Growth of optical-quality anthracene crystals doped with dibenzoterrylene for controlled single photon production

Dibenzoterrylene (DBT) molecules within a crystalline anthracene matrix show promise as quantum emitters for controlled, single photon production. We present the design and construction of a chamber in which we reproducibly grow doped anthracene crystals of optical quality that are several mm across and a few μm thick. We demonstrate control of the DBT concentration over the range 6-300 parts per trillion and show that these DBT molecules are stable single-photon emitters. We interpret our data with a simple model that provides some information on the vapour pressure of DBT.


I. INTRODUCTION
Individual photons provide an excellent way to encode quantum information, 1 and interference between photons offers a practical way to process that information. 2For this purpose, one would like a source of photons that are available on demand at short notice and are indistinguishable so that they will interfere with each other.A number of single-photon sources already exist, but none of them satisfies all of these requirements. 3The workhorse method is still spontaneous parametric downconversion, 4 which produces pairs of photons, the detection of one being used to herald the presence of the other. 5Because the pairs are produced at random times, this method is limited to very low repetition rates to ensure one photon at a time.By contrast, quantum dots 6 excited by short laser pulses fluoresce promptly and can produce photons rapidly, as desired.However, the dots switch off from time to time when the light randomly drives them into a long-lived dark state and, further, the emission linewidth is greatly broadened by phase noise.Defects in diamond also show promise as photon sources. 7The nitrogen-vacancy defect benefits from having the natural, lifetime-limited linewidth, but has a low yield (5%) at the main emission frequency, the "zero-phononline," due to red-shifted vibronic sidebands.The siliconvacancy defect emits a high fraction of photons (88%) on the zero-phonon line, 8 but the radiative decay probability is low.
An excellent alternative is provided by the dibenzoterrylene (DBT) molecule in a host matrix of crystalline anthracene. 9When cooled to low temperatures, these molecules emit with high (34%) efficiency on the zero-phonon line, and with lifetime-limited linewidth. 10,11The molecules are exceptionally photostable, even at room temperature, and their optical dipole moments align naturally with one of the axes of the anthracene crystal. 12Toninelli et al. have shown coupling of DBT to an optical cavity 13 and Hwang and Hinds have a) kyle.major11@imperial.ac.uk b) ed.hinds@imperial.ac.uk discussed the exchange of radiation between the molecular dipole and other optical structures. 14They have shown that the radiative coupling to an integrated optical waveguide can be very efficient.These properties have motivated us to develop a method of producing anthracene crystals of high optical quality, suitable for insertion into photonic devices and to explore how the crystals may be doped in a controlled way to achieve the desired density of DBT molecules.
There is already a body of research on growing crystals of conjugated hydrocarbons, such as anthracene, 15 motivated mainly by the field of organic semiconductors.Single crystals have been made with high purity, and with a low density of defects, using the physical vapour growth method. 16A common design is a closed chamber with the molecular powder heated at one end and a cooled collection plate at the other. 17rystals can be grown quickly when the temperature difference between the ends is large.Without active control of the wall temperature, one finds in practice that the growth is not reproducible.To address this, Karl introduced a cylindrical chamber with closely spaced ends for sublimation and collection. 18With the uncontrolled side walls far away and small, this method reproducibly grows many crystals of the same shape and size, but the convective coupling between the closely spaced top and bottom only permits a small temperature difference, and the growth runs typically last for days.Laudise et al. popularised the physical vapour transfer method, where the chamber is a horizontal tube, with a temperature gradient along its length to provide a continuous distribution of the growth conditions. 19The molecular powder is placed at the hot end, and the crystals are collected throughout the tube.The growth can be fast and does not require stringent temperature control of many surfaces.If there are impurities, it is a benefit of this method that they normally grow at a different position from the crystals of interest.Initial investigations of DBT in Ac 9,10 used a co-sublimation method after Laudise with powder heated at one end and collected at the other.The DBT was introduced with another heated crucible, but with little quantitative control of the DBT concentration.
For our application, it is important to control the concentration of DBT molecules within the crystal, and this is not easily achieved with the methods outlined above because the concentration is mainly sensitive to the temperatures of individual crucibles.In this paper, we describe a vapour deposition chamber that produces DBT-doped anthracene crystals of optical quality, with good control over the DBT concentration.Active cooling ensures that the collection plate is the coolest surface, and stabilises the temperature to ensure that the crystals grow with the desired morphology.Crucibles for the anthracene and DBT are separately heated to achieve the desired DBT concentration and fast growth rate.

II. APPARATUS
Figure 1 shows the apparatus.The main body of the chamber is a 304 stainless steel cylinder, 30 cm tall and 30 cm in diameter, with ISO DN250 flanges welded at each end.The end plates, sealed to this by high temperature Polymax Viton O-rings encapsulated in FEP (fluorinated ethylene propylene), are easily removed to gain access to the interior of the chamber.A vacuum port (E) allows the chamber to be pumped out and admits dry nitrogen gas used during crystal growth.This port also brings electrical wiring into the chamber.Identical crucibles (A and B) and a collection plate (C) are mounted in the end plates.The crucibles anthracene (Ac) and DBT powders, separately heated to control the ratio of Ac:DBT:N 2 .The crystals grow on the collection plate, whose temperature gives us control over the growth conditions.We view the crystals through a 2 3 /4-in.Pyrex Conflat viewport (D), sealed by a standard copper gasket.The chamber is heated by tapes (OMEGA Engineering, FGS0031-010) wrapped around the outside of the body to prevent the deposition of Ac and DBT on the walls, thereby improving the reproducibility of the growth runs.Outer layers (not shown) of fibreglass wool, and then aluminium foil, provide thermal insulation.The vacuum port connects the chamber to a pressure gauge (OMEGA Engineering, DPG1000B-30V100G), a 15 pin D-sub KF50 electrical feedthrough (Kurt J. Lesker, IFDGG151058C), and a valve (Swagelok, SS-8BK).When the valve is open, the chamber is also connected to a 20 l/s rough vacuum pump and a nitrogen cylinder.As the pressure gauge, electrical feedthrough, and valve are well separated from the chamber body, they do not need thermal O-rings.All O-rings are greased with high vacuum silicone grease (Dow Corning, 976).
The collection plate is temperature-controlled in two stages, as detailed in Fig. 2. The first, coarse, stage is cooled by a nut (F) wrapped in copper piping and plumbed into a cold water supply to ensure adequate sinking of the heat load when the temperature needs to drop.This is particularly important for the collection plate, because it has to be the coldest point.The nut screws onto a cylindrical holder (H) that contains a cartridge heater (Omegalux CSH-202250).PEEK (Polyether ether ketone) spacers (G), sealed with high temperature encapsulated Viton O-rings, provide thermal isolation from the body of the chamber.We choose PEEK thermoplastic because, unlike Teflon, it does not creep when heated to 200 • C. Active control of the cartridge heater current stabilises the reading on a Pt100 resistance thermometer, inserted into a small hole in the nut.
The second stage of temperature control is a thermoelectric component (European Thermodynamics, GM250-49-45-30) (J), responsible for quick and accurate temperature regulation.This particular model was chosen because of its ability to handle a large heat flow.The TEC is sandwiched between the base of the cartridge holder and the collection plate (K) and thermally linked to both by high temperature paste (Jelt, 6018).The collection plate is a 0.5 mm-thick, 80 mmdiameter copper disk, held by another PEEK retaining ring.A second Pt100 resistance thermometer is sandwiched between the Peltier heater and the collection plate.With each of the two stages under active proportional integral differential (PID) control, the collection plate thermometer changes by less than 0.1 • C when the crucibles running at full power are switched off.The change in the other, coarse stage, thermometer is typically 3 • C.
The same scheme is used to control the crucible temperatures, except that the copper disk is replaced by an aluminium crucible.

III. CRYSTAL GROWTH
To grow crystals, we put 1 g of Ac powder into one of the aluminium crucibles, and a 2.3 mg lump of DBT powder into the other.The chamber is then purged three times with dry nitrogen before pumping down to a gauge pressure of 255 mbar.The actual nitrogen pressure is uncertain by about ±20 mbar because of atmospheric pressure variations, which we do not monitor.The heaters for the chamber body and the collection plate are turned on and allowed to stabilise for an hour, bringing the walls to 180 • C and the plate to 102 • C. The temperature of the Ac crucible is then raised to 162 • C, while the DBT is heated to a temperature in the range 120-200 • C, depending on the required doping level.At this point, the pressure is approximately 350 mbar.The crystals grow over the next hour.During cool-down, the collection plate is held at 102 • C until all the other elements in the chamber have cooled to the same temperature (∼20 min); then, that heater is also switched off and the whole chamber is allowed to cool completely before opening.
This recipe produces the thin, transparent platelets we desire, such as the one shown in Figure 3 from the collection plate and are barely visible.There are also some thicker, more obvious, striated crystals, which we do not use.The good crystals are harvested with tweezers and placed on glass microscope slips, which have been cleaned by oxygen plasma etching to ensure good adhesion and low background fluorescence.At the specified temperatures, crystals of the desired morphology grow over a wide range of nitrogen pressure.With the starting pressure below 150 mbar, the growth becomes sparse and rocky, while above 350 mbar it becomes very dense and powdery.Our preferred starting pressure is in the middle of this range.The temperature of the collection plate is not particularly critical.At higher temperatures, the crystals do not grow to a large enough size within the growth time to be harvested.Below 80 • C, the crystals that grow begin to become dendritic in habit.
In Fig. 3, the dark regions of the crystal are those that have adhered well to the slip.The large surface is the 001 plane and the typical size of the crystals we use is 2-3 mm.The edges of the crystal clearly reflect the fundamental structure, illustrated in the inset, 20 and the angles immediately show us which direction is the b-axis, along which the DBT molecules are aligned. 12The crystal is too thin to discern which planes are exposed on the sides.An atomic force microscope (Nanoscope Dimension 3100) imaged the surface profiles.Any one crystal is uniformly thick, but those harvested from a given batch vary in thickness, typically between 0.7 µm and 3 µm.The microscope records a surface roughness that also varies from one crystal to another in the range 1-3 nm.

IV. COUNTING DBT MOLECULES
The DBT molecules within a crystal are detected using 779 nm light to excite them from the electronic ground state S 0 to the first excited state S 1 .The molecules relax rapidly to the bottom of the S 1 band then fluoresce, with essentially 100% quantum yield, at a redshifted wavelength of ≥785 nm.It is the fluorescence that we detect.The excitation light is produced by an external cavity diode laser (TEC 100 Littrow -Lynx, Sacher Lasertechnik).This light is filtered spatially through a single-mode optical fibre (Thorlabs PI-780A-FC-1), then spectrally by a band-pass filter (Thorlabs FL 780-10), tilted to cut off any light at wavelengths ≥785 nm, either from the laser or from fluorescence in the fibre.The light is steered by galvo plates into a confocal microscope that images the beam onto the crystal through a 60 × oil objective (Nikon Apo Plan, 60 × N.A. 1.4).The minimum of the first Airy ring is at 530 nm radius, and the spot can be scanned over a square region of 15 µm × 15 µm.Light returning through the microscope is filtered by two successive long-pass filters (Chroma LP800HQ) that block the scattered excitation light, but transmit the DBT fluorescence.This is coupled into a multimode optical fibre (Thorlabs, GIF 625) and sent to a single-photon counting module (Perkin-Elmer SPCM-AQRH-15-FC) for detection.A raster scan of the light spot produces images such as the one in Fig. 4, where we see 237 bright fluorescence spots.On cooling to 4 K, the number of molecules fluorescing is expected to decrease by a factor of ≈40, giving 6 molecules in the field of view.For the photonic systems we have in mind this is an ideal density.In order to check that these are single DBT molecules, we split the output of the multimode fibre between two detectors and record the distribution of time delays τ between a photon arriving at one detector and the next photon arrival at the other.Figure 5 shows a plot of this probability distribution, known as g (2) (τ), for the light from a typical bright fluorescence spot.We see the deep dip that is characteristic of a single quantum source. 21This occurs at zero delay time, because a single molecule, having emitted one photon, cannot emit a second one until it has had time to be re-excited and decay again.Making this measurement on many of the bright spots, we find that they all exhibit the characteristic dip, with a visibility in the range 75%-91% providing unambiguous proof that these are single quantum emitters.The g (2) (τ) curves have a width of 3.1-3.5 ns, which is characteristic of the DBT spontaneous emission lifetime.We take these curves as proof that we are looking at individual DBT molecules.
The 237 DBT molecules shown in Fig. 4 are distributed throughout the thickness of the crystal (0.77 µm).Since FIG. 5.An example of a second order correlation curve g 2 (τ) of the emission from a single molecule.The red line shows a fit with a lifetime of 3.11 ns.In a simple model, the concentration is just the ratio of the arrival rates at the collection plate for the two species: R DBT /R Ac .Since the walls of the chamber are heated, we can reasonably equate these rates to the rates of production at the two crucibles.These are both proportional to pA/(T m) 1/2 , where p is the vapour pressure above the source material (Ac or DBT) of area A and temperature T and m is the mass of the molecule.The masses and temperatures are known, as is the vapour pressure of Ac. [23][24][25] We have estimated the ratio of DBT:Ac surface area as 6% by photographing the crucibles and comparing the areas covered by the chemicals.This is likely to be quite unreliable as it takes no account of the microscopic structure of the surfaces, but whatever the error, it is a constant factor as the same sources were used for all the data shown in Fig. 6.That leaves the vapour pressure of DBT, which does not seem to have been measured.Let us suppose that it has the basic Antoine form Log 10 p DBT = A − B /T, with A and B being constants, and with the DBT vapour pressure being specified in Torr and the temperature in Kelvin.A fit of this simple model to the data in Fig. 6 yields the coefficients A = 1 ± 1 and B = 4100 ± 700.In principle, there should be small corrections to allow for deposition of DBT on the AC crucible at high DBT temperatures and for deposition of Ac on the DBT crucible at low DBT temperatures.Further, there could be some reduction of concentration as a result of DBT molecules failing to find a suitable lattice site in the Ac crystal, and this might well depend on the growth rate.Despite these uncertainties, it seems likely that this measurement provides This article is copyrighted as indicated in the article.Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 155.198.12.147On: Fri, 28 Aug 2015 15:34:04 a reasonable estimate of the coefficient B, which (by contrast with A) does not depend on the ratio of areas or on the probability of the DBT finding a lattice site.

V. CONCLUSION
We have discussed the design, construction, and performance of a crystal growth oven, which makes large, thin anthracene crystals of optical quality, doped with a controlled number of photo-stable DBT molecules.Two-stage heating and cooling elements on separate Ac and DBT crucibles, and on the collection plate, make it possible to grow crystals with large, stable temperature differences and hence with short growth times of only a few hours.The independent control of the DBT crucible temperature permits well controlled production of doped anthracene crystals, with concentrations in the range 6 to 300 parts per trillion.Using a simple model for the variation of concentration with crucible temperatures, we have obtained first experimental information on the vapour pressure of DBT.

FIG. 1 .
FIG. 1.The design of the crystal growth chamber in section showing cartridge heaters with stabilising TEC heater/coolers.A: Ac crucible.B: DBT crucible.C: Collection plate.D: Viewport.E: Vacuum port.See text for detail.

FIG. 2 .
FIG. 2. Exploded view of one of the heating elements shown with collector plate.F: Water-cooled nut.G: PEEK spacers.H: Cartridge heater holder.I: High temperature O-rings.J: Peltier heater.K: Collection plate.See text for detail.This article is copyrighted as indicated in the article.Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 155.198.12.147On: Fri, 28 Aug 2015 15:34:04 FIG. 3. Photograph of a doped anthracene crystal placed on a clean glass cover slip.The image is dark in regions where the crystal has adhered well to the slip.Inset: diagram of the Ac crystal structure.Angle A is 109.757• and B is 125.122• .The b axis of the crystal is indicated by an arrow.

FIG. 6 .
FIG. 6. Ratio of DBT to Ac molecules in the cosublimated crystals as a function of DBT crucible temperature (Ac crucible at 120 • C).Points: measured concentration, with 1 σ error bars.We have good control of DBT concentration over two orders of magnitude.Line: a simple model in which the vapour pressure of DBT is fitted to the form of an Antoine equation (see text).