Amine-Functionalized ZIF-8 as a Fluorescent Probe for Breath Volatile Organic Compound Biomarker Detection of Lung Cancer Patients.

The highly thermally and chemically stable imidazole framework ZIF-8 samples were separately postmodified with amine groups by using N,N'-dimethylethylenediamine (MMEN) and N,N-dimethylaminoethylamine (MAEA), which had the same molecular formula but different structures. The modified ZIF-8 samples (ZIF-8@amine) were thoroughly characterized, including powder X-ray diffractometry, Fourier-transformed infrared spectroscopy, and physical adsorption at 77 K by nitrogen, thermogravimetric analysis, and photophysical characterization. Results showed that after modification, the Brunauer-Emmett-Teller surface area and total pore volume both increased, almost one time higher than those of the original ZIF-8 sample, and followed the order: ZIF-8-MMEN > ZIF-8-MAEA > ZIF-8. Furthermore, the N-H group was successfully grafted into the modified ZIF-8 samples. To examine the sensing properties of the modified ZIF-8@amine samples toward the breath biomarkers of lung cancer, five potential volatile organic compound biomarkers were used as analytes. ZIF-8-MMEN and ZIF-8-MAEA revealed a unique capacity for sensing hexanal, ethylbenzene, and 1-propanol with high efficiency and sensitivity. The three samples all did not show sensing ability toward styrene and isoprene. In addition, ZIF-8, ZIF-8-MMEN, and ZIF-8-MAEA all can sense hexanal with a detection limit as low as 1 ppb.

Introduction

Nowadays, periodic computed tomography and bronchoscopy, which are based upon invasive procedures accompanied by the culture of the pathogen are the principle techniques used for lung cancer.[1−3] Compared to these traditional methods, analysis of exhaled breath has shown increasing potential to provide a highly informative, specific, and noninvasive alternative diagnostic tool. Volatile organic compounds (VOCs) and other low molecular weight compounds (such as carbon monoxide and ammonia) can supply a fingerprint of metabolic products and can be released by various bodily routes, such as through skin, urine, and breath. Currently, exhaled breath VOCs have been extensively studied for pollutant exposure and disease diagnosis.[2,4−7]

Lung cancer is the foremost cause of cancer-related deaths, causing approximately 1.8 million fatal casualties per year worldwide. The danger arrives from the unrestrainable nature of abnormal cells that begin in one or both lungs and are prone to spread to other parts of the human body rapidly. Monitoring of specific cancer biomarkers present in the exhaled VOCs is of particular interest for human safety and quality of life because of the severity of lung cancer. Many early lung cancers are curable, and the cure rate of orthotopic lung cancer in situ is close to 100%. Nevertheless, the 5-year survival rate after surgical resection of stage III patients is only 5–20% while stage I patients was up to 60–90%. Therefore, if proper methods for early-stage disease diagnosis can be found, especially by developing novel material-based breath sensor approaches with highly selective and ultrasensitive capabilities,[8−14] biomarker monitoring can be greatly improved. However, the screening of lung cancer breath biomarkers has still been a challenge because of the absence of unified sampling and analyzing methods, small sample size, and individual differences in patients. Saalberg and Wolff[15] reviewed the recent studies (/publications from 1985 to 2015) on VOCs which were considered lung cancer biomarkers for breath analysis diagnosis and concluded that the six most frequently emerging biomarkers were 2-butanone, 1-propanol, isoprene, ethylbenzene, styrene, and hexanal. For the purpose of looking at the fingerprint of breath VOCs of patients with lung cancer, various types of electronic noses have been used for breath VOC detection.[16−20] Among the previous studies, many gas sensor systems have been applied to lung cancer diagnosis using breath gas detection,[10,21−23] such as polymer/carbon composites,[17] surface acoustic wave gas sensors,[24] quartz crystal microbalance gas sensors,[25] conducting polymer gas sensors,[26] colorimetric sensor arrays,[27] and metal oxide gas sensors.[28] Nevertheless, although many efforts have been done to develop novel sensors, a more practical, robust technology for early diagnosis of lung cancer is still nonexistent today.

Because many recent research studies have focused on high mobility, as well as low cost, and low power solutions to the trace breath VOCs, the development and application of sensing materials or composites have become a hot issue with considerable research interest. Metal–organic frameworks (MOFs) have been proved to be an appropriate candidate for a variety of biosensing systems because of their remarkable physical–chemical properties and infinite network structures which can be directly self-assembled through the coordination of metal ions/clusters with organic linkers.[29−34] MOFs exhibit many interesting characteristics such as high specific surface areas, small crystal density, various types, adjustable pore size, and tunable internal surfaces. These properties make MOFs be extensively explored in applications such as gas storage,[31,33−38] separation,[39,40] catalysis,[41,42] drug delivery,[43,44] and chemical sensors[45−51] over the past decades. Especially, MOFs have inherent advantages to be fluorescent sensors for the detection and recognition of various small molecules, anions, cations, and VOCs as well based upon their fluorescence response.

In spite of many advantages, MOFs have also shown some drawbacks, for example, low thermal stability and high moisture susceptibility.[52] These shortcomings have restricted their applications in certain areas. Postsynthetic modification strategy has been expected to modify the pore structures and help in generating functional properties that are not directly accessible from conventional MOF synthetic procedures and have been demonstrated as a powerful tool for introducing specific functional groups into MOFs.[13,53−60] The postsynthetic modification retains the intact crystal structure, nevertheless the physicochemical properties of the materials can be altered. The various advantages of postsynthetic modification are highlighted, which make these modification methods play a key role in development and application of MOF materials.[61] The imidazole framework ZIF-8 samples are one of the most classic and highly stable MOFs. Since its discovery in 2002,[62] it has been extensively investigated in the areas of synthetic methods and potential applications.[63−67]

Herein, a postsynthetic modification strategy was used to modify ZIF-8 with N,N′-dimethylethylenediamine (MMEN) and N,N-dimethylaminoethylamine (MAEA) as the functional groups. MMEN and MAEA had the same molecular formula but different structures, and they were introduced into the framework of ZIF-8 samples. The modified ZIF-8-MMEN and ZIF-8-MAEA were applied to fluorescence detection of lung cancer VOC biomarkers. Herein, hexanal, 1-propanol, ethylbenzene, isoprene, and styrene are chosen to be the objective breath biomarkers of lung cancer in this study based on the former studies. As far as we know, this is the first time that postmodified MOFs are employed as fluorescent probes for sensing breath VOC biomarkers of lung cancer.

Experimental Section

Materials

Zinc nitrate hexahydrate (99.9% purity) was purchased from Alfa Chemicals. 2-Methylimidazole (98% purity), N,N-dimethylformamide (DMF; 99.9% purity), methanol (99.9% purity), MMEN (97% purity), MAEA (99% purity), and n-hexane (95% purity) were purchased from J&K Chemical. Chloroform was purchased from Samhoo Trading Co., LTD.

Physical Methods

The FTIR spectra of the materials were recorded using KBr pellets on a Bruker VERTEX70 spectrometer within the range 4000–400 cm–1. The thermogravimetric analyses (TGA) was measured with a Netzsch TG209F3-ASC TG analyzer at a heating rate of 5 K min–1, which were carried out under a N2 atmosphere. The power X-ray diffraction (PXRD) patterns were obtained on a Bruker D8 Advance automated diffractometer for structure analysis using Cu Kα radiation (40 kV and 40 mA) with a scan range of 2θ from 5 to 60° at a scan speed of 2°/min.

Preparation and Modification of ZIF-8 Materials

Preparation of ZIF-8 was carried out following the reported procedures.[35,67] The modification of ZIF-8 crystals was performed using the following steps. First, 60 mL n-hexane was added to 0.5 g of the ZIF-8 sample which was placed in a 100 mL single-mouth flask. The flask was placed in a self-made oil bath which was placed on a magnetic stirring vessel at 320 rad/min. Second, N,N′-dimethylethylenediamine (denoted as MMEN) was added into the flask dropwise. After that, the condensation reflux was conducted and the temperature of the magnetic stirring vessel was set at 85 °C, lasting for 12 h. Finally, the mixture was filtered and the sample was dried at 110 °C for 12 h. The modification method of N,N-dimethylaminoethylenediamine (denoted as MAEA) was the same as that of MMEN. The concentration gradient of the experiments and the name of the obtained samples can be found in Table S1. The molecular structure of MMEN and MAEA are depicted in Figure . In the study, the MMEN-modified samples with different concentrations were named ZIF-8-MMEN1, ZIF-8-MMEN2, .... The MAEA-modified samples with different concentrations were named ZIF-8-MAEA-1, ZIF-8-MAEA-2, .... The modified samples, which were used for the fluorescence study, were the best among all the samples, and they were named ZIF-8-MMEN and ZIF-8-MAEA, respectively.

Figure 1

Molecular structures of MMEN and MAEA used in the study. (a) MMEN. (b) MAEA.

Pore Structure Analysis

An autosorb IQ2 gas adsorption analyzer equipped with commercial software of calculation and analysis was used to characterize the specific surface area and pore volume of the ZIF-8 and modified ZIF-8 materials. Prior to measurement of the N2 adsorption–desorption experiments, the ZIF-8 samples were degassed at 150 °C for 10 h. The Brunauer–Emmett–Teller (BET) and Langmuir specific surface area were calculated from the isotherms using the BET and Langmuir equations, respectively. The total pore volume was calculated through converting the adsorption amount at p/p0 to a volume of liquid adsorbate, and the micropore volume was calculated by using the t-plot method of Lippens and de Boer to the adsorption data. Finally, the pore size distribution was determined by the density functional theory (DFT) equation.

Fluorescence Experiment

The fluorescence properties of the modified ZIF-8 samples were examined in the solid state in organic solvent suspensions. The photoluminescence excitation and emission spectra were recorded using a Hitachi F-7000 spectrophotometer. A 150 W xenon lamp was equipped as the excitation source. Prior to the measurements, the solid ZIF-8-modified samples were ground into powder and used for sensing experiments. Powder samples (6 mg) of ZIF-8, ZIF-8@MMEN, and ZIF-8@MAEA were dispersed into 3 mL hexanal, styrene, isoprene, 1-propanol, and ethylbenzene, respectively. Suspensions were treated with ultrasonication for 5 min prior to fluorescence measurements.

Results and Discussion

PXRD of the Modified ZIF-8 Samples

Figure presents the powder X-ray single-diffraction (PXRD) patterns of the modified ZIF-8@MMEN and ZIF-8@MAEA samples. As it can be seen from Figure , the main peaks of the modified ZIF-8 materials match very well with the ZIF-8 sample, revealing that the modified ZIF-8 samples with different concentrations of MMEN and MAEA maintain integrated well after the modification. Furthermore, the main peaks of the original ZIF-8 material match well with the simulated XRD pattern of the ZIF-8 crystal.[62]

Figure 2

(a) PXRD patterns of ZIF-8 as made and ZIF-8@MMEN samples. (b) PXRD patterns of ZIF-8 as made and ZIF-8@MAEA samples.

Textural Properties

Figure a,c presents the N2 adsorption–desorption isotherms on the ZIF-8 and the modified samples. It can be noted that the N2 isotherms on the modified samples all are of type-I isotherms, illustrating that there are mainly micropores in the modified ZIF-8 samples. Figure b,d shows the pore size distribution profiles of the modified ZIF-8 samples, calculated using the DFT. It was noted that the average pore size of ZIF-8 as made was mainly distributed at 10.868 Å; however, after modification, the average pore size of the modified samples obviously decreased. For ZIF-8-MAEA samples, the pore size were mainly distributed at 9.26 Å, while the average pore size of ZIF-8-MMEN was distributed at 8.89 Å. The decreased pore size was because of the steric hindrance effect causing by the introduction of the N–H group from MMEN and MAEA.

Figure 3

(a) N2 adsorption–desorption isotherm on the ZIF-8 sample as made and ZIF-8-MMEN samples. (b) DFT pore size distribution of the ZIF-8 as made and ZIF-8-MMEN samples. (c) N2 adsorption–desorption isotherms on the ZIF-8 sample as made and ZIF-8-MAEA samples. (d) DFT pore size distribution of the ZIF-8 as made and ZIF-8-MAEA samples.

The textural properties obtained from the N2 adsorption–desorption isotherms are summarized in Table . It can be found that the micropore volume and total pore volume of the modified samples were much larger than those of the original ZIF-8, indicating that the guest molecules were removed completely and formation of new pores. After modification with MMEN, the BET surface area and Langmuir surface area both obviously increased, and followed the order: ZIF-8-MMEN2 > ZIF-8-MMEN1> ZIF-8-MMEN3 > ZIF-8. The total pore volume of the MMEN-modified samples obviously increased, almost one time higher than that of the original sample, and they followed the order: ZIF-8-MMEN1 > ZIF-8-MMEN2 > ZIF-8-MMEN3 > ZIF-8. After the concentration of MMEN reached a specific value, the BET surface area are basically the same, indicating that the modifications reached saturation. For the ZIF-8-MAEA samples, the BET surface area and total pore volume all increased notably, following the order: ZIF-8-MAEA2 > ZIF-8-MAEA3 > ZIF-8-MAEA1 > ZIF-8. Interestingly, it can be noted that the BET surface area and total volume of ZIF-8-MAEA3 and ZIF-8-MAEA1 were very close. The pore structure parameters of modified samples by MAEA show a trend of increasing first and then decreasing. Above all, although MMEN and MAEA have the same molecular formula, but the steric hindrance resulting from the straight chain of MMEN is much smaller than that from the branch chain of MAEA. Furthermore, the dramatically increased micropore volume and total pore volume of ZIF-8-MMEN and ZIF-8-MAEA samples can be ascribed to the removed guest molecules and formation of new pores with the introduction of N–H groups. Thereafter, the ZIF-8-MMEN2 and ZIF-8-MAEA2 can be regarded as the two best samples among the modified samples, and they would be denoted as ZIF-8-MMEN and ZIF-8-MAEA, respectively, in the following fluorescence sensing experiments.

Table 1

Pore Structure Parameters of ZIF-8 as Made and Modified ZIF-8 Samples

sample	BET (m2/g)	Langmuir (m2/g)	micropore volume (cm3/g)	micropore area (m2/g)	total pore volume (cm3/g)
ZIF-8	1038	1089	0.370	1019	0.390
ZIF-8-MMEN1	1963	1944	0.674	1931	0.809
ZIF-8-MMEN2	1993	1982	0.649	1890	0.782
ZIF-8-MMEN3	1899	1886	0.657	1874	0.700
ZIF-8-MAEA1	1390	1376	0.476	1367	0.539
ZIF-8-MAEA2	1838	1815	0.631	1813	0.690
ZIF-8-MAEA3	1435	1468	0.478	1349	0.580

FTIR Characterization of Modified ZIF-8 Materials

Figure shows the FTIR spectra of the modified ZIF-8 materials. As it can be noted, the characteristic peaks of the modified ZIF-8-MMEN were slightly different from original ZIF-8 in that: (1) the peak at 3134 cm–1 which was assigned to the band of N–H was more pronounced on modified ZIF-8 than those on the original ZIF-8 samples, indicating that the N–H group was introduced into ZIF-8 successfully; (2) the peak that appeared at 2930 cm–1 was more evident and it can be attributed to −CH2 antisymmetric stretching vibration for the modified samples; (3) two peaks of ZIF-8-MMEN1 and ZIF-8-MMEN2 observed at 2963 cm–1 were more evident which was related to the antisymmetric stretching vibration of −CH3 introduced by modification of MMEN; (4) the spectrum of ZIF-8-MMEN3 also revealed the formation of a new peak at 2972 cm–1 because of −CH3 stretching vibration. For the ZIF-8@MAEA samples, there were two main differences as shown in Figure c and Figure d. First, it can be found that the N–H stretching vibration peak at 3136 cm–1 after modification was obviously enhanced, and the peak intensity increased with the increase of the concentration of MAEA, indicating that the N–H bond was successfully introduced into ZIF-8. Second, the −CH2 antisymmetric stretching vibration peak at 2930 cm–1 was more evident than before. Thereafter, the N–H groups was successfully introduced into the modified ZIF-8 samples with MMEN and MAEA.

Figure 4

(a) FTIR spectra of ZIF-8@MMEN samples between 4000 cm–1–400 cm–1. (b) Enlarged FTIR spectra of ZIF-8@MMEN samples between 3800 cm–1–2800 cm–1. (c) FTIR spectra of ZIF-8@MAEA samples between 4000 cm–1–400 cm–1. (d) Enlarged FTIR spectra of ZIF-8@MAEA between 3800 cm–1–2800 cm–1.

Thermogravimetric Analysis

Figure S1 shows the thermogravimetric profiles of the modified ZIF-8 samples. The weight loss of the modified ZIF-8 samples was also listed in Tables S2 and S3. It can be noted that there were two distinct weight loss steps. The first step of ZIF-8 was corresponded to the loss of guest molecules, and the second weight loss step was because of the collapse of frameworks. For the first stage of modified ZIF-8 samples, the weight loss was attributed to release of guest molecules and introduced groups. During the second stage, the weight loss was mainly ascribed to decomposition of frameworks. The initial temperature of the first stage of ZIF-8@MAEA samples was lower than that of ZIF-8, indicating that the thermal stability was slightly decreased after MAEA modification. However, the initial temperature of the first stage of the ZIF-8@MMEN sample was higher than that of the ZIF-8 sample, indicating that MMEN modification can enhance the thermal stability of the samples.

Fluorescent Performance

The fluorescence of original ZIF-8, ZIF-8-MMEN, and ZIF-8-MAEA have been used for sensing the potential biomarkers of lung cancer patients (i.e., hexanal, styrene, isoprene, 1-propanol, and ethylbenzene) bearing electron-donating or -withdrawing groups, which can quench or enhance the fluorescence of ZIF-8. To demonstrate the applicability of the modified ZIF-8 crystals for sensing VOCs, the modified ZIF-8 samples were activated by vacuum at 100 °C for 6 h and then soaked at room temperature in various liquid VOCs. Figure presents the fluorescence intensity changes of ZIF-8, ZIF-8-MMEN, and ZIF-8-MAEA when exposed to the potential biomarkers of lung cancer patients (hexanal, ethylbenzene, and 1-propanol). As it was noted, in comparison with ZIF-8, ZIF-8-MMEN and ZIF-8-MAEA showed turn-off effects toward hexanal and ethylbenzene, while they all illustrated turn-on effects toward 1-propanol. This may be because of the fact that the transfer of electrons from the electron-donating framework to high electron deficiency one can occur upon excitation, causing fluorescence quenching. For the other two VOC biomarkers of lung cancer patients, styrene and isoprene, there was no fluorescence for ZIF-8, ZIF-8-MMEN, and ZIF-8-MAEA. Table lists the changes in fluorescence intensity and emission frequency shift of ZIF-8 and the modified ZIF-8 samples toward the selected VOCs. Interestingly, an emission frequency shift can be noticed for both ZIF-8-MMEN and ZIF-8-MAEA toward hexanal and ethylbenzene. This is because of the fact that after the introduction of the N–H group, the binding affinity between the analytes and the frameworks became much stronger. The strong binding affinity may be responsible for the emission frequency shift.[68] As other references have been stated, this property is very appealing and useful as far as signal transduction is concerned: the evolution of peak placement at a specific wavelength is easily tracked and monitored, such as hexanal and ethylbenzene.

Figure 5

Fluorescence intensity changes of ZIF-8, ZIF-8-MMEN, and ZIF-8-MAEA when exposed to the potential biomarkers of lung cancer patients (hexanal, ethylbenzene, and 1-propanol).

Table 2

Fluorescence Parameters and Changes of the Selected VOCs on ZIF-8-MMEN and ZIF-8-MAEA Samples As Compared to ZIF-8

parameters	compounds	ZIF-8	ZIF-8-MMEN	ZIF-8-MAEA
excitation/emission wavelength (nm)	hexanal	356/420	356/426	356/425
	ethylbenzene	330/384	330/389	330/390
	1-propanol	382/434	382/384	382/434
emission frequency shift (nm)	hexanal		6 nm/red shift	5 nm/red shift
	ethylbenzene		5 nm/red shift	6 nm/red shift
	1-propanol
change in emission intensity (%)	hexanal		–11.3%	–21.6%
	ethylbenzene		–37%	–43%
	1-propanol		+27.5%	+10%

Furthermore, the fluorescence detection experiments of the modified ZIF-8 samples were performed to examine the sensing sensitivity to a series of hexanal. To examine the sensing sensitivity toward hexanal in more details, a batch of suspensions of the modified ZIF-8 samples were prepared by dispersing them in a DMF solution, meanwhile gradually increasing the contents to monitor the emissive response. The fluorescence intensities increase with an increase of the concentration of the hexanal added to DMF, indicating that the detection limit of hexanal on all the three samples were lower than 1 ppb as seen from Figure .

Figure 6

(a) Emission spectra of ZIF-8 in different concentrations of hexanal. (Ex = 370 nm). (b) Emission spectra of ZIF-8-MMEN sample in different concentrations of hexanal. (Ex = 387 nm). (c) Emission spectra of ZIF-8-MAEA sample in different concentrations of hexanal. (Ex = 387 nm).

Regeneration Ability

Besides high sensitivity and a low detection limit, regeneration ability is another crucial factor in the performance of sensors. To study the recyclability of the modified ZIF-8 samples, the synthesized ZIF-8, ZIF-8-MMEN, and ZIF-8-MAEA were regenerated in vacuum at 100 °C for 6 h after exposure to hexanal. Remarkably, it was observed that the fluorescence intensities of ZIF-8, ZIF-8-MMEN, and ZIF-8-MAEA were regenerated completely (Figure ), and the emission intensities of these three samples almost maintain its original intensity even after four cycles of exposure to hexanal and vacuum treatment at 100 °C, suggesting a high reversibility of the modified ZIF-8 samples for potential applications. The PXRD patterns of the recovered ZIF-8 sample after four cycles showed no change compared to those of the as-synthesized samples and simulated patterns, indicating high stability of the framework (Figure S2). The good regeneration ability of the modified ZIF-8 samples illustrates the feasibility toward the detection of breath biomarkers of lung cancer.

Figure 7

Fluorescence intensities of ZIF-8 and ZIF-8@amine samples after four cycles. (a) ZIF-8, (b) ZIF-8-MMEN, (c) ZIF-8-MAEA.

Conclusions

The ZIF-8 crystals were modified by postsynthetic methods by introducing MMEN and MAEA into pores. After modification, the BET surface area and pore volume were both increased obviously. Especially, the BET surface area and total pore volume of the ZIF-8-MMEN sample were both almost one time higher than that of the ZIF-8 sample as made. Although, MMEN and MAEA has the same molecular formula, the modified ZIF-8-MMEN and ZIF-8-MAEA samples showed different fluorescence properties toward the selected potential biomarkers of lung cancer patients because of the steric hindrance effects. In comparison with ZIF-8, ZIF-8-MMEN and ZIF-8-MAEA showed turn-off effects toward hexanal and ethylbenzene, while they all illustrated turn-on effects toward 1-propanol. This may be because of the fact that the transfer of electron from the electron-donating framework to high electron deficiency can occur upon excitation, causing fluorescence quenching. However, for the other two VOC biomarkers of lung cancer patients, styrene and isoprene, there was no fluorescence for ZIF-8, ZIF-8-MMEN, and ZIF-8-MAEA. For the fluorescence sensitivity toward hexanal, the fluorescence intensities increase with an increase of the concentration of the hexanal added to DMF, and the detection limit of hexanal on all the three samples were lower than 1 ppb. Moreover, the modified ZIF-8 samples all showed good reproducibility.