SERRS. In Situ Substrate Formation and Improved Detection Using

Department Pure and Applied Chemistry, University of Strathclyde, Thomas Graham Building, 295 Cathedral Street,. Glasgow G1 1XL, U.K., and Department ...
0 downloads 0 Views 215KB Size
Download PDF
Anal. Chem. 2002, 74, 1503-1508

SERRS. In Situ Substrate Formation and Improved Detection Using Microfluidics Ruth Keir,† Eishi Igata,‡ Martin Arundell,‡ W. Ewen Smith,*,† Duncan Graham,† Callum McHugh,† and Jonathan M. Cooper‡

Department Pure and Applied Chemistry, University of Strathclyde, Thomas Graham Building, 295 Cathedral Street, Glasgow G1 1XL, U.K., and Department Electronics and Electrical Engineering, University of Glasgow, Rankine Building, Oakfield Avenue, Glasgow G12 8LT, U.K.

Surface-enhanced resonance Raman scattering (SERRS) of a model derivative of TNT was detected using a microflow cell designed within the framework of the labon-a-chip concept, using only the analyte and readily available reagents. The SERRS substrate, silver colloid, was prepared in situ, on-chip, by borohydride reduction of silver nitrate. The silver colloid was imaged within the chip using a white light microscope in either transmission or, due to the high reflectivity of the colloid, reflection mode. A fine stream of colloid ∼30 µm in width was formed in a 250-µm-wide channel at the point where the colloid preparation reagents met. The chip was designed to produce a concentrated stream of colloid within a laminar regime, such that particles did not readily disperse into the fluid. One result of this was to reduce the effective volume of analysis. Attempts to deliberately disrupt this stream with microstructured pillars, fabricated in the fluidic channels, were unsuccessful. The chip was also designed to have the appropriate dimensions for detection using a modern Raman microscope system, which collects scattering from a very small volume. A dye derived from TNT was used as a model analyte. Quantitative behavior was obtained over 4 orders of magnitude with a detection limit of 10 fmol. This performance is between 1 and 2 orders of magnitude better than that achieved using a macroflow SERRS cell. The technique has the added advantage that both reagent consumption and effluent production are greatly reduced, leading to reduced operating costs and a decreased environmental impact. Surface-enhanced resonance Raman scattering (SERRS)1-4 is an extremely powerful analytical tool, which not only yields information about the molecular structure of the analyte in the form of a vibrational spectrum but also gives sensitivities com* Corresponding author: (e-mail) [email protected]; (fax) +44(0) 141 552 0876. † University of Strathclyde. ‡ University of Glasgow. (1) Moskovits, M. J. Chem. Phys. 1978, 69, 4159-4161. (2) Fleischmann, M.; Hendra, P. J.; McQuillan, A. J. J. Chem. Phys. Lett. 1974, 26, 163-166. (3) Albrecht, M. G.; Creighton, J. A. J. Am. Chem. Soc. 1977, 99, 5215-5217. (4) Jeanmaire, D. L.; Van Duyne, R. P. J. Electroanal. Chem. 1977, 84, 1-20. 10.1021/ac015625+ CCC: $22.00 Published on Web 03/05/2002

© 2002 American Chemical Society

parable to that achieved using fluorescence spectroscopy.5-7 The use of resonance enhancement, where the frequency of the excitation source is tuned to be in resonance with an analyte chromophore, results in an increase in sensitivity and also enhances the selectivity of the system.8 This enables discrimination of the analyte of interest from any contaminants that may be present in the system. One of the main problems associated with using SERRS as a quantitative analytical technique is the difficulty associated with producing reproducible SERRS substrates. Variations in the morphology of SERRS substrates can give rise to large variations in the SERRS intensities produced.9 One common SERRS substrate is silver colloid,10-13 which can be easily prepared by reduction of aqueous solutions of silver salts.11,14,15 However, as with other substrates, there is still a problem associated with the production of stable, reproducible silver colloids.12,16 It has also been found that, in order to achieve the maximum enhancement from silver colloid, it must be aggregated17-20 (due to the greatly enhanced electromagnetic field between aggregated particles when compared with that at the surface of a single colloidal (5) Nie, S. M.; Emery, S. R. Science 1997, 275, 1102-1106. (6) Kneipp, K.; Wang, Y.; Dasari, R. R.; Feld, M. S. Appl. Spectrosc. 1995, 49, 780-784. (7) Rodger, C.; Smith, W. E.; Dent, G.; Edmondson, M. J. Chem. Soc., Dalton Trans. 1996, 5, 791-799. (8) Hildebrandt, P.; Stockburger, M. J. Phys. Chem. 1984, 88, 5935-5944. (9) Yang, X. M.; Ajito, K.; Tryk, D. A.; Hashimoto, K.; Fujishima, A. J. Phys. Chem. 1996, 100, 7293-7297. (10) Laserna, J. J.; Torres, E. L.; Winefordner, J. D. Anal. Chim. Acta 1987, 200, 469-480. (11) Cermakova, K.; Sestak, O.; Matejka, P.; Baumruk, V.; Vlckova, B. Collect. Czech. Chem. Commun. 1993, 58, 2682-2694. (12) Schneider, S.; Halbig, P.; Grau, H.; Nickel, U. Photochem. Photobiol. 1994, 60, 605-610. (13) Vlckova, B.; Solecka-Cermakova, K.; Matejka, P.; Baumruk, V. J. Mol. Struct. 1997, 408, 149-154. (14) Lee, P. C.; Meisel, D. J. Phys. Chem. 1982, 86, 3391-3395. (15) Munro, C. H.; Smith, W. E.; Garner, M.; Clarkson, J.; White, P. C. Langmuir 1995, 11, 3712-3720. (16) Shirtcliffe, N.; Nickel, U.; Schneider, S. J. Colloid Interface Sci. 1999, 211, 122-129. (17) Campbell, M.; Lecomte, S.; Smith, W. E. J. Raman Spectrosc. 1999, 30, 37-44. (18) Sanchez-cortes, S.; Garciaramos, J. V.; Morcillo, G.; Tinti, A. J. Colloid Interface Sci. 1995, 175, 358-368. (19) Blatchford, C. G.; Campbell, J. R.; Creighton, J. A. Surf. Sci. 1982, 120, 435-455. (20) Jones, J. C.; McLaughlin, C.; Littlejohn, D.; Sadler, D. A.; Graham, D.; Smith, W. E. Anal. Chem. 1999, 71, 596-601.

Analytical Chemistry, Vol. 74, No. 7, April 1, 2002 1503

particle21). Aggregation can be induced using a wide range of reagents, although this is a dynamic process, and as different SERRS intensities are accumulated from aggregates of different morphologies,13,22 it is desirable to be able to control the process in order to achieve a reproducible signal. One method of improving the reproducibility of signals obtained using silver colloid as the SERRS substrate is to use a flow cell for analysis.23-27 In this type of system, the colloid, analyte, and aggregating agent are pumped into a cell where they are allowed to mix. The signal is then accumulated from the flowing stream as it passes through a laser beam. As the sample solution is continually flowing during the time of signal accumulation, there is an averaging of the SERRS intensities, accumulated from different aggregates. This results in increased analytical precision and good quantitation over several orders of magnitude. Flow cells also give greater control over the rates of addition of reagents and the state of aggregation of the colloid at the point of signal accumulation. Similar systems have also been developed in which the colloid is prepared on-line in macroscale systems,10,23,28 thereby eliminating the need to prepare stable, reproducible colloid batches. The development of a microflow cell for SERRS analysis (“onchip”), where the silver colloid is prepared in situ by borohydride reduction of silver nitrate, is described in this paper. This technique has several advantages over similar macroflow cells in that there is reduced reagent consumption and waste production. The system is also ideal for analysis where only small quantities of analyte are available, i.e., clinical analysis.29,30 One important advantage of the chip-based SERRS system is that the design of the chip influences the Reynolds number, such that colloid produced can be constrained within a laminar region of flow. The result is to further reduce the “analytical volume” detected, while maintaining low back-pressure flow in the chip. The combination of the small analyte volume used in the microflow cell and the use of a Raman microscope system, which collects the signal from a very small volume, means that the Raman scattering can be accumulated from a larger proportion of the sample than in a macroflow system. It is also common when developing small portable Raman instruments to use smaller, lower powered lasers and increase the power density at the sample by focusing the laser into a small volume. This means that the development of portable Raman instruments will be compatible with the use of micro-SERRS lab-on-a-chip devices. (21) Xu, H. X.; Aizpurua, J.; Kall, M.; Apell, P. Phys. Rev. E 2000, 62, 43184324. (22) Stockman, M. I.; Shalaev, V. M.; Moskovits, M.; Botet, R.; George, T. F. Phys. Rev. B 1992, 46, 2821-2830. (23) Laserna, J. J., Berthod, A., Winefordner, J. D. Microchem. J. 1988, 38, 125136. (24) Cabalin, L. M.; Ruperez, A.; Laserna, J. J. Anal. Chim. Acta 1996, 318, 203-210. (25) Sheng, R. S.; Ni, F.; Cotton, T. M. Anal. Chem. 1991, 63, 437-442. (26) Freeman, R. D.; Hammaker, R. M.; Meloan, C. E.; Fateley, W. G. Appl. Spectrosc. 1988, 42, 456-460. (27) Taylor, G. T.; Sharma, S. K.; Mohanan, K. Appl. Spectrosc. 1990, 44, 635640. (28) Berthod, A.; Laserna, J. J.; Winefordner, J. D. Appl. Spectrosc. 1987, 41, 1137-1141. (29) Shi, Y. N.; Simpson, P. C.; Scherer, J. R.; Wexler, D.; Skibola, C.; Smith, M. T.; Mathies, R. A. Anal. Chem. 1999, 71, 5354-5361. (30) Jacobs, E.; Vadasdi, E.; Sarkosi, L.; Coman, N. Clin. Chem. 1993, 39, 10691074.

1504

Analytical Chemistry, Vol. 74, No. 7, April 1, 2002

The chip-based SERRS system has been demonstrated using an azo dye as a model analyte. This dye is a derivative of trinitrotoluene (TNT) and a potential analyte for the development of methods for explosives detection. An analytical chip was constructed in glass using standard microfabrication procedures for photolithography, pattern transfer, wet etching, and bonding. The flow characteristics of the colloid and the dye through the chip-based system are investigated using white light microscopy. The effects of different experimental protocols involving changing the order of addition of the reagents are also investigated, and finally, the detection limit for the TNT-derived dye is determined. EXPERIMENTAL SECTION Device Fabrication. The fluid channel was fabricated using photolithographic techniques adapted from the semiconductor industry. S1818 photoresist (Shipley Europe, Coventry, U.K.) was spin coated onto an acid-cleaned 1.5-mm-thick soda lime glass substrate (Soda Lime glass, Nanofilm) at 4000 rpm for 30 s. The glass was baked on a 90 °C hot plate for 3 min, followed by UV exposure for 8 s through an acetate sheet lithography mask in order to transfer the channel pattern into the photoresist. The photoresist was developed in a mixed solution of 1 part Microposit Developer Concentrate (Shipley Europe) and 1 part RO ultrapure 18 MΩ resistance water (Millipore RO 120 and Millipore Super Q systems, Bedford, MA) for 35 s and dried with nitrogen. The substrate was then baked on a 90 °C hot plate for 15 min to ensure all the solvent had evaporated and to harden the photoresist. Finally, the glass was wet etched in a mixed solution of 1 part hydrofluoric acid and 4 parts RO water for 15 min, resulting in 30-µm-deep and 250-µm-wide channels. After the etching procedure, the glass was ultrasonicated in acetone to remove remaining photoresist. A glass cover plate was fabricated, to seal the chip. Holes, for the inlet and outlet reservoirs, were drilled using a 1.5-mm diamond engraving tool (RS, Corby, Northants, U.K.). The etched substrate and cover plate were then prepared for bonding. Successful bonding relies on ultraclean surfaces. Therefore, the glass was cleaned in piranha solution (H2SO4:H2O2 ) 1:7) for at least 10 min, followed by ultrasonication in acetone for 5 min. The surfaces were then rinsed in RO water and thoroughly dried with nitrogen. The etched substrate (with the microfluidic network) was aligned with the input and output interconnects within the cover plate and placed in a steel clamp between two Macor plates (RS, Corby). The clamp was then placed in a furnace at 60 °C, which was increased at a constant ramp to 500 °C over 60 min and maintained at this temperature for 1 h. The temperature was then increased to 570 °C and held at this temperature for a further 5 h. Finally, the furnace was cooled to 60 °C and left for 12 h. The thermal bonding protocol produced a watertight and permanent bond between the etched substrate and the cover plate. A diagram of the flow system used is shown in Figure 1. Azo Dye. The chosen analyte was an azo dye, 5-(2′-methyl3′,5′-dinitrophenylazo)quinolin-8-ol, the structure of which is shown in Figure 2. This was synthesized as part of a program to detect TNT by derivatization methods.31 SERRS. SERRS spectra were accumulated using a Renishaw Mark I system 2000 (Gloucs, England) with a Spectra Physics (31) McHugh, C.; Keir, R. L.; Graham, D.; Smith, W. E. Submitted.

Figure 1. Diagram showing the design of the flow system. (1-4) marks the reagent inlets, and (5) the outlet. (A-I) are the points of signal accumulation.

Figure 3. Image of the point of colloid formation within the flow system channels collected using a white light microscope in reflection mode. Figure 2. Structure of 5-(2′-methyl-3′,5′-dinitrophenylazo)quinolin8-ol.

361C 15-mW argon ion laser working at 514 nm, as the excitation source. Sodium borohydride (99%), silver nitrate (99.9999%), and sodium hydroxide (97%) were purchased from Aldrich (Dorset, England). An aqueous solution of silver nitrate (10 µL, 2.6 × 10-3 M) and a solution of sodium borohydride (10 µL, 1.1 × 10-3 M) in sodium hydroxide solution (0.1 M) were introduced into inlets 1 and 2 of the chip (see Figure 1), respectively, using a micropipet (thereby priming the chip). No additional aggregating agent was added as this concentration of sodium hydroxide is sufficient to aggregate the colloid. The azo dye solution (10 µL), which was prepared in methanol, was introduced into inlet 3 and distilled water (10 µL) was introduced through inlet 4. All solutions were drawn through the system simultaneously using a syringe drive. Spectra were accumulated by focusing the laser on the colloid stream using a 10× objective lens with an accumulation time of 10 s. In the experiments involving premixing of the dye solution with the borohydride solution, a mixture of 1:1 sodium borohydride and dye solution (1 × 10-6 M, 10 µL) was introduced into inlet 2, silver nitrate was introduced into inlet 1, and distilled water was introduced into inlets 3 and 4. The solutions were then drawn through the system simultaneously, using the syringe. Spectra were accumulated as described previously. Silicon standards were run prior to each replicate. The calculated intensities of the SERRS spectra were based on the height of the peak at ∼1375 cm-1, which was normalized against the intensity of the appropriate silicon standard. Images. Images of the flow system were obtained using a white light microscope (Olympus) with a color CCD camera (Polnex). Images were collected using a 5× microscope objective (NA ) 0.10).

RESULTS AND DISCUSSION Flow Characteristics. Introduction of silver nitrate and sodium borohydride solutions into inlets 1 and 2 of the microflow cell (Figure 1) resulted in the formation of a thin stream of colloid within the chip. The colloid was imaged using a white light microscope in either transmission or reflection mode. An image of the point of colloid formation within the channel is shown in Figure 3. The diameter of the colloid line is ∼30 µm, which is compatible with the use of a Raman microscope system, which can focus down to a 1-2-µm spot when a 50× objective lens is used. This provides a SERRS detection system where a large proportion of the stream is analyzed by the instrument as it flows past the interrogation point. To investigate whether mixing within the channels was possible, pillars were introduced into the flow system at two points; see Figure 1. By imaging the colloid flow within the chip, it could be seen that, as expected, these features did not disrupt the colloid flow (Figure 4). To show whether there was diffusion of colloid within the channels when the flow was stopped, silver nitrate, sodium borohydride, and distilled water were introduced into inlets 1, 2, and 3, respectively, and these solutions were drawn through the flow system once. Flow was then stopped. Images of the point of colloid formation were taken immediately after colloid formation and subsequently at various time intervals over a 70min period. The images taken immediately after colloid formation and after 70 min are shown in Figure 5. Over the period of investigation, there is a slight darkening of the colloid stream, suggesting that there is diffusion of the silver nitrate and sodium borohydride toward the center of the channel where there is immediate formation of more colloid. However, the colloid stream itself does not disperse and no colloid is formed beyond the point of confluence of the two solutions. Imaging over the same 70-min period showed there was no change in the appearance of the colloid stream at the point in the flow system where there were pillared structures. The absence Analytical Chemistry, Vol. 74, No. 7, April 1, 2002

1505

Figure 6. Images taken at the point of colloid formation (a) after colloid formation, (b) after distilled water has been passed through the system, and (c) after cleaning with nitric acid and distilled water. Images taken in transmission mode.

Figure 4. Image of the colloid stream as it passes, in laminar flow, through the part of the flow system containing pillar structures, collected using a white light microscope in reflection mode.

Figure 7. Normalized SERRS intensities of 5-(2′-methyl-3′,5′dinitrophenylazo)quinolin-8-ol accumulated from points D to I in the flow system after introduction of the dye solution through inlet 3.

Figure 5. Images taken at the point of colloid formation (a) immediately after colloid formation and (b) 70 min after the flow had stopped. Both images were taken using a white light microscope in transmission mode. These images also show the effect of the presence of an impurity within the channels.

of any further darkening with time suggested that all of the colloid had been formed by the time the stream had reached this part of the system. Again, there was no further diffusion of the colloid particles once the flow had stopped. The lack of diffusion of the colloidal particles may be due to the particles interacting with the glass walls of the chip. To determine the extent of this interaction, colloid was formed in the system in the usual way and an image was taken after colloid formation and again after distilled water had been drawn through the system (Figure 6a and b). Although some of the colloid has been washed through the system by the distilled water, a significant amount remains, suggesting an interaction with the glass surface, which would partly account for the lack of diffusion of the colloidal particles. Further washing with distilled water did not remove any more colloid from the glass surface. The remaining silver was easily removed from the system by washing with nitric acid followed by distilled water (Figure 6c). Although it is possible to collect SERRS spectra of the dye from the silver colloid that remains in the system after washing with water, there is no evidence of memory effects after 1506 Analytical Chemistry, Vol. 74, No. 7, April 1, 2002

thorough cleaning of the system with nitric acid, distilled water, and methanol and reintroduction of silver colloid into the system (indicating that the chip can be reused, if so required). SERRS. SERRS spectra of a dye 5-(2′-methyl-3′,5′-dinitrophenylazo)quinolin-8-ol, were accumulated using this system by introducing silver nitrate, sodium borohydride, and dye solution into inlets 1-3, respectively. To determine the sampling point from which the maximum signal intensity could be achieved, spectra were accumulated from various points (marked D-I, Figure 1) in the chip based system. At points A-C, only the sodium borohydride and silver nitrate solutions were present in the system, and therefore, no spectra were accumulated from these points. To eliminate any possible effects of the time at which the spectra were collected relative to each other, four replicates were carried out and the order in which the spectra were taken was randomized. To conduct this experiment, the flow was stopped for a short period, during which time the intensities of the spectra varied, thereby significantly increasing the error. The intensities of the spectra accumulated from points D to I are shown in Figure 7. The most intense spectra were collected from points D to F where there is more colloid formation compared with points G-I. This may be due to rapid colloid formation within the first half of system and a lack of flow of the colloidal particles throughout the remainder of the system. The higher concentration of silver in these parts of the system gives rise to an increase in the amount of silver surface available for dye attachment, therefore leading to an increase in signal intensity in these areas. It was expected that the intensity at points H and I would be lower than that at the other points in the system as these points are after inlet 4,

Figure 8. SERRS spectra of 5-(2′-methyl-3′,5′-dinitrophenylazo)quinolin-8-ol accumulated using the flow system. Concentrations of dye within the flow system are (a) 10 pmol, (b) 1 pmol, (c) 0.1 pmol, and (d) 10 fmol.

which introduces distilled water into the system. This dilutes both the colloid and the dye solution resulting in lower signal intensities. As it was found that there was an increase in the signal intensity at the start of the flow system (points D-F), it was thought that premixing of the dye with either the sodium borohydride or silver nitrate solutions may lead to an increase in the signal intensity, as the dye would be present at the point in the system where the colloid is formed. In all cases, when the dye was premixed with silver nitrate, this resulted in no colloid formation, due to the dye complexing with the silver prior to mixing with sodium borohydride solution, thereby preventing colloid formation. Spectra were also collected from points A to I in the system by introducing silver nitrate into inlet 1 and a mixture of sodium borohydride and dye solution into inlet 2. Again the accumulation of spectra from the different points was randomized in order to eliminate any potential effects of the time of signal accumulation. As expected, the most intense signals were accumulated from point A. All other points showed intensities similar to that obtained from point H in the previous experiment (Figure 7). There was, however, no significant difference between the signal intensities achieved from point A after premixing of the dye with sodium borohydride solution and those from points D to F after introducing the dye through inlet 3. The absence of any diffusion of the colloid line once the flow has stopped makes this chip-based system suitable for analysis using a stopped-flow system. This would enable accumulation of the SERRS signal over an extended period of time, thereby decreasing the detection limits. Initial experiments carried out using stopped flow showed that there was considerable variation in signal intensity with time. This is likely to be due to variation in the state of aggregation of the colloid therefore causing a shift in the surface plasmon resonance and hence variation in the signal intensity. Despite this, it was still possible to collect a clear dye signal after 1 h. To determine the improvement in sensitivity of detection for the chip-based system, when compared with a macroflow cell, dye solutions of concentrations ranging from 10-6 to 10-9 M were

Figure 9. Log SERRS intensity of 5-(2′-methyl-3′,5′-dinitrophenylazo)quinolin-8-ol measured using the peak at ∼1375 cm-1 against Log dye concentration.

introduced into the chip through inlet 3. The colloid was formed in the system as described above and 10 µL of each of the dye solutions was introduced into the system, resulting in amounts of dye ranging from 10 pmol to 10 fmol. The spectra were accumulated by focusing the laser on the colloid stream at the point where the dye solution met the colloid (point D, Figure 1). The spectra accumulated are shown in Figure 8 and are characteristic of this dye.31 There is no evidence of contamination, but there is some variability in the spectra at the higher concentrations. Spectrum a was taken at a dye concentration that is close to monolayer coverage, and hence, the changes may be due to packing effects. Here it can be seen that it is possible to detect 10 fmol of the dye using this system. This represents a 20-fold increase in sensitivity over that achieved using a macroflow cell.32 A plot of SERRS intensity against dye concentration is shown in Figure 9. Thus, the results indicate semiquantitative behavior and a low detection limit. The error on the experiment is dependent on the control of the flow rate and the length of time for which the flow is stopped. Therefore, improvements in the experimental variability should be achieved by increasing the control over the flow rates and further optimizing the analytical protocol. Ulti(32) Keir, R. L. Unpublished work, 2001.

Analytical Chemistry, Vol. 74, No. 7, April 1, 2002

1507

mately, since increasing the length of time for which the flow is stopped increases the error, a compromise must be found between the detection limit required and the accuracy. Concentrationdependent results indicate saturation at between 10-7 and 10-6 M. This is expected due to complete surface coverage as only dye directly adsorbed on the first layer is expected to be effective.33 CONCLUSIONS On-chip SERRS has been successfully implemented in a microfluidic system. It has been shown that, using this system, it is possible to detect 10 fmol of an azo dye derived from TNT, which represents an improvement of between 1 and 2 orders of magnitude over similar macrofluidic systems. In situ, on-chip, colloid preparation results in a thin line of colloid which remains confined to the center of the channel over extended periods of time. The dimensions of the colloid line (∼30-µm width) make this system compatible with signal accumulation using a Raman microscope, which collects signal from a small volume. The longterm stability of the colloid stream means that, using this system in a stopped-flow format, it is possible to accumulate signal over extended periods of time, thereby further increasing the sensitivity (33) McLaughlin, C.; Smith, W. E. Submitted.

1508

Analytical Chemistry, Vol. 74, No. 7, April 1, 2002

of the system. In addition, this system has the advantages inherent to all on-chip devices of reduced reagent consumption and waste production resulting in lower running costs and decreased environmental impact. Now that SERRS has been shown to be advantageous for onchip analysis, it is expected that it will prove valuable in a wide range of fields, including clinical analysis. The advantages would include in situ, fast identification, low detection limits, and small volumes of reagents. Preliminary results carried out on analytes such as mitoxantrone and dye-labeled DNA are promising and will be further developed soon. ACKNOWLEDGMENT The authors thank EPSRC, EPSRC/DTI Lab on a chip forsight program and the Police Scientific Branch of the Home Office for help in enabling this project.

Received for review September 14, 2001. Accepted January 15, 2002. AC015625+