- Email: [email protected]
The development of diamond tracking detectors for the LHC
ARTICLE IN PRESS
Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
The development of diamond tracking detectors for the LHC W. Adama, E. Berdermannb, P. Bergonzoc, W. de Boerd, F. Boganie, E. Borchif, A. Brambillac, M. Bruzzif, C. Colledanig, J. Conwayh, P. D’Angeloi, W. Dabrowskij, P. Delpierrek, J. Doroshenkoh, W. Dulinskig, B. van Eijkl, A. Fallouk, P. Fischerm, F. Fizzottin, C. Furettai, K.K. Gano, N. Ghodbanep, E. Grigorievd, G. Hallewellk, S. Hano, F. Hartjesl, J. Hrubeca, D. Hussong, r . H. Kagano,*, J. Kaplonq, C. Karlp, R. Kasso, M. Keilm, K.T. Knopfle , T. Koethh, M. Krammera, A. Logiudicen, R. Lun, L. mac Lynneh, C. Manfredottin, R.D. Marshallc, D. Meierq, D. Menichellif, S. Meuserm, M. Mishinas, L. Moronii, J. Noomenl, A. Ohq, L. Pererah, H. Perneggerq, M. Pernickaa, P. Polesellon, R. Potenzat, J.L. Riesterg, S. Roeq, A. Rudgeq, S. Salai, M. Sampietrou, S. Schnetzerh, S. Sciortinof, H. Stelzerb, R. Stoneh, C. Suterat, W. Trischukv, D. Tromsonc, C. Tuvet, B. Vincenzot, P. Weilhammerq, N. Wermesm, M. Wetsteinh, W. Zeunerp, M. Zoellero . Institut fur Akademie d. Wissenschaften, Vienna, Austria . Hochenergiephysik der Osterr. b GSI, Darmstadt, Germany c LETI (CEA-Technologies Avancees) DEIN/SPE-CEA Saclay, Gif-Sur-Yvette, France d Universitat . Karlsruhe, Karlsruhe, Germany e LENS, Florence, Italy f University of Florence, Florence, Italy g LEPSI, IN2P3/CNRS-ULP, Strasbourg, France h Rutgers University, Piscataway, NJ, USA i INFN, Milano, Italy j Faculty of Physics and Nuclear Techniques, UMM, Cracow, Poland k CPPM, Marseille, France l NIKHEF, Amsterdam, The Netherlands m Universitat . Bonn, Bonn, Germany n University of Torino, Italy o The Ohio State University, Columbus, OH, USA p II.Inst. fur . Exp. Physik, Hamburg, Germany q CERN, Geneva, Switzerland r MPI fur . Kernphysik, Heidelberg, Germany s FNAL, Batavia, USA t University of Roma, Italy u Polytechnico Milano, Italy v University of Toronto, Toronto, ON, Canada
a
The RD42 Collaboration *Corresponding author. E-mail address: [email protected] (H. Kagan). 0168-9002/$ - see front matter r 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2003.08.086
ARTICLE IN PRESS 80
W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
Abstract Chemical vapor deposition diamond has been discussed extensively as an alternate sensor material for use very close to the interaction region of the LHC where extreme radiation conditions exist. During the last few years diamond devices have been manufactured and tested with LHC electronics with the goal of creating a detector usable by all LHC experiment. Extensive progress on diamond quality, on the development of diamond trackers and on radiation hardness studies has been made. Transforming the technology to the LHC specific requirements is now underway. In this paper we present the recent progress achieved. r 2003 Elsevier B.V. All rights reserved. PACS: 29.40.n; 29.40.Gx; 29.40.Wk; 29.40.Ym Keywords: CVD diamond; Radiation hardness; Charged particle tracking
1. Introduction Detectors and radiation monitors of future experiments will be employed in environments with several orders of magnitude more radiation than that of any current detector. In this environment, the components, both sensor material and electronics, will need to be exceedingly radiation hard in order to provide the high precision information required for physics analyses. At the present time detectors for radiation monitoring and tracking close to the interaction region are based on the mature silicon technology which functions very well in relatively low radiation environments (e.g., the LEP experiments, CLEO, etc.). Silicon is by its nature not radiation hard and great efforts are being made at ‘‘engineering in’’ the necessary hardness. Much progress in this area has been achieved, especially with oxygenated silicon. However, the practical limits on the radiation hardness of silicon still falls short of what is required for many future experiments. The radiation hardness of chemical vapor deposition (CVD) diamond has been investigated by RD42 up to fluences of 3 1015 particles=cm2 [1–3]. Present results indicate that diamond detectors can be operated with sufficiently high efficiency and spatial resolution close to the interaction region for many years of LHC operation at design luminosity. The availability of very radiation hard detector material and electronics will be of great importance in view of possible future luminosity upgrades for the LHC [4].
Diamond has two additional properties that are of interest. Its small dielectric constant yields, for a given geometry, a low detector capacitance and thereby, low-noise performance of the associated front-end electronics. In addition, even though diamond is an electrical insulator, it is an excellent thermal conductor with a thermal conductivity exceeding that of copper by a factor of five at room temperature. This is important since a common problem with large detector systems is the management of the thermal load generated by the on-detector electronics used in the detector readout. The handling of this thermal load would be simplified if the detectors were constructed from diamond since the diamond would act as a heat spreader. The discovery of the growth of diamond using the Chemical Vapor Deposition (CVD) process allowed the possibility of large scale use of diamond detectors. In this process, a hydrocarbon gas, such as methane, is mixed with a large concentration of molecular hydrogen gas. The gas mixture is then excited by an energy source. The resulting reactive gas mixture is brought into contact with a substrate where carbon-based radicals are reduced and the carbon atoms link together with single bonds (sp3 hybridized orbitals) forming a diamond lattice. Raman spectroscopy and X-ray diffraction unambiguously show that the CVD films are diamond and polycrystalline in structure. Although diamond appears ideal in many respects it does have a limitation: the large band
ARTICLE IN PRESS W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
2. CVD diamond detectors 2.1. Principles of diamond detectors
Amplifier Charged Particle
Diamond
250 8000 200 6000 150 4000
100
2000
50
0
e-h Creation
0 0
Vbias Electrodes Fig. 1. A schematic view of a diamond detector.
Mean Charge (e)
In Fig. 1, we show the basic principle of using diamond as a particle detector. A voltage is applied across a layer of diamond a few hundred microns thick. When a charged particle traverses the diamond, atoms in the crystal lattice sites are ionized, promoting electrons into the conduction band and leaving holes in the valence band. On average, 3600 electron–hole pairs are created per 100 mm of diamond traversed by a minimum ionizing particle. These charges drift across the diamond in response to the applied electric field producing a signal that can be measured. A feature of polycrystalline CVD diamond sensors is that their signal size increases with exposure to radiation for exposures up to about 1 krad: This ‘‘pump-up’’ effect is due to an increase of carrier lifetime caused by passivation
of deep traps. Because diamond has such a large band gap there may exist traps more than 1 eV from the valence or conduction bands. Exposure to radiation fills these traps with electrons (holes) produced by the radiation. If the traps are far enough from the conduction (valence) bands, the rate of thermal ionization of these traps will be slow and they will remain passivated for long times. Polycrystalline CVD diamonds kept in the dark remained pumped for at least 3 months. Exposure to white light ionizes the traps and depumps the diamond. In Fig. 2, we show the collected charge as a function of electric field, using a 90 Sr source after ‘‘pumping’’. To obtain this distribution the diamond is metallized with solid electrodes on each side, an external voltage is applied across the diamond and the collected charge is measured. Polycrystalline CVD diamonds all have the characteristic that the collected charge reaches a plateau at an electric field of approximately 1 V=mm: The measured pulse height distribution, using a 90 Sr source, of two samples as-received from the manufacturer is shown in Fig. 3. Operating at an electric field of 1 V=mm the charge distribution has the shape of a Landau separated from zero so that the charge collection is symmetric with applied voltage. The most probable charge is E6000e and 99% of the distribution is above 2250e: These
Collection Distance (µm)
gap (5:5 eV) which produces many of its outstanding properties also means that its signal size is at most half that of silicon for a given detector thickness in radiation lengths. Thus CVD diamond has three issues that need to be resolved before diamond detectors are viable. Foremost is to understand the effects of the material being polycrystalline. Second is to understand the implications of the signal size. Finally, the material must be reproducible in its electronic properties, especially charge yield and uniformity from production reactors.
81
0.2
0.4
0.6
0.8
1
1.2
Electric Field (V/µm) Fig. 2. The collected charge signal versus electric field from a polycrystalline diamond sample measured using a 90 Sr source in the laboratory illustrating the saturation of the collected charge at an electric field of B1 V=mm:
ARTICLE IN PRESS W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
82
CD-S87 - Both Sides - Pumped (Sr-90 source)
200
Number per 200 Electrons
Number per 200 Electrons
CD-S85 - Both Sides - Pumped (Sr-90 source) 250 Collection Distance: Side A: 210 µm
150
Side B: 198 µm
100 50 0
0
5000
10000
15000
20000
250 200
Collection Distance: Side A: 207 µm
150
Side B: 203 µm
100 50 0
0
5000
10000
15000
20000
Signal Size (Electrons)
Signal Size (Electrons)
Fig. 3. Charge signal from production reactor diamond samples measured using a 90 Sr source in the laboratory. The two curves on each figure (side A/side B) are the results for positive and negative applied voltages, respectively. No cuts to the data have been applied.
/QS½e d% ¼ 36e=mm
ð1Þ
where 36e=mm is the average number of electron– hole pairs generated by a minimum ionizing particle along 1 mm in diamond. Since the collection distance is usually smaller than the thickness of the polycrystalline CVD material, the collection distance serves as a useful tool to compare diamonds. The charge collection distance of 210 mm in CDS-85 corresponds to a mean charge of 7560e: 2.2. Progress on the improvement of CVD diamond Over the last few years, the RD42 Collaboration have worked closely with De Beers Industrial Diamonds1 to achieve major improvements in the charge collection distance and uniformity of polycrystalline CVD diamond. Each year members of our collaboration typically characterize 30 new diamond samples from production reactors. In previous years the diamond from production reactors was not of the same quality as that from research reactors. 1 De Beers Industrial Diamond Division Ltd., Charters, Sunninghill, Ascot, Berkshire, SL5 9PX UK.
Charge Collection in DeBeers CVD Diamond Collection Distance (microns)
results are quite typical of very good polycrystalline diamonds at around 2001. From the mean value, /QS; of the signal spectrum one derives the charge collection distance, the average distance the electrons and holes move apart in the diamond:
200
RD42 Goal
100
0 1994
1995
1996
1997
1998
1999
2000
2001
2002
Time (year)
Fig. 4. Charge collection distance in De Beers CVD diamond samples since 1995.
Fig. 4 shows the charge collection distance of De Beers polycrystalline CVD diamond samples since 1995. In the course of our work with De Beers the charge collection distance improved from B10 mm to over 200 mm in production reactors. The original goal of the RD42 program was to attain 200 mm charge collection distance. In order to improve the quality of CVD diamond a research program was started in 2001 to understand the carrier drift length across the diamond bulk. Fig. 5 shows the charge collection distance measured in two polycrystalline CVD diamond samples as a function of the thickness after thinning on the growth side and on the nucleation side. One observes that material removal from the growth side decreases the charge collection distance. The measurements from successive material removals on the growth side can
ARTICLE IN PRESS W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
be fit by a straight line indicating that the carrier drift length increases linearly from the nucleation side to the growth side. The intersection at zero gives the carrier drift length on the nucleation side
Material Removal Experiment 0.25
Collection Distance (mm)
0.2
0.15
0.1
0.05
0
0
0.1
0.2
0.3
0.4 0.5 0.6 Thickness (mm)
0.7
0.8
0.9
1
Fig. 5. Thinning of two CVD diamond samples. The charge collection distance measured as a function of the thickness after thinning on the growth side (solid) and on the nucleation side (open).
83
of the diamond. One also finds that material removal from the nucleation side improves the charge collection distance. However, charge collection distance must decrease again when a certain amount of material is removed from the nucleation side. Hence there is a thickness where the charge collection distance reaches a maximum. This study proves that post-processing of polycrystalline CVD by removing material from the substrate will improve the quality of the diamond. In Fig. 6, we show the collected charge of a recent diamond which resulted from the research program. The gain of the system used here was 124e=mV: We observe a most probable (MP) charge of 7600e and an average charge of 9800e: For this diamond 99% of the charge distribution is above 3000e indicating a clear separation of the charge signal from the pedestal. The mean charge corresponds to a collection distance of 275 mm: This currently available diamond material is of sufficiently high quality that the RD42 collaboration has undertaken studies of its performance as a particle detector. The results of beam tests at CERN for these devices yielded position resolu-
Fig. 6. Charge signal from a recent diamond sample measured using a 90 Sr source in the laboratory. The histogram is taken for each scintillator trigger (upper trace). An individual single diamond pulse is shown in the second trace and the average of all pulses is shown in the third trace.
ARTICLE IN PRESS W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
84
Signal from Irradiated Diamond Tracker
Diamond CDS-69 at 0.9 V/µm
800
700
before irradiation mean 57, most prob. 41 FWHM 54
700 600
600 500
400
100 0 1000
2
200
CDS-69
100 0 0
50 100 150 200 2 strip transparent signal to single strip noise [ ]
2-strip center of gravity method
200
after 2.2 E 15 protons/cm and re-metalization mean 47, most prob. 35 FWHM 36
300
before irradiation σ = 11.5 µm
300
after 1 E15 protons/cm mean 49, most prob. 35 FWHM 41
500
Diamond CDS-69
400
2
entries/2µm [ ]
entries/bin [1/100]
Residual Distributions, Proton Irradiated Diamond 800
250
800 600
after 1E15 p/cm
2
σ = 9.1 µm
400 200 0 1200
Fig. 7. Transparent 2-strip charge signal-to-noise distributions before (solid line), after proton irradiations with 1 1015 p=cm2 (dashed line) and after 2:2 1015 p=cm2 (dotted line).
1000 800 600 400
CDS-69 after 2.2 E 15 p/cm 2 and re-metalization σ = 7.4 µm
200
tion slightlypbetter ffiffiffiffiffi than that expected for the pitch divided by 12; the digital resolution. In the best devices, the measured efficiency of the diamond detector peaked at 95% and had an average efficiency of approximately 94%. 2.3. Radiation hardness studies of CVD diamond trackers In Fig. 7, we show the collected charge of a polycrystalline diamond strip detector after irradiation with a fluence of 24 GeV protons of 1 1015 particles ðpÞ=cm2 and after 2:2 1015 p=cm2 : The diamond which was irradiated was one of the best from a research reactor. While the strip contacts before and after irradiation with fluences of 1 1015 p=cm2 were unchanged the contacts were replaced after a fluence of 2:2 1015 p=cm2 and then characterized in the test beam. At 1 1015 p=cm2 we observe that the shape of the signalto-noise distribution is narrower than before irradiation and entries in the tail of the distribution appear closer to the most probable signal. At 2:2 1015 p=cm2 and after re-metallization we observe essentially the same signal-to-noise distribution as at 1 1015 p=cm2 indicating that very little further damage occured to the diamond bulk.
0
-100 -80
-60
-40 -20 0 20 40 residual, u h -u t , [µm]
60
80
100
Fig. 8. Residual distributions measure before and after proton irradiations.
The most probable signal-to-noise was 41 before irradiation and 35 at 1 1015 p=cm2 and also at 2:2 1015 p=cm2 : We find a reduction of maximum 15% in the most probable signal-to-noise after irradiation with 2:2 1015 p=cm2 : The noise was measured to remain constant at each beam test. Since the beam test with the detector irradiated with a fluence of 2:2 1015 p=cm2 used new contacts the observed decrease of 15% is attributed to damage in the diamond bulk. This work has spurred research in the development of radiation hard diamond contacts. Diamond strip detectors with such contacts will be irradiated with neutrons and protons in 2003. Fig. 8 shows residual distributions before and after irradiation at 1 1015 p=cm2 and 2:2 1015 p=cm2 : We observe that the spatial resolution improves from ð11:570:3Þ mm before irradiation to ð9:170:3Þ mm at 1 1015 p=cm2 and to ð7:470:2Þ mm at 2:2 1015 p=cm2 : At present, the explanation for this effect is that the irradiated
ARTICLE IN PRESS W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
*
*
*
*
*
*
Polycrystalline CVD diamond from production reactors now ‘regularly’ exceeds 200 mm charge collection distance. Research on lapping and uniformity of material provided the understanding necessary to improve the material quality with processing. Research pursued by RD42 in conjunction with De Beers yielded polycrystalline CVD diamonds with 270 mm collection distance. This appears to be the limit of the polycrystalline CVD process. Polycrystalline CVD detectors require externally applied electric fields of B1 V=mm for operation. The typical position resolution with a polycrystalline CVD detector pffiffiffiffiffi is slightly better than the pitch divided by 12: After irradiation of E1015 the most probable charge decreases slightly while the position resolution improves.
071415 - Both Sides - Pumped (Sr-90 source) Number per 400 Electrons
material is more uniform in the sense that the Landau distribution is narrower. The spatial resolution of nearly 7 mm with a detector of 50 mm strip pitch is comparable to results obtained with silicon detectors. A summary of the present status of polycrystalline CVD diamond is listed below:
85
400 300 200 100 0
0
10000
20000
30000
40000
50000
Signal Size (Electrons)
Fig. 9. The pulse height distribution of the first scCVD diamond operating at 500 V: The two curves are for positive and negative voltage applied.
tion. The FWHM/MP for this single-crystal CVD diamond is approximately 0.3, about one-third that of polycrystalline CVD diamond. This indicates that the energy resolution of this new diamond should be quite good. Based on our first test results, this new material seems extraordinary. The single-crystal CVD diamond may resolve many if not all of the issues associated with polycrystalline material. We are looking forward to measuring more of this material.
4. Future plans 3. Single-crystal CVD diamond—the new development In September 2002 De Beers announced that it had successfully produced a single-crystal diamond by the CVD process [5]. The samples they produced were synthesized with a microwave plasma-assisted CVD reactor using a specially prepared /100S oriented single-crystal synthetic diamond substrate. The RD42 group was asked to evaluate this diamond. For the evaluation of this first sample we investigated charge collection. In Fig. 9 we show the pulse height spectrum observed from this first single-crystal CVD diamond. This diamond is 450 mm thick. We observe a most probable charge of 14 000e; a FWHM of 4000e and a 10 000e separation between the pedestal and the beginning of the charge distribu-
The RD42 collaboration plans this year to continue its work on improving the quality of polycrystalline CVD diamond. A new goal of 300 mm collection distance may be achieved by identification of the particular trapping centers which limit the collection distance. In the coming year RD42 will test the newest polycrystalline and single-crystal material in strip detectors and pixel detectors with the recently available LHC electronics. Moreover, the radiation hardness studies will be repeated up to fluences of 5 1015 particles=cm2 :
References [1] W. Adam, et al., (RD42 Collaboration), Nucl. Instr. and Meth. A 447 (2000) 244.
ARTICLE IN PRESS 86
W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
[2] D. Meier, et al., (RD42 Collaboration), Nucl. Instr. and Meth. A 426 (1999) 173. [3] W. Adam, et al., (RD42 Collaboration), Development of diamond tracking detectors for high luminosity experiments at the LHC, Status Report/RD42, CERN/LHCC 2002-010, LHCC-RD-001, 2002.
[4] First Workshop on Radiation Hard Semiconductor Devices for Very High Luminosity Colliders, CERN, Geneva, Switzerland, 2001, pp. 28–30. [5] J. Isberg, et al., Science 297 (2002) 1670.
Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
The development of diamond tracking detectors for the LHC W. Adama, E. Berdermannb, P. Bergonzoc, W. de Boerd, F. Boganie, E. Borchif, A. Brambillac, M. Bruzzif, C. Colledanig, J. Conwayh, P. D’Angeloi, W. Dabrowskij, P. Delpierrek, J. Doroshenkoh, W. Dulinskig, B. van Eijkl, A. Fallouk, P. Fischerm, F. Fizzottin, C. Furettai, K.K. Gano, N. Ghodbanep, E. Grigorievd, G. Hallewellk, S. Hano, F. Hartjesl, J. Hrubeca, D. Hussong, r . H. Kagano,*, J. Kaplonq, C. Karlp, R. Kasso, M. Keilm, K.T. Knopfle , T. Koethh, M. Krammera, A. Logiudicen, R. Lun, L. mac Lynneh, C. Manfredottin, R.D. Marshallc, D. Meierq, D. Menichellif, S. Meuserm, M. Mishinas, L. Moronii, J. Noomenl, A. Ohq, L. Pererah, H. Perneggerq, M. Pernickaa, P. Polesellon, R. Potenzat, J.L. Riesterg, S. Roeq, A. Rudgeq, S. Salai, M. Sampietrou, S. Schnetzerh, S. Sciortinof, H. Stelzerb, R. Stoneh, C. Suterat, W. Trischukv, D. Tromsonc, C. Tuvet, B. Vincenzot, P. Weilhammerq, N. Wermesm, M. Wetsteinh, W. Zeunerp, M. Zoellero . Institut fur Akademie d. Wissenschaften, Vienna, Austria . Hochenergiephysik der Osterr. b GSI, Darmstadt, Germany c LETI (CEA-Technologies Avancees) DEIN/SPE-CEA Saclay, Gif-Sur-Yvette, France d Universitat . Karlsruhe, Karlsruhe, Germany e LENS, Florence, Italy f University of Florence, Florence, Italy g LEPSI, IN2P3/CNRS-ULP, Strasbourg, France h Rutgers University, Piscataway, NJ, USA i INFN, Milano, Italy j Faculty of Physics and Nuclear Techniques, UMM, Cracow, Poland k CPPM, Marseille, France l NIKHEF, Amsterdam, The Netherlands m Universitat . Bonn, Bonn, Germany n University of Torino, Italy o The Ohio State University, Columbus, OH, USA p II.Inst. fur . Exp. Physik, Hamburg, Germany q CERN, Geneva, Switzerland r MPI fur . Kernphysik, Heidelberg, Germany s FNAL, Batavia, USA t University of Roma, Italy u Polytechnico Milano, Italy v University of Toronto, Toronto, ON, Canada
a
The RD42 Collaboration *Corresponding author. E-mail address: [email protected] (H. Kagan). 0168-9002/$ - see front matter r 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2003.08.086
ARTICLE IN PRESS 80
W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
Abstract Chemical vapor deposition diamond has been discussed extensively as an alternate sensor material for use very close to the interaction region of the LHC where extreme radiation conditions exist. During the last few years diamond devices have been manufactured and tested with LHC electronics with the goal of creating a detector usable by all LHC experiment. Extensive progress on diamond quality, on the development of diamond trackers and on radiation hardness studies has been made. Transforming the technology to the LHC specific requirements is now underway. In this paper we present the recent progress achieved. r 2003 Elsevier B.V. All rights reserved. PACS: 29.40.n; 29.40.Gx; 29.40.Wk; 29.40.Ym Keywords: CVD diamond; Radiation hardness; Charged particle tracking
1. Introduction Detectors and radiation monitors of future experiments will be employed in environments with several orders of magnitude more radiation than that of any current detector. In this environment, the components, both sensor material and electronics, will need to be exceedingly radiation hard in order to provide the high precision information required for physics analyses. At the present time detectors for radiation monitoring and tracking close to the interaction region are based on the mature silicon technology which functions very well in relatively low radiation environments (e.g., the LEP experiments, CLEO, etc.). Silicon is by its nature not radiation hard and great efforts are being made at ‘‘engineering in’’ the necessary hardness. Much progress in this area has been achieved, especially with oxygenated silicon. However, the practical limits on the radiation hardness of silicon still falls short of what is required for many future experiments. The radiation hardness of chemical vapor deposition (CVD) diamond has been investigated by RD42 up to fluences of 3 1015 particles=cm2 [1–3]. Present results indicate that diamond detectors can be operated with sufficiently high efficiency and spatial resolution close to the interaction region for many years of LHC operation at design luminosity. The availability of very radiation hard detector material and electronics will be of great importance in view of possible future luminosity upgrades for the LHC [4].
Diamond has two additional properties that are of interest. Its small dielectric constant yields, for a given geometry, a low detector capacitance and thereby, low-noise performance of the associated front-end electronics. In addition, even though diamond is an electrical insulator, it is an excellent thermal conductor with a thermal conductivity exceeding that of copper by a factor of five at room temperature. This is important since a common problem with large detector systems is the management of the thermal load generated by the on-detector electronics used in the detector readout. The handling of this thermal load would be simplified if the detectors were constructed from diamond since the diamond would act as a heat spreader. The discovery of the growth of diamond using the Chemical Vapor Deposition (CVD) process allowed the possibility of large scale use of diamond detectors. In this process, a hydrocarbon gas, such as methane, is mixed with a large concentration of molecular hydrogen gas. The gas mixture is then excited by an energy source. The resulting reactive gas mixture is brought into contact with a substrate where carbon-based radicals are reduced and the carbon atoms link together with single bonds (sp3 hybridized orbitals) forming a diamond lattice. Raman spectroscopy and X-ray diffraction unambiguously show that the CVD films are diamond and polycrystalline in structure. Although diamond appears ideal in many respects it does have a limitation: the large band
ARTICLE IN PRESS W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
2. CVD diamond detectors 2.1. Principles of diamond detectors
Amplifier Charged Particle
Diamond
250 8000 200 6000 150 4000
100
2000
50
0
e-h Creation
0 0
Vbias Electrodes Fig. 1. A schematic view of a diamond detector.
Mean Charge (e)
In Fig. 1, we show the basic principle of using diamond as a particle detector. A voltage is applied across a layer of diamond a few hundred microns thick. When a charged particle traverses the diamond, atoms in the crystal lattice sites are ionized, promoting electrons into the conduction band and leaving holes in the valence band. On average, 3600 electron–hole pairs are created per 100 mm of diamond traversed by a minimum ionizing particle. These charges drift across the diamond in response to the applied electric field producing a signal that can be measured. A feature of polycrystalline CVD diamond sensors is that their signal size increases with exposure to radiation for exposures up to about 1 krad: This ‘‘pump-up’’ effect is due to an increase of carrier lifetime caused by passivation
of deep traps. Because diamond has such a large band gap there may exist traps more than 1 eV from the valence or conduction bands. Exposure to radiation fills these traps with electrons (holes) produced by the radiation. If the traps are far enough from the conduction (valence) bands, the rate of thermal ionization of these traps will be slow and they will remain passivated for long times. Polycrystalline CVD diamonds kept in the dark remained pumped for at least 3 months. Exposure to white light ionizes the traps and depumps the diamond. In Fig. 2, we show the collected charge as a function of electric field, using a 90 Sr source after ‘‘pumping’’. To obtain this distribution the diamond is metallized with solid electrodes on each side, an external voltage is applied across the diamond and the collected charge is measured. Polycrystalline CVD diamonds all have the characteristic that the collected charge reaches a plateau at an electric field of approximately 1 V=mm: The measured pulse height distribution, using a 90 Sr source, of two samples as-received from the manufacturer is shown in Fig. 3. Operating at an electric field of 1 V=mm the charge distribution has the shape of a Landau separated from zero so that the charge collection is symmetric with applied voltage. The most probable charge is E6000e and 99% of the distribution is above 2250e: These
Collection Distance (µm)
gap (5:5 eV) which produces many of its outstanding properties also means that its signal size is at most half that of silicon for a given detector thickness in radiation lengths. Thus CVD diamond has three issues that need to be resolved before diamond detectors are viable. Foremost is to understand the effects of the material being polycrystalline. Second is to understand the implications of the signal size. Finally, the material must be reproducible in its electronic properties, especially charge yield and uniformity from production reactors.
81
0.2
0.4
0.6
0.8
1
1.2
Electric Field (V/µm) Fig. 2. The collected charge signal versus electric field from a polycrystalline diamond sample measured using a 90 Sr source in the laboratory illustrating the saturation of the collected charge at an electric field of B1 V=mm:
ARTICLE IN PRESS W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
82
CD-S87 - Both Sides - Pumped (Sr-90 source)
200
Number per 200 Electrons
Number per 200 Electrons
CD-S85 - Both Sides - Pumped (Sr-90 source) 250 Collection Distance: Side A: 210 µm
150
Side B: 198 µm
100 50 0
0
5000
10000
15000
20000
250 200
Collection Distance: Side A: 207 µm
150
Side B: 203 µm
100 50 0
0
5000
10000
15000
20000
Signal Size (Electrons)
Signal Size (Electrons)
Fig. 3. Charge signal from production reactor diamond samples measured using a 90 Sr source in the laboratory. The two curves on each figure (side A/side B) are the results for positive and negative applied voltages, respectively. No cuts to the data have been applied.
/QS½e d% ¼ 36e=mm
ð1Þ
where 36e=mm is the average number of electron– hole pairs generated by a minimum ionizing particle along 1 mm in diamond. Since the collection distance is usually smaller than the thickness of the polycrystalline CVD material, the collection distance serves as a useful tool to compare diamonds. The charge collection distance of 210 mm in CDS-85 corresponds to a mean charge of 7560e: 2.2. Progress on the improvement of CVD diamond Over the last few years, the RD42 Collaboration have worked closely with De Beers Industrial Diamonds1 to achieve major improvements in the charge collection distance and uniformity of polycrystalline CVD diamond. Each year members of our collaboration typically characterize 30 new diamond samples from production reactors. In previous years the diamond from production reactors was not of the same quality as that from research reactors. 1 De Beers Industrial Diamond Division Ltd., Charters, Sunninghill, Ascot, Berkshire, SL5 9PX UK.
Charge Collection in DeBeers CVD Diamond Collection Distance (microns)
results are quite typical of very good polycrystalline diamonds at around 2001. From the mean value, /QS; of the signal spectrum one derives the charge collection distance, the average distance the electrons and holes move apart in the diamond:
200
RD42 Goal
100
0 1994
1995
1996
1997
1998
1999
2000
2001
2002
Time (year)
Fig. 4. Charge collection distance in De Beers CVD diamond samples since 1995.
Fig. 4 shows the charge collection distance of De Beers polycrystalline CVD diamond samples since 1995. In the course of our work with De Beers the charge collection distance improved from B10 mm to over 200 mm in production reactors. The original goal of the RD42 program was to attain 200 mm charge collection distance. In order to improve the quality of CVD diamond a research program was started in 2001 to understand the carrier drift length across the diamond bulk. Fig. 5 shows the charge collection distance measured in two polycrystalline CVD diamond samples as a function of the thickness after thinning on the growth side and on the nucleation side. One observes that material removal from the growth side decreases the charge collection distance. The measurements from successive material removals on the growth side can
ARTICLE IN PRESS W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
be fit by a straight line indicating that the carrier drift length increases linearly from the nucleation side to the growth side. The intersection at zero gives the carrier drift length on the nucleation side
Material Removal Experiment 0.25
Collection Distance (mm)
0.2
0.15
0.1
0.05
0
0
0.1
0.2
0.3
0.4 0.5 0.6 Thickness (mm)
0.7
0.8
0.9
1
Fig. 5. Thinning of two CVD diamond samples. The charge collection distance measured as a function of the thickness after thinning on the growth side (solid) and on the nucleation side (open).
83
of the diamond. One also finds that material removal from the nucleation side improves the charge collection distance. However, charge collection distance must decrease again when a certain amount of material is removed from the nucleation side. Hence there is a thickness where the charge collection distance reaches a maximum. This study proves that post-processing of polycrystalline CVD by removing material from the substrate will improve the quality of the diamond. In Fig. 6, we show the collected charge of a recent diamond which resulted from the research program. The gain of the system used here was 124e=mV: We observe a most probable (MP) charge of 7600e and an average charge of 9800e: For this diamond 99% of the charge distribution is above 3000e indicating a clear separation of the charge signal from the pedestal. The mean charge corresponds to a collection distance of 275 mm: This currently available diamond material is of sufficiently high quality that the RD42 collaboration has undertaken studies of its performance as a particle detector. The results of beam tests at CERN for these devices yielded position resolu-
Fig. 6. Charge signal from a recent diamond sample measured using a 90 Sr source in the laboratory. The histogram is taken for each scintillator trigger (upper trace). An individual single diamond pulse is shown in the second trace and the average of all pulses is shown in the third trace.
ARTICLE IN PRESS W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
84
Signal from Irradiated Diamond Tracker
Diamond CDS-69 at 0.9 V/µm
800
700
before irradiation mean 57, most prob. 41 FWHM 54
700 600
600 500
400
100 0 1000
2
200
CDS-69
100 0 0
50 100 150 200 2 strip transparent signal to single strip noise [ ]
2-strip center of gravity method
200
after 2.2 E 15 protons/cm and re-metalization mean 47, most prob. 35 FWHM 36
300
before irradiation σ = 11.5 µm
300
after 1 E15 protons/cm mean 49, most prob. 35 FWHM 41
500
Diamond CDS-69
400
2
entries/2µm [ ]
entries/bin [1/100]
Residual Distributions, Proton Irradiated Diamond 800
250
800 600
after 1E15 p/cm
2
σ = 9.1 µm
400 200 0 1200
Fig. 7. Transparent 2-strip charge signal-to-noise distributions before (solid line), after proton irradiations with 1 1015 p=cm2 (dashed line) and after 2:2 1015 p=cm2 (dotted line).
1000 800 600 400
CDS-69 after 2.2 E 15 p/cm 2 and re-metalization σ = 7.4 µm
200
tion slightlypbetter ffiffiffiffiffi than that expected for the pitch divided by 12; the digital resolution. In the best devices, the measured efficiency of the diamond detector peaked at 95% and had an average efficiency of approximately 94%. 2.3. Radiation hardness studies of CVD diamond trackers In Fig. 7, we show the collected charge of a polycrystalline diamond strip detector after irradiation with a fluence of 24 GeV protons of 1 1015 particles ðpÞ=cm2 and after 2:2 1015 p=cm2 : The diamond which was irradiated was one of the best from a research reactor. While the strip contacts before and after irradiation with fluences of 1 1015 p=cm2 were unchanged the contacts were replaced after a fluence of 2:2 1015 p=cm2 and then characterized in the test beam. At 1 1015 p=cm2 we observe that the shape of the signalto-noise distribution is narrower than before irradiation and entries in the tail of the distribution appear closer to the most probable signal. At 2:2 1015 p=cm2 and after re-metallization we observe essentially the same signal-to-noise distribution as at 1 1015 p=cm2 indicating that very little further damage occured to the diamond bulk.
0
-100 -80
-60
-40 -20 0 20 40 residual, u h -u t , [µm]
60
80
100
Fig. 8. Residual distributions measure before and after proton irradiations.
The most probable signal-to-noise was 41 before irradiation and 35 at 1 1015 p=cm2 and also at 2:2 1015 p=cm2 : We find a reduction of maximum 15% in the most probable signal-to-noise after irradiation with 2:2 1015 p=cm2 : The noise was measured to remain constant at each beam test. Since the beam test with the detector irradiated with a fluence of 2:2 1015 p=cm2 used new contacts the observed decrease of 15% is attributed to damage in the diamond bulk. This work has spurred research in the development of radiation hard diamond contacts. Diamond strip detectors with such contacts will be irradiated with neutrons and protons in 2003. Fig. 8 shows residual distributions before and after irradiation at 1 1015 p=cm2 and 2:2 1015 p=cm2 : We observe that the spatial resolution improves from ð11:570:3Þ mm before irradiation to ð9:170:3Þ mm at 1 1015 p=cm2 and to ð7:470:2Þ mm at 2:2 1015 p=cm2 : At present, the explanation for this effect is that the irradiated
ARTICLE IN PRESS W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
*
*
*
*
*
*
Polycrystalline CVD diamond from production reactors now ‘regularly’ exceeds 200 mm charge collection distance. Research on lapping and uniformity of material provided the understanding necessary to improve the material quality with processing. Research pursued by RD42 in conjunction with De Beers yielded polycrystalline CVD diamonds with 270 mm collection distance. This appears to be the limit of the polycrystalline CVD process. Polycrystalline CVD detectors require externally applied electric fields of B1 V=mm for operation. The typical position resolution with a polycrystalline CVD detector pffiffiffiffiffi is slightly better than the pitch divided by 12: After irradiation of E1015 the most probable charge decreases slightly while the position resolution improves.
071415 - Both Sides - Pumped (Sr-90 source) Number per 400 Electrons
material is more uniform in the sense that the Landau distribution is narrower. The spatial resolution of nearly 7 mm with a detector of 50 mm strip pitch is comparable to results obtained with silicon detectors. A summary of the present status of polycrystalline CVD diamond is listed below:
85
400 300 200 100 0
0
10000
20000
30000
40000
50000
Signal Size (Electrons)
Fig. 9. The pulse height distribution of the first scCVD diamond operating at 500 V: The two curves are for positive and negative voltage applied.
tion. The FWHM/MP for this single-crystal CVD diamond is approximately 0.3, about one-third that of polycrystalline CVD diamond. This indicates that the energy resolution of this new diamond should be quite good. Based on our first test results, this new material seems extraordinary. The single-crystal CVD diamond may resolve many if not all of the issues associated with polycrystalline material. We are looking forward to measuring more of this material.
4. Future plans 3. Single-crystal CVD diamond—the new development In September 2002 De Beers announced that it had successfully produced a single-crystal diamond by the CVD process [5]. The samples they produced were synthesized with a microwave plasma-assisted CVD reactor using a specially prepared /100S oriented single-crystal synthetic diamond substrate. The RD42 group was asked to evaluate this diamond. For the evaluation of this first sample we investigated charge collection. In Fig. 9 we show the pulse height spectrum observed from this first single-crystal CVD diamond. This diamond is 450 mm thick. We observe a most probable charge of 14 000e; a FWHM of 4000e and a 10 000e separation between the pedestal and the beginning of the charge distribu-
The RD42 collaboration plans this year to continue its work on improving the quality of polycrystalline CVD diamond. A new goal of 300 mm collection distance may be achieved by identification of the particular trapping centers which limit the collection distance. In the coming year RD42 will test the newest polycrystalline and single-crystal material in strip detectors and pixel detectors with the recently available LHC electronics. Moreover, the radiation hardness studies will be repeated up to fluences of 5 1015 particles=cm2 :
References [1] W. Adam, et al., (RD42 Collaboration), Nucl. Instr. and Meth. A 447 (2000) 244.
ARTICLE IN PRESS 86
W. Adam et al. / Nuclear Instruments and Methods in Physics Research A 514 (2003) 79–86
[2] D. Meier, et al., (RD42 Collaboration), Nucl. Instr. and Meth. A 426 (1999) 173. [3] W. Adam, et al., (RD42 Collaboration), Development of diamond tracking detectors for high luminosity experiments at the LHC, Status Report/RD42, CERN/LHCC 2002-010, LHCC-RD-001, 2002.
[4] First Workshop on Radiation Hard Semiconductor Devices for Very High Luminosity Colliders, CERN, Geneva, Switzerland, 2001, pp. 28–30. [5] J. Isberg, et al., Science 297 (2002) 1670.