting of peanuts: Comparison of inactivation of Sal- determination of pyrazines in perilla seed oils: Impact monella surrogate Enterococcus faecium, color, vol- of roasting on the pyrazines in perilla seed oils. J. Ag- atiles, flavor, and lipid oxidation. J. Food Sci. 79, ric. Food Chem. 61, 8514-8523（2013）. S1584-S1594（2014）. 21） Jiao, S.S.; Zhu, D.D.; Deng, Y.; Zhao, Y.Y. Effects of hot 33） Kallio, H. Development of volatile aroma compounds air-assisted radio frequency heating on quality and in arctic bramble, Rubus arcticus L. J. Food Sci. 41, shelf-life of roasted peanuts. Food Bioprocess. Tech- 563-566（1976）. nol. 9, 308-319（2016）. 34） Cortés-Delgado, A.; Sánchez, A.H.; Castro, A.D.; 22） Ling, B.; Yang, X.M.; Li, R.; Wang, S.J. Physicochemical López-López, A.; Beato, V.M.; Montaño, A. Volatile properties, volatile compounds, and oxidative stability profile of Spanish-style green table olives prepared of cold pressed kernel oils from raw and roasted pista- from different cultivars grown at different locations. chio（Pistacia vera L. Var Kerman）. Eur. J. of Lipid Food Res. Int. 83, 131-142（2016）. Sci. Technol. 118, 1368-1379（2016）. 35） Shahidi, F.; Aishima, T.; Abou-gharbia, H.A.; Youssef, 23） Gracka, A.; Jeleń, H.H.; Majcher, M.; Siger, A.; Kacz- M.; Shehata, A.Y. Effect of processing on flavor precur- marek, A. Flavoromics approach in monitoring chang- sor amino acids and volatiles of sesame paste（Tehi- es in volatile compounds of virgin rapeseed oil caused na）. J. Am. Oil Chem. Soc. 74, 667-679（1997）. by seed roasting. J. Chromatog. A 1428, 292-304 36） Cai, J.; Zhu, B.Q.; Wang, Y.H.; Lu, L., Lan, Y.B.; Reeves, （2016）. M.J.; Duan, C.Q. Influence of pre-fermentation cold 24） Lasekan, O. Headspace solid-phase microextraction maceration treatment on aroma compounds of Caber- gas chromatography–mass spectrometry（HS-SPME- net Sauvignon wines fermented in different industrial GC–MS）determination of volatile compounds in roast- scale fermenters. Food Chem. 154, 217-229（2014）. ed plantains（French sombre and Dwarf kalapua）. 37） Chandrasekara, N.; Shahidi, F. Oxidative stability of LWT-Food Sci. Technol. 46, 536-541（2012）. cashew oils from raw and roasted nuts. J. Am. Oil 25） Yang, J.H.; Pan, Z.L.; Takeoka, G.; Mackey, B.; Bingol, Chem. Soc. 88, 1197-1202（2011）. G.; Brandl, M.T.; Garcin, K.; McHugh, T.H.; Wang, H. 38） Lee, Y.C.; Oh, S.W.; Chang, J.; Kim, I.H. Chemical com- Shelf-life of infrared dry-roasted almonds. Food position and oxidative stability of safflower oil pre- Chem. 138, 671-678（2013）. pared from safflower seed roasted with different tem- 26） Kim, S.J.; Yoon, H.N.; Rhee, J.S. The effects of roasting peratures. Food Chem. 84, 1-6（2004）. temperatures on the formation of headspace volatile 39） Ministry of Health of the People’s Republic of China compounds in perilla seed oil. J. Am. Oil Chem. Soc. and Standardization Administration of the People’s 77, 451-456（2000）. Republic of China. Hygienic standard for edible vege- 27） Hojjati, M.; Lipan, L.; Carbonell-Barrachina, Á. Effect table oil. Beijing, China（2005）. of roasting on physicochemical properties of wild al- 40） Özkan, G.; Kiralan, M.; Karacabey, E.; Çalik, G.; Öz- monds（Amygdalus scoparia）. J. Am. Oil Chem. demir, N.; Tat, T.; Bayrak, A.; Ramadan, M.F. Effect of Soc. 93, 1211-1220（2016）. hazelnut roasting on the oil properties and stability 28） Agila, A.; Barringer, S.A. Volatile profile of cashews under thermal and photooxidation. Eur. Food Res.

9 J. Oleo Sci. F. Jin, J. Wang, J. M. Regenstein et al.

Technol. 242, 2011-2019（2016）. C1278（2012）. 41） Alasalvar, C.; Pelvan, E.; Topal, B. Effects of roasting 46） Jung, M.Y.; Bock, J.Y.; Back, S.O.; Lee, T.K.; Kim, J.H. on oil and fatty acid composition of Turkish hazelnut Pyrazine contents and oxidative stabilities of roasted varieties（Corylus avellana L.）. Int. J. Food Sci. soybean oils. Food Chem. 60, 95-102（1997）. Nutr. 61, 630-642（2010）. 47） Park M.H.; Seol, N.G.; Chang, P.S.; Yoon, S.H.; Lee, J.H. 42） Kim, I.H.; Kim, C.J.; You, J.M.; Lee, K.W.; Kim, C.T.; Effects of roasting conditions on the physicochemical Chung, S.H.; Tae, B.K. Effect of roasting temperature properties and volatile distribution in perilla oils（Pe- and time on the chemical composition of rice germ oil. rilla frutescens var. japonica）. J. Food Sci. 76, J. Am. Oil Chem. Soc. 79, 413-418（2002）. C808-C815（2011）. 43） Elleuch, M.; Besbes, S.; Roiseux, O.; Blecker, C.; Attia, 48）“ Flavornet and human odor space”. http://flavornet. H. Quality characteristics of sesame seeds and by- org. Accessed Apr 20, 2017. products. Food Chem. 103, 641-650（2007）. 49） Durmaz G.; Karabulut I.; Topcu, A.; Asiltürk, M.; Kutlu, 44） Durmaz, G.; Gökmen, V. Changes in oxidative stability, T. Roasting-related changes in oxidative stability and antioxidant capacity and phytochemical composition antioxidant capacity of apricot kernel oil. J. Am. Oil of Pistacia terebinthus oil with roasting. Food Chem. Chem. Soc. 87, 401-409（2010）. 128, 410-414（2011）. 50） Meng, F.X.; Zhang, H.J. Suggestions about limited 45） Zhao, T.T.; Hong, S.I.; Lee, J.; Lee, J.S., Kim, I.H. Im- amount indexes of cyanide on the bitter apricot oils. pact of roasting on the chemical composition and oxi- Food Res Dev. 34, 100-102（2013）. dative stability of perilla oil. J. Food Sci. 77, C1273-

10 J. Oleo Sci.