Chapter 2

Growth and Characterisation of the Diamond Films


2.0       Outline


·            This chapter describes the diamond growth method used in this study and explains why it was chosen.


·            The main components of the diamond growth apparatus are described.


·            The benefits of four different types of electrical contact are compared.


·            Scanning electron microscopy, optical microscopy and Raman spectroscopy were used to characterise the films and the results are presented here.



2.1       The Choice of Growth Technique


Hot filament chemical vapour deposition (HFCVD) was used to grow diamond films for these studies. HFCVD diamond was first produced in 1952. 61,62 The technique is now established and has been well described in a number of review articles. 11,17,18 Extensive work has previously been performed on similar HFCVD set-ups at the University of Bristol. 63-65


The chemical vapour deposition (CVD) process had a combination of properties that made it the method of choice for fabricating the diamond films in this study:


·          The use of gaseous reagents was easily controlled and allowed the composition of the reaction mixtures to be varied.


·          The technique produces continuous films which, if grown on planar substrates, were macroscopically flat. These films were ideal for use as planar electrodes. The technique could also be extended to coat metal wires with diamond, 66 which could be used in a range of alternative electrode geometries. 67


·          Diamond films were produced which covered a surface area of several square centimetres. While certain other techniques could coat significantly larger areas, much work on the electrochemistry of diamond had previously studied smaller single crystal diamonds and homoepitaxial layers grown onto single crystals. The larger surface area available with CVD diamond allowed for a range of electrodes to be fabricated that were ideal for laboratory experiments. 


HFCVD was used in preference to more sophisticated forms of CVD for a variety of reasons:


·          The apparatus was sufficiently simple to allow thorough cleaning of the inside of the reaction chamber. This was necessary to remove boron contamination.


·          The relative simplicity of the design allowed for modifications to the chamber to be introduced as required.


·          The components necessary to assemble a HFCVD chamber are relatively inexpensive.


·          The HFCVD technique has been studied extensively and there is a reasonable understanding of the processes involved. The growth apparatus used in this study was a modified version of a well characterised HFCVD chamber and this significantly reduced the work necessary to characterise and optimise the growth conditions.


The use of HFCVD applied a number of restrictions on work that could be performed:


·          HFCVD does not give the high rates of growth available from certain other techniques, such as plasma torches, 19,68 and so thick films could not be grown. If sufficiently thick diamond films had been grown then the silicon substrate could have been etched away in a bath of hydrofluoric acid (HF) and nitric acid (HNO3). 69 This would have left a free-standing diamond film which can be used in a range of applications.


·          The use of hot-filament CVD prevented the use of any oxygen containing gas species. The presence of species such as oxygen (O2), carbon monoxide (CO), carbon dioxide (CO2), trimethylborate (TMB, B(OCH3)3) or water (H2O) would have caused the rapid burn out of the tantalum (Ta) filaments.


·          Metallic contamination of the diamond films from the filament was a potential problem. The possibility of surface or bulk tantalum (Ta) affecting the electrochemical or electrical properties had to be considered. 70,71 Any possible effect of tantalum could be reduced if the diamond films were treated in acids before use. Refluxing in hot nitric acid could be used to remove any graphitic (sp2) content in the surface layer of the films and a short (ten second) dip in warm chromic acid (H2Cr2O7) could be used to oxidise the surface of the films.

Figure 2.1 - A schematic diagram of the gas lines which fed the diamond CVD chamber

2.2       The Reaction Gases


Figure 2.1 shows a schematic diagram of the experimental set-up of the gas lines which fed the reaction chamber.


Hydrogen (H2), methane (CH4) and diborane (B2H6) gases (BOC Speciality Gases) were used in the growth process. See section 2.10 for details of the reaction conditions.


Laboratory standard “high purity hydrogen” and methane were used. The purities of the hydrogen and methane were 99.995% (“N4.5”) and 99.5% (“N2.5”) respectively.


Diborane was chosen as the boron source because it held advantages over most other sources. Table 2.1 summarises these advantages and disadvantages of commonly used sources.



Boron Source









·    simple molecular chemistry
(no C or O atoms)


·    gas source provides excellent controllability and variability of concentrations


·    availability



·          highly toxic


·          highly explosive


·          highly reactive

72 - 75

trimethyborate (TMB)



·    non-toxic




·          liquid source requires dilution in acetone (CH3)2CO or methanol CH3OH


78 - 84



(rod or powder)




·    solid added directly to chamber - easy to store


·          lack of controllability and variability of concentration


67, 85, 86

boron trioxide



·    readily available




·          solid source requires dissolution in acetone (CH3)2CO or methanol CH3OH


87 - 91

ex-situ ion implantation



·    can dope any diamond sample


·    good controllability


·    achieve low doping levels



·          non-uniform doping profile


·          damage to diamond sample due to high energy bombardment



Table 2.1

A summary of the advantages and disadvantages of various boron sources


The diborane was a special order gas which was supplied as a 4.75% premix of diborane in hydrogen. Diborane is a highly explosive, toxic and reactive gas. Great care was taken to ensure that all gas lines were leak tested before use. The gas cylinder, the dilution cylinder and the associated piping were housed in a specially designed fume cupboard to provide some secondary protection in the event of an accident. The reactivity of diborane made it necessary to avoid the use of many materials including brass and nitrile rubber.


The source gases were stored in gas cylinders with nominal pressures up to 200 bar at 15 °C. Regulators reduced the cylinder pressures to the level of one to two bar above atmospheric pressure. Mass flow controllers then set the flow rates into a gas manifold where the gases mixed before they flowed along a common gas line into the reaction chamber.


The regulator on the diborane gas cylinder was a stainless steel single stage design which could be pumped down to vacuum and contained a purge line. This also allowed the dilution procedure to be performed (see section 2.4). The hydrogen and methane gas cylinders used standard single stage brass regulators.


Low levels of impurities in the hydrogen gas could be significant because of the relative flow rates of the source gases. A typical gas mixture may use as many as seven orders of magnitude more hydrogen than diborane. Therefore an impurity species in the hydrogen with a concentration of only one hundred thousandth of one percent (0.00001%) would have a concentration in the mixture of gases comparable to that of the diborane present. Fortunately, nitrogen (N2), the main impurity present, did not effect the growth process or the electrical characteristics of the diamond films. Nitrogen is not incorporated into diamond during growth as rapidly as boron 94 and it tends to occupy electrically inactive sites. 95-97 The effects of nitrogen on the growth rate and surface morphology of the diamond also require higher concentrations than present in the reaction mixture. 98-99

2.3       The Mass Flow Controllers


A bank of three mass flow controllers (MFCs) was used to set the flow rates of the gases. Each MFC controlled the flow of gas through a gas line and into a manifold where the gases were mixed. The MFCs were manufactured by Tylan General and their specifications are detailed in Table 2.2.


The MFCs were controlled electronically and could be adjusted to give a wide range of flow rates. They performed less well if the rate of flow was reduced below five percent of the maximum flow rate and so alternative methods was needed to obtain low flow rates. For this reason, the diborane MFC was replaced with a lower capacity model when boron doping levels below 3000 p.p.m. were to be achieved.


The MFCs could be set to 115% of their nominal maximum flow rate to enable rapid pumping down of the system. However, this was not sufficient to allow the diborane line to be pumped down in a reasonable time and so a bypass valve was fitted beside the MFC. The diborane MFCs used viton seals and were compatible with the gas.


MFC number



Range (s.c.c.m.)




Hydrogen (H2)


0 - 200




Methane (CH4)


0.0 - 10.0




Diborane (B2H6)
in a carrier gas of hydrogen (H2)

the main component of the mixture

0.0 - 10.0
0.00 - 1.00





not used





Table 2.2 - Specifications of the Mass Flow Controllers


A fourth MFC was available to allow more gases to be added to the reaction mixture. This facility was not used in this study. However, it has been proposed that adding hydrogen sulphide (H2S) or ammonia (NH4) would enable co-doping of the diamond film with sulphur (S) or nitrogen (N) respectively.


The mixing manifold had a second output to allow the MFCs to be used to supply other experiments with hydrogen or methane when the CVD chamber was not running. In particular, the hydrogen line was used to supply a prototype liquid CVD reactor.


2.4       Dilution of the Diborane Gas


The diborane gas was supplied pre-mixed with hydrogen. The nominal concentration was five percent diborane in hydrogen. This proved far too concentrated for the deposition of low doped diamond films and so a dilution system was constructed which allowed the premix to be further diluted with hydrogen. This process could be performed a number of times to allow the levels of diborane in the reaction mixture to be reduced by several orders of magnitude. A lecture bottle with a capacity of 0.4 litres was used as a mixing vessel. Once the diborane was diluted, the lecture bottle acted as a reservoir which could supply gas for approximately twenty hours at a typical flow rate equivalent to 4 × 10-5 s.c.c.m. of pure diborane. A photograph of the dilution assembly is shown in figure 2.2.



                         Figure 2.2 -A photograph of the diborane dilution assembly


The dilution method was a highly effective technique for obtaining low concentrations of diborane but it added significantly to the error in the measurement. The pressures used in the dilution procedure were read from the pressure gauge on the diborane regulator. This was not sufficiently sensitive to give a highly precise reading and so the random errors were significant. The calculation below shows the effect of a triple dilution on the error in the measurement of diborane concentration. The physical displacement of the gauge from the lecture bottle also gave rise to the possibility of a systematic error.


The calculation below shows the random error for a triple diborane dilution and using a formula given in reference 100. The errors used for the readings are half a scale division.


Typical values for a triple diborane dilution:

molar fraction of diborane in premix, R = (0.0475 ± 0.00005)

pressure of first fill of diborane mixture, pD1 = (0.500 ± 0.125) bar

pressure of first fill of hydrogen dilutant, pH1 = (2.6 ± 0.1) bar

pressure of second fill of diborane mixture, pD2 = (0.500 ± 0.125) bar

pressure of second fill of hydrogen dilutant, pH2 = (2.6 ± 0.1) bar

pressure of third fill of diborane mixture, pD3 = (0.500 ± 0.125) bar

pressure of third fill of hydrogen dilutant, pH3 = (2.6 ± 0.1) bar

molar fraction of fully diluted diborane = r


\ molar fraction of fully diluted diborane, r = 0.0034 ± 0.0006


The random error in the measurement of the gas phase composition can be up to nineteen percent.


2.5       The Deposition Chamber


A photograph of the deposition chamber is shown in figure 2.3.



                              Figure 2.3 -A photograph of the deposition chamber


The deposition chamber consisted of a single-walled stainless steel six-way cross with welded joints and bolted flanges.


The top flange incorporated three pairs of electrical contacts made from tungsten (W) rods. Glass coatings were used to electrically insulate the rods as they fed through the flange. Vacuum seals were made with epoxy resin and silicone rubber. One pair of contacts was used to supply power to the substrate heater and the other two supplied power to the filaments. The deposition chamber was electrically earthed by a cable attached to the top flange. The casing of the rotary pump provided a second route to ground.


The front flange housed a glass window which provided a viewing port to allow for visual inspection of the reaction. The window also allowed for measurement of the temperature of the hot filament using an optical pyrometer.


Three gas lines fed into the reaction chamber via the rear flange: the reaction gas inlet leading from the mixing manifold; a connection to a capacitance diaphragm pressure gauge (Baratron) which was used to measure the pressure in the chamber during diamond growth runs; and an air vent valve which allowed the chamber to be brought up to atmospheric pressure before it was opened.


The bottom flange led to a two-stage rotary pump via two stainless steel gas lines. A large bore line, with an internal diameter (int. Æ) of approximately 20 mm, allowed the vacuum system to be pumped down rapidly. A narrow bore line, int. Æ 3 mm, fitted with a needle valve allowed for greater control of pressure required for diamond deposition. The openings to both these lines were protected from debris by a copper (Cu) gauze. The exhaust line from the rotary pump was safely routed to an outlet on the roof of the building.


The flange on the left-hand side of the chamber was blank while the right-hand one incorporated a spare pair of electrical feed-throughs.


The top flange could be readily removed for access to the chamber. The filament holders and substrate heater assembly were attached to the underside of this flange, to allow the new filaments to be fitted and the substrates to be loaded before the assembly was loaded into the deposition chamber.


The other flanges could be removed to allow for thorough cleaning of the system. This was required whenever the level of boron contamination became too high to maintain reproducible levels of doping in the diamond films. In practice, this procedure was performed when the required boron doping level in the diamond films was reduced significantly.


During operation, the deposition chamber was air cooled by four fans.



2.6       The Substrate Heater


A substrate heater was made from a coil of nichrome wire [nickel-chromium (Ni‑Cr) 24 SWG, standard wire gauge]. The wire was threaded through insulating ceramic beads and then coiled. The coil of wire was then encased in fire cement and gently heated to allow the cement to set. It was important to drive off any water that could be released into the chamber in order to prevent oxidation of the hot filaments. A molybdenum (Mo) plate was placed onto the top of the cement block to provide a flat surface on which the substrates could be placed. Molybdenum was chosen because it has a high melting point, it is inert under the reaction conditions and is effective in evenly spreading the heat to the base of the substrates.


The substrate heaters were operated by passing a current of 4 A through the nichrome wire. A potential of 15 to 20 V was required to achieve this current depending on the length of wire used in the construction of the heater. From previous studies on similar systems, it is estimated that the hot filaments reached 2400 °C and provided additional radiative heating which raised the substrate temperature to 900 °C. 64 The maximum operating temperature of the substrate heater was limited by the breakdown of the fire cement at approximately 1000°C which would cause water vapour to be released into the chamber and lead to burn out of the filaments.


The substrate heaters had an operational lifetime in the order of eight hundred hours, after which the resistance wires would become too brittle and break. This allowed for approximately ninety growth runs during the life of the heater. The periodic replacement of the heater was also important to reduce residual boron levels in the chamber as the heating block could not be cleaned and the fire cement was believed to be a significant source of contamination due to its porous nature.


Electrical connections between the heating block and the tungsten feed-throughs were made with standard copper flex insulated by ceramic beads.


An Iso-tech DC power supply was used for the substrate heater and this was protected from overheating by an external fan behind the heat exchanger at the rear of the unit.



2.7       The Filaments


The CVD reactor used tantalum (Ta) filaments which were prepared by coiling a length of wire around a 4 mm rod. The shaft of a screwdriver was routinely used for this purpose. The wire used had a diameter of 0.25 mm. Previous work by the group had used six-turns of wire which spanned approximately one centimetre. The chamber used for this study was designed for larger scale deposition. It was found that ten turns of wire spanning 2 cm was the optimum scale for the system as longer filaments tended to sag during operation which resulted in poor diamond growth.



                    Figure 2.4 - A photograph of the deposition chamber showing the

                       internal assembly after a deposition run - the heating stage and

                                        a drooping tantalum filament are visible


Previous studies had used single filament set-ups. The chamber used for this experiment was a double filament arrangement. This allowed for either a larger deposition area to be covered or for two or more substrates to be coated simultaneously. The two filaments were connected in series with the connection being made outside the chamber. The apparatus could be readily reconfigured to allow single filament operation using either the front or the rear filament. As well as allowing for planned runs with single filaments, runs in which one of the two filaments failed could be continued after a quick rewiring of the connections.


The filaments were attached to tungsten (W) rods by stainless steel clamps, and the power supply was provided by a Variac variable resistor with an external voltmeter and ammeter. The current was maintained at 6.75 A throughout the diamond deposition as this provided an optimum filament temperature of approximately 2400 °C. 64


The filaments could not be operated at very low pressures as the evaporation of tantalum would be too great. Under normal operating conditions, the methane in the gas mixture caused carburization of the filaments as the malleable silver-coloured tantalum reacted to form brittle gold-coloured tantalum carbide. As the filament reacts, the resistance changes and so the potential had to be varied from 20 to 30 V to maintain a current of 6.75 A. The filaments were replaced after each growth run.


The tantalum wire (Aldrich) used had a purity of greater than 99.9%, the levels of impurity are listed in table 2.3.





Aluminium, Al


Copper, Cu


Tin, Sn


Nickel, Ni


Chromium, Cr


Vanadium, V


Magnesium, Mg



Table 2.3 - Trace elements in the tantalum wire

Values taken from Supplier’s Certificate of Analysis

2.8       The Substrates


Silicon (Si) was used as a substrate in all the films studied for the work described in this thesis, although boron doped diamond films were also grown on tungsten (W) substrates using the same reaction conditions.


The surface-area of the growth side of the silicon substrates varied from 1 cm2 (1 cm ´ 1 cm) to 4 cm2 (2 cm ´ 2 cm). Typically two 2 cm2 (2 cm ´ 1 cm) substrates would be placed next to each other on the heating stage to allow two diamond films to be fabricated per growth run. This allowed for better control experiments and increased the production rate of the electrodes. The substrates were obtained by cutting or breaking pieces of silicon from silicon wafers reclaimed from the microchip fabrication industry. The thickness of these wafers varied from ½ mm to 3 mm. The surface of the substrate requires pre-treatment to enhance nucleation of diamond growth sites at the start of growth process. The pre-treatment used was manual abrasion of the surface by rubbing with 2-3 μm diamond powder followed by cleaning in 2-propanol (IPA). A two stage cleaning process was employed. An initial cleaning with IPA soaked cotton buds was followed by a 15 minute ultrasonic bath in IPA.


If the diamond film was to be doped throughout and electrical connections made through the silicon substrate then the substrate would be dipped in hydrofluoric acid (HF) and rinsed in 18.2 MW cm ultrapure deionised water (Millipore) immediately before loading into the vacuum chamber. This was to reduce the effect of any insulating layer of silicon oxide (SiO2) on the film.


For the majority of experiments, electrical conduction through the substrate was not desirable. This was prevented by the use of undoped silicon substrates (as undoped silicon exhibits high electrical resistivity) and a variable doping profile in the diamond film (with a higher resistance region in the diamond adjacent to the substrate).


2.9       Electrical Contacts to the films


Four types of electrical contact were used.


2.9.1    Indium/Gallium Eutectic


Early work with highly doped films ([C]/[B] ratio of 28,000 to 33,000 p.p.m.) in the gas phase) used uniformly doped diamond films grown on heavily doped, conductive silicon (Si) substrates. Electrical contacts were made to the reverse side of the silicon using a gallium (Ga) /indium (In) eutectic. The silicon was roughened by scratching with a diamond scribe and then the liquid eutectic was applied. A contact was made by pressing a steel backing strip onto the eutectic. This technique was satisfactory for highly conductive diamond films but was limited by the need to consider the effects of the liquid contact and the interface between the silicon and the diamond.


2.9.2    Silver Loaded Epoxy Resin


A simpler contact could be made via the top of the film. For some highly doped films ([C]/[B] ratio of 3,000 to 33,000 p.p.m.), a satisfactory contact could be made by attaching a wire to the surface using a silver-loaded epoxy resin (‘silver dag’). The silver dag was often physically protected by a layer of stronger epoxy resin adhesive (Araldite Rapid) applied over the contact.


It should be noted that hydrofluoric acid is an extremely hazardous material and the use of direct contacts to diamond, rather than through the substrate, removed the need for hydrofluoric acid treatment of the silicon substrates.


The use of top contacts allowed greater flexibility in the doping profile of the diamond films. Films could be grown without the addition of diborane during the early stages of growth. This would allow nucleation to occur in the absence of boron containing species. It is not believed that the boron could diffuse significantly through the film and so the lower portion of the films could be made more resistive than the upper section. The path of current flow through the films would therefore avoid the nucleation side of the film and the substrate material. This was advantageous because the nucleation side may be considered to be of lower quality. The nucleation side may be contaminated by species present in the substrate material and it has an increased number of grain boundaries due to the nature of the columnar growth method, where diamond growth starts with many small crystallites which coalesce and form fewer larger domains as the film grows. This effect could be seen in the cross-sectional SEM image of an industrial diamond film shown in figure 2.28.


Silver dag contacts did not give Ohmic behaviour for diamonds with low doping levels ([C]/[B] ratio below 250 p.p.m.) due to a Schottky barrier forming at the metal-semiconductor interface. 101 Some research groups have argued that sufficiently electronegative metals such as gold (Au) overcome this problem 101‑103 but their work remains less than conclusive.


2.9.3    Three Layer Metalisation


A more reliable way of forming an Ohmic top contact was to use a “three layer metalization” technique where three layers of metal were deposited on the surface of the diamond. The bottom layer was a titanium (Ti) spot which could be annealed to produce a titanium carbide (TiC) layer between the metal and the diamond. This chemical bonding reduced the Schottky barrier at the metal-semiconductor interface and gave the required Ohmic behaviour. A gold (Au) top layer was required to prevent the oxidation of the titanium during the annealing step and, more slowly, over the lifetime of the device as titanium reacts readily with any traces of oxygen at elevated temperatures. Finally, a layer of platinum (Pt) was required as a barrier layer between the titanium and gold layers to prevent the interdiffusion of the metals during annealing. Wires could be attached to the surface of the gold spot using silver dag, as before. A schematic diagram of the design of the contacts can be found in figure 2.5. Further details of the formation of Ohmic contacts can be found in chapter 3.


The titanium and gold layers were deposited using a Edwards evaporator which was modified to incorporate a heating stage. To prevent oxidation of the titanium, the substrate was heated in vacuum to remove oxygen containing species from the surface. The titanium evaporation was then performed on a clean surface. After the deposition, the sample was allowed to cool down to room temperature before being exposed to air.


The evaporator worked by passing a current through a tungsten (W) wire basket containing the source metal (titanium pellets or short strips of gold). A current of approximately 40 A was sufficient to evaporate the source metal. The substrate, which was positioned under the basket, became coated by the metal as it resolidified.


Platinum possesses a high boiling point and could only be evaporated with difficulty. Therefore, a sputter coater was used to deposit the platinum spots.





Melting Point, qC,m / °C


Boiling Point, qC,b / °C


Gold (Au)






Titanium (Ti)






Platinum (Pt)






Tungsten (W)






Table 2.4 - Transition temperatures of selected pure metals at standard pressure

Values taken from reference 12 and based upon IUPAC(1983) standards


Figure 2.5 - A schematic diagram of a three layer metal top contact (not to scale)

Note that the plan view shows the metal layers in reverse order for illustrative purposes. The layers from top to bottom are Ag, Au, Pt & Ti



The three layer metalization described above is the best technique used to form good Ohmic contacts to diamond electrodes. However, the need for three separate metal depositions and an annealing step reduce the practicality of the technique significantly. The extensive processing may also effect the surface properties of the film.

2.9.4    Titanium Contacts


A novel method was developed to form Ohmic contacts for the low doped diamond samples in this study. A single strip of titanium was deposited onto the surface of the silicon substrate before the diamond deposition. Selective diamond growth on top of this metal layer was achieved by leaving a smooth section of silicon which provided few nucleation sites. As the diamond formed at raised temperatures, a metal carbide would form without the need for a separate annealing step after the growth process and, since the film was being grown in the absence of an oxygen containing species, no protective metals were needed to prevent oxidation of the titanium. A schematic diagram of the design of the contacts can be found in figure 2.6. Full details of the technique can be found in chapter 3.


This technique greatly simplified the processing required to fabricate a working electrode. The platinum and gold deposition steps and the annealing process were no longer required. The titanium deposition could be performed on the substrate before the diamond deposition. For the three layer metalisation technique, all four of these steps involved heating the samples above room temperature and so had the potential to change the surface conditions of the diamond films.


The reduction of the number of steps significantly reduced the time required to produce a working electrode and the reliability of the overall fabrication process was dramatically increased.


The use of a back contact gives the maximum possible surface area for use as an electrode but requires the film to be conductive throughout its bulk.


The presence of the titanium strip made the use of post-treatments more complicated. Procedures such as refluxing in nitric acid (HNO3) or dipping into chromic acid (sat. soln. of K2Cr2O7 in conc. H2SO4) could not be easily performed. This problem could be avoided by the use of PTFE cells to mark out an area of the diamond film for selective post-treatment.

Figure 2.6

A schematic diagram of a single layer metal bottom contact (not to scale)

Zone A is a region of diamond growth on bare silicon.

Zone B is a region of diamond growth on titanium coated silicon.

Zone C is a region of titanium coated silicon where the growth of a continuous diamond film has been prevented by not abrading the silicon prior to the titanium deposition.

2.9.5    Summary


The advantages and disadvantages of the four types of contact are summarised below in table 2.5.



Type of Contact






·       does not effect diamond film
- no special treatments needed


·       flexibility with post- and pre-treatments


·       relatively convenient


·       no reduction in available surface area



·       need to consider diamond-silicon interface


·       liquid contact - less reliable


·       HF dip required

Silver Dag


·       very convenient


·       does not affect diamond film
 - no special treatments needed


·       flexibility with post- and pre-treatments



·       Schottky barriers

Three Layer Metalisation


·       reliable Ohmic contacts formed


·       diamond film must be heated


·       time consuming


·       complicated - risk of failure


·       limits on possible





·       reliable Ohmic contacts formed


·       does not affect diamond film
- no special treatments needed


·       no reduction in available surface area



·       restricted post-treatments


·       need to conduct through the entire depth of diamond sample
(insulating layer not possible)


Table 2.5 - Summary of the types of electrical contacts used in this study

2.10     Typical Growth Conditions


Table 2.6 summarises the typical growth conditions for the diamond films. Appendix B gives full details of the growth conditions for the films used in these studies.





20 Torr


Hydrogen flow rate


200 s.c.c.m.


Methane flow rate


1.4 s.c.c.m.


Diborane flow rate (range)


5 ´ 10-6 s.c.c.m. to 5 ´ 10-2 s.c.c.m.


Diborane flow rate (typical of low end)


4 ´ 10-5 s.c.c.m.


Substrate temperature


900 ºC


Filament temperature


2400 ºC


Filament/substrate separation


4 mm


Deposition Time


7 hours to 27 hours




Si - see appendix B for specifications




two ten-turn Ti coils


Table 2.6 - Typical deposition conditions for the hot-filament CVD reactor


2.11     Secondary Ion Mass Spectroscopy


Secondary Ion Mass Spectroscopy (SIMS) was used to detect the presence of boron in the films and to determine the level of doping.


SIMS involved the bombardment of the sample with high energy ions in an ultra-high vacuum (UHV) chamber. On impact with the sample, the incident ions caused the ionisation of the surface. The ions released from the surface entered an electric field which drew them into a time-of-flight mass spectrometer (ToF‑MS). The time taken for the ions to reach the detector was proportional to the ratio of the mass of the ion to its electronic charge (m/z).


SIMS analysis of the diamond films grown for this study was performed at two institutions: the University of Bristol Interface Analysis Centre (IAC) and Millbrook Instruments Limited (Lancashire).


The IAC produced positive and negative ion spectra (figures 2.7 and 2.8 respectively). The positive ion spectrum showed peaks at m/z values of 10 and 11 daltons. This corresponded to singly charged ions of the two isotopes of boron (10B+ & 11B+) and indicated the presence of boron in the films. Unfortunately, without knowledge of the relative sensitivity of the technique to the different species present, the data can not be used for quantitative analysis of the boron concentration.



                           Figure 2.7 - A positive ion time-of-flight SIMS spectrum



                          Figure 2.8 - A negative ion time-of-flight SIMS spectrum


Four diamond samples were sent to Millbrook Instruments for SIMS analysis. Table 2.7 shows the relevant characteristics of the films.



Code Number



of boron in the gas phase

[B]/[C] in p.p.m.



Resistance (W)





grown before first use of boron containing species


2.5 ´ 106





boron contamination from chamber






8.8 ´ 10-5







5.3 ´ 10-3





Table 2.7 - Samples analysed by Millbrook Instruments using SIMS


The resistance values given were determined using a digital voltmeter (DVM) and were intended for use only as rough guidance.


Millbrook Instruments provided a laboratory standard graphite sample with a known concentration of boron (50 p.p.m.). This allowed for calibration of the spectra.


Figures 2.9 to 2.12 show the raw spectra for the four diamond samples and figure 2.13 shows the corresponding figure for the graphite control. The processed data are shown in figure 2.14. The analysis conditions are summarised in table 2.8 below.



Incident Ion






Beam Intensity



6 keV, 20 nA



Raster Area



250 mm ´ 250 mm



Mass Range



6 - 16 m/z (10 channels per mass)



Dwell Time



1 second per channel






10 scans



Acquisition Time (per spectrum)



1000 seconds


Table 2.8 - A Summary of the SIMS Experimental Conditions


Figure 2.9 - A positive ion time-of-flight SIMS spectrum of sample (a)

Figure 2.10 - A positive ion time-of-flight SIMS spectrum of sample (b)

Figure 2.11 - A positive ion time-of-flight SIMS spectrum of sample (c)

Figure 2.12 - A positive ion time-of-flight SIMS spectrum of sample (d)

Figure 2.13 - A positive ion time-of-flight SIMS spectrum of the graphite control


Figure 2.14 - SIMS spectra plotted relative to the undoped control sample (a)

Adsorbed oxygen and water enhanced the B+ signal by an order of magnitude. However, long count times had to be employed to detect the low levels of boron and the surface was significantly eroded. Therefore, the enhanced signal intensity at the surface did not give a reduction of the detection limit.


As expected, samples (b), (c) and (d) showed boron peaks at m/z values of 10 and 11. The ratio of the intensities of the two boron peaks ranged from 80:20 to 70:30 which is in agreement the ratio of natural abundances of the boron isotopes 10B and 11B (81.3 : 18.7). 12


The peaks at m/z values of 12, 13, 14 and 15 corresponded to CHx+ ions where x = 0 to 3.


Samples (a), (b) and (d) all showed a low level contaminant peak at m/z = 7. This corresponded to lithium ions (Li+). The intensities of the peaks indicated that the lithium contamination was at the sub p.p.m. level.


The results suggested that the level of doping in the films is broadly proportional to the gas phase concentrations. Previous published work using SIMS to analyse levels of boron doping in CVD diamond had been inconclusive. 79,104,105


As SIMS is a locally destructive technique, it may be used to produce depth profiles where the concentrations of the species are plotted against the depth of erosion into the sample. Depth profiling requires long measurement times and so proved impractical due to the hardness of diamond and its resistance to erosion. However, in the experiments detailed here, the low levels of boron in the films necessitated the sampling of a sufficiently large quantity of material in order to produce reliable results that the values may be considered to apply to the bulk rather than the surface.


It should be noted that SIMS measures the total boron content in the films. This boron does not necessarily have to occupy electrically active substitutional sites in the diamond crystal.

2.12     Scanning Electron Microscopy and Optical Microscopy


Scanning electron microscopy (SEM) and optical microscopy were the main techniques used to characterise the films.


The SEM work was performed on a Hitachi Model 3200 Scanning Electron Microscope operating with an accelerating potential of 25 keV. A Mamiya camera loaded with rolls of 6 ´ 7 cm film was used to capture images.


SEM requires samples that are electrically conductive. In order to increase the surface conductivity of insulating samples, a layer of gold (Au) was deposited with an Edwards S150A Sputter Coater. This treatment was not required for the boron doped diamond samples as the doping made them sufficiently conductive.


The electron microscope was fitted with an energy-dispersive X-ray (EDX) spectrometer designed to perform an elemental analysis. The EDX detector was protected by a beryllium (Be) window. This prevented the detection of X-rays from elements with low atomic numbers such as carbon (C) and boron (B) and so the EDX spectra were of limited use in the analysis of the boron doping of diamond.


A Zeiss Axiolab Optical Microscope fitted with Zeiss Epiplan lenses was routinely used to view the samples. Objective lenses of 5 ´, 10 ´, 50 ´ and 100 ´ coupled with an 10 ´ eyepiece to give a range of magnifications from fifty to one thousand.


Images were captured with a JVC TK-1280E Colour Camera attached to an Olympus BH2 Optical Microscope fitted with a 10 ´ eyepiece and 4 ´, 10 ´, 20 ´ and 50 ´ objective lenses.


Optical microscopy was particularly helpful in monitoring the deposition of titanium. Titanium layers that had become oxidised were often blue in colour.


SEM and optical microscopy showed the diamond films to be continuous and composed of well formed crystallites. Initial experiments were performed in which the growth conditions were varied. These confirmed that the optimum growth conditions outlined in previous studies gave the best results. Figures 2.15 to 2.18 show SEM images of an undoped diamond film grown under the typical conditions listed in table 2.6. Figures 2.15 to 2.17 show images of the top surface at various levels of magnification. Figure 2.18 shows a image of the cross-section of the film. The film was uniform and has well defined, predominantly square faceted crystallites.


Figures 2.19 and 2.20 show diamond films grown with double and quadruple the normal concentration of methane respectively. The films exhibited a much less well defined structure, characteristic of growth with a high carbon to hydrogen ratio in the gas mixture. 14,17


Figures 2.21 and 2.22 show cross-sectional SEM images typical of those used to measure the thickness of the diamond films. The rate of deposition, calculated by dividing the thickness of the diamond by the growth time, varied between films from zero to 0.7 mm/hr.


Within each growth run, the rate of deposition was expected to vary. Initial nucleation had to occur before a continuous film could start to form. Growth would then progress on different crystal faces at different rates. The gradual reaction of the hot filaments or any other changes in the reaction conditions may have added to any variations in the rate of growth.


The thickness of thirty samples, which had grown beyond the nucleation phase and formed continuous films, was measured. The mean growth rate was 0.3 ± 0.2 mm/hr. No significant difference in the rate of deposition was detected between boron doped and undoped diamond films.


                       Figure 2.15 - The surface of a HFCVD diamond film showing

                              square facets and a twinned crystallite [sample B13]



                   Figure 2.16 - The surface of a HFCVD diamond film [sample B13]


                             Figure 2.17 - The surface of a HFCVD diamond film

                         showing uniform coverage over the substrate [sample B13]



                       Figure 2.18 - A cross-section of a diamond film [sample B13]


         Figure 2.19 - A diamond film grown in an atmosphere containing 1.4% methane

                                  (double the normal concentration) [sample B6]



         Figure 2.20 - A diamond film grown in an atmosphere containing 2.8% methane

                                (quadruple the normal concentration) [sample B8]


           Figure 2.21 - A cross-sectional view of a diamond film grown with quadruple

                    the normal concentration of methane in the gas phase [sample B8]



               Figure 2.22 - A cross-sectional view of the results of an early growth run

                                                            [sample B1]

Figures 2.23 to 2.25 show images of a sample where the diamond growth had been stopped after 5 hours and 35 minutes. Figure 2.23 shows the surface at the centre of the sample where the diamond film was continuous with well defined facets. Figures 2.24 and 2.25 show the surface at the periphery of the sample where the growth has not progressed far beyond the nucleation stage, leaving individual crystallites on the silicon substrate.


The quality and morphology of the films was comparable to that obtained in industrially grown samples. Figures 2.26 to 2.28 show an industrial free-standing diamond film that was grown for an optical application in the aerospace industry using microwave plasma assisted CVD (MPACVD). 106 The industrial film was grown continuously for about ten days 107 and is therefore much thicker (175 mm) than the HFCVD films and it exhibits larger facet sizes (~50 mm).


Figures 2.29 and 2.31 show various views of a thin diamond film.



                          Figure 2.23 - A continuous diamond film [sample B142a]


           Figure 2.24 - Incomplete surface coverage at the edge of a thin diamond film.

         The silicon substrate was visible behind the diamond crystallites [sample B142a]



          Figure 2.25 - The reaction stopped at the stage where the individual crystallites

                   have started to coalesce to form a continuous film [sample B142a]


                            Figure 2.26 - Industrial MPACVD diamond (undoped)




                            Figure 2.27 - Industrial MPACVD diamond (undoped)



                          Figure 2.28 - A cross-section of a free-standing industrial

                                                   MPACVD diamond film



             Figure 2.29 - SEM image showing large crystallites formed around 2-3 mm
         diamond crystals remaining from the substrate abrasion process [sample B141b]


                     Figure 2.30 - SEM image showing growth of a thin diamond film

                                                         [sample B141b]



               Figure 2.31 - SEM image showing a cross-section of a thin diamond film

                                                         [sample B141b]

The titanium contacts described in section 2.9.4 required three distinct zones of diamond growth.


·          The main section of the sample had to comprise of a continuous film of high quality diamond grown directly over an insulating silicon substrate. This zone would form the electrode surface in the electrochemical experiments.


·          An exposed titanium strip was required at the edge of the sample to allow electrical connections to be made.


·          Finally, a middle zone is required comprising of diamond grown over titanium. This zone would form the Ohmic contact between the metal and the semi‑conducting diamond.


The standard electrode design had a single strip of titanium (as shown in figure 2.6). This was ideal for the electrochemical studies as the design maximised the surface area of usable diamond. The single strip design did not, however, readily facilitate studies of the contact itself. The absence of a second Ohmic contact complicated measurements of the current-voltage characteristics. To avoid this problem, a diamond sample was made with two titanium strips, on opposing edges of the sample, as shown in figure 2.32. The electrical properties of this film were investigated as detailed in chapter 3.


Figures 2.33 to 2.60 show SEM and optical images of the double ended sample (B147a). The sample has good coverage where required (zones 2 to 4) and incomplete coverage where the titanium needs to be exposed (zones 1 and 5). The two end zones were prepared in the same way: a coating of titanium was applied to a clean, smooth, unabraded silicon surface. However, the extent of diamond growth on these zones varied considerably. Zone 5 was typical of growth on unabraded substrates: isolated crystallites formed with minimal overall coverage. Zone 1 was unusual in that there was much greater coverage than normal. The growth had progressed beyond the nucleation stage and the crystallites had started to coalesce. The increased coverage did not, however, provide complete coverage and good electrical contacts could still be made to the exposed underlying titanium.

Figure 2.32 - A schematic diagram of sample B147a (not to scale)

Zone 3 is a region of diamond growth on bare silicon.

Zones 2 & 4 are regions of diamond growth on titanium coated silicon.

Zones 1 & 5 are a regions of titanium coated silicon where the growth of a continuous diamond film has been prevented by not abrading the silicon prior to the titanium deposition.


Figures 2.33 to 2.36 show four optical microscopy images of the central section of the surface (zone 3) taken with a 50 ´ objective lens. The images were taken at well spaced points across zone one and show the uniformity of the film. Figure 2.37 shows a smaller scale image taken with a 20 ´ objective lens. SEM images of zone 3 are shown in figures 2.38 and 2.39.


As described in section 2.8, 2-3 mm diamond grit was used in the preparation of the substrates. The majority of the diamond grit was removed from the silicon surface before the substrate was loaded into the CVD chamber. Any remaining diamond crystals that survived the cleaning process acted as nucleation sites for diamond growth. Figure 2.28 shows a SEM image of diamond crystallites formed from such growth. Optical microscopy clearly showed these crystallites as any facets orientated normal to the incident light beam would shine brightly as they reflected the light back towards the optical system. A number of isolated square and triangular facets can be seen in figures 2.33 to 2.35, while 2.36 shows a conglomeration of them. Figures 2.34 and 2.37 show complete crystallites, highlighted by red circles, in which none of the facets are perpendicular to the normal illumination.


Sample B147a contained larger numbers of these crystallites than were normally seen in the diamond films. The presence of two titanium strips on the substrate made the abrasion and cleaning process more complicated. It was necessary when cleaning the surfaces of the abraded regions (zones 2 to 4) to keep the surfaces of the end regions (zones 1 & 5) free from diamond grit and so, in order to prevent contamination, a less rigorous cleaning procedure was performed.

Figures 2.33 & 2.34 - optical microscopy images of the central section (zone 3),

50 ´ objective lens, red circle highlights a crystal (see page 67) [sample B147a]


Figures 2.35 & 2.36 - optical microscopy images of the central section (zone 3)

taken with a 50 ´ objective lens [sample B147a]

Figure 2.37 - optical microscopy image of the central section (zone 3),

         20 ´ objective lens, red circle highlights a crystal (see page 67) [sample B147a]


Figure 2.38 - SEM image of the central section (zone 3) [sample B147a]


Figure 2.39 - SEM image of the central section (zone 3) [sample B147a]


As expected, the growth of diamond on zones 1 and 5 did not totally cover the underlying titanium.


Zone 5 was typical of the films grown on smooth titanium covered silicon, in that there was only isolated diamond growth. Figure 2.40 shows an optical image of the sample taken with a 50 ´ objective lens. The lower left hand side showed some coalescing of the individual cystallites but the majority of the area was covered only by well spaced, isolated crystallites. Figure 2.41 shows a similar image, in which a continuous narrow strip of diamond could be seen, probably formed along a microscopic scratch on the surface of the silicon.


Figure 2.42 shows an area of denser coverage nearer the adjacent abraded region (zone 4) and figure 2.43 shows an optical image taken at a lower resolution (20 ´ objective lens). Figure 2.44 and 2.45 show SEM images of zone 1 of the film. The detail of an individual crystallite can be seen in figure 2.45.


The opposite end of the sample (zone 1) exhibited unusually good diamond coverage, comparable to the most extensive growth seen on any of the unabraded titanium strips in this study. Figures 2.46 and 2.47 show optical images of zone 1 taken with a 50 ´ objective lens and figures 2.48 and 2.49 show the region viewed with a 20 ´ objective lens. Figure 2.50 shows an SEM image of zone 1. The images show that although there was considerable diamond coverage, there were still areas of exposed titanium. Good electrical contact could therefore be made to this region.


Figures 2.40 & 2.41 - optical microscopy images of one of the end sections

(zone 5) taken with a 50 ´ objective lens

showing limited diamond growth [sample B147a]


Figure 2.42 - optical microscopy image of taken with a 50 ´ objective lens

showing partial diamond growth [sample B147a]


Figure 2.43 - optical microscopy image of one of the end sections (zone 5)

taken with a 20 ´ objective lens [sample B147a]

Figure 2.44 - SEM image of one of the end sections (zone 5) [sample B147a]


Figure 2.45 - SEM image of one of the isolated diamond crystallites

[sample B147a]



Figures 2.46 & 2.47 - optical microscopy images of one of the end sections

(zone 1) taken with a 50 ´ objective lens showing partial diamond growth

[sample B147a]



Figures 2.48 & 2.49 - optical microscopy images of one of the end sections

(zone 1) taken with a 20 ´ objective lens showing partial diamond growth

[sample B147a]

                       Figure 2.50 - SEM image of one of the end sections (zone 1)

                                 showing partial diamond growth [sample B147a]


The boundaries between the growth zones can be seen in figures 2.51 to 2.57 (optical microscopy) and figures 2.58 to 2.60 (SEM).


Figures 2.51, 2.52 and 2.58 show the clearly defined boundary between zones 4 and 5.


Figures 2.51 and 2.58 show the same region viewed by optical and electron microscopy respectively. Measurement of the distances between distinctive features visible in both the images allowed an accurate scale to be applied to the optical images presented in this chapter.


Figures 2.53 and 2.54 show the boundary at the opposite end of the sample (between zones 1 and 2). There was less difference between the diamond coverage in these two zones but the boundary could still be identified.


Figure 2.55 shows the boundary between zones 3 and 4. There is complete diamond coverage over both regions. The transparency of the diamond film allowed light to pass through the sample. Zone 4 appeared brighter than zone 3 as the titanium beneath zone 4 reflected more light than the silicon beneath zone 3.


Figures 2.56 and 2.57 show the boundary between zones 2 and 3. As for the boundary between zones 3 and 4, the diamond coverage was complete and reflection from the underlying titanium could be seen.


Figures 2.59 and 2.60 show SEM images of the boundary between zones 1 and 2. Zone 2 can be seen at the top of the images with zone 5 beneath it. The dark areas at the bottom of the images showed the silver dag which had been used to make electrical contacts to zone 1.

Figure 2.51 - the boundary between zones 4 and 5 (´ 20 lens) [sample B147a]


Figure 2.52 - the boundary between zones 4 and 5 (´ 10 lens) [sample B147a]


Figure 2.53 - the boundary between zones 1 and 2 (´ 10 lens) [sample B147a]


Figure 2.54 - the boundary between zones 1 and 2 (´ 10 lens) [sample B147a]


Figure 2.55 - the boundary between zones 3 and 4 (´ 20 lens) [sample B147a]


Figure 2.56 - the boundary between zones 2 and 3 (´ 20 lens) [sample B147a]


Figure 2.57 - the boundary between zones 2 and 3 (´ 10 lens) [sample B147a]


                    Figure 2.58 - SEM image of the boundary between zones 4 and 5

                                                         [sample B147a]


                    Figure 2.59 - SEM image of the boundary between zones 1 and 2

                                                         [sample B147a]



                    Figure 2.60 - SEM image of the boundary between zones 1 and 2

                                                         [sample B147a]

2.13     Laser Raman Spectroscopy


Raman Spectroscopy is a well established diagnostic technique for determining the quality of diamond films. 108-112 Spectra of diamond show a characteristic peak at a shift of 1332 cm-1 relative to the excitation wavenumber. The peak is the highest energy Raman peak for diamond and corresponds to the zero phonon excitation in the crystal. In good quality diamond samples, this peak is sharp. However, if the crystal structure is distorted then the spread of the peak, as measured by the full-width half-maximum, will increase and the peak may be shifted up or down the spectrum.


As well as the diamond peak at 1332 cm-1, a number of other Raman peaks may be seen in poor quality diamond films. Graphite produces peaks at 1580 cm-1 and 1355 cm-1 and disordered sp2 carbon produces an asymmetrical broad band from 1100 cm-1 to 1600 cm-1 with a maximum at around 1520 cm-1.


Raman spectroscopy can be a very sensitive technique in detecting small amounts of graphitic content in diamond films. When the incident laser beam with an excitation wavelength of 488 nm is chosen, the technique is fifty times more sensitive to sp2 structures than to sp3 structures.


In addition to Raman peaks, the laser illumination can also lead to photoluminescence (PL) peaks due to impurities and structural defects.


Raman spectroscopy provides an excellent tool to gain qualitative information on the quality of diamond films.


Several studies have reported changes to the Raman spectrum due to the presence of boron in the diamond films. These changes require higher levels of doping than can be those present in the films presented here

(50 p.p.m., 9 ´ 1018 B atoms/cm3).


Raman spectra were taken with a Renishaw Raman Imaging Microscope and Spectrometer with a spectral resolution of 1 cm-1. An argon gas laser with a wavelength of 488 nm (green light) was used to provide the incident laser beam.


Figure 2.61 shows a Raman spectrum of a natural type IIb diamond crystal.* A single sharp peak can be seen which was used to calibrate the spectrometer.


Figure 2.62 shows a Raman spectrum of an poor quality undoped diamond film grown by MPACVD which provides an example of some of the features that may be seen in diamond films. The 1332 cm-1 diamond peak is visible but is not particularly sharp and not larger than the other features. A broad peak centred around 1355 cm-1 due to graphite and disordered sp2 carbon and a broad band centred at around 1500 cm-1 can be seen. The broad band shows some fine structure that may be due to impurities or defects (such as nitrogen centres). The fine structure can not be readily assigned from a spectrum taken at room temperature.


Figures 2.63 to 2.68 show Raman spectra of diamond films grown for this study. While there was some variation in the spectra, they all displayed a sharp 1332 cm-1 diamond peak. There was some evidence of a broad band due to disordered sp2 carbon but this was weak and therefore confirmed that there was little sp2 carbon present in the films.

Figure 2.61 - Raman spectrum of a type IIb natural diamond showing a sharp

                               zero‑phonon peak and no other significant features

Figure 2.62 - Raman spectrum of a poor quality undoped MPACVD

                                                            diamond film

Figure 2.63 - Raman spectrum of boron doped diamond [sample B128a]

Figure 2.64 - Raman spectrum of boron doped diamond [sample B128b]


              Figure 2.65 - Raman spectrum of boron doped diamond [sample B129a]


Figure 2.66 - Raman spectrum of boron doped diamond [sample B130b]

Figure 2.67 - Raman spectrum of boron doped diamond [sample B140a]

Figure 2.68 - Raman spectrum of boron doped diamond [sample B140b]

2.14     Summary


CVD apparatus has been assembled which can deposit boron doped polycrystalline diamond films with a wide range of doping levels.


SIMS, SEM and optical microscopy have been used to show the films to be continuous and of high quality.


Diamond films have been deposited selectively onto patterned silicon wafers. The use of patterned silicon wafers with a partial titanium covering provided a novel method to fabricate boron doped diamond films with good Ohmic contacts.


Chromic acid was prepared immediately before use by making a saturated a saturated solution of potassium dichromate (K2Cr2O7) in concentrated sulphuric acid (H2SO4). 134

The manufacturer of the diborane gas used in these studies required approximately six months to produce and deliver a standard order of diborane.  Supplies of more esoteric gases, such as the trimethylboron (B(CH3)3) used in references 76 and 77 were not as readily available.


Vaporisation has been used as an alternative to dissolution for solid B2O3 sources 92 but problems with controllability remain.

* Type II is a classification established in 1934 for diamonds transparent to 8 mm infra-red radiation. It was subsequently discovered that the diamonds that are opaque to 8 mm radiation, and hence classified as type I, contain nitrogen impurities. 5 Type IIb signifies a very low concentration of nitrogen.